UNT Digital Library/67531/metadc693815/m2/1/high_re… · Lawrence Livermore National Laboratory This is an informal report intended primarily for internal or limited external distribution

Lawre

nce

Liver

more

National

Labora

tory

This is an informal report intended primarily for internal or limited externaldistribution. The opinions and conclusions stated are those of the author and mayor may not be those of the Laboratory.

Work performed under the auspices of the U.S. Department of Energy by theLawrence Livermore National Laboratory under Contract W-7405-Eng-48.

Chemistry & Materials ScienceProgress Report

Weapons Research and Developmentand Laboratory Directed Researchand Development

FY96

March 1997

UCID-20622-96

DISCLAIMER

This document was prepared as an account of work sponsored by an agency of the United StatesGovernment. Neither the United States Government nor the University of California nor any of theiremployees, makes any warranty, express or implied, or assumes any legal liability or responsibilityfor the accuracy, completeness, or usefulness of any information, apparatus, product, or processdisclosed, or represents that its use would not infringe privately owned rights. Reference herein to anyspecific commercial product, process, or service by trade name, trademark, manufacturer, or other-wise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by theUnited States Government or the University of California. The views and opinions of authorsexpressed herein do not necessarily state or reflect those of the United States Government or theUniversity of California, and shall not be used for advertising or product endorsement purposes.

This report has been reproduceddirectly from the best available copy.

Available to DOE and DOE contractors from theOffice of Scientific and Technical Information

P.O. Box 62, Oak Ridge, TN 37831Prices available from (615) 576-8401, FTS 626-8401

Available to the public from theNational Technical Information Service

U.S. Department of Commerce5285 Port Royal Rd.,

Springfield, VA 22161

Recycled Recyclable

iC&MS PROGRESS REPORT—FY96

LAWRENCE LIVERMORE NATIONAL LABORATORYUniversity of California • Livermore, California • 94550

Chemistry & Materials ScienceProgress Report

Weapons Research and Developmentand Laboratory Directed Researchand Development

FY96

ii C&MS PROGRESS REPORT—FY96

iiiC&MS PROGRESS REPORT—FY96

Contents

Weapons Research and DevelopmentAtomic Site Physics and Chemistry of Uranium Hydriding: Water VaporAdsorption on and Reaction with Uranium .......................................................................... 1M. Balooch, A. V. Hamza, W. J. Siekhaus

Modeling Hydride Formation .................................................................................................. 4P. A. Sterne

Synthesis of New Explosives ................................................................................................... 7P. F. Pagoria, A. R. Mitchell, R. D. Schmidt, L. E. Fried

Ab Initio Study of Energetic Material Decomposition Mechanisms ............................. 12L. E. Fried, C. J. Wu

First Positron Annihilation Lifetime Measurement of Plutonium ................................. 15C. Colmenares, R. H. Howell, D. Ancheta, T. Cowan, J. Hanafee, P. Sterne

Microstructural Evolution in Welds ...................................................................................... 18J. W. Elmer, J. Wong

Deformation Mechanisms of U–6Nb .................................................................................... 21G. Gallegos, A. Schwartz, E. Li

Photothermal Radiation during Laser-Induced Damage in Bulk KDP ......................... 26M. Yan, M. Staggs, M. Runkel, J. De Yoreo

Large Capsule Development for National Ignition Facility Targets .............................. 31S. A. Letts, K. E. Hamilton, S. R. Buckley, E. M. Fearon, D. Schroen–Carey, R. C. Cook

Preparation of High-Beryllium-Content Plasma Polymer Coatings UsingCyclopentadienylberyllium Methyl as a Precursor ........................................................... 38R. Brusasco, R. Cook, G. Wilemski, M. Saculla

iv C&MS PROGRESS REPORT—FY96

Laboratory-Directed Research and Development,Exploratory ResearchFundamental Studies of Matrix-Assisted Laser Desorption/Ionization, UsingTime-of-Flight Mass Spectrometry to Identify Biological Molecules ............................ 43D. Eades, D. Wruck, H. Gregg

Damage Evolution in Low-Energy Ion-Implanted Silicon: STM Experimentsand Atomistic Computer Simulations .................................................................................. 47P. Bedrossian, M.–J. Caturla, T. Diaz de la Rubia

Bonding and Structure of Nanocrystalline Thin-Films .................................................... 50L. J. Terminello

A Novel Approach to SiC Film Production for Micro-Mechanical andElectronic Components ............................................................................................................ 53A. Hamza, M. Balooch

Giant Magnetoresistance Materials with Novel Spacer Layers ...................................... 55A. Chaiken

Studies of Meso-Structural Features in High-Tc Superconducting MaterialsUsing Pair Distribution Function Analysis of Neutron/X-ray Scattering Data ............ 57G. H. Kwei

Superplasticity in Aluminum Alloys ................................................................................... 59T. G. Nieh

Molecular Scale Investigation of Crystal Growth from Solutions ................................. 64T. Land, J. De Yoreo

Isotope Measurements for Innovative Groundwater Management ............................... 69G. B. Hudson, M. L. Davisson

Selenium Isotope Geochemistry: A New Approach to Characterizing theEnvironmental Chemistry of Selenium ............................................................................... 71A. Volpe, B. Esser

Studies in the Region of Enhanced Nuclear Stability AroundN = 162 and Z = 108 .................................................................................................................. 72J. F. Wild, R. W. Lougheed, K. J. Moody, N. J. Stoyer

Weapons Research and Development

ATOMIC SITE PHYSICS AND CHEMISTRY OF URANIUM HYDRIDING

1C&MS PROGRESS REPORT—FY96

ATOMIC SITE PHYSICS AND CHEMISTRY OF

URANIUM HYDRIDING: WATER VAPOR ADSORPTION

ON AND REACTION WITH URANIUM

M. Balooch, A. V. Hamza, W. J. Siekhaus

Introduction

We investigated the fundamental propertiesof the interaction of water vapor with uranium bycombining modulated molecular beam scatteringwith temperature programmed desorption tomeasure the initial sticking probability, hydrideformation probabilities, and desorption kineticson clean uranium.

Experimental Procedure

The experiments were performed in twoultra-high vacuum apparati, both of which havebeen described in detail elsewhere.1,2 The first is amodulated molecular beam apparatus; the secondis equipped with a differentially pumped quadru-pole mass spectrometer (QMS) for temperatureprogrammed desorption (TPD), a cylindricalmirror analyzer for Auger electron spectroscopy(AES), and an argon ion gun for sputter cleaning.Triply distilled, deionized (greater than 18 MΩresistivity) water was used, purged by ultra-highpurity argon gas and further purified by repeatedfreeze-thaw cycles under vacuum. The uraniumsample was cleaned by repeated argon ion sputter-ing for a few hours with 5 keV ions (normalincidence) and annealing afterward. With thebase pressure at 4 × 10–10 Torr, the surface becomesoxidized overnight. However, the oxide thick-ness is thin enough to be removed in 1 to 2 min ofsputtering at ~10 µA/cm2.

Modulated Water Vapor Beam onClean Uranium

A modulated beam of water vapor is directedat the surface with an equivalent pressure of~1.5 × 10–6 Torr, above the onset pressure for

hydride formation,3 assuming the water vapordecomposes. Since the reaction rate of water withoxidized uranium is low [see Eq. (1)], then thereaction probability of water vapor on cleanuranium at a particular surface temperature isequal to one minus the water vapor signal fromclean uranium divided by the water vapor signalfrom oxidized uranium

εH OH O

H OK2

2

2K

Ton

Ton

off

S

S

T[ ] = −[ ][ ]

1

300300

12 , (1)

where ε is the reaction probability, S is the reflectedsignal, and T is the surface temperature. On andoff refer to argon sputtering; the surface is oxidizedduring the off cycle. The water reaction has twobranches—dissociative adsorption followed bysurface recombination to form dihydrogen orfollowed by hydride formation. The recombinationprobability is determined from the H2 signal, from

εσ

σHH

H O300K

H

H O

H O

H

H

2 2K2

2 2 2

2

2

1

21

2

300T

on

T

on

off

S

S

M

M

T[ ] [ ][ ]

=

−

, (2)

where σ is the ionization cross section of thedetector. σwater/σdihydrogen = 2.45. The hydrideformation probability is the difference betweenthe water reaction and the H2 production probability,

ε ε εUH H O H3 2 2

Ton

Ton

Ton[ ] = [ ] − [ ] . (3)

Figure 1 shows the water reaction probabili-ties for the clean and oxidized surface at roomtemperature (i.e., the reaction probability is about8 × 10–2, the hydride formation probability isabout 1 × 10–2 at an equivalent water vaporpressure of 1.5 × 10-6 Torr, and the dihydrogenproduction probability is about 6–7 × 10–2).


2 C&MS PROGRESS REPORT—FY96

Figure 1. Plot of the water reaction probabilityand the H2 production reaction probability vstime. The squares represent the water reactionprobability and the circles represent the H2production probability. At 1.75 min, the ionsputter gun was on at 1 keV and 7 × 1014 ions/cm2/s and the oxygen coverage was reduced, butbecause of the water flux could not be madenegligible.

0 1 2 3 4 50

0.02

0.04

0.06

0.08

Ion off

εH2

PH2O = 1.5 × 10–6 Torr

Eion = 1 keV

Iion = 7 × 1014 ions/cm2-s

εUH3

εH2O

Ion on

0.1

t (min)

Rea

ctio

n pr

obab

ility

Water Vapor on Clean Uranium(TPD)

Prior to each exposure to water vapor, theuranium surface is sputter cleaned and annealedto 700K. Figure 2 shows the TPD of dihydrogenfrom a water vapor dosed clean uranium surfaceat room temperature for dihydrogen/water vaporcoverages (0.7 to 4.6 × 1018 molecules/m2).Desorption of dihydrogen occurs between 320and 670K. The most striking features are the twodesorption peaks observed at coverages of less than4 × 1014 molecules/cm2. The first feature isobserved at a peak temperature of 360K and thesecond is observed at 495K. A third feature, ashoulder initially, grows in with a peak tempera-ture for desorption of 460K and appears only as

coverages approaching a monolayer (one watervapor or dihydrogen molecule/uranium atom)are reached. As exposure is increased, the thirdfeature dominates the spectrum and continues togrow without apparent bounds. Since the secondfeature (peak temperature 495K) at coverages lessthan 4 × 1014 molecules/cm2 is relatively free fromoverlapping desorption, the peak is analyzed toextract kinetic parameters. The analysis stronglysuggests a first-order desorption process. Theactivation energy for desorption is 23 ±1 kcal/moland the pre-exponential factor is ~1010 s–1. Bothparameters are independent of the coverage.

Plotting the coverage of surface withdihydrogen vs exposure of the surface to watervapor affords the determination of the initialsticking probability. The initial slope is linear andgives an initial sticking probability of 0.7–0.8.

Discussion

Colmenares and co-workers4 suggest thatwater vapor dissociatively adsorbs as Hads andOHads. In this case, dihydrogen could evolvefrom the reaction of two Hads or from reaction oftwo OHads or from the reaction of one Hads withone OHads, giving rise to possibly three desorption

H2 /H2O/Uranium

0.7 × 1018 molec/m2

1.3 × 1018 molec/m2

2.0 × 1018 molec/m2

3.8 × 1018 molec/m2

4.6 × 1018 molec/m2

300 400 500 600 700 8000

20

40

60

80

100

120 × 1015

Surface temperature (K)

Hyd

roge

n de

sorp

tion

rate

(m

olec

ules

/m2 /s

)

Figure 2. Temperature programmed desorption ofdihydrogen from water vapor exposed uraniumat room temperature. Coverages are determinedby integrating the desorption flux. Coveragesincrease from 0.7 to 4.6 × 1018 molecules/m2.



states. We observe two reactions: the first (lowtemperature) at a surface temperature lower thanHads recombination on clean uranium, and thesecond (high temperature) at ~500K, ~70L2∏igherthan Hads recombination on clean uranium. Sincethe OH bond is stronger than the UH bond, it islogical to assign the high-temperature peak to thedecomposition of OHads. This assignment issupported by the fact that the high-temperaturestate is the only desorption state observed on theoxidized uranium surface. The first-order depen-dence of this feature on coverage suggests thatOH bond breaking is the rate limiting step in theOH decomposition to two adsorbed oxygens anddihydrogen. The low-temperature state is assignedto recombination of Hads on the remaininguncovered uranium surface. The presence ofadsorbed OH has affected the electronic structureof the remaining uranium such that the bindingenergy of the adsorbed H is lower than on theclean uranium. Other factors that favor thisassignment include the saturation of this state atlow coverage and the absence of the low-tempera-ture state on the oxidized surface. Because of thehigher binding energy of the hydrogen to oxygenor oxide on the uranium surface, a likely hydridenucleation site could be at the edge of oxygen oroxide islands on surfaces or at interfaces such asgrain boundaries.

References

1. M. A. Schildbach and A. V. Hamza, Phys. RevB 45, 6197 (1992).

2. M. Moalem, M. Balooch, A. V. Hamza,W. J. Siekhaus, and D. R. Olander, Journal ofChemical Physics 99, 4855 (1993).

3. G. L. Powell, W. L. Harper, and J. R. Kirkpatrick,Journal of Less-Common Metals 172–174, 116(1991).

4. K. Winer, C. A. Colmenares, R. L. Smith, andF. Wooten, Surface Science 183, 67 (1987).

Publication

1. M. Balooch and A. V. Hamza, “Hydrogen andWater Vapor Adsorption on and Reaction withUranium,” J. Nuclear Materials 230, 259 (1996).

Invited Presentation

1. M. Balooch, A. V. Hamza, and W. J. Siekhaus,”Hydrogen and Water Vapor Adsorption onand Reaction with Uranium,” The 20thCompatibility, Aging and Stockpile StewardshipConference, Allied–Signal Federal Manufactur-ing & Technologies, April 30–May 2, 1996,Kansas City, MO.

MODELING HYDRIDE FORMATION



P. A. Sterne

The goal of this work is to use electronicstructure calculations to address the interactionbetween hydrogen gas and the uranium surface toform uranium hydride. Although the mechanismand site and defect sensitivity of the hydrideformation are unknown, we hope to establish akinetic model to describe the hydrogen adsorp-tion, diffusion, and subsequent hydride formationon the uranium surface.

A complete model of hydride formationrequires knowledge of energy differences betweenhydrided and non-hydrided phases, energybarriers for diffusion, and a representation of theinfluence of both point defects, e.g., impurities onthe uranium surface, and extended defects, suchas grain boundaries. The approach adopted hereis to determine parameters from first principlescalculations where possible and to correlate theresulting values with experimental data. Thiscross-validation is essential to strengthen ourunderstanding of the mechanisms contributingto the hydride formation, thereby validating themodeling procedure.

Uranium hydride forms in two separate phases.(1) The metastable alpha UH3 phase has the cubicA15 structure with an 8-atom unit cell in whichthe uranium atoms forming a body-centered-cubiclattice and the hydrogen atoms occupy tetrahedralsites. (2) This transforms to the more complicated32-atom UH3 beta phase, which is also cubic. Theuranium atoms in this structure are located on thesites of an A15 lattice with the hydrogen atomsarranged in lower symmetry positions.

In this initial stage, electronic structurecalculations have been performed for the alphaand beta uranium hydride phases. For compari-son, calculations have also been performed on theorthorhombic-alpha phase and high-temperaturebcc phase of pure uranium. The calculations havebeen performed using the Linear Muffin Tin Orbitalmethod and the Local Density Approximation toelectron–electron interactions. All calculationsare fully relativistic and include the sizable effectsdue to spin–orbit coupling.

Figure 1 shows the densities of states for thepure uranium phases. The features in the densities

Figure 1. Densitiesof states (states/Rydberg/atom) foruranium in the bodycentered cubicstructure (upperpanel) and theorthorhombic alpha-Uranium structure(lower panel).Calculations wereperformed at theexperimental vol-umes in both cases.

0

50

100

150

-10 -5 0 50

50

100

150

bcc Uranium

alpha Uranium

Energy (eV)

D.O

.S.



of states are similar for the bcc and orthorhombic-alpha phase, although the alpha phase does showmore structure due to its lower symmetry. Thedensities of states are dominated by f-electron stateswith a bandwidth of about 3.2 eV. These f-stateslie in a much broader density of states associatedwith the more delocalized s–d valence electrons.

The densities of states for the UH3 phasescontrast sharply with those of the pure phases.Figure 2 shows both alpha- and beta-phase UH3densities of states. The states around 9 eV belowthe Fermi energy are hydrogen 1s states. Thehydrogen atoms essentially form H2– ions in thelattice, removing the s–d valence electrons fromthe uranium atoms. This interaction lowersf-electron bands below the delocalized s–d bandsand significantly narrows the uranium f-electrondensity of states from 3.2 eV in the pure phase toabout 2 eV in the hydride, increasing the localiza-tion of the uranium f-electrons in the hydride.

The localization of the f-electrons is driven inpart by the larger lattice constant of the hydride.The alpha UH3 phase has the same arrangementof uranium atoms as bcc uranium, but the latticeconstant is 17% larger. The hydrogen appears toplay two roles in the hydride. (1) It removeselectrons from the uranium, on average removing0.5 electrons per hydrogen. (2) It spaces theuranium atoms apart, reducing the f-electronbonding between them. This second effect is

Figure 2. Densitiesof states (states/Rydberg/formulaunit) for uraniumhydride in the UH3alpha phase (upperpanel), and the UH3beta phase (lowerpanel). Note theadditional H-1sfeature around 9 eVbelow the Fermienergy and thesignificant narrow-ing and lowering ofthe f-electron statesat the Fermi energywhen comparedwith the calculationsfor pure uranium.

0

100

200

300

400

-10 -5 0 50

100

200

300

400

alpha UH3

beta UH3

Energy (eV)

D.O

.S.

responsible for the narrowing of the f-electronbands in the hydride, which is indicative ofincreased f-electron localization. This localizationeffect is similar in nature to the f-electron localiza-tion observed in the alpha–delta transition inplutonium, where there is a similarly pronouncedvolume change. Theoretically, both effects aredifficult to treat and are the subject of extensiveongoing research.

The calculations indicate that there aredramatic differences in the electronic structureupon hydriding uranium. These differencesshould be readily observable spectroscopically,either by looking for the characteristic hydrogen1s state about 9 eV below the Fermi energy or byfocusing on the significant changes in the densityof states around the Fermi energy. There, thef-electron states narrow significantly and dropbelow the bottom of the s–d valence electrons inthe hydride.

Energetic comparisons between the phasesare complicated by the difficulties in correctlydescribing the electron–electron interaction incorrelated systems with f-electrons. The calcu-lated equilibrium lattice constants for the hydridedphases are about 10% smaller than the experimen-tally observed values. Although this is a largedifference, it is consistent with other calculationson f-electron systems which routinely calculatelattice constants 6–12% smaller than experiment.



This is a well-known anomaly in local densityfunctional theory when applied to localizedf-electron systems. In contrast, calculations ontransition metals generally give parameterswithin 1–2% experiment.

A considerable effort has been made to extractestimates of energy differences between phasesfrom these calculations. However, we have beenunable to establish a reliable set of energy differ-ences. The large discrepancy in equilibrium latticeconstants prohibits us from finding a reliablebasis for comparison of the total energies from thepresent calculations. It is hoped that subsequentfull-potential calculations may be able to circum-vent this problem.

Full potential methods are also being used tocalculate the most stable position for a hydrogenatom on a uranium surface. These calculationsare still at a preliminary stage, and results havenot yet been obtained.

SYNTHESIS OF NEW EXPLOSIVES



P. F. Pagoria, A. R. Mitchell, R. D. Schmidt, L. E. Fried

Introduction

Our major effort has been the development of anew synthesis of 1,3,5–triamino–2,4,6–trinitrobenzene(TATB), an insensitive high explosive (IHE) whichis a critical component to the enduring stockpile.We developed a two-step process that incorporatesa new aminating reagent discovered at LLNL,which should be superior to the current synthesisof TATB. We are also investigating the synthesisof new energetic materials comprised of nitro- andamino-substituted heterocycles. We divide thisresearch into two main areas: (1) insensitiveenergetic materials which approach HMX in energy,and (2) new energetic materials that calculate tohave more energy than CL–20. We synthesizedthe new, insensitive, energetic heterocyles, 2,6–diamino–3,5–dinitropyrazine–1–oxide (PZO) and4–amino–3,5–dinitropyrazole (ADNP) with 80–85%the energy of HMX and excellent thermal stability.We have begun an investigation of a series ofnitro-substituted, bicyclic heterocycles, which ourcalculations predict to have more energy than CL–20.We are conducting this study, collaboratively,with our theoretical modeling group—the struc-tures of the target molecules were generated byL. Fried and the synthetic schemes were devel-oped by the synthesis group. A continuing effortin our program is the synthesis of insensitiveenergetic materials by the Vicarious NucleophilicSubstitution (VNS) of hydrogen, which incorpo-rates 1,1,1–trimethylhydrazinium iodide (TMHI)

as an aminating reagent for nitro-substitutedaromatic and heterocyclic compounds.

TATB Synthesis

Currently, TATB is not being manufacturedin the United States. The traditional manufactur-ing technique is a multi-step process that involveschlorinated aromatic species and reaction steps,which require high temperatures and pressures.Our new synthesis is only a two-step reactionsequence starting from an inexpensive startingmaterial (4–nitroaniline) used extensively incommercial dye synthesis (Fig. 1). The two reactionsteps use ambient conditions and incorporatereactants, which are available from surplusmunitions, e.g., uns-dimethylhydrazine (UDMH),a surplus propellant ingredient from the formerSoviet Union currently being targeted for demili-tarization. The synthesis is based on the VicariousNucleophilic Substitution (VNS) of hydrogen,1 aninnovative method for the synthesis of insensitiveenergetic materials by the addition of amino- groupsto known explosives. The key amination step usesTMHI, first recognized in our group as an aminat-ing reagent in the VNS of hydrogen. Thus far, ourprocedure has been performed at the laboratory-scale, although plans are in place to produce 1 kgof material. It is superior to the traditional proce-dure in that it involves no chlorinated species, thereaction sequence is done at room temperature,and preliminary results indicate that the product

Figure 1. New synthesis of TATB developed at LLNL.

NH2

NO2

NH2

O2N NO2

NO2

NH2

O2N NO2

H2N NH2

NO2

KNO3

r.t.H2SO4

TMHI

DMSONaOMe

Picramide TATB 4-Nitroaniline



is a high-quality material, with virtually no organicimpurities. This new method of synthesis maylead to a significant decrease in the overall cost ofTATB, which may facilitate expansion of the useof TATB in industry. A few of our industrialaffiliates expressed interest in our synthesis inresponse to inquiries they had for the use of TATBin deep-hole mining operations because of itsextraordinary thermal stability. The detonatordevelopment department at LLNL has alsoshown interest in our new TATB because it mayhave advantages over the crash-precipitatedmaterial in corner-turning ability and symmetryof detonation.

This year, we spent time on research anddevelopment of the VNS process for synthesizingTATB, employing some new cost-saving measureswhich make the process even more attractive. Wehave optimized the conditions for the synthesis ofTATB from picramide, using TMHI in the presenceof sodium methoxide. We have performed thissynthesis on the 10 g scale providing 95% yield ofTATB. The material shows no organic impurities,although the onset of the decomposition point isslightly lower than wet-aminated TATB. Wehave optimized the synthesis of picramide from4–nitroaniline, allowing us to use the crudereaction product directly in the next step, saving apurification step for the picramide. We have alsodemonstrated that we can generate the TMHI inthe dimethylsulfoxide (DMSO) solution directly

(from UDMH and methyl iodide) and then runthe TMHI amination. This would eliminate theisolation and purification step of TMHI, thusmaking the process more attractive, economically.

Other Research

As stated, we divided our synthetic work intotwo main areas: (1) synthesis of new energeticcompounds that outperform HMX without increasein sensitivity, and (2) synthesis of new IHE’s withmore energy than TNT. From this latter effort, wesynthesized PZO,2 an insensitive energetic materialwith 30% more energy than TNT. The energycontent and thermal stability of this materialmake it very interesting for several applications,including insensitive boosters and detonators.We investigated two new approaches to thesynthesis of PZO, which are more environmentallyfriendly and have a shorter reaction sequencethan the previous method.2 (1) The first methodinvolves the condensation of alloxan monohydrateand 5,6–diaminouracil in water to yield 2,4,5,7–tetrahydroxy[5,4–g]pyrimidopteridine3 whichwas treated with aqueous sodium hydroxide at170˚C in a Parr pressure vessel to yield 2,6–diamino-3,5–dicarboxypyrazine.4 (2) We attemptedto nitratively decarboxylate this material to 2,6–diamino–3,5–dinitropyrazine (ANPZ), but havethus far been unsuccessful (Fig. 2). We alsoinvestigated the synthesis of PZO by the synthetic

Figure 2. Alternative approach to the synthesis of ANPZ and PZO.

OHOH

OH

HO

OH

OO

O

O

H

OHN

N

NNN

NNN

H

N N N

NN

N

H2O

HNO3

H2SO4

H2ONaOH170˚C

H2N

H2N

CO2HNN

NNH2N NH2

HO2CNO2NN

NNH2N NH2

O2N

+

N N

NN

N

N



Figure 3. New Approach to the synthesis ofANPZ (and PZO).

Cl

N

N

NNN

NN N

NaOCH3

CH3OH

NH4OH

∆

∆

H2SO4

SO3

Cl Cl

NO2NN

NNCl OCH3

O2NNO2

OCH3

HNO3

NN

NNH2N NH2

O2N

Figure 4. Synthesis of the new VNS aminatingreagent, 1,1,1–trimethylhydrazinium iodide(TMHI).

UDMH TMHI

CH3X(CH3)2N-NH2

THF(CH3)3N+-NH2 I-

scheme shown in Fig. 3. We completed thesecond step and are preparing to aminate the2–chloro–6–methoxy–3,5–dinitropyrazine. Thissynthesis looks very encouraging and should beeasily scaled-up.

We have had some excellent results using ourVNS5 approach to the synthesis of new insensitiveenergetic materials. The VNS approach incorpo-rates the use of an aminating agent in the presenceof base to introduce amino groups onto electro-philic aromatic rings. This is analogous to aFriedl–Crafts reaction in that the amino groupformally replaces a hydrogen on the aromaticring. In FY94, we successfully converted TNT to2,5–diamino–2,4,6–trinitrotoluene (DATNT), aninsensitive explosive of interest to the Departmentof Defense, in one-step at ambient temperatureusing 4–amino–1,2,4–triazole (ATA).6 This is asignificant improvement over the previouslypublished procedures for the synthesis of thiscompound which required four to five steps andan expensive starting material such as 3,5–dichloroanisole.7 As previously mentioned, wealso used our VNS approach for the synthesis of1,3,5–triamino–2,4,6–trinitrobenzene and 1,3–diamino–2,4,6–trinitrobenzene (DATB) frompicramide,5

using both TMHI and ATA as the

aminating reagents.In FY95, we developed a new VNS aminating

reagent, TMHI, which may be synthesized byalkylation of UDMH (see Fig. 4). It may also besynthesized directly from the inexpensive reagent,hydrazine, by alkylation with methyl iodide in thepresence of base. This reagent acts analogously toATA, although FY96 studies have shown it to besignificantly more reactive. In a comparison test,TMHI converts picramide to TATB by stirring

overnight at room temperature (although actualtime to completion is probably around 3 hr) whileATA takes at least 30 hr for complete conversion.In FY96, we also demonstrated the use of TMHIin the synthesis of DATNT from TNT in one stepin 50% yield.

We have synthesized the previously unknown4–amino–3,5–dinitropyrazole (ADNP), a poten-tially insensitive energetic material with 80% theenergy of HMX, using the VNS method withTMHI. This is significant because ADNP couldnot be synthesized by ordinary methods. Duringour study of the scope and limitations of TMHI asa VNS aminating reagent, we found that thenumber of amino groups which may be added tothe electrophilic aromatic ring is equal to thenumber of nitro groups present on the substrate.This observation led us to investigate the aminationof 3,5–dinitropyrazole (DNP), which carries anacidic hydrogen, to give ADNP. We reasonedthat the acidic proton on DNP would initiallyreact with one equivalent of base to form a stablenitronate anion, leaving the second nitro-groupavailable to participate in the VNS amination. Wefound the reaction of DNP with TMHI in thepresence of excess potassium tert-butoxide gaveADNP in 70% yield (Fig. 5). This allowed thesynthesis of ADNP without the need of a protectinggroup for the pyrazole proton and demonstrateda rare example of a nucleophilic substitutionreaction on an aromatic ring already bearing anegative charge. The structure of ADNP was

Figure 5. Synthesis of 4–amino–3,5–dinitropyrazole (ADNP).

NN

NNDMSO

TMHI

t-BuO- K+NO2

H H

NH2O2N

NN

NNNO2

O2N



confirmed by x-ray crystallographic analysis,isolated as a 1:1 complex with dimethylsulfoxidewith a crystal density of 1.608 g/cc.8 We foundthat refluxing in water breaks up the complex andcrystallizes ADNP as white cubes. An x-raycrystallographic analysis showed this material tobe a 1:1 complex of ADNP/H2O with a crystaldensity of 1.72 g/cc.8 We are in the process ofevaluating the safety and performance character-istics of this new compound.

We have continued our investigation of thesynthesis of several nitro- and amino-substitutedazoles. Calculations predict that these targetcompounds have 80–95% the energy of HMX andto be relatively insensitive materials. The synthe-ses are based on VNS chemistry in which knownpolynitroazoles will be aminated with TMHI. Wehave been concentrating on the synthesis of aseries of amino-substituted polynitropyrroles(Fig. 6), using both protected and non-protected

Figure 6. Amination of polynitropyrroles withTMHI.

NN

NN NN

NN NN

NN

NN

NN

O

O

+

+–

–

O2N

NO2

NO2

NNNN

NNNN NN

NN

NN NN

NO2 NO2

NO2O2N

NN

NNNO2

H H

NH2O2N

O2N

O2N

O2NNN

NO2

O2N

NNDMSO

TMHI

t-BuO- K+NO2

H H

NH2O2N

NNNO2

O2N

NNNO2

H H

NH2

O2NNN

NO2

Figure 7. Targetmolecules withcalculatedenergy greaterthan CL–20.

polynitropyrroles as substrates for aminationwith TMHI. We have attempted the amination ofboth 1–t-butyl–2,4–dinitropyrrole and 1–t-butyl–2,3,4–trinitropyrrole, yielding only recoveredstarting material. We have synthesized both the2,5– and 2,4–dinitropyrrole and are currentlyinvestigating the amination of these materials.

We have begun an investigation of the synthesisof several energetic heterocycles, which ourcalculations predict will have more energy thanCL–20 (see Fig. 7). These target molecules are allplanar, high nitrogen, zero-hydrogen nitroazoles.They are oxygen balanced to CO2 and should havea significant energy contribution from their highheat of formation. We have outlined syntheticroutes to each of these based on literature prece-dence. We concentrated our initial efforts on thesynthesis of 4,5–diamino–2–phenyl[1,2,3]triazole,9

an important intermediate to two of our targetmolecules. We investigated new methods for thesynthesis of this material because reported methodsuse the highly toxic gas, cyanogen, as the startingmaterial. We have chosen to use thiooxamide,10 areadily available, inexpensive starting material, asa cyanogen surrogate and to follow proceduressimilar to those used with cyanogen. We havealso been investigating the synthesis of 3–bromo–4–formyl–1–tosylpyrazole, as an intermediate tothe bipyrazole target compound. Work on thesetwo syntheses continues.

Conclusions

We developed a new synthesis of TATB whichmay be superior to the current manufacturingprocess. We have synthesized a series of insensitiveenergetic materials with good thermal stabilityand have begun a study of the synthesis of newenergetic heterocyles that are predicted to be moreenergetic than the industry standard, CL–20.



References

1. M. Makosza and J. Winiarski, Acc. Chem. Res.20, 282–89 (1987).

2. D. S. Donald, U.S. Patent 3,808,209, Apr. 30,1974.

3. E. C. Taylor, C. K. Cain, and H. M. Loux,J. Am. Chem. Soc. 76, 1874 (1954).

4. E. C. Taylor, H. M. Loux, E. A. Falco, andG. H. Hitchings, Ibid. 77, 2243 (1955).O. Baudisch and D. Davidson, J. Biol. Chem.71, 497 (1927).

5. (a) P. F. Pagoria, A. R. Mitchell, andR. D. Schmidt, “Vicarious Nucleophilic Substi-tution to prepare 1,3–diamino–2,4,6–trinitrobenzene or 1,3,5–triamino–2,4,6–trinitrobenzene,” U.S. Patent 5,569,783, Oct. 29,1996. (b) P. F. Pagoria, A. R. Mitchell, andR. D. Schmidt, “New Aminating Reagents forthe Synthesis of 1,3,–triamino–2,4,6–trinitrobenzene (TATB) and Other InsensitiveEnergetic Materials,” International Symposiumon Energetic Materials Technology, ADPAMeeting #680, Phoenix, AZ, Sept. 24–27, 1995.

6. A. R. Katritzky and K. S. Laurenzo, J. Org.Chem. 51, 5040–41 (1986).

7. A. P. Marchand and G. M. Reddy, Synthesis,261–2 (1992). S. Iyer, J. Energetic Mat. 2, 151(1984).

8. X-ray crystallographic analysis performed byR. Gilardi, Naval Research Laboratory,Washington, D.C.

9. J. Thiele and K. Schleussner, Ann. der Chem.295, 129–72 (1897).

10. N. V. Koshkin, Analit. Khim. 20, 534–9 (1965).

Publications

1. P. F. Pagoria, A. R. Mitchell, and R. D. Schmidt,J. Org. Chem. 61, 2934 (1996).

2. P. F. Pagoria, A. R. Mitchell, R. D. Schmidt,C. L. Coon, and E. S. Jessop, “New Nitrationand Nitrolysis Procedures in the Synthesis ofEnergetic Materials,” in Nitration: RecentLaboratory and Industrial Developments;L. F. Albright, R. V. C. Carr, R. J. Schmitt,eds., ACS Symposium Series 623; AmericanChemical Society: Washington, DC, 1996.

3. M. F. Foltz, D. L. Ornellas, P. F. Pagoria, andA. R. Mitchell, J. Mat. Sci. 31, 1893–1901 (1996).

4. P. F. Pagoria, A. R. Mitchell, and E. S. Jessop,J. Prop., Expl., Pyrotech 21, 14–18 (1996).

Presentations

1. P. F. Pagoria, A. R. Mitchell, and R. D. Schmidt,“New Aminating Reagents for the Synthesis ofTATB and other Insensitive Energetic Materi-als,” presented at the International Symposiumon Energetic Materials Technology, ADPAMeeting #680, Phoenix, AZ, Sept. 24–27, 1995.

2. P. F. Pagoria, A. R. Mitchell, and R. D. Schmidt,“Vicarious Amination of Nitroarenes withTrimethylhydrazinium Iodide,” presented atthe 211th American Chemical Society NationalMeeting, New Orleans, LA, March 24–28, 1996.

Patents

1. P. F. Pagoria, A. R. Mitchell, and R. D. Schmidt,“Vicarious Nucleophilic Substitution toPrepare 1,3–diamino–2,4,6–trinitrobenzene or1,3,5–triamino–2,4,6–trinitrobenzene,” U.S.Patent 5,569,783, Oct. 29, 1996.

2. Patent Application IL-9628. “Vicarious Nucleo-philic Substitution using 4–amino–1,2,4–triazole,hydroxylamine or O–alkylhydroxylamine toprepare 1,3–diamino–2,4,6–trinitrobenzene or1,3,5–triamino–2,4,6–trinitrobenzene.”

AB INITIO STUDY OF ENERGETIC MATERIAL DECOMPOSITION MECHANISMS


AB INITIO STUDY OF ENERGETIC MATERIAL

DECOMPOSITION MECHANISMS

L. E. Fried, C. J. Wu

Introduction

Science-based stockpile stewardship requiresLLNL to predict the chemical behavior and stabilityof energetic materials over a wide range of timescales and environments. Nevertheless, little isknown regarding the detailed molecular mecha-nisms of energetic material decomposition. As awidely used high-explosive and monopropellant,hexahydro–1,3,5,–trinitro–1,3,5,–triazine (RDX) isone of the most thoroughly studied energeticcompounds. Therefore, we have chosen RDX asan initial test case to develop methods to calculatethe decomposition pathways of energetic materials.

There are many suggested initial unimolecularsteps in the thermal decomposition of RDX,including N–NO2 bond rupture, concerted ringfission to three CH2N2O2, C–N bond cleavage,C–H bond dissociation, and transfer of an oxygenfrom the NO2 to an adjacent CH2. Among them,the most supported mechanism for condensedphase decomposition is N–NO2 bond rupture,

shown in Fig. 1(a), as Path I. The most recentsupporting evidence for this mechanism wasgiven by the transient infrared (IR) laser pyrolysisexperiments of Wight and Botcher.1–3

Another competing mechanism, which hasconvincing experimental evidence, is the con-certed symmetric ring fission to three CH2N2O2molecules shown in Fig. 1(b) as Path II. Using IRmultiphoton dissociation (IRMPD) of RDX in amolecular beam, Zhao and Lee4 suggested thatthe RDX molecule dissociates via both paths (I andII). They concluded that the dominant channel isthe symmetric ring-fission, not the N–NO2 bondcleavage. However, no conclusive evidence forthe ring-fission has been reported from otherstudies of liquid and solid RDX.

This paper presents a detailed study ofunimolecular dissociation of RDX via paths I andII, using several recent gradient-corrected densityfunctional theory (DFT) methods. Our resultsprovide a direct comparison with the gas phaseRDX experiment,4 and therefore enhance ourunderstanding of this issue.

Figure 1. (a) Calculated transition states for N–NO2 bond breaking (Path I), and (b) concerted ringdissociation (Path II).

(a) Path I (b) Path II



Computational Details

All calculations were performed using theGaussian 94/DFT package with spin-polarizedgradient-corrected exchange and correlationfunctionals. The procedures are based on the spinunrestricted Kohn–Sham formalism. We chosefour widely adopted and promising functionals:B-PW91, B3-PW91, B-LYP, and B3-LYP. B refers toBecke’s 1988 gradient-corrected exchange func-tional5 derived from the two-particle density matrixof the helium atom, and B3 is denoted for Becke’shydrid method of mixing Hartree–Fock (HF)exchange energy into the exchange functional.PW91 and LYP are the gradient-corrected correla-tion functionals of Perdew–Wang 91 and Lee,Yang, and Parr.6 Four basis sets were used forthe Kohn–Sham orbital expansion, including thecontracted Gaussian-type Dunning’s Double-ζvalence (D95V), D95V plus polarization functionson all atoms [D95V(d)], D95V plus diffuse func-tion, and Dunning’s most recent correlationconsistent polarized valence double zeta basissets (cc-pVDZ).7

N–NO2 Bond Cleavage

Using the zero point energy correction, wepredict that the N–NO2 bond energy D0(N–NO2)is 36.5 ± 2.3 kcal/mol, significantly smaller thanthe previous estimate8 of ~48 kcal/mol. There aretwo possible reasons for the smaller D0(N–NO2)value in RDX—either geometric or electronicrelaxations occur through the RDX ring fragment(H6C3N5O4). We have calculated that RDXgeometric stabilization is between (2–6 kcal/mol);thus, electronic relaxation is the dominant mecha-nism, accounting for 6–10 kcal/mole.

To calculate the N–NO2 bond dissociationbarrier, we mapped the potential energy curve (seeFig. 2) along the reaction coordinate using B-PW91functionals and the D95V basis set. The potentialenergy profile confirms the previous assumptionthat the backward reaction has a zero barrier (typicalcase for a radical recombination reaction). Con-sidering zero point energy correction, the barrierfor breaking the N–N bond is 36.8 ± 2.3 kcal/mol,slightly larger than D0 (N–NO2).

Figure 2: Predictedenergetics of N-NO2bond breaking inRDX.

1

0

10

20

30

40

50

2 3 4 5

R(Angstroms)

E (

kcal

/mol

e)

BPW91/D95VB3LYP



Concerted Symmetric Ring Fission

Using the largest basis set cc-pVDZ, fourfunctionals give D0(C–N), ranging from 66.0 to45.5 kcal/mol, with an average value of55.2 ± 10.2 kcal/mol. Using the zero point energycorrection of 9.2 kcal/mol, our best estimate ofthe heat of reaction for the concerted ring fissionis 46.0 ± 10.2 kcal/mol. We then mapped thepotential energy profile of the ring fission toidentify the transition state. At each C–N separa-tion, we optimized all other degrees of freedom atboth B-PW91/D95V and B-PW91/D95V+ level.

Summary

We investigated the mechanism of the gasphase unimolecular decomposition of RDX, apply-ing a range of basis sets and modern gradientcorrected density functional methods to thisproblem. We found that the activation barrier forconcerted ring fission is roughly 12 kcal/molgreater than that for N–NO2 bond rupture. Thisvalue holds qualitatively for a variety of compu-tational methods, suggesting that thermal gasphase decomposition at temperatures signifi-cantly under 12 kcal/mol (7200K) most likelyproceeds via N–NO2 bond rupture. Of course,detailed comparision requires the reliable evalua-tion of the reaction prefactors—a problem notaddressed here. Our results could profitably beused in constructing improved classical potentialsfor RDX, allowing a more detailed comparision ofthe branching ratios of the two mechanisms as afunction of temperature and excitation mechanism.

References

1. C. A. Wight and T. R. Botcher, J. Am. Chem.Soc. 114, 8303 (1992).

2. T. R. Botcher and C. A. Wight, J. Phys. Chem.97, 9149 (1993).

3. T. R. Botcher and C. A. Wight, J. Phys. Chem.98, 5541 (1994).

4. E. J. H. X. S. Zhao and Y. T. Lee, J. Chem. Phys.88, 801 (1987).

5. A. D. Becke, J. Chem. Phys. 88, 1053 (1988).6. C. Lee, W. Yang, and R. G. Parr, Phys. Rev. B

37, 785 (1988).7. J. T. H. Dunning, J. Chem. Phys. 90, 1007 (1989).8. C. F. Melius and J. S. Binkley, 21st Symposium

(International) on Combustion, The CombustionInstitute 1953 (1984).

Publication

1. C. J. Wu and L. E. Fried, “Ab Initio Study ofRDX Decomposition Mechanisms,” LawrenceLivermore National Laboratory, Livermore,CA, UCRL-JC-125718 (1997); submitted to J.Phys. Chem.

Presentation

1. L. E. Fried, “LLNL Energetic Material DesignProgram,” invited, NRL Workshop onEnergetic Materials, Annapolis, MD, Dec. 1996.

FIRST POSITION ANNIHILATION LIFETIME MEASUREMENT OF PLUTONIUM


FIRST POSITRON ANNIHILATION LIFETIME

MEASUREMENT OF PLUTONIUM

C. Colmenares, R. H. Howell, D. Ancheta, T. Cowan, J. Hanafee, P. Sterne

We made the first measurement of defects inan aged sample of δ phase, gallium-stabilizedplutonium, using positron annihilation lifetimespectroscopy. This measurement validates theprocedure necessary to perform measurements onthis highly toxic material and to obtain datarepresentative of sample conditions. Comparisonof the results of positron annihilation lifetimeanalysis with calculated values suggests thathelium-filled vacancies or vacancy clusters domi-nate the defect population. Such defects are thenecessary precursor to void growth and swelling.

Introduction

The evolution of defects, resulting from theradioactive decay of plutonium during its life inthe stockpile, is one of the unknown quantitiesaffecting our confidence in predictions of the limiton stockpile components. Radiation damage leadsto changes in the size and strength of metals studiedfor reactor and accelerator use and similar effectsmay be expected in plutonium. The evolution ofradiation-produced vacancies into larger voidstructures and accompanying macroscopicswelling may occur in plutonium at some age. Adetailed understanding of the defects in self-irradiated plutonium is required to predict thetimescale of void swelling and related radiationeffects. We have made the first measurement ofdefects in an aged sample of δ-phase plutonium,using positron annihilation lifetime spectroscopy.

Positron annihilation lifetime spectroscopy isan established tool for the analysis of vacanciesand voids in metals and compounds. There aremany techniques that can identify the species andsize of impurities. There are few, however, thatcan detect open volume defects such as atomicvacancies or voids. All positron annihilationspectrographic methods are particularly sensitiveto determining low concentrations of the class ofdefects including vacancies, voids, and negativelycharged defects. This sensitivity stems from theattractive interaction between these defects andpositrons which often binds the positron to the

defect site. Using positron annihilation lifetimespectroscopy determines the concentration andsize or charge of defects in metals, semiconductors,and molecular or organic compounds.

Positron annihilation lifetime spectroscopymeasures the electron density at the annihilationsite. Since the electron density is sensitive to thedefect volume there is a distinct correlationbetween the positron lifetime and defect size. Inmetals and some compounds, the correlationbetween lifetime and defect size can be calculatedfrom first principles with sufficiently high accu-racy to differentiate between major defect classessuch as vacancies, vacancy clusters, and voids.Defect concentrations can be determined from thefraction of positrons annihilating at the defect site.


A newly developed high energy positronbeam and lifetime spectrometer, located in afacility capable of containing plutonium materialenabled our measurements. Our high energybeam is derived from a 100 mCi 22Na sourcemoderated by a 2 µm-thick tungsten single-crystalfoil positioned in the terminal of a 3 MV Pelletronelectrostatic accelerator. This beam contains acurrent of 5 × 105 e+s–1 at 3 MeV. LLNL is theonly U.S. facility to offer this capability.

To perform measurements, a sample of agedplutonium was vacuum-encapsulated in theLLNL plutonium facility in a holder with two1-mil SS windows. After transportation to themeasurement facility, positrons accelerated in abeam to 3 MeV were introduced into the capsulethrough the thin overlaying windows. Timingsignals to form a positron annihilation lifetimedistribution were obtained from the positrons asthey were implanted and after annihilation.

We performed calculations and experimentsto design the sample capsule and to determinepotentially useful configurations of beam anddetectors for the lifetime measurement. Lifetimemeasurements with the Pelletron were conductedon uranium and other metals held in a capsule



similar to the plutonium capsule to determine thecharacteristics of the lifetime system and to obtain apreliminary design for the placement of the lifetimedetectors. Experiments using both dummy samplesand the encapsulated plutonium sample wereperformed to determine the effects of backgroundcomponents and to optimize the detector-samplegeometry to obtain high quality data.

During these measurements, we continued torefine our spectrographic technique. Initial datashowed unambiguous features related to defects inthe material, but detailed analysis demonstratedthat cleaner data and additional measurements werenecessary to provide accurate values of defectparameters. Improved detectors and changes indetector and sample geometry resulted in spectrawith lower contributions from backgrounds andartifacts, therefore, additional measurements wereperformed. Refinement of the experiment throughgeometric improvements and better detectors willcontinue, and we are storing our samples in anarchive for additional measurements.

Results and Discussion

A first measurement of aged plutonium ontwo separate 21-yr-old, δ-phase samples from thesame source has been performed. Figure 1 showsthe time distribution of the annihilation of implantedpositrons. Reduction of these data into annihilationrates is performed by least-squares fit of a spectrummodel that includes one or more separate annihi-lation rates, a background from uncorrelatedgamma rays, and the time resolution of thedetector system. Analysis of the plutonium dataproduces only two identified annihilation life-times. A single lifetime of 184 ps is found fromthe plutonium sample and a small contribution,~1%, is identified by measurements of othermaterials as background from positrons annihilat-ing in the wrapping of the detector system with alifetime of 800 ps.

The results of the lifetime analysis show thatall positrons are annihilating in the same condition.

Figure 1. Time distribution of positron annihilations in aged δ-phase plutonium. Inset lines arecalculated annihilation rates for defect free material, vacancies, and helium-filled vacancies.

100

101

102

103

104

105

Cou

nts

-500 0 500 1000 1500 2000

t (ps)

2500 3000

Defect Free

Empty Vacancy

He filled vacancy

25 year old delta Pu

Lifetime: 184 ps 100% intensity



If all the positrons are trapped, then the trapconcentration must be of the order of 1 part per1000 or higher. If two separate trapping defectssuch as single vacancies and voids were compet-ing for the positron, then we would see separatecontributions in the lifetime spectra from the twodefects. Also, the relatively short value of 184 pssuggests that the traps are not large empty voids.So these data are showing a relatively strongconcentration of defects with lifetimes too smallto be empty voids.

To identify the annihilation state and thedefect species in plutonium requires knowledgeof the annihilation rates for each kind of defect.Many of these values can be obtained by ab initiocalculation. First calculations for the values ofpositron lifetimes were performed for defect-freeplutonium and specific plutonium defects. Defect-free, δ-phase plutonium was predicted to have alifetime of 139 ps and a single vacancy in anundeformed lattice 249 ps. Adding helium to avacancy will add electrons and lower the annihi-lation lifetime. Initial calculations for vacancies

containing helium were predicted to have annihi-lation lifetimes between 167 ps and 193 ps,depending on the number and location of thehelium atoms. Comparison of these values to theanalysis of our data at 184 ps suggests a highconcentration of helium-stabilized defects in agedmaterial. This conclusion must be regarded astentative since several defect-helium combinationsremain to be calculated and we have not yet madeverifying measurements of defect free material.

A high concentration of helium-stabilizedvacancies or vacancy clusters is a necessarycondition for void growth and swelling. Vacanciesproduced by the recoiling atoms during theradioactive decay are lost to recombination withinterstitial atoms, grain boundaries, impurities,and other microstructrual features unless they arestabilized. All models of void swelling proceedfrom migration of vacancies that have been stabilizedby helium. Thus, in 21-yr-old plutonium we havefound the precursor to void swelling, but noevidence of voids in the samples measured.

MICROSTRUCTURAL EVOLUTION IN WELDS



J. W. Elmer, J. Wong

Spatially resolved x-ray diffraction (SRXRD) isbeing used for in-situ mapping of phases duringwelding where severe temperature gradients, high-peak temperatures, and rapid thermal fluctuationsoccur as the heat source passes through the material.These temperature fluctuations non-uniformly alterthe microstructure and properties of the material tocreate a heat-affected-zone (HAZ) adjacent to theweld fusion zone. Solid state phase transformationsoccur in the HAZ and create a gradient of bothmicrostructure and mechanical properties of thewelded material that is important to the integrity ofthe resulting weld.

To adequately predict the microstructuralgradients that form in the HAZ, the kinetics ofphase transformations are required, but transforma-tion diagrams under steep thermal transients rarelyexist. For commercially pure titanium, both par-tially transformed and fully transformed regionshave been observed to exist within the HAZ.1 In thepartially transformed region, the peak temperaturesare above the β-Ti transition temperature, but thetime above this temperature is insufficient tocompletely transform the low-temperature α-Tiphase to the high-temperature β-Ti phase. In thecompletely transformed region of the HAZ, thetemperature/time history of the titanium is suchthat α-Ti transforms completely to β-Ti during theweld heating cycle, a condition which allows largegrains to form and the weld joints to weaken.

Conventional Experiments

Conventional isothermal heat treating experi-ments were conducted to simulate the partially andfully transformed HAZ regions in Grade 2 titanium.These experiments were performed by heating to afixed temperature, isothermal holding at thattemperature for 30 min, and quenching the sample inwater. At temperatures up to 920°C, the sampleshowed no significant grain growth but did exhibitsome signs of phase transformation, suggesting thatthe α-Ti phase only partially transformed to β-Ti. At930°C and above, the sample showed extensive graingrowth, indicating that the sample had completelytransformed to β-Ti and suggesting that the tempera-ture was above the β-Ti transus. Figure 1 shows

Figure 1. Microstructures of the Grade 2tatanium. (a) As received, (b) 920˚C for 30 min,showing partial transformation to the bcc phase,and (c) 930˚C for 30 min, showing the results ofcomplete transformation to the bcc phase.

(b)

920° 100×

(a)

0° 100×

(c)

930° 100×



these results and indicates the changes inmicrostructure and grain size in these simulatedweld microstructures.

Isothermal experiments such as these are usefulfor generating microstructures that are similar tothose that occur after the weld is complete; however,experiments of these types do not determine thephases that were present during welding, and theycan only be used to a limited extent in modelingthe kinetics of phase transformations in real weldsthat occur under highly non-isothermal conditions.For these reasons, SRXRD is now being used toexperimentally determine the phases that arepresent during welding and under the actualtemperature gradients imposed by the weldingheat source.

Synchrotron-Based Experiments

Point-by-Point Phase Mapping

To modify our SRXRD experimental setup,1we enclosed it in an environmental chamber toprevent oxidation of the titanium surface duringwelding.2 We then performed SRXRD experimentsto map the α-Ti plus β-Ti two-phase HAZ regionaround an arc weld in Grade 2 titanium. Figure 2shows the results of these experiments where theSRXRD measurements of the bcc + hcp two-phasefield were made along numerous rows of points at0.33-mm spacing perpendicular to the weld traveldirection. These data are plotted as the dark-shaded

Figure 2. SRXRD data comparing the experi-mentally measured bcc + hcp region (bar chart)of the HAZ in a G-2 Ti weld with the calculatedbcc/hcp isotherm (dotted line).

bars in the bar graph and indicate that the bcc + hcptwo-phase field grows more steeply on the leading(heating) side of the weld than it decays on thetrailing (cooling) side of the weld.

For comparison, superimposed on this bargraph is a line graph that overlays the calculatedα-Ti/β-Ti isotherm temperature of 920°C.1 Themodel used to calculate the isotherm1 can only beconsidered approximate since it does not take intoaccount fluid convection in the weld pool, but itdoes serve as a first-order estimate to the tempera-ture field surrounding the weld. This isothermpredicts the thermodynamic start locations for theα-Ti to β-Ti transformation on the leading side ofthe weld and the reverse transformation on thetrailing side of the weld (i.e., the calculated isothermpredicts the first occurrence of bcc on the leadingside of the weld and the first occurrence of hcpfrom the completely transformed region on thetrailing side of the weld). Comparing the calculatedisotherm with the SRXRD data shows that if theisotherm were expanded outward by approxi-mately 1.5 mm, then there would be goodagreement between the isotherm and the experi-mentally measured onset of transformation onboth the leading and trailing sides of the weld.Coupled thermal and fluid flow numerical model-ing of the weld will later be used to provide abetter estimate of the actual bcc/hcp isothermlocation through our collaboration with ProfessorDebRoy at Penn. State University.

Row-by-Row Phase Mapping

Preliminary work is being performed on atechnique that will allow the collection of diffrac-tion patterns simultaneously from a row of pointsto reduce the number of measurements requiredto map the weld HAZ from n2 to n. In this mode,a new beam geometry and detection scheme willbe used. The synchrotron beam is unfocused to aslit beam (0.25-mm vertical × 15-mm horizontal).Diffraction patterns are recorded on a large(20 × 20 cm) imaging storage plate (Fuji-type) andspatial resolution is achieved by use of a micro-Soller slit assembly positioned vertically in frontof the imaging plate. Results so far indicate thatthis row-by-row method yielded a progressiveimprovement in spatial resolution of 1 mm in ourMarch 1996 run to 0.5 mm in our August 1996 run.Figure 3 shows a row of data collected from aroom-temperature sample that consisted of a thinvanadium foil on top of bulk titanium to simulatea bcc/hcp boundary in the HAZ of a weld. The

bcc or liquid bcc+hcp Calculated isotherm

–25 –20 –15 –10 10–5 5

8

6

4

10

12

0Distance from electrode (mm)

Dis

tanc

e fr

om c

ente

rlin

e (m

m)



Figure 3. Row-by-row SRXRD patterns using animaging plate and a Soller slit assembly. Aresolution of 0.5 mm was achieved on a room-temperature sample that consisted of athin-vanadium foil on top of bulk Ti to simulatea bcc/hcp boundary in the HAZ of a weld.

bcc V hcp Ti

1 mm

References

1. J. W. Elmer, J. Wong, M. Fröba, P. A. Waide,and E. M. Larson, Metall. Mater. Trans. A 27A(3),775 (1996).

2. J. Wong, M. Fröba, J. W. Elmer, P. A. Waide,and E. M. Larson, J. Mater. Sci. 31 (1996).

Invited Talks

1. “In-Situ Phase Mapping and Real TimeChemical Dynamics using a Novel SRXRDTechnique,” IUCr Synchrotron RadiationSatellite Meeting, Advanced Photon Source,ANL, August, 1996.

2. “Real-Time Chemical Dynamics, PhaseMapping and Structure of Solids usingSRXRD and YB66,” Keynote Lecture, 13thInternational Symposium of the Reactivity ofSolids, Hamburg, Germany, September, 1996.

3. “Analysis of Heat Affected Zone PhaseTransformations Using Spatially ResolvedX-Ray Diffraction with Synchrotron Radia-tion,” INTSOL96 Conference, Trivandrum,India, November, 1996.

Other Presentations

1. “Direct Observations of HAZ Phase Transfor-mations in Titanium using Spatially ResolvedX-Ray Diffraction,” BES Center of Excellencefor the Synthesis and Processing of AdvancedMaterials—Materials Joining Workshop,Chicago, IL. April, 1996.

Soller slit will be redesigned to achieve 0.25 mmspatial resolution and will be tested at thisresolution during the next synchrotron run.

Future Work

Additional SRXRD data will be gathered ontitanium to map the complete transformation fromthe hcp-to-bcc on the leading side of the weld, andthe bcc-to-hcp reverse transformation on thetrailing side of the weld. The coupled thermal/fluid flow model will then be used to provide thetime temperature profiles for all points in the HAZ.The combined results of numerical modeling ofthe temperature field plus the SRXRD experimen-tal data of the location of the bcc + hcp two-phasefield will be used to determine the kinetic param-eters for the α-Ti to β-Ti transformation. Thesekinetic parameters can then be used to predictmicrostructural evolution in the HAZ of titaniumfor a wide range of conditions through the develop-ment of a non-isothermal nucleation and growthmodel of the transformation kinetics of titanium.

DEFORMATION MECHANISMS OF U–6NB



G. Gallegos, A. Schwartz, E. Li

Introduction

During FY96, we extended the type of experi-ments to better understand the deformationmechanism of U–6Nb. Previous work on thisproject, and input from other investigators,suggested that the alloy will undergo shape-memory at lower strains with detwinning as thedominant mechanism. At larger strains, it isassumed that slip accounts for the observedplasticity. During FY95, we performed torsionexperiments and a transmission electron micros-copy (TEM) microstructural study to verify thatslip was the operative mechanism. Results fromthese studies were inconclusive. Additionaltesting this past year has provided insight intobetter understanding the overall deformationbehavior of this alloy.

Interrupted Tensile Testson U-6Nb

Figure 1 shows the stress-stain behavior of a“typical” shape-memory alloy. After initial elasticloading, the onset of martensite detwinning results

in measurable strain with little or no additionalload. This is followed by an elastic deformation ofthe detwinned martensite, and eventually theonset of slip (or dislocation motion) occurs.

Some shape memory alloys, including U–6Nb,also exhibit a pseudoelastic behavior that isobserved on unloading at various strain levels asreported by Vandameer et al.1 Figure 2 shows thiseffect on a tensile specimen of U–6Nb that hasbeen solution treated and water quenched. TheU–6Nb specimen was loaded and unloadedseveral times. The nonlinear unloading behaviorand hysteresis loop form in a constant mannerwhen the loading occurs in the “slip” region ofthe stress-strain curve. This recovery is explainedas twinning that can occur with the removal ofapplied stress.

Examination of the specimens’ microstructurewas performed for various strain levels: Thespecimens were electropolished and then loadedin a tensile machine, followed by unloading andexamining by scanning electron microscopy (SEM)and then reloading again. Figure 3 shows themicrostructures deformed at the various strainlevels—the twinning evidence is clear, and thedirections and extent clearly changes with increas-ing strain. The interpretation of the microstructuralchanges studied is not clear at the highest strain level.

Bauschinger Tests

A strain reversal, or a Bauschinger test, wasperformed to verify that slip is operative beyondthe yield point. In a Bauschinger test, a load in thereverse direction results in lower flow stressesthan the original loading because of dislocationpile ups that occur during initial loading. Thesedislocations are easily moved when the loading isin the opposite direction, resulting in lower flowstresses. Figure 4 shows the result for a torsionalloading, reverse loading, and reloading of speci-mens. These results provide strong evidence ofdislocation motion being responsible in the largerstrains of this alloy, by reason of “softening”being evident at the larger strains on deformation,subsequent to the initial loading.

Figure 1. Typical stress-strain curve for atwinned martensitic material shows two distinctelastic regions and two distinct plasticityplateaus—the first due to twin motion, and thesecond due to slip.

Elastic deformation of detwinned martensite

Fully detwinned martensite

Onset of martensite detwinning

Onset of slip

Strain

Stre

ss



1000

800

Eng

inee

ring

Str

ess(

MPa

)

600

400

200

00

shape memory hysteresis

5

A

A'B' C' D' E' F'

B C

DE F

10Engineering Strain (%)

15 20

Figure 2. Interrupted tensile test on U–6Nb.

Figure 3. U–6Nb tensile test.

0

100

200

300

400

500

600

700

800

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4

U-6Nb Tensile Test

Ten

sile

str

ess

(MPa

)

Strain (mm/mm)300

18% strain

7.5% strain

1.5% strain



Figure 4. U–6Nb Bauschinger torsion test—comparing forward and reverse segments.

A, Forward Strain B, Reverse Strain C, Forward (2d cycle)

0 0.05 0.150.10

1

2

3

4

5

6

7

8

Thete (radians)

Forw

ard

and

Rev

erse

Tor

que

(abs

olut

e va

lue,

kN

-cm

)

For lower strains (below the onset of slip) itappears that the stresses required to produce theequal amount of deformation are also lower onreloading than on initial loading, presumably thisis because the deformation twinning has occurredin the previous loading cycle. The subsequentreloading showed minimal change. Analysis ofthis effect in this small strain region will continue.

Work Hardening and Strain-Rate Sensitivity

Accurate determination of shear-stress, shear-strain data is highly dependent on establishingproper values for n*, the work hardening exponent,and m*, the strain rate sensitivity. Tests wereconducted at room temperature for three differentstrain rates to determine these effects. The resultsin Table 1 show that the values of n* fall withinthe expected range over the range of strain ratesstudied. The strain rate sensitivity m* was foundto be negligible over the rates tested, whichgreatly simplifies the calculations for shear stress.

Microstructural Examination

TEM and diffraction including in situ strainingwithin the TEM have been used to characterizethe microstructure at various levels of deformationin the U–6Nb alloys. The as-quenched microstruc-ture consists primarily of the monoclinic α” phasein addition to regions of tetragonal γ˚, niobiumrich regions, and uranium and niobium carbides.The α” phase, which can be viewed as a mono-clinic distortion of the orthorhombic α-uranium,exhibits a characteristic martensitic microstructure

Table 1. Determination of n*.

Strain rate (rpm) Slope n*

0.2 0.59

2.0 0.64

10.0 0.68

*Slope = log of (Torque)/log of (twist angle)



of finely twinned variants, as shown in Fig. 5. Ourapproach has been to employ conventional TEMto characterize the as-quenched microstructure, themicrostructure near failure, and at various pointsalong the stress-strain curve. Due to the fact thatthe as-quenched microstructure is relativelycomplex, small changes in microstructure associ-ated with small strains become difficult to

differentiate. For this reason, in situ straining wasemployed to observe the microscopic deformationprocesses occurring in these alloys.

Due to the fact that at room temperature U–6Nbis a shape memory alloy, one would expect toobserve elastic deformation followed by twinboundary migration, and then by traditionaldislocation motion. The stress distribution during insitu straining is not well-defined due to the specimengeometry which results from the electropolishingrequired for specimen preparation. However, it ispossible to obtain qualitative observations regard-ing the deformation mechanisms.

A primary factor in determining whether ornot dislocation motion or twinning will beobserved is the specific grain orientation relativeto the straining axis. Due to the limited thin areain the U–6Nb specimens, the probability of afavorably oriented grain is relatively low. Inaddition, it is often observed that cracks form atthe edge of the specimen rendering the experimentunsuccessful. In spite of the experimental difficul-ties concerning generating sufficiently thin areasand avoiding cracks in these thin areas, a numberof in situ experiments have been successful. Asreported in FY95, dislocations have been observedto move across larger untwinned areas. Thegeneral observation is that dislocations aregenerated in thin regions near the hole, andprogress toward thicker regions. Recently,deformation twinning has been observed for thefirst time, as shown in Fig. 5.

In Fig. 5, the top image shows a small sectionof the starting microstructure consisting ofmultiple variants of the monoclinic martensite, aswell as numerous twins within the variants. Afterapproximately 1 sec of straining, a deformationtwin (labeled A) is formed and rapidly growsbetween two twin plates. During this time period,the nearly vertical twin (C) continues to growdownward until it intersects twin A. This inter-section prevents any further growth of twin (C).After a period of time of approximately 10 sec,additional deformation twins appear and slowlygrow parallel to A. (Because of the dimensionsand geometry it is not possible to determine thestrain rate that is applied leading to these obser-vations). These deformation twins continue togrow as the straining progresses until theyintersect twin C, at which time growth ceases.

Figure 5. Captured video images during in situstraining of U–6Nb. The top image is representa-tive of the starting as-quenched microstructurewhich consists of multiple variants of the mono-clinic α” martensite. The middle imageillustrates the microstructure after approximately1 sec of straining and shows a deformation twin(A) which rapidly formed and grew until inter-section with a variant boundary. At the sametime, twin (C) grew until intersection with twin(A). The bottom image was captured approxi-mately 10 sec later and shows the formation andgrowth of additional deformation twins (B).

CC

150 nm150 nm

AA

AABB



Summary

Both dislocation glide and deformationtwinning have been observed as deformationmechanisms in U–6Nb. It is interesting to note,however, that detwinning or twin boundarymigration has not been observed, as would beexpected for the early stages of deformation in ashape-memory alloy.

Reference

1. R. A. Vandameer, J. C. Ogle, and W. G.Northcutt Jr., “A Phenomenological Study ofthe Shape Memory Effect in PolycrystallineUranium–Niobium Alloys,” MetallurgicalTransactions A 12A, 733 (May 1981).

PHOTOTHERMAL RADIATION DURING LASER-INDUCED DAMAGE IN BULK KDP


PHOTOTHERMAL RADIATION DURING LASER-INDUCED

DAMAGE IN BULK KDPM. Yan, M. Staggs, M. Runkel, J. De Yoreo

Introduction

The mechanism of nonlinear interactionbetween high field intensity pulsed laser beamsand transparent solids has been studied for manyyears.1 For laser pulses longer than a few tens ofpicoseconds, bulk damage to transparent materi-als involves the heating of free electrons andtransfer of energy into the lattice, causing break-down of the material.2 Generation of free electronshas been discussed1 by either electron–avalancheimpact ionization or multiphoton–absorption tostates above the band gap. Here, we investigatethe fundamental mechanisms of nanosecondlaser-induced damage in KDP and attempt tounderstand the role of extrinsic defects in causingsuch material breakdown. The morphology ofdamage sites caused by nanosecond laser pulsesin KDP crystals usually exhibits features reminis-cent of local melting and fracture,3 which suggeststhat local heating is involved in the breakdown ofthe material. Also, the fluence at which laser-induceddamage first occurs—the damage threshold—scales roughly with the square root of pulse width,τ1/2(Ref. 4). This behavior is expected for transferof electron kinetic energy into the lattice, followedby thermal diffusion during the laser pulse.

In this report, we describe measurements ofphotoemission during laser-induced damagewhich address the issue of local heating in KDPduring illumination by high-intensity laser pulses.Light emission during and after illumination isfound to be spatially coincident with subsequentdamage sites, and the spectrum and dynamics ofthe photoemission exhibits the features of high-temperature blackbody radiation. In addition,radiation due to heating is observed at fluencesbelow the damage threshold of KDP and thedecay of the emission at high fluence is similar tothat observed during photoluminescence fromdefects resonantly excited at 266 nm.


It has been shown that a high-density plasmais formed during laser-induced damage at thesurface of a solid5 with light emission from theplasma corresponding to the ionization of nitro-gen in the air. During laser-induced damage inthe bulk of KDP, we found that light was emittedfrom sites where local material breakdown occurs.To investigate this process, we constructed asystem in which the third harmonic of a Nd:YAGlaser beam incident on a 10-mm-thick crystal wasco-linear with the beam of an argon–ion laseroperating at 532 nm3. A charge-coupled device(CCD) camera was used to record both theemission of light during damage and scatteredlight from the resulting damage sites. Figure 1(a)shows an image of visible light emission duringultra-violet (UV) laser irradiation in a region of aKDP crystal with no visible, pre-existing defects.Figure 1(b) is a subsequent image of the sameregion obtained, using side-scattered light whichshows the presence of damage sites. Notice thedamage sites are coincident with the locations oflight emission during damage. The results fromhundreds of such measurements show that thecorrelation between sites of light emission andthose of laser-induced damage is over 98%.

To understand the nature of this light emission,the emission spectra were measured during andafter illumination by single laser shots using aspectrograph and a CCD camera array. As Fig. 2shows, a typical emission spectrum consists of abroad spectral feature without distinct spectrallines. (The 700 nm peak in the figure is anexperimental artifact due to the second order ofthe 355 nm laser scattered from the damage site.)This broad feature can be fitted with the onset ofa blackbody emission curve. The heavy gray linesin Fig. 2 are the blackbody radiation spectra for4000, 6000, 8000, and 10,000˚K radiation. The



blackbody radiation at 6,000˚K gives the best fit tothe shape of the emission spectrum. The black-body radiation curves shown in Fig. 2 correspondto the thermal radiation at a given temperature.Because our spectra are collected at millisecondtime intervals, they give the integrated lightemission during temperature decay. However,the decay lifetime of the light emission is rapid(see the discussion below) and, because thethermal radiation varies as the fourth power ofthe temperature, the intensity of the thermalradiation also decreases rapidly with decreasingtemperature. We calculate that the spectrum ofintegrated intensity should be about the same asthat obtained instantaneously at the peak tempera-ture. Therefore, our measured spectra onlyrecord the maximum temperature during thewhole heating and decay process.

The high temperature observed during laserdamage suggests that thermal breakdown is the

fundamental mechanism responsible for laser-induced damage. But a few thousand degrees oftemperature is well above the tetragonal to mono-clinic phase transition temperature of 180°C inKDP. Consequently, this temperature is unlikelyto correspond to the lattice temperature, and webelieve that it is associated with a free electronplasma formed in the solid due to the high-intensitylaser pulse. This conclusion is supported bymeasurements of the time-resolved light emis-sion. Figure 3 shows a series of emission decaycurves collected at laser fluences both below andabove the damage threshold as determined bylight scattering. These curves are measured at thecenter of the emission band of 600 nm via a fastphotomultiplier tube. As Fig. 3 shows, below thethreshold, weak light emission is observed andthe decay lifetime is within our instrumentresolution of 25 ns. In contrast, above the thresh-old, the emission exhibits a much longer decay

Figure 1. (a) Emission of light from damage sites during 355 nm illumination. (b) Scattered light fromdamage sites after 355 nm illumination.

(a)

(b)



Figure 2. Spec-trum of emissionintensity duringlaser-induceddamage using355 nm illumina-tion. Gray curvesare calculatedblackbodyspectra at 4000,6000, 8000, and10,000˚K.

400 500 600 7000

1

2

3

4

5

8000K

Bla

ckbo

dy r

adia

tion

Em

issi

on in

tens

ity

0.0

0.1

0.2

0.3

0.4

0.5

4000K

6K

10000K

6000K

Wavelength (nm)

Figure 3. Time dependence of the emissionintensity (arb. units) at fluences both below andabove the damage threshold.

0 0.2 0.4 0.6 0.8

0.2

0

0.4

Above threshold

Time (µsec)

Rel

ativ

e In

tens

ity

lifetime (>25 ns) and during catastrophic damageof KDP, the emission decay lifetime is 20 µs.

Figure 4 demonstrates that the integratedintensity increases monotonically as the laserfluence increases. At a certain fluence, theintensity of the light emission saturates, with theknee of the curve occurring near the unconditionedlaser-damage threshold. The light emissionintensity increases rapidly after laser-induceddamage is observed. We believe that the lightemission observed in the latter instance consistsof both bulk emission and emission from thedamage sites. At laser fluences below the damagethreshold, most of the emitted light comes fromhot free electrons generated in the bulk material.But these hot electrons do not reach the criticalplasma density to cause material breakdown.When the laser fluence reaches the threshold ofdamage, the critical density of the hot electrons isreached, causing local material breakdown and along thermal decay lifetime. During catastrophicdamage, a local, high-density hot-electron plasmais generated which is in some way responsible forthe light emission with a decay lifetime of 20 µs.

As Fig. 5 shows, a similar 20 µs decay lifetimeis observed during photoluminescence by reso-nantly excited defects with an absorption bandcentered at 266 nm. Such a long decay of the



Figure 4. Integratedarea (arb. units)under emissioncurves such as thoseof Fig. 3 vs 355 nmlaser fluence.

Inte

grat

ed a

reas

of d

ecay

cur

ves

0 15105 20 25 30 35 40 45 50

1

2

4

5

Fluence (J/cm2)

Run #1 Run #2

Figure 5. Intensityof photolumines-cence (arb. units) vstime for four differ-ent total fluences.Inset shows depen-dence ofexponentialprefactor on totalfluence at 0.14 J/cm2

per pulse.

0 20 40 60 80 1001

t (µsec)

Phot

olum

ines

cenc

e 21 µs

10

100

Exp

onen

tial P

refa

ctor

0 2×103 3×103 5×103 7×103 8×1030

2

4

6

8

10

Total fluence (J/cm2)

pulse fluence = 0.14 J/cm2



photoluminescence is characteristic of “colorcenters,” electron-trapping defect states in crystals.We also note that the magnitude of the photolu-minescence increases with increasing number oflaser pulses (see inset to Fig. 5). We hypothesizethat similar color centers are generated by UVlight emission from the locally dense electronplasma generated during laser damage. The long-decay lifetime light emission that is observed isgenerated by the electron-trap states—a hypoth-esis that requires further experimental proof.

Conclusion

In summary, we have investigated the emissionof light from KDP crystals during laser irradiationboth above and below the damage threshold. Thedetection of photothermal properties at sub-damagethreshold laser fluences provides a technique forinvestigating the defects associated with laser-induced damage in KDP. For the first time, wehave provided experimental evidence of hot-electron plasma formation in the bulk of thecrystals. Our image of thermally induced damageis consistent with the other experimental observa-tions such as damage morphology and laser pulselength dependence. Future research will focus oninvestigating local (micrometer scale) photothermalproperties and their correlation with laser-induceddamage. In addition, we will study the origin ofhot electron generation to correlate damage withgrowth defects in the crystals, such as local strainand electronic impurities.

References

1. S. Jones et al., Optical Engineering 28, 1039(1989).

2. B. C. Stuart et al, Phys. Rev. B. 53, 1749 (1996).3. F. Rainer, L. J. Atherton, J. J. De Yoreo, in

Laser Induced Damage in Optical Materials:1992, H. E. Bennett, L. L. Chase, A. H.Guenther, B. E. Newnam, and M. J. Soileau,eds. (SPIE 1848, 1992).

4. R. W. Wood, Laser Damage in OpticalMaterials (Adam Hilger, Bristol and Boston,1986) p. 131.

5. W. J. Siekhaus, L. L. Chase, and D. Milam, inLaser Induced Damage in Optical Materials:1985, H. E. Bennett, A. H. Guenther, D.Milam, and B. E. Newnam, eds. (NBS SpecialPublication 746, 1985).

LARGE CAPSULE DEVELOPMENT FOR NIF TARGETS


LARGE CAPSULE DEVELOPMENT FOR

NATIONAL IGNITION FACILITY TARGETS

S. A. Letts, K. E. Hamilton, S. R. Buckley, E. M. Fearon, D. Schroen–Carey,* R. C. Cook

Introduction

The National Ignition Facility (NIF) willrequire 2-mm-diam targets with specifications onsurface finish and uniformity more rigorous thanthose for currently produced capsules.1 Thecurrent technology that uses a heated drop towerto blow 0.5-mm shells from polymer solution doesnot scale to the larger NIF requirement.2 Therefore,we are exploring several methods for producinglarge capsules, including: a depolymerizing mandreltechnique,3–5 microencapsulation,6,7 ballisticfurnace blowing technologies,8 and the techniquedescribed here of interfacial polycondensation.

The interfacial polycondensation method wasfirst investigated for ICF capsules as a method forproducing a vapor barrier skin on foam shells.9 Inthis project we explored the possibility of usinginterfacial polycondensation to produce NIF-scaleshells. We envisioned interfacial polycondensationas a method that could possibly produce a full-thickness shell in one step. This would avoidvacuole and nonuniformity problems encounteredin microencapsulation and would not be limitedto producing thin shells as is the case with droptower and ballistic furnace methods. By workingin a neutrally bouyant solution, we hoped to beable to produce perfectly spherical shells. In itsfirst phase in FY95, we found that good sphericityand uniformity could be achieved, although theshell thickness was limited to about 30 µm andthe surface roughness was not acceptable.10

During this phase, we focused on understandingthe shell growth mechanism from which wehoped to better assess the potential for increasingand controlling wall thickness and improvingsurface finish. Our approach was to measure therate of wall growth as a function of processconditions, starting reaction concentrations, oil-phase solvent, and reaction time. The experimentalparameters we examined were concentrations of

reactants IPC and PVP, solvent, and temperature.The shells were characterized for structure bymeasuring permeability and by observing thesurface and fracture morphology using scanningelectron microscopy (SEM). Composition andinformation on the cross-linking of the shell wasmeasured by infrared (IR) spectroscopy.

Shell Growth

Figure 1 shows the interfacial reaction schemeused for making large shells. Droplets of oilcontaining a reactive component are suspendedin an immiscible aqueous solution of a secondreactive component. By matching the density ofthe two immiscible phases, gravitational deforma-tion forces are minimized allowing surface tensionto drive the droplet to sphericity. Because shear-induced deformations are potentially problematic,minimal levels of agitation were used to preservethe sphericity during the reaction. The oil phasecontains isophthaloyl dichloride (IPC) and theaqueous phase contains poly(vinylphenol) (PVP),which is deprotonated in the basic solution. Atthe interface PVP reacts with IPC forming esterlinkages. Some fraction of the ester links lead tocrosslinking of the PVP. The crosslinked polymerbecomes an insoluble phase, resulting in deposi-tion of a polymer skin on the droplet. Our initialwork, which covered a limited range of solutionconcentrations, was able to produce shells up to30-µm thick with rough surface finish. We expandedour investigation to include a wider range of IPCconcentrations as well as examining other solvents.Figure 2 compares the wall thickness for shellsproduced, using 10% IPC in two different solvents:diethylphthlate (DEP) and 1,6-dichlorohexane(DCH). In DEP the thickness grows to a limit ofabout 30-µm, while in DCH the thickness growsto about 90 µm. This discovery showed that the30 µm barrier could be broken, using a modified

*W. J. Schafer Associates



reaction media. Shells were generated using arange of concentrations from which we foundthat the shell wall growth rate and ultimate wallthickness were a function of IPC concentration.We found that the shell was limited in its growthby the amount of IPC contained inside the shell,although it is not known if complete consumptionof the IPC occurred.

Figure 1. Reactionscheme used forshell generationwith interfacialpolycondensation.A drop of the oilphase (composedof IPC in eitherDEP or DCH)reacts with thesurroundingaqueous phase(containing thedeprotonated PVPchains).

Oil phase

Aqueous phase

pH 13

Oil

Aqueous

+

in diethyl phthalate (DEP)

or 1,6 dichlorohexane (DCH)

Crosslinked PVP

PVPIPC

C Cl

O

C Cl

O

O- Na+

(CHCH 2)n

+ 2 HCl

Interfacial

reaction

(CHCH 2)n

O

C O

C OO

(CHCH2)n

t (m)

Argonpermeation

Slow

Fast

Wal

l thi

ckne

ss (

µm)

0

50

100

150

0 10 20 30 40 50 60

DCH

DEP

Figure 2. Wallthickness as afunction of reactiontime (10 wt% IPCand 4 wt% PVP).Both membranedeposition rate andmaximum wallthickness are largerfor the DCH solventthan the DEPsolvent at a constantIPC concentration.

Surface Roughness Analysis

Dried shells were characterized for surfacestructure using SEM. Figure 3(b) shows a typicalsurface which appears to have a rough texture.We further investigated the shell structure byfracturing. Occasionally, we observed crystals onthe inner surface when the exchange processing



was incomplete. However, in general, we foundthe inner surface, shown in Fig. 3(a), to be verysmooth in comparison with the outer surface.Further investigation of the fracture surface showsthat the inner smooth region is a film only about100-nm thick. Beyond this smooth skin, the shellmorphology is uniformly granular. The observedgranules were approximately 50 nm in diameterwhile the radius of gyration of the PVP used wasapproximately 8 nm. The structure maintains thisgranular texture consistently to the outer surface.This structure suggests that the shell first forms athin film by a homogeneous reaction mechanism.Subsequently, material deposits via a reactionlimited by diffusion of the IPC through the growingfilm to the outer surface where it is able to reactwith the PVP. We believe the reaction proceedsup to a critical level at which point the reactedpolymer becomes insoluble in the interfacial regionand phase separates to form a particle aggregate.This behavior is consistent with the conversion ofa soluble polyelectrolyte to insoluble ionomers.The degree of reaction reached before precipitationis dependent on the reaction conditions, includingorganic phase solvent and temperature. Furtherconfirmation of this mechanism will be presentedin the section on IR composition measurement ofthe shells.

The surface roughness of the dried shells wasmeasured using an atomic force microscope (AFM).By measuring surface roughness as a function of

reaction parameters we hoped to be able to optimizethe processing to improve the surface smoothness.Shells made in DEP have a roughness of approxi-mately 20 nm rms that seem to become smootherwith reaction time. The smoother surface may alsobe a result of better crosslinking (see IR section formore detail) at longer reaction. This makes thepolymer less susceptible to swelling in the subse-quent solvent washes and supercritical drying. Incontrast, the surface roughness of the polymershells made in DCH show increasing roughnesswith thickness. These shells grow more rapidlypossibly because the IPC is able to diffuse throughthe growing polymer shell faster. The shells madein DCH may also be less crosslinked and as a resultare more swollen by the supercritical dryingprocess. This is consistent with analysis of shellcomposition by FTIR (shown later).

To produce a smooth outer surface we invertedthe phases by dispersing aqueous PVP droplets inan organic outer phase containing the IPC. Shellsmade using inverted phases have an outer surfaceroughness of 2 nm rms; however the inner surfacehas the same bumpy structure we previouslyobserved on the outside. Clearly, we did not alterthe physical processes which control the develop-ment of surface structure by merely reversing thephases. However, it does show that the surfacestructure is consistently smooth on the oil phaseside of the interface.

Figure 3. At veryshort reactiontimes, a very thinskin (~100 nm) isformed at the innershell surface asviewed by thecross-sectional,fracture surface (a).All subsequentshell deposition hasa granular nature toit. Shell walls arefairly uniform asviewed from theexterior surface (b).

4% IPC / DEP

SurfaceInterior

Inner skin

Growth direction

1.5 µm 1.5 µm1.5 µm 1.5 µm



calibrating the absorbance using phenol benzoate,which has a very similar structure to the productof the reaction between IPC and PVP. For a shellof a particular mass, we were thus able to deter-mine the mass of IPC incorporated in the shell. Bydifference (mass balance) we were able to calcu-late the mass of PVP. From the masses of eachcomponent we were able to calculate the molarratio (ratio of moles of IPC to moles of PVP repeatunits) of the two starting materials in the finalshell. For shells made in DEP, we experimentallymeasured a molar ratio of 1.0. This indicates thatthe PVP has completely reacted with the IPC. InDCH we found that the molar ratio was 0.33,indicating that one third of the PVP hydroxygroups reacted with IPC. There are three impor-tant implications of the IR analysis regarding thegrowth of shells, the morphology, and the role ofthe oil phase solvent. First, shell walls growthicker in DCH because less IPC (the limitingreactant) is consumed per mass of shell. Only onethird as much IPC is needed in DCH comparedwith DEP to produce a given thickness of shell.Second, as the PVP reacts with IPC at the interfacea critical degree of reaction is reached at whichpoint the polymer becomes insoluble and phaseseparates. The precipitated polymer becomesunreactive to further condensation. Third, solubil-ity controls the extent of reaction. The polymer isvery soluble in DEP and fully reacts with theavailable IPC. The polymer is less soluble in DCHand precipitates when only one third of theavailable sites are reacted.

Permeation

The transport of small molecules through thevarious shell materials can be an important measureof the nature and structure of the membrane.Permeation is a measure of transport, and it is theproduct of the two driving forces—diffusion rateand solubility. This can be measured directlywith mass uptake/sorption experiments in whichthe polymeric test sample is subjected to saturatedvapor of various permeates (in this case ethanolor tetrahydrofuran, THF). The experiments canbe run in either adsorption (increasing mass fromneat material) or desorption mode (decreasingmass from saturated material). For Fickiandiffusion the following approximation applies toadsorption at short experimental times:11

M MM M L

Dtt −−

=∞

o

o

4π

, (1)

IR Composition Analysis

IR was used to characterize interfacial poly-condensation shells to determine the effect ofreaction conditions on composition. Figure 4compares the IR spectrum of the two startingmaterials, PVP and IPC, with the copolymerformed at the interface. Several absorbancespotentially could be used to make structure andcomposition measurements on the shell. Thecarbonyl stretching vibration at 1740 cm–1 isparticularly attractive because of the absence ofinterferences in the spectrum. The carbonylabsorption in IPC occurs as a doublet at 1720 and1700 cm–1. The IPC reacts with PVP by formingester linkages that incorporates carbonyl groupsin the copolymer film. From measurements of themagnitude of the carbonyl absorption we wereable to determine the IPC concentration in thepolymer shell. The carbonyl absorption per unitmass (proportional to IPC concentration) in thecopolymer film was found to be independent ofreaction time for shells made in either DEP or DCH.

To go beyond the relative concentrations(absorbance per mass), we needed to calibrate theIR absorption measurements. This was done by

Figure 4. IR spectroscopy used to measure thecomposition of the interfacial polymer. IRspectra of (a) PVP (b) PVP and IPC, and (c) IPC.

400

IPC

PVP+IPC

PVP

1800 1400 1000cm-1

C=O

%Τ

1740



where Mo is the original mass of the dry membranematerial, Mt is the mass of the material at anytime, t, and L is the characteristic diffusion length;for broken shell materials with diffusant vaporexposure on both sides, L is half of the actualwall thickness and D is the diffusivity. A numberof adsorption and desorption measurements wereobtained for shells generated from both the DCHand DEP solvents. In general, small moleculestended to move through the shells made with DCHquicker than those made with DEP.

It is possible to apply the same principle, usedto examine the permeation of the solvent throughshell membranes, to examine the mechanism ofshell growth. Instead of observing the sorption ofan “inert” solvent, we can observe the transportof the key reactants. Because we have postulatedthat the observed interfacial reaction is limited bydiffusion, we have utilized the same Fickiananalysis, described above for the motion of thesmall molecules (THF and ethanol) through themembrane material, to characterize IPC motionthrough the forming membrane. It is presumedthat the reaction predominantly occurs at the

outermost surface of the forming membrane, i.e.,at the interface of IPC / solvent oil-phase andPVP-loaded aqueous phase. Thus, an IPC mol-ecule must diffuse from the core fluid interfacethrough the forming membrane to the exteriorsurface where it is met by a deprotonated repeatunit of the PVP chain. It is possible that thetransition of PVP chain from a solvated/partiallyionized chain to solid membrane material occurs inpart due to a change of solubility with the additionof the singly (possibly crosslinked) reacted IPCmolecule and partly due to a composite/ statisticalchange in the ionized character of the PVP chain.

If the wall thickness data for various IPCconcentrations in a singular solvent (includingdata from Fig. 2) are normalized by a characteris-tic length scale, L, (which is taken to be the finalwall thickness of the shell) and plotted on aL-normalized time scale as suggested by Fickiandiffusion and Eq. (1), a similar type of behavior isobserved (see Fig. 5). With this rescaling, weobserved excellent overlap of the data into onecoherent trend (similar to solvent sorption data)which corroborates the suggestion that this

Figure 5. Rescaledshell wall thicknessdata for 2 wt% IPC,4 wt% IPC, 10 wt%IPC, 20 wt% IPC inDCH solvent. Thecharacteristic wallthickness, L, in theexpression for Fickiandiffusion is taken tobe Lmax for each IPCconcentration. Thesolid line represents abest fit to the Osakamodel.9,10

0

0.2

0.4

0.6

0.8

1

1.2

0 0.1 0.2 0.3 0.4 0.5 0.6

t1/2 / Lmax (min1/2 / µm)

L(t

) / L

max Osaka fit

20% IPC

10% IPC

4% IPC

2% IPC



formation. Wall thickness was found to be afunction of two reaction parameters: concentrationof IPC (the limiting reactant) in the inner phase,and the choice of oil phase solvent. We believe thegrowth mechanism proceeds as follows. At theinterface, a thin (100-nm-thick) film first forms.All subsequent material is formed by IPC, diffus-ing through the shell wall to react with the PVP atthe outer surface of the shell. The IPC reacts untila critical level is reached where the polymerbecomes insoluble and precipitates forming agranular particle. The solvent choice determinesthe level of reaction reached prior to precipitation.The resulting wall morphology is composed of aninner smooth skin with a uniform granular layerbuilt up by diffusion limited growth. Permeabilitymeasurements showed that the shell structurechanged with the oil phase solvent. Using a growthmodel based on Osaka University work, combinedwith our permeability measurements, we wereable to derive a generalized expression for thegrowth of shells. With this new understanding ofthe interfacial polycondensation reaction, it maybe possible to obtain shells with the desiredthickness and surface features.

References

1. S. W. Haan et al., Phys. Plasmas 2, 2480 (1995).2. R. Cook, “Production of Hollow Microspheres

for Inertial Confinement Fusion Experiments,”in Hollow and Solid Spheres and Microspheres–Science and Technology Associated with TheirFabrication and Application, D. L. Wilcox et al.,eds. (Materials Research Society, Pittsburgh,PA, 1995) pp. 101–112.

3. S. A. Letts, E. M. Fearon, S. R. Buckley,M. D. Saculla, L. M. Allison, and R. Cook,“Preparation of Hollow Shell ICF TargetsUsing a Depolymerizing Mandrel,” in Hollowand Solid Spheres and Microspheres—Science andTechnology Associated with Their Fabrication andApplication, D. L. Wilcox et al., eds. (MaterialsResearch Society, Pittsburgh, PA, 1995)pp. 125–130.

4. S. A. Letts, E. M. Fearon, L. M. Allison, andR. Cook, J. Vac. Sci. Technol. A14, 1015 (1996).

5. S. A. Letts, E. M. Fearon, S. R. Buckley,M. D. Saculla, L. M. Allison, and R. Cook,Fusion Technology 28, 1797 (1995).

6. M. Takagi, T. Norimatsu, T. Yamanaka, andS. Nakai, J. Vac. Sci. Technol A9(3), 820 (1991).

Technique Diffusant PDCH / PDEP

Wall thickness IPC 12X-ray fluorescence Argon 8Sorption study THF 14

Ethanol 5

Table 1. Comparison of permeation values.

From the agreement for each of the techniques(we consider the 5–14 range in reasonable agree-ment for diffusants used given the typical rangeobtained for a diffusion experiment), it is possibleto conclude that the two shell materials havedistinct transport and shell growth properties. Itis apparent that the diffusant traverses thematerials fabricated in DCH about 10-fold timesas fast as the materials generated from DEP,regardless of the diffusant chosen (IPC, argon,THF, or ethanol).

Summary

Shells were made by interfacial polyconden-sation in the-2-mm-diam range needed for NIF.Much of our work in this phase of the studyfocused on understanding the mechanism of shell

interfacial reaction scheme is in fact limited by thediffusion of IPC. Furthermore, these data can befitted to a model, incorporating both aspects ofthe diffusion and reaction kinetics, 9,10 and a bestfit to their expression is provided as the solidcurve in Fig. 5.

With this protocol, we can calculate the diffusioncoefficient for IPC in two shell materials examinedin the solvent sorption experiments. The calculateddiffusion coefficients for IPC through a formingshell using either DEP or DCH as solvents are2.30 × 10–8 cm2/sec and 2.82 × 10–7 cm2/sec,respectively. Both values are of the correct orderof magnitude one would expect for small mol-ecule motion through a rubbery polymer.11 Whenadjusted for solubility, these values give a ratio of12 for the permeation of IPC through the two typesof membranes, PDCH/PDEP. This value agrees (inboth direction and magnitude) with both thesorption measurements and the x-ray fluorescencemeasurements for the two materials (Table 1).



7. Y. A. Merkuliev, A. A. Akunets, V. S. Bushuev,V. M. Dorogotovtsev, A. I. Gromov,A. I. Isakov, A. I. Nikitenko, S. A. Startsev,S. M. Tolokonnikov, and R. C. Cook, “Studyof Production and Quality of Large (1–2 mm)Polystyrene Hollow Microspheres,” in Hollowand Solid Spheres and Microspheres—Science andTechnology Associated with Their Fabrication andApplication, D. L. Wilcox et al., eEds. (Materi-als Research Society, Pittsburgh, PA, 1995)pp. 119–124.

8. T. Boone, L. Cheung, D. Nelson, D. Soane,G. Wilemski, and R. Cook, “Modeling ofMicroencapsulated Polymer Shell Solidifica-tion,” in Hollow and Solid Spheres andMicrospheres—Science and Technology Associ-ated with Their Fabrication and Application,D. L. Wilcox et al., eds. (Materials ResearchSociety, Pittsburgh, PA, 1995) pp. 193–198.

9. M. Takagi, M. Ishihara, T. Norimatsu,T. Yamanaka, Y. Izawa, and S. Nakai, J. Vac.Sci. Technol. A11(5), 2837 (1993).

10. S. Buckley, G. Wilemski, D. Vu, D. Schroen–Carey, S. Letts, and R. Cook, “InterfacialPolymerization Shells Progress Report,” TAT96-102.003, Feb 9, 1996.

11. Crank, J., Park, G.S., Diffusion in Polymers,(New York: Academic Press, Inc.), 1968.

Publications

1. S. Letts, K. Hamilton, S. Buckley, E. Fearon,D. Schroen–Carey , and R,. Cook “The Roleof Reactant Transport in Determining theProperties of NIF Shells Made by InterfacialPolymerization,” LLNL, Livermore, CA,UCRL-JC-125125 (1996); submitted to FusionTechnology.

Presentations

1. S. Letts, S. Buckley, E. Fearon, K. Hamilton,D. Schroen-Carey and R. Cook, “The Role ofReactant Transport in Determining theProperties of NIF Shells Made by InterfacialPolymerization,” 11th Target FabricationSpecialists’ Meeting, Orcas Island, WA,September 8–12, 1996.

2. S. Buckley, S. Letts, E. Fearon, G. Wilemski,D. Vu, R. Cook, and D. Schroen–Carey,“Processing and Properties of NIF ScaleTargets Using Interfacial Polycondensation,”11th Target Fabrication Specialists’ Meeting,Orcas Island, WA, September 8–12, 1996.

3. D. Schroen–Carey, S. Letts, and S. Buckley,“Analysis of the Effect of pH on Poly(vinylphenol),” 11th Target Fabrication Specialists’Meeting, Orcas Island, WA, September 8–12,1996.

PREPARATION OF HIGH-BERYLLIUM-CONTENT PLASMA POLYMER COATINGS


PREPARATION OF HIGH-BERYLLIUM-CONTENT

PLASMA POLYMER COATINGS USING

CYCLOPENTADIENYLBERYLLIUM METHYL AS A PRECURSOR

R. Brusasco, R. Cook, G. Wilemski, M. Saculla

Introduction

Beryllium-doped plasma-polymer coatingswere prepared using a novel precursor material,cyclopentadienylberyllium methyl (CpBeMe).Although this compound is more suitable forvapor-phase processing, due to its increased thermalstability and higher vapor pressure, the coatingscontinue to exhibit oxidation upon exposure to airand are characterized by gravimetry and FourierTransform Infrared Spectroscopy to furtherunderstand the oxidation behavior. A simplecorrosion model fits the temporal weight gain data.

Previous work describes investigations intothe doping of plasma-polymer coatings withberyllium via the use of diethylberyllium as aprecursor.1,2 Although this coating would beconsidered as a candidate ablator for fusiontargets in the National Ignition Facility (NIF)laser, diethylberyllium was found to be unsuit-able for preparing coatings due to its low thermalstability and low vapor pressure and vapor recoveryproperties. Therefore, this report describes theresults obtained using cyclopentadienylberylliummethyl (CpBeMe) as the precursor. Our goalswere to find and use a more suitable precursor toprepare CHBe, then to investigate more closelythe properties of the resulting beryllium dopedcoating. Specifically, the structure of the CHBecoating and the details of the oxidation of thecoating upon exposure to air were some of ourprimary concerns.


Deposition experiments were conducted in ahelical resonator plasma polymerization coaterthat has also been used for previous work onberyllium-doped coatings. CpBeMe has a vaporpressure of 24 Torr at room temperature, allowingone to use the vapor without supplemental heating.The vapor pressure of the CpBeMe precursor was

monitored with an absolute capacitance manometer(MKS Corporation). For vibrational spectroscopywork, coatings were prepared on 1-cm-squaresilicon wafers, glass microscope slides, and KBrcrystals as supplied by International CrystalLaboratory (Garfield, NJ).

To handle coatings without oxidation, a specialload lock and sample transport device wasconstructed to fit onto the existing coating apparatus.Substrates could be introduced into the coater,removed, and transported to other lab apparatuswithout exposure to the air.

CpBeMe (CAS No. 36351-95-8) was preparedby Dr. Donald Gaines and Dr. Dovas Shaulys atthe University of Wisconsin, Madison, WI, andwas used as received.

After deposition of the beryllium-dopedplasma-polymer coatings, the deposition rate wasdetermined by obtaining thickness measurements,using a Tencor stylus profilometer and knowingthe deposition time. The relative amounts ofberyllium, carbon, and oxygen in the films weredetermined using x-ray photoelectron spectros-copy (XPS).

Gravimetry on the coatings was performedusing a Cahn Microbalance located inside a purgedglove box to study the oxidation behavior of theorganoberyllium films. Fourier transform infra-red spectroscopy (FTIR) analysis of coatingsdeposited on the KBr crystals was done using aPerkin–Elmer Spectrum 2000 instrument. CoatedKBr crystals were characterized without exposureto the air by use of the previously described loadlock, coupled with a gas tight infrared (IR)transmission cell.


In contrast to diethylberyllium, CpBeMe is amuch more stable organoberyllium precursor.Also, the vapor pressure was found to be close tothe expected value of 24 Torr throughout theruns, thus showing markedly improved stability.



Given the more stable nature of the precursorand the higher operating vapor pressure, theprecursor flow rate could be increased to a valuemore typical for plasma polymerization. At a flowrate of 0.2 sccm, the nominal coating rate of CHBewas found to be 1.35 µm hr–1. This value can becompared with typical deposition rates of plainplasma polymer (0.7 µm hr–1) and germaniumdoped plasma polymer (0.8 µm hr–1).

XPS analysis of the coatings prepared usingCpBeMe show the beryllium-to-carbon ratio to benominally 1:7. This is much lower than the nearone-to-one behavior of the coatings prepared usingdiethylberyllium. However, it was discovered thatthe beryllium content was a function of the deposi-tion rate. A coating prepared at a rate of 0.5 µmhr–1 using CpBeMe had a beryllium-to-carbon ratioof 4:1. Therefore, since the coatings prepared withdiethylberyllium were all prepared at a slow rate,it could well be that the beryllium content in thoseruns was influenced by the deposition rate.

Beryllium-doped plasma-polymer coatingsprepared using CpBeMe will oxidize slowly uponexposure to air. An 8-µm-thick coating on a 1-cm-square silicon wafer was allowed to oxidize indry air for a period of 24 hr. By the end of thisperiod, no further weight gain was observed. Thecoating had increased in weight by 120 µg. If one

assumes that Be simply substitutes for one eighthof the carbon atoms in the plasma polymer and isthen fully oxidized to BeO we can calculate thatthe weight gain should be about 129 µg, certainlyconsistent with the measurements.

Additional samples were characterized gravi-metrically with special attention paid to thetemporal oxidation behavior. We observed asquare root dependence of weight gain over time,which suggested a simple corrosion model toexplain the oxidation behavior. Therefore, wedeveloped a model to fit the observed behaviorand discovered that the data fit well to the modelpredictions and that the diffusion characteristicsof oxygen diffusion in CHBe are qualitativelysimilar to polyethylene. This behavior also fitsthe observation that the film shows no cracking orcrazing upon oxidation.

In an attempt to further understand the micro-structure of the coating and the changes wroughtby oxidation, FTIR spectroscopy of CHBe coatingswere obtained before, during, and after exposureto dry air. Figure 1 shows the spectrum of plainplasma polymer and CHBe, as deposited beforeoxidation and after 1 hr of exposure to dry air.

The major difference between plain plasmapolymer and CHBe is the broad doublet centeredat about 1000 cm–1. All other features of the

Figure 1. Compari-son of the infraredspectra of plainplasma polymer asdeposited CHBe andoxidized berylliumdoped polymer.

39000

10

% T

rans

mis

sion

20

30

40

50

60

70

80

90

CH

CHBe

Oxidized CHBe

3400 2900 2400 1900Wavenumber (cm-1)

1400 900 400



spectra of plain plasma polymer and CHBe arethe same. The doublet consists of two peaks—onecentered at 1032 cm–1 and another at 887 cm–1.Upon oxidation, a broad feature grows into thespectrum at about 3500 cm–1. This peak ischaracteristic of hydroxyl groups in the coating.Since the coating is exposed to dry air, the hydro-gen for the formation of OH groups must be comingfrom the coating itself.

Upon oxidation, the 1032 cm–1 peak decreasesmarkedly while the 887 cm–1 peak increasessomewhat in intensity and shifts slightly to 875 cm–1.Thus, the 1032 cm–1peak must be associated withberyllium incorporated into the polymer networkin the form of beryllium–carbon bonds. This isconsistent with the infrared spectra taken by Bartke.3A comparison of the vapor phase spectra shows apeak at approximately 1020 cm–1 only for theCpBeMe (C5H5BeCh3) compound and not theCpBe (C5H5BeH) hydride compound. This peakcould be attributed to the beryllium–carbon sigmabond. The 1032 cm–1 peak in CHBe disappearsupon oxidation, attributable to bond breakingbetween beryllium and the polymer network.

Walrafen and Samanta report the major peakin the infrared spectrum of amorphous BeO to belocated at approximately 868 cm–1.4 This fact mightexplain the peak at 875 cm–1 in oxidized CHBe ifit were not for the fact that it appeared in theunoxidized spectrum as well. This difficulty couldbe explained if one takes into account the natureof the starting precursor. The cyclopentadienylgroup exhibits a peak at approximately 815 cm–1

in ferrocene. This peak is seen to shift to higherwavenumbers and become considerably broadenedwhen the five-membered ring has other aliphaticsubstituents attached. We postulate that the 887 cm–1

peak in unoxidized CHBe is due to the existenceof cyclopentadienyl groups in the polymer networkwhich were not dissociated in the plasma duringfilm formation. These groups would not be expectedto undergo significant change during oxidationof the beryllium. The shift to slightly lowerwavenumber could be due to the formation ofamorphous BeO, which has a characteristic bandat 868 cm–1.

Conclusions

CpBeMe is a useful precursor for the preparationof beryllium-doped plasma-polymer films. It ispossible to prepare coatings with a deposition rateof approximately 1.35 µm hr–1 and to maintain aberyllium-to-carbon ratio of 1:7. The berylliumcontent is a function of the deposition rate, andthis dependence is likely due to the details of thetransport mechanisms within the coater. Thecoating suffers from oxidation upon exposure tothe ambient. The oxidation behavior follows a t0.5

behavior very well, supported by a simple corrosionmodel to account for this behavior. FTIR resultsare consistent with a material where the berylliumis formally bonded to the polymer network.Upon oxidation, these bonds are broken andamorphous BeO is formed.

Acknowledgments

The authors gratefully acknowledge theassistance of Stephen Letts for technical discus-sions and Leslie Allison and Steven Buckley fortechnical assistance.

References

1. R. Brusasco, S. Letts, R. Cook, P. Miller, andM. Saculla, “High Beryllium Content Capsulesfor the National Ignition Facility,” Chemistry& Materials Science Progress Report, Weapons–Supporting Research and LDRD, LLNL,UCID-20622-95 (April, 1996).

2. R. M. Brusasco, S. Letts, P. Miller, M. Saculla,and R. Cook, J. Vac. Sci. Technol. 14(3), 1019(1996).

3. T. C. Bartke, “The Synthesis and Character-ization of Some Cyclopentadienyl Compoundsof Beryllium,” Ph.D. Thesis, University ofWyoming, October, 1975.

4. G. E. Walrafen and S. R. Samanta, Appl.Spectrosc. 33(5), 524 (1979).

Laboratory-Directed Research andDevelopment, Exploratory Research

FUNDAMENTAL STUDIES OF MATRIX-ASSISTED LASER DESORPTION/IONIZATION


FUNDAMENTAL STUDIES OF MATRIX-ASSISTED LASER

DESORPTION/IONIZATION, USING TIME-OF-FLIGHT MASS

SPECTROMETRY TO IDENTIFY BIOLOGICAL MOLECULES

D. Eades, D. Wruck, H. Gregg

Matrix-assisted laser desorption/ionization(MALDI) mass spectrometry (MS) is a relativelynew technique in analytical MS. This potentiallyrevolutionary method was developed in responseto the needs of the biotechnical research communitywho require molecular weight information onsmall quantities (picomole to femtomole levels) ofhigh-mass and thermally labile macromolecules.While most other analytical MS ionization techniquescause fragmentation, decomposition, or multiplecharging, MALDI efficiently places intact macro-molecules into the gas phase with little fragmentationand/or rearrangement.

This project had three primary objectives:(1) establish the MALDI capability at LLNL,(2) perform fundamental studies of the MALDIprocess to better understand analyte–matrixinteractions, and (3) apply the technique as apowerful tool for biochemical research. To meetthese objectives, a retired time-of-flight (TOF)instrument was adapted for MALDI analyses,relevant parameters influencing the MALDIprocess were identified for further study (i.e., matrixmolar absorptivity, sample crystal preparation),and collaborations were established with severalresearch groups in the Biology and BiotechnologyResearch Program (BBRP) at LLNL.

The MALDI technique is accomplished bymixing the macromolecule of interest with a high-molar excess (1:100 to 1:10,000) of an organic matrix,which readily absorbs energy at the wavelengthcorresponding to a ultraviolet laser.1,2 The mixtureis dried to form crystals, and upon laser irradia-tion, the matrix absorbs the majority of the energy,causing it to desorb from the surface and gentlyrelease the macromolecule into the gas phase withlittle or no fragmentation. Once in the gas phase,ion-molecule reactions between the excited matrixand the neutral macromolecules generate ionizedanalyte species which then can be focused into aMS for detection. Due to the compatibility withpulsed techniques (i.e., lasers) and an “infinite”mass range, a TOF mass analyzer typically iscoupled to a MALDI source. The resulting

MALDI-TOF/MS spectra consist of abundant lowmass-to-charge (m/z) matrix ions and characteris-tic molecular weight ions, corresponding to themacromolecule of interest. Although MALDI hasproven to be an efficient means of generatingintact gas phase macromolecular ions, the exactmechanisms responsible for its success remainunclear. One current explanation is that themechanism responsible for the MALDI processesconsists of reactions occurring in both the solidand gas phase. In the solid phase, the process ofsample crystal formation (and choice of organicmatrix) appears to significantly influence analyte–ion production.3 Once desorbed and in the gasphase, ion–molecule reactions appear to beresponsible for ionization, and the efficiency ofthe ionization appears to depend on both the matrixand analyte species.4 Due to the uncertainty inthe fundamental MALDI mechanism, the processof selecting appropriate matrices for specificapplications remains a trial and error approachand therefore warrants further study.

In FY95, a retired surface science instrument(a laser ionization mass analyzer, model LIMA–3from Cambridge Mass Spectrometry LTD,Cambridge, England) was adapted for use as aMALDI-TOF/MS to study a series of syntheticpeptides (from BBRP) and 10 different matrices.Interestingly, there was no clear correlation betweenthe matrices with the highest molar absorptivityat 266 nm (i.e., the fourth harmonic of the Nd:YAGlaser employed) and the MALDI ion signalintensities. These results suggested that factorsother than the absorption of laser energy at agiven wavelength contribute significantly to thesuccessful generation of MALDI peptide spectra.Other factors which may explain the observationsare the method of crystal preparation and thecompatibility of the analyte and matrix in thesolid phase.

The study of this peptide series also served tocharacterize the system for sensitivity, useful massrange, and mass resolution. Results from thesestudies indicated peptide sensitivities of between



50–100 picomoles, a useful mass range of up to2000 Da, and a mass resolution of approximately100 (i.e., m/∆m = 100, where m = mass-to-chargeratio of ion). However for the MALDI techniqueto be viable for most biological applications,increased sensitivity and mass resolution wouldbe needed over a mass range to over 100 kDa.Therefore, the instrument was reconfigured forimproved MALDI capabilities. The improvementsincluded converting the flight path to a lineardesign, modifying the source to a high-voltagetwo-stage acceleration region with improved ionoptics, inserting deflection plates to remove theabundant, interfering low-mass matrix ions, andupgrading the data acquisition and detectionsystems. Results from repeating the characteriza-tion studies with the peptide series on the redesignedsystem showed significantly higher sensitivity(5–10 picomoles), a mass resolution comparablewith other linear TOF systems (i.e., m/∆m = 200),and the nearly complete removal of the low-massmatrix ions. However, the mass range remainedlimited to less than 5000 Da.

The instrumental modifications continued intoFY96. In addition to the Nd:YAG laser supplied

with the original instrument, a new nitrogen laserwas added which yielded a different excitationwavelength (337 nm vs 266 nm), a more stablepulse-to-pulse power output (3% vs ~20%), and alarger spot size (300 × 500 µm) than that from theNd:YAG laser (50 µm diam). A precise beamattenuator and power meter were also added. TheN2 laser was directed toward the sample stage viaan alternate vacuum flange thus allowing indepen-dent use of each laser system. In a side-by-sidecomparison, excitation using the nitrogen laseryielded better ionization efficiency for the variouscombinations of matrices and peptides investigated.Figure 1 is a schematic of the redesigned instrument.

Matrix materials (e.g., nicotinic acid, vanillicacid, sinapic acid, and gentisic acid) as well as anumber of sample preparation techniques werere-evaluated. The sample preparation techniquesinvestigated included deposition from water:acetonitrile solutions on unheated or heatedsample holders, and fast deposition of the matrixfrom acetone followed by deposition of the analytefrom a drop of aqueous solution placed on top ofthe matrix. For the peptide and protein samplesexamined, the highest ion currents were observed

Figure 1. Current MALDI-TOF/MS system schematic.

30 kVAcceleration

PC Computer

DSO

Quartz

Mirror

Mirror

FocusingLens

PowerMeter

BeamAttenuator

2-Stage Acceleration(25 kV)

Camera750 X Mag

Nd:YAGLaser(266 nm)

Duel MicroChannelPlate Detector

X

+400V

A-to-D Card

486/66Gateway

DeflectionPlates

SteeringPlates

Y 3-D Sampling StageZ

##

Pre-Amp

N2 LASER(337nm)



using sinapic acid as the matrix, the nitrogen laser,and the preparation technique where the samplewas deposited onto an unheated sample holderfrom water:acetonitrile solution. The mass spectrapresented here are positive ion spectra recordedwith 15 kV accelerating voltage (10 kV on the“extraction” lens). Mass calibration was madeusing the equation (m/z)1/2 = At + B, where A andB were determined by known low-mass ions inthe matrix [23Na+, 39K+, (M-OH)+ at 207, (M+H)+

at 225, and (M+Na)+ at 247, where M = 224.2, themolecular weight of sinapic acid]. In addition, asmall amount of gramicidin S was added to theprotein samples to give an additional calibrationpoint at m/z = 1165 [(gramicidin S + Na)+]. Thiscalibration procedure extrapolates from low-masscalibrant ions to high-mass analyte ions, and theprocedure typically results in a mass assignmenterror of ±0.1%. Higher accuracy, on the order of ±0.01%,can be obtained by adding an internal calibrantwith a mass comparable to the analyte mass.5

Figure 2 shows a mass spectrum obtained forcytochrome c (sum of 25 laser pulses, no smoothing).The molecular weight of cytochrome c is 12,327,the mass resolution is about 200, and matrix ionadducts are seen as shoulders on the main peaks.Figure 3 is a mass spectrum obtained for bovinetrypsin (sum of 80 laser pulses, no smoothing). Themolecular weight of trypsin is 23,311. The shoulderat 24,500 may correspond to (trypsin + gramicidinS + Na)+. The mass resolution for the shown peaksis 40 to 45, but the resolution of the gramicidin Scalibrant peak was about 200 for this sample. The

Mass to charge ratio (m/z)

Rel

ativ

e in

tens

ity

4000 6000 8000 10000 1400012000

1800016000

(M+H)+ 12330±30

(M+2H)2+ 6160±10

250

200

150

100

3200

3000

2800

2600

2400

2200

2000

Rel

ativ

e in

tens

ity

10000 15000 20000 25000 30000 35000Mass to charge ratio (m/z)

(2M+3H)3+ 15150±200

(M+2H)1+ 11640±150

(M+H)+ 23290±300

-24500

low resolution at high m/z values may arise fromthe formation of a large number of unresolvedmatrix ion adducts by the protein molecules; evenwith matrix ion deflection, the sample produced alarge ion signal at low m/z values.

Several unsuccessful attempts were made torecord a mass spectrum for bovine albumin(molecular weight 66,267). It is likely that thebovine albumin samples, which were from amolecular weight marker kit for gel electrophoresis,contained significant quantities of sodiumdodecyl sulfate (SDS). SDS is known to suppressthe ion signal in MALDI-TOF/MS.6

Future Work and Conclusions

It should be possible to improve the massresolution of the MALDI-TOF/MS by employingtime-lag focusing.7 In this technique, the regionbetween the sample holder and the first grid is keptfield-free for a short time (roughly 0.3 to 3 µs) afterthe laser pulse, then the extraction field is rapidlyapplied. Resolution in the range 800 to 1000 hasbeen demonstrated for proteins in a linearMALDI-TOF instrument, using time-lag focusing.8

A MALDI-TOF/MS system has been con-structed from an existing instrument and spectrafrom peptides and proteins have been successfullyacquired. The system currently allows sensitivedetection of biomolecules with masses to at least30 kDa. This system can be used as an analyticaltool for BBRP programs, as well as for determin-ing molecular weight distributions for polymers.

Figure 2. MALDI-TOF mass spectrum ofcytochrome c sample.

Figure 3. MALDI-TOF mass spectrum of bovinetrypsin sample.



References

1. M. Karas and F. Hillenkamp, Anal. Chem. 60,2299–2301 (1988).

2. F. Hillenkamp, M. Karas, R. C. Beavis, andB. T. Chait, Anal. Chem. 63, 1193A–1202A (1991).

3. M. George, J. M. Y. Wellemans, R. L. Cerny,M. L. Gross, K. Li, E. L. Cavalieri, J. Am. Soc.Mass Spectrom. 5, 1021–1025 (1994).

4. D. H. Russell, presented at the 42nd ASMSConference on Mass Spectrometry and AlliedTopics, May 29–June 3, 1994, Chicago, IL, p. 1.

5. R. C. Beavis and B. T. Chait, Anal. Chem. 62,1836 (1990).

6. O. Vorm, B. T. Chait, and P. Roepstorff, 41stASMS Conference Proceedings, 621a (1994).

7. W. C. Wiley and I. H. McLaren, Rev. Sci.Instrum. 26, 1150 (1955).

8. R. M. Whittal and L. Li, Anal. Chem. 67, 1950(1995).

DAMAGE EVOLUTION IN LOW-ENERGY ION-IMPLANTED SILICON


DAMAGE EVOLUTION IN LOW-ENERGY

ION-IMPLANTED SILICON: STM EXPERIMENTS

AND ATOMISTIC COMPUTER SIMULATIONS

P. Bedrossian, M.–J. Caturla, T. Diaz de la Rubia

The Semiconductor Industry Association(SIA) projects that the semiconductor chips usedin personal computers and scientific workstationswill reach five times the speed and ten times thememory capacity of the current pentium-classprocessor by the year 2007. However, 1 GHz on-chip clock speeds and 64 Gbit/Chip DRAMtechnology will not be simple to achieve orinexpensive to produce. Such technologies willrequire scaling the minimum feature size of CMOSdevices (the transistors in the silicon chip) downto 100 nm from the current 350 to 500 nm. Thisrequirement has profound implications for devicemanufacturing—existing processing techniquesmust increasingly be understood quantitativelyand modeled with unprecedented precision.Indeed, revolutionary advances in the develop-ment of physics-based process simulation toolswill be required to achieve the goals for costefficient manufacturing and to satisfy the needs ofthe defense industrial base. These advances willnecessitate a fundamental improvement in ourbasic understanding of microstructure evolutionduring processing.

To cut development time and costs, the semi-conductor industry makes extensive use of simplemodels of dopant implantation and of phenom-enological models of defect annealing and diffusion.However, the production of a single device oftenrequires more than 200 processing steps, and thecumulative effects of the various steps are far toocomplex to be treated with these models. The lackof accurate process modeling simulators is provingto be a serious impediment to the development ofthe next generation of devices. New atomic-levelmodels are required to describe the point defectdistributions produced by the implantation process,and the defect and dopant diffusion resultingfrom rapid thermal annealing steps. The tempera-ture-dependent diffusivities of point defects insilicon have been a matter of controversy for manyyears. A seemingly consistent picture has emergedfor the interstitial (I) contribution to self diffusion,CIDI, where CI and DI are the equilibrium Iconcentration and diffusivity, respectively.

However, attempts to determine DI with variousexperimental conditions have resulted in discrep-ancies as large as nine orders of magnitude at700°C. Recent molecular dynamics (MD) simula-tions with empirical potentials1 and tight bindingmethods show that the vacancy diffuses fasterthan the interstitial at elevated temperatures.These calculations find activation energies forvacancy migration that range from 0.2 to 0.4 eV ingood agreement with experiments, and activationenergies for I migration that range from 0.9 eV to1.5 eV. Understanding point defect clusteringand cluster stability during annealing is also acomplex problem. While little experimental dataexist, recent MD simulations have shown thatbinding energies for vacancy clusters are lowerthan those for interstitials, with differences of atleast 1 eV for any given cluster size. The presentpoor state of understanding of the properties ofdefects in silicon not only poses a challenge forfundamental semiconductor physics, but alsopresents a formidable obstacle for the developmentof predictive models of silicon bulk processing.

In our work, we are developing a new combinedexperimental and simulation approach to defectkinetics in silicon. Experimentally, we are usingatomically clean Si(111) surfaces and the scanningtunneling microscope (STM) as a monitor for thediffusion of implantation-induced vacancies andinterstitials. The arrival of a vacancy (interstitial)at the surface would cause the disappearance(reappearance) of a surface atom. We measurenet arrival rates of vacancies and interstitialsdirectly by using the STM to count the number ofatoms populating the surface layer after variousstages of annealing at different temperatures.

The experiments were performed in an ultra-high vacuum (UHV) system with base pressurebelow 10–10 torr. Si(111) samples, which were cutfrom commercial wafers, exhibited the 7 × 7reconstruction in both low-energy electrondiffraction (LEED) and STM after annealing at1250˚C for 30 sec and cooling to room temperature.The STM reveals that step bunching results intypical terrace lengths exceeding 1 µm, with



occupation of over 99.7% of the atomic sites in theadatom layer, the outermost atomic layer, and thatwhich is imaged with the STM. Such atomicallyclean surfaces were then exposed briefly toirradiation by 5 keV Xe, with a total dose of~1.5 × 1013 cm–2, and then transferred in situ tothe tunneling microscope. After subsequentannealing cycles, the sample was always allowedto cool to room temperature before imaging in theSTM. Sample temperatures during annealing wererecorded with a pyrometer which was calibratedwith a thermocouple and are expected to beaccurate within 20˚C.

We find that after room-temperature irradia-tion of Si(111) – 7 × 7, annealing at 350˚C results ina decrease in the atomic population of the adatomlayer, while subsequent annealing at 500˚C restoresthe population of that layer. While the irradiationhas resulted in the disappearance of 8% of theatoms from the adatom layer, the long-rangeorder of the surface is still present in both STMand LEED.

STM images were also obtained followingprogressive stages of annealing of the surfaceirradiated with xenon ions. While we are unable toimage exactly the same location after each anneal-ing cycle, we believe (on the basis of measurementsat various positions on the surface) that regionswhich are imaged generally represent the morpho-logical characteristics of the sample. It is evidentfrom the images that annealing at 350°C leads todisappearance of the adatoms from the surface.While those remaining still assume binding sitesconsistent with the (7 × 7) superlattice, the percent-age of sites actually occupied drops from 92%immediately after irradiation to 35% after 5 minof annealing, and to 28% after 4 hr of annealing.Despite the depopulation of the adatom layer at350°C, the 7 × 7 periodicity remains, so the crystallineorder in the layers immediately below the adatomlayer has not been disrupted by the annealing.Annealing of smooth Si(111) – 7 × 7, which hasnot been exposed to ion irradiation, does not leadto the disappearance of adatoms.

Annealing the same surface at 500°C for 2 minresults in repopulation of 73% of the adatom layer.The strong coherence of the reconstruction observeddemonstrates that it was not disrupted by eitherthe irradiation or the subsequent lower-tempera-ture anneal. We do not observe at any stage eitherthe formation of adatom islands or of single-atomic-height terraces at the base of step bunches.

From a simulation viewpoint, we have devel-oped techniques to span the complete length andtimescale relevant to the silicon doping problem.

We use a combination of MD and kinetic MonteCarlo (KMC) simulations to show that the experi-mental results can be explained by different ratesof arrival to the surface of the vacancies andinterstitials produced in the bulk during ionirradiation. MD simulations provide a three-dimensional representation of the location of allthe defects induced by the implantation processwithin the nanosecond timescale. These resultsare used as input to the KMC simulations, whichdescribe defect evolution over timescales compa-rable with the experiments. Classical MDsimulations, with the Stillinger–Weber potentialfor silicon and the Ziegler–Biersack–LitmarkUniversal pair potential for silicon–xenon, incomputational boxes with up to 106 atoms areused to model the prompt (10–11 sec) displace-ment cascade process. This process gives rise tothe primary state of damage during implantation.We find that the average ion range is ≈80Å, withmost of the damage located close to the surface.Our observation of the concentration of disorderin amorphous pockets and an average sputteringyield of 2.25 atoms/ion are consistent with previousreports. The final ion dose of 1.5 × 1013 ions/cm2

is accumulated by incorporating 152 ion trajecto-ries into a box of dimensions (0.32 × 0.32 × 5 µm),which is used for the KMC simulations.

The defects produced by implantation evolveduring annealing through point defect diffusion,defect clustering, and cluster evaporation. Wefirst study the evolution of the damage producedby 5 keV Xe during annealing at 350°C. The KMCsimulation starts by recombining all Frenkel (V–I)pairs within one nearest-neighbor. The number ofvacancies reaching the surface after 4 hr annealingat 350°C is greater than the number of interstitials.The density of vacancies at the surface at this timeis 0.8 × 1014 vacancies/cm2, in reasonable agree-ment with the experimental observation of 1.2 × 1014

vacancies/cm2. This, in turn, represents that sixexcess vacancies per ion reach the surface, greatlyexceeding the sputter yield. The evolution ofdefects in the bulk proceeds in two stages: (1) Whenboth free vacancies and free interstitials arepresent, the process is governed by the differentmobilities of the defects. At the temperatureconsidered here, 350˚C, the diffusivities of V andI are similar. Both V and I diffuse and recombine,and some reach the surface. After 10 sec, onlydefect clusters remain in the bulk. (2) In thissecond stage, corresponding to the experimen-tally measurable times, the evolution of thedefects is governed by the binding energies of theclusters. Interstitial clusters dissociate at a much



lower rate than vacancies. Vacancies released fromclusters, having low binding energies, recombineeither at the surface, increasing the total populationof vacancies at the surface, or with interstitials inclusters, or other vacancies in clusters. The averagecluster size for interstitials is <3, while vacancyclusters grow to an average size of six.

After annealing for 4 hr, 42% of the defectscreated initially have recombined in the bulk,leaving only 5 V/ion and 9 I/ion in clusters.Increasing the temperature to 500˚C inducesdissociation of both vacancy and interstitialclusters. All vacancies quickly disappear from thebulk, leaving only interstitial clusters, whosegradual dissolution liberates interstitials thatmigrate to the surface and cause the experimen-tally observed recovery of the adatom layerwithin a few minutes.

The combination of experimental results andsimulations leads naturally and consistently to apicture of surface recombination of the bulk defectsinduced by initial irradiation. The experimentallyobserved depopulation of Si (111) – 7 × 7 adatomlayer under annealing at 350˚C, following initialroom-temperature irradiation, therefore indicatesthe net surface accumulation of bulk vacanciescreated by the irradiation and results from thelower binding energy of vacancy clusters relativeto interstitial clusters. The repopulation of theadatom layer observed experimentally uponsubsequent annealing at 500˚C indicates the netarrival of bulk intersitials, which graduallyevaporate from the remaining clusters in the bulkafter the less-stable vacancy clusters have evapo-rated. The combination of the results of annealingat the two temperatures would not be explainedby either a simpler picture of redistribution ofdefects created by sputtering solely at the surfaceor other recent models of ion-induced damage ofSi (111) – 7 × 7 surfaces. These results haveimportant consequences for understanding ionimplantation and annealing and for the develop-ment of process modeling.

References

1. H. Huang, T. Diaz de la Rubia, and M. J. Fluss,Mater. Res. Soc. Symp. Proc. 428, 177 (1996).

Publications

1. M. Jaraiz, G. H. Gilmer, J. M. Poate, andT. Diaz de la Rubia, Appl. Phys. Lett. 68, 409 (1996).

2. J. Zhu, T. Diaz de la Rubia, L. Yang, andC. Mailhiot, Phys. Rev. B 54(7), 4741 (1996).

3. T. Diaz de la Rubia and M. J. Caturla,Electroch. Soc. 96, 429 (1996).

4. M. Tang, L. Colombo, and T. Diaz de la Rubia,Mater. Res. Soc. Symp. Proc. 396, 33 (1996).

5. L. A. Marques, M. J. Caturla, H. Huang, andT. Diaz de la Rubia, Mater. Res. Soc. Symp.Proc. 396, 201 (1996).

6. M. J. Caturla, T. Diaz de la Rubia, L. Marques,and G. H. Gilmer, Phys. Rev. B 54, 16683 (1996).

7. L. Marques, M. J. Caturla, T. Diaz de la Rubia,and G. H. Gilmer, J. Appl. Phys. 80, 6160 (1996).

8. M. Tang, L. Colombo, and T. Diaz de la Rubia,“Intrinsic Point Defects in Crystalline Silicon:Tight Binding Molecular Dynamics Studies ofSelf Diffusion, Interstitial-Vacancy Recombi-nation and Formation Volumes,” LLNL,Livermore, CA, UCRL-JC-124739 (1996);accepted for publication in Phys. Rev. B (Nov1996).

9. T. Diaz de la Rubia, Ann. Rev. of Mater. Sci. 26,613 (1996).

10. P. Bedrossian, M. J. Caturla, andT. Diaz de la Rubia, Appl. Phys. Lett. 70, 176(1997).

Related Invited Presentationsin 1996

1. International Workshop on Radiation DamageCorrelations: Theory and Experiments,Davos, Switzerland, Oct 1996.

2. International Conference on Ion Beam Modifi-cation of Materials (IBMM 96), Albuquerque,NM, Sept 1996.

3. International Conference on Materials Modifi-cation with Ion Beams, Darmstadt, Germany,Sept 1996.

4. Ion Implantation Technology Conference (IIT96). Austin, TX, June 17–21, 1996.

5. 187th Annual Meeting of the ElectrochemicalSociety, Los Angeles, CA, May 6–10, 1996.

6. Workshop on “New Directions in TCAD forSilicon Processing,” NASA Ames, MountainView, CA, April 1996.

7. Workshop on Applications of AtomisticSimulation Tools to Semiconductor Processing.Paris, France, January 1996.

8. Kiritani Symposium on Defects in AdvancedMaterials, Nagoya, Japan, Dec 1996.

9. XIII SLAFES Solid State Physics Conference,Gramado, Brazil, Nov 1995.

BONDING AND STRUCTURE OF NANOCRYSTALLINE THIN-FILMS


BONDING AND STRUCTURE OF

NANOCRYSTALLINE THIN-FILMS

L. J. Terminello

The focus of this project is to establish anddevelop core-level soft x-ray photoabsorption intoan analytic tool to determine the bonding andstructure of incoherent, or amorphous, thin-filmswith nanoscale (<100Å) ordering or domains—inparticular, boron nitride, boron carbide, and carbonthin-films. These films are of great interest to thecoatings and microelectronics industries and areused as tribological, passivation, and hardeningmaterials within defense-related programs.Unfortunately, conventional structure determin-ing tools that are used to characterize mostmaterials are inaccurate when applied to films orclusters that have long-range order, or domainsizes, less than the coherence length of atomicstructure probing x-rays (<100Å) or the wave-length of bond sensitive probes (infrared andRaman < 1000Å). Only a local atomic structuralprobe-like core-level photoabsorption can mea-sure the actual bonding and structure in theseincommensurate-phase materials.

In FY96, we completed the project to determinethe bond order of carbon and diamond thin-films,experimenting in collaboration with C. Zuker,D. Gruen and A. Krausse at Argonne NationalLaboratory (ANL). The carbon 1s near edge x-rayabsorption fine structure (NEXAFS) from thesefilms were compared with the photoabsorptionfrom carbon standards: graphite (representativeof pure sp2 bonding) and diamond (representativeof sp3). The results of this comparison helped todetermine which growth conditions lead to filmsof the most desirable mechanical, electronic, andmorphological properties. For example, typicalfilms grown using magnetron sputtering exhib-ited an equal amount of sp2 and sp3 bonding, butfilms grown at ANL, using a microwave dischargeto extract reactive carbon dimers from C60 molecules,produced material that was pure sp3 (diamond)in character. In fact, these films grown frombuckyballs have nanometer-scale grain sizes, andstructure measurements on these materials usingx-ray and Raman techniques gave indeterminateresults. Our measurements provided conclusiveevidence that these materials were true diamondfilms that had unique nanoscale dimensions.1–3

Using core level photoabsorption, we charac-terized several light emitting MEH–PPV polymerfilms that were prepared by H. Pakpaz, H. Lee,and G. Fox at LLNL and then evaluated thedestructive process that leads to material failure.With this work, we were able to ascertain that theorbitals of the bonding backbone are disrupted bya photo-oxidation pathway. We found thatexposure of the polymers to both light andoxygen is necessary to cause damage to the film.This type of information is useful in understand-ing why the current technology for makingmetallic electrical contacts to the film interactswith the polymer, thereby damaging the light-emitting polymer panel.4

During the year, we completed our workcharacterizing boron–nitride and boron–carbidematerials, finding new results on the determina-tion of stoichiometric defects in BN and B4C thinfilms. This work was done in collaboration withA. Jankowski, LLNL, and G. L. Doll, GeneralMotors Corporation. In earlier work, we devel-oped x-ray absorption into an analytic techniquefor characterizing the bonding in incoherent filmsthat were grown under varying conditions. Wefound that the sp2 and sp3 content of BN filmscould be extracted from the photoabsorption, andthus, the technologically important cubic phasecould be ascertained. Our recent results showedthat nitrogen defects in h–BN films manifestedthemselves as a series of extra peaks in the π*structure in the boron 1s photoabsorption. Wewere also able to show that hexagonal BN filmsgrown on silicon substrates by pulsed laserdeposition could be converted to sp3 (cubic)bonding with subsequent ion implantation.5,6

We observed similar defect content in B4Cfilms that were grown with sputter deposition,where the matrix of B4C boron icosohedra werenot entirely made up of boron, but that some ofthe interstitial carbon was incorporated into thecages. This would affect the high temperaturestability of the material. We also observed thatsputtered films of the material were a differentcomposition of carbon and boron than the bulkmaterial (starting sputter targets). This would



have a significant impact on the behavior of thecryo-panels being built for the National IgnitionFacility. The bond order for these incommensuratefilms could only be probed with the elementallyand chemical state specific probe NEXAFS.

This work was performed at the AdvancedLight Source at Lawrence Berkeley Laboratory(LBNL). Collaboratively, IBM, University ofTennessee, Tulane University, LBNL, and LLNLare working as a research team to characterize thebonding and morphology of incoherent thin filmsand have recently begun more bulk sensitivemeasurements that relate the structure of novelmaterials to features in the x-ray fluorescence andthe core-level photoabsorption. In particular, weinitiated the probing of some new materials withsoft x-ray fluorescence.7 With this new tool wewere able to probe the valence structure ofdiamond thin-films.

References

1. D. Gruen, A. R. Krauss, R. Csencsits, C. D. Zuiker,J. A. Carlisle, I. Jimenez, D. G. J. Sutherland,L. J. Terminello, D. K. Shuh, W. M. Tong, andF. J. Himpsel, Appl. Phys. Lett. 68, 1649 (1996).

2. C. D. Zuiker, A. R. Krauss, D. M. Gruen,J. A. Carlisle, L. J. Terminello, S. A. Asher, andR. W. Bormett, “Characterization of DiamondFilms by Core-Level Photoabsorption,” inMRS Proceedings of the 1996 Spring Meeting ofthe Materials Research Society 437, 211, Apr 8–12,1996, San Francisco, CA.

3. F. L. Coffman, R. Cao, P. A. Pianetta, S. Kapoor,M. Kelly, and L. J. Terminello, Appl. Phys. Lett.69, 568 (1996).

4. D. G. J. Sutherland, K. Pakpaz, L. J. Terminello,S. C. Williams, P. Elliker, G. Fox, T. W. Hagler,H. W. Lee, T. A. Callcott, J. A. Carlisle,D. L. Ederer, F. J. Himpsel, J. J. Jia, I. Jimenez,D. K. Shuh, and W. M. Tong, Appl. Phys. Lett.68, 2046 (1996).

5. I. Jimenez, D. G. J. Sutherland, W. M. Tong,D. K. Shuh, J. A. Carlisle, A. Jankowski,L. J. Terminello, G. L. Doll, and F. J. Himpsel,Appl. Phys. Lett. 68, 2816 (1996).

6. A. F. Jankowski, I. Jimenez, J. P. Hayes,D. K. Shuh, D. G. J. Sutherland, J. A. Carlisle,L. J. Terminello, and F. J. Himpsel, “Near EdgeX-Ray Absorption Fine Structure Examinationof Chemical Bonding in Sputter DepositedBoron and Boron–Nitride Films,”in MRSProceedings of the 1996 Spring Meeting of theMaterials Research Society 437, 237, Apr 8–12,1996, San Francisco, CA.

7. J. J. Jia, T. A. Callcott, J. A. Carlisle, L. J. Terminello,A. Asfaw, D. L. Ederer, F. J. Himpsel, andR. C. C. Perera, Phys. Rev. Lett. 76, 4054 (1996).

Presentations

1. D. G. J. Sutherland, F. J. Himpsel, L. J. Terminello,J. A. Carlisle, D. K. Shuh, W. M. Tong,R. C. C. Perera, T. A. Callcott, J. J. Jia, andD. L. Ederer, “Core-Level Spectroscopy ofThin Oxides and Oxynitrides Grown withN2O on Si (100) Surfaces,” The 42nd NationalSymposium of the American Vacuum Society,October 17, 1995, Minneapolis, MN.

2. R. P. Winarski, D. L. Ederer, A. Moewes,T. A. Callcott, L. Zhou, J. J. Jia, J. A. Carlisle,L. J. Terminello, A. Asfaw, F. J. Himpsel, andR. C. C. Perera, “Inelastic and ResonantScattering in Metal Diborides and Hexaborides,”The 1996 March Meeting of the AmericanPhysical Society, St. Louis, MO, March 18, 1996.

3. D. G. J. Sutherland, I. Jimenez, D. K. Shuh,W. M. Tong, L. J. Terminello, J. A. Carlisle,K. Pakbaz, H. W. Lee, G. Fox, H. Radousky,P. Elliker, T. W. Hagler, S. C. Williams, andF. J. Himpsel, “Bond Order and DamageMechanism in Light Emitting MEH–PPVPolymer Films Determined with NEXAFS,”The 1996 March Meeting of the AmericanPhysical Society, St. Louis, MO, March 20, 1996.

4. I. Jimenez, D. G. J. Sutherland, W. M. Tong,D. K. Shuh, J. A. Carlisle, L. J. Terminello,J. J. Jia, T. A. Callcott, D. L. Ederer, C. Zuiker,A. R. Krauss, and D. M. Gruen, “Character-ization of Nanocrystalline Diamond FilmsUsing Soft X-Ray Absorption and Fluores-cence,” The 1996 March Meeting of the AmericanPhysical Society, St. Louis, MO, March 21, 1996.



5. D. G. J. Sutherland, K. Pakbaz, L. J. Terminello,S. C. Williams, I. Jimenez, J. A. Carlisle,T. A. Callcott, J. J. Jia, and D. L. Ederer, “Core-Level Photoabsorption Spectroscopy ofMEH–PPV,” The 1996 Spring Meeting of theMaterials Research Society, San Franciso, CA,April 9, 1996.

6. I. Jimenez, W. M. Tong, D. K. Shuh,D. G. J. Sutherland, A. Jankowski, J. A. Carlisle,L. J. Terminello, G. L. Doll, L. V. Mantesc, andF. J. Himpsel, “NEXAFS Characterization ofBoron–Nitride Thin Films,” The 1996 SpringMeeting of the Materials Research Society, SanFranciso, CA, April 9, 1996.

7. W. Tong, D. K. Shuh, I. Jimenez, J. A. Carlisle,D. G. J. Sutherland, B. C. Holoway, P. Pianetta,D. Berns, and M. A. Cappelli, “Investigationof Carbon Nitride Films by NEXAFS and SoftX-Ray Photoemission,” The 1996 SpringMeeting of the Materials Research Society, SanFranciso, CA, April 10, 1996.

8. A. F. Jankowski, J. A. Carlisle, I. Jimenez, andL. J. Terminello, “NEXAFS Examination ofChemical Bonding in Sputter DepositedBoron and Boron Nitride Films,” The 1996Spring Meeting of the Materials Research Society,San Franciso, CA, April 10, 1996.

9. C. D. Zuiker, D. M. Gruen, A. R. Krauss,J. A. Carlisle, and L. J. Terminello, “Character-ization of Diamond Films by Core LevelPhotoabsorption,” The 1996 Spring Meeting ofthe Materials Research Society, San Franciso, CA,April 10, 1996.

A NOVEL APPROACH TO SIC FILM PRODUCTION


A NOVEL APPROACH TO SIC FILM PRODUCTION FOR

MICRO-MECHANICAL AND ELECTRONIC COMPONENTS

A. Hamza, M. Balooch

We developed a new method for selectivegrowth of stoichiometric, crystalline SiC filmsfrom fullerene precursors that grow epitaxially atlow substrate temperature (<1100 K). Conven-tional chemical vapor deposition techniquesrequire reactor and substrate temperatures ashigh as 1700K. Deposition of the SiC film can bemade selective, because of the very differentchemical interaction of fullerenes with silicon andsilicon–dioxide surfaces. This interaction elimi-nates the need to etch the SiC film to manufacturedevices, which is difficult to do, because of SiC’sextreme resistance to chemical attack. Thisprocess enables SiC to be incorporated intoimproved devices.

The scope of the application of SiC films isextremely broad. The mechanical and chemicalproperties of bulk SiC suggest that micromachinesand microtransducers that operate at high tempera-tures and/or in extremely corrosive environmentscan be fabricated from SiC films. Thesemicromachines will provide enormous applicationsin microchemistry and microbiology, as well asautomotive and aerospace engineering, where thereis a need for performance at high temperatures.

Problem of Interface VoidFormation

SiC film growth via reaction of sublimedfullerenes with a silicon substrate proceeds by thediffusion of substrate silicon to the surface of thegrowing film. The sources of silicon for growthare defects in the substrate and/or interface siliconin the case of growth on SiO2/Si patterned wafers.Screw dislocations are known sources of substratematerial. During growth, voids form at the defectsand the Si/SiC interface. As growth continues,the voids grow and coalesce until there is little orno contact between the film and the substrate, atwhich point growth stops. This process is usefulfor releasing SiC microcomponents of variousthicknesses by controlling defect density. However,

a more desirable process would allow the voids tobe controlled to the point of elimination.

Co-Sublimation of Silicon

We have found that growth of SiC films, viareaction of sublimed fullerenes, can proceed withco-sublimed silicon on silicon substrates (see Fig. 1).By varying the flux of co-sublimed silicon, voidformation can be controlled at the Si/SiC interface.By growing in excess fullerene flux and at lowsubstrate temperature (950K) co-sublimation ofsilicon can produce void free, smooth Si/SiCinterfaces for both blanket deposition and onpatterned wafers. It is important to maintain anexcess fullerene flux during co-sublimation so thatfilm stoichiometry, Si:C, can be maintained at 1:1.On a highly defected surface, 105 defects/cm2, aSiC film grown 2000 Å thick will not adhere to thesubstrate due to substantial void formation. Byco-sublimation of silicon at the same low sub-strate temperature, we grew a 6000 Å SiC filmwith no interface voids.

Figure 1. Blanket deposition of SiC on SiO2/Sivia reaction of co-sublimed fullerenes andsilicon. Interface voids are eliminated.

A NOVEL APPROACH TO SIC FILM PRODUCTION


Control of MicrocomponentThicknesses

Microcomponent thickness can easily becontrolled by varying the co-sublimed silicon fluxrelative to the substrate silicon diffusion. Sub-strate silicon diffusion is controlled by thesubstrate temperature and the silicon concentrationgradient. The silicon concentration gradient isfixed by the conditions for maintaining a 1:1 Si:Cstoichiometry. Thus the sublimed silicon flux andthe substrate temperature are easily manipulatedvariables that allow the growth of SiC films atnearly any thickness at release from the surface.

Liftoff of Co-Sublimed Film

Co-sublimation of silicon does eliminate oneof the advantages of the fullerene-based growthtechnique—selective growth on silicon substrates(no growth on SiO2 substrates). We have foundthat patterned wafers can still be achieved withoutadding any processing steps via a liftoff technique.While the co-sublimation method does produceblanket deposition, HF acid stripping of the oxidelifts off the SiC film deposited on SiO2 regions ofthe substrate. A discontinuity in the film at SiO2features is required so that the HF can attack theoxide features. This co-sublimation liftoff processis significant in that undercut of features, due todiffusion of interface silicon, is eliminated.

Improvement of CantileverProperties

We have also measured the mechanicalproperties of the SiC films grown by co-sublimationof silicon and fullerenes. Silicon microcantileverswere fabricated and the elastic properties of thebeam were determined by a LLNL-developedatomic force microscope-based deflection technique.The microcantilevers were then coated with a1-µm-thick SiC film. The original silicon cantileverwas 35-µm thick, 80-µm wide, and 500-µm long.

A 5% increase in the thickness of the film lead to a100% increase in the stiffness of the coated cantile-ver, revealing the importance of SiC coatings toimprove microcomponent performance.

Elastic Modulus and Momentof Inertia of the Beam

The increase in the SiC coated cantileverstiffness is due to the factor-of-four increase in theelastic modulus of the SiC vs silicon and theincrease in the moment of inertia of the coatedcantilever beam. These components could evenbe lightened by etching away of the substrate.This result suggests the possibility of makingextremely robust microcomponents for structuralapplications (i.e., disk drive supports). In addition,the increased stiffness may improve the perfor-mance of accelerometers (e.g., airbag triggers).

SiC Membranes

Our first devices were produced this year. Inaddition to the microcantilever beams, we alsomanufactured 0.5- to 1-µm-thick SiC membranes.These membranes are produced simply byblanket deposition of SiC films via reaction offullerenes with co-sublimed silicon or the siliconsubstrate. The back of the substrate is thenmasked with silicon–nitride and the wafer isetched from the back with KOH. The SiC filmacts as a natural etch stop because of its chemicalinertness. These SiC membranes could findapplications as pressure sensors in harsh environ-ments (i.e., high temperature).

These results and the results from FY95 suggestSiC thin films can be used in microelectromechanicalsystems (MEMS) to enhance their wear propertiesand to improve their operation. We have demon-strated the use of this growth technique tofabricate SiC microcomponents. The mechanicalproperties of these SiC films are more robust thanthose for silicon that is currently used in MEMS.Therefore, SiC films grown from C60 may presentan opportunity for MEMS applications that havenot been previously considered.

GIANT MAGNETORESISTANCE MATERIALS WITH NOVEL SPACER LAYERS


GIANT MAGNETORESISTANCE MATERIALS WITH NOVEL

SPACER LAYERS

A. Chaiken

Although the phenomenology of the giantmagnetoresistance (GMR) effect is well-known,the physical origin is still incompletely explained.Greater insight into the GMR will not only enrichour understanding of electron transport in magneticmaterials, it will also help us to fully exploit theeffect for practical applications such as landminedetection and non-destructive inspection of welds.

Giant magnetoresistance refers to the observa-tion of a large change in the electrical resistance ofa ferromagnet/non-magnet multilayer that isplaced in an applied magnetic field. Accompa-nying the large change of resistance (as great as80% at room temperature) is a change in themagnetic state of the sample. Existing theories ofthe magnetoresistance treat only multilayers withmetallic spacer layers. Are the magnetic propertiesof these ferromagnet/semiconductor multilayerssimilar to those of ferromagnet/metal multilayers,and if so will we be able to exploit the effect incompletely novel ferromagnet–semiconductorintegrated circuits? This work investigatessemiconductor spacer layers.

During the first two years of the project, weestablished conclusively that the magneticproperties of iron/silicon multilayers are muchlike those of more commonly studied GMRmaterials like iron/chromium multilayers. In aseries of publications, we detailed evidence thatthe magnetization curves of the iron/siliconmultilayers depend on whether film depositionconditions favor the formation of a metallic ironsilicide phase, which plays a role analogous tothat of chromium in the iron/chromium multilayers.Figure 1 shows spectroscopic evidence thatdemonstrates the identification of the iron sili-cides as metals, acquired by LLNL collaboratorsat the Advanced Light Source.

In the last year of the project, we emphasizedbasic materials science and electrical engineeringstudies that are directly relevant to DOE-mission-relevant applications. In particular, we studied

the phenomenon of exchange biasing in nickel–iron/nickel–oxide bilayers, which are importantmaterials for high-performance GMR-basedmagnetic sensors. Exchange biasing refers to thephenomenon whereby an antiferromagnet incontact with a ferromagnet may apply an effectivefield to the ferromagnet, thereby forcing theferromagnetic film to its point of maximumsensitivity to external fields. Understanding thisphenomenon is crucial for the manufacture of GMRsensors with acceptable sensitivity and yield. Aseries of two papers describe this research, compar-ing the exchange bias effect in polycrystallinebilayers with that in single-crystal epitaxial bilayers.We found that, surprisingly, the exchange biaseffect is actually smaller in the single-crystal films.

Figure 1. X-ray spectroscopic data taken at theAdvanced Light Source. The valence andconduction bands of the silicon wafer do notcross because silicon, a semiconductor, has anenergy gap. The crossing of the energy bandsin the iron/silicon sample shows that the ironsilicide in this film is metallic.

85.0 90.0 95.0 100.0 105.0

Silicon

Fe/Si

Fluo

resc

ence

Yie

ld

Photon Energy (eV)

GIANT MAGNETORESISTANCE MATERIALS WITH NOVEL SPACER LAYERS


Publications

1. A. Chaiken, R. P. Michel, and M. A. Wall,Phys Rev B53, 5518 (1996).

2. A. Chaiken, R. P. Michel, and M. A. Wall,Journal of Appl Phys 79, 4472 (1996).

3. J. A. Carlisle, A. Chaiken, R. P. Michel,L. J. Terminello, J. J. Jia, T. A. Calcott, andD. L. Ederer, Phys Rev B53, R8824 (1996).

4. R. P. Michel, A. Chaiken, Y. K. Kim, andL.E. Johnson, “NiO Exchange Bias LayersGrown by Direct Ion-Beam Sputtering of aNickel–Oxide Target,” IEEE Transactions onMagnetics 32, 4651 (1996).

5. R. P. Michel, A. Chaiken, L. E. Johnson, andC. T. Wang, “Comparison of ExchangeAnistropy in Polycrystalline and Epitaxial(001)- Oriented NiO/NiFe Bilayers Grown byIon-Beam Sputtering,” Lawrence LivermoreNational Laboratory, Livermore, CA, UCRL-JC-124721 (1996); submitted to Phys Rev B.

Presentations

1. R. P. Michel, “Exchange Anisotropy inEpitaxial and Polycrystalline NiO/NiFeBilayers Grown by Ion Beam Sputtering,”contributed talk, MMM Meeting, Atlanta, GA,November 1996.

2. A. Chaiken, “Interlayer Exchange Coupling inFe/Si Multilayers,” invited talk, AmericanVacuum Society Meeting, Philadelphia, PA,October 14, 1996.

3. A. Chaiken, “Interlayer Exchange Coupling inFe/Si Multilayers,” invited talk, AmericanVacuum Society Meeting, Denver, CO,August 22, 1996.

4. R. P. Michel, “NiO Exchange Bias LayersGrown by Direct Ion Beam Sputtering of aNickel Oxide Target,” Intermag Meeting,Seattle, WA, April 1996.

STUDIES OF MESO-STRUCTURAL FEATURES IN HIGH-TC SUPERCONDUCTING MATERIALS


STUDIES OF MESO-STRUCTURAL FEATURES IN HIGH-TCSUPERCONDUCTING MATERIALS USING PAIR DISTRIBUTION

FUNCTION ANALYSIS OF NEUTRON/X-RAY SCATTERING DATA

G. H. Kwei

The objectives and goals of this program areto study the local structure of superconductingcompounds, colossal magnetoresistant (CMR)materials, and plutonium alloys to better under-stand the relationship between the microscopicstructure of materials and their macroscopicproperties. In particular, we plan to explore thedifferences between the true local structure andthe long-range crystallographic structure and toexamine how these differences might betterexplain some of the unusual physical propertiesof these materials. In addition, we began a studyto apply these techniques for use as a possible newtool for protein structure determination.

During FY96, we conducted several traditionalstudies of materials to determine if they were goodcandidates for further powder diffraction file (PDF)studies. Our study of disordered structures inCe2Pt6Ga15 led to new techniques in the analysisof diffuse scattering in single-crystal diffraction tosort out the detailed structure. Local structuralchanges in phase transition in high-Tc supercon-ductors are often different from the crystallographicchanges.1 We also studied changes in local struc-ture with temperature in the CMR materialLa1–xCaxMnO3, which generally established theimportance of polaron formation in theseperovskite materials.2 We have also embarked ona series of x-ray absorption fine structure (EXAFS)experiments to study the local structure inTl2Mn2O7,3 where the sample was too small forneutron scattering studies. This material alsodisplays CMR, but instead has a pyrochlore

structure. The EXAFS data show that polaronformation is not important and confirms theearlier expectation that CMR results from acompletely different mechanism.4

In our study of the local structure of Pu–2at.%Gaalloys, we find it to be in reasonable agreementwith that of other fcc metals, such as Ni; however,there is an important difference—distortions and/oranistropic thermal parameters that develop fromthe system’s cubic symmetry. This behavior isunderstandable in the context of the extremestructural instability of plutonium, and what wehave is the beginning of a soft-mode transition tothe next structural phase.

We have been successful in using resonantdiffraction, which can be used to provide differen-tial pair distribution functions in biologicalstructure studies. Therefore, we have begun thiswork with EXAFS and differential PDF studies ofsome model compounds that mimic peptides andproteins in which a central Cd ion is bound tocysteins and histidines. The differential PDF (ob-tained by going on and off the Cd resonance) willprovide radial distributions of atoms about the Cdion which can then be used to determine thetertiary structure. We plan to study small Cdbinding peptides (phytochelatins) and proteins(metallothionein), which play an important roleof occupying heavy metals in plants and humans,respectively. This work is important becausethere are only two other techniques for studyingbiological structures—high-resolution nuclearmagnetic resonance and single-crystal x-raydiffraction.

STUDIES OF MESO-STRUCTURAL FEATURES IN HIGH-TC SUPERCONDUCTING MATERIALS


References

1. S. J. L. Billinge and G. H. Kwei, J. Phys. Chem.Solids 57, 1057 (1996).

2. S. J. L. Billinge, R. G. diFrancesco, G. H. Kwei,J. J. Neumeier, and J. D. Thompson, Phys. Rev.Lett. 77, 715 (1996).

3. G. H. Kwei, C. H. Booth, F. Bridges, andM. A. Subramanian, Phys. Rev. B 55 (Jan 1, 97).

4. M. A. Subramanian, B. H. Toby, A. P. Ramirez,A. W. Sleight, W. J. Marshall, and G.H. Kwei,Science 273, 81 (1996).

Publications

1. B. Morosin, J. D. Jorgensen, S. Short, G. H. Kwei,and J. E. Schirber, Phys. Rev. B 52, 1675 (1996).

2. G. H. Kwei, J. D. Jorgensen, J. E. Schirber, andB. Morosin, Fullerene Science and Technology 4(1996).

3. R. L. Harlow, G. H. Kwei, R. Suryanarayanan,and M. A. Subramanian, Physica C 256, 125(1996).

4. G. H. Kwei and B. Morosin, J. Phys. Chem. 100,8031 (1996).

5. J. E. Schirber, L. Hansen, B. Morosin, J. E. Fischer,J. D. Jorgensen, and G. H. Kwei, Physica C 260,173 (1996).

6. G. H. Kwei, A. C. Lawson, B. Morosin,A. C. Larson, E. A. Larson, and P. C. Canfield,Acta Cryst B 52, 580 (1996).

7. D. Klesnar, B. Morosin, G. H. Kwei,A. C. Lawson, and T. L. Aselage, J. AlloysCmpds. 241, 180 (1996).

8. J. M. Lawrence, G. H. Kwei, J. L. Sarrao, Z. Fisk,D. Mandrus, and J. D. Thompson, Phys. Rev. B54, 6011 (1996).

9. J. M. Lawrence, Y.–C. Chen, G. H. Kwei,M. F. Hundley, and J. D. Thompson, “Structureand Magnetism in CePt2+x,” Phys. Rev. B 55(in press).

SUPERPLASTICITY IN ALUMINUM ALLOYS



T. G. Nieh

Microstructure

The microstructure of a sample before testingconsists of a cellular structure with the cell sizeranging from 100 nm to as large as 2 µm. Thecellular structure was readily recovered uponannealing. Figure 1 shows a TEM dark-fieldimage from the grip region of a sample tested at350°C at a strain rate of 10–5 s–1. The microstruc-ture has a strong <011> rolling texture and containslow-angled grains (i.e., subgrains). The 350°Cannealing (~16 h) apparently causes recovery andresults in the formation of subgrains from theinitial cellular structure (where the averagesubgrain size is about 1 µm). Examination of thegage region (dynamically annealed) of the testedsample indicates that, except for having a slightlylarger subgrain size (~2 µm), the microstructure isvirtually the same as that in the grip region. Theoverwhelming presence of subgrain boundaries isattributable to the fact that the alloying of magne-sium to aluminum greatly reduces its stackingfault energy. Dislocation recovery is expected tobe difficult in an alloy with a low stacking faultenergy and, thus, enhances the formation ofsubgrains in the alloy. The microstructure also

The most effective way of increasing automo-bile mileage while decreasing emissions is toreduce vehicle weight. Therefore aluminum,which has only about one–third the density ofsteel, is being considered as a suitable substitute.Also, most commercial aluminum alloys possesssubstantially higher specific strength comparedwith steel, and aluminum is readily recyclable.However, several obstacles need to be addressed,including: the relatively high cost of aluminum ascompared with steel, and the forming limits ofaluminum, which are significantly lower thanthose for steel. Also, there is a growing interest insuperplasticity, and particularly high-rate super-plastic forming to create autobody components.1

This technique has the advantages of deliveringexceptional formability, potentially giving gooddimensional tolerance and delivering rapid time–to–market. The general objective of this researchis, therefore, to develop a basic understanding ofhigh strain rate superplasticity in aluminum alloys.

Scandium is the only alloying element to forma thermally stable, coherent L12 phase, Al3Sc, inaluminum (analogous to in Ni-based superal-loys). In fact, Al3Sc is the most potent strengthener,on an equal atomic fraction basis, known inaluminum-base systems,2 and its precipitate isunusually resistant to coarsening. As a result,Al3Sc precipitates are extremely effective instabilizing substructures, thus allowing the use ofstrain-hardening and grain-boundaries strength-ening to enhance the strength of aluminum alloys.In addition, the effectiveness of the Al3Sc precipi-tate in pinning grain boundaries can be used toproduce fine-grained aluminum for superplasticity.

In the present research, we demonstrate theeffectiveness of Al3Sc in stabilizing the substructure/structure in an Al–Mg–Sc alloy (composition inwt%: Al–5.76Mg–0.32Sc–0.3Mn–0.1Fe–0.2Si–0.1Zn)and to relate the microstructural evolution to theformability of this alloy. Tensile tests were con-ducted in air at temperatures between 300 and500°C and at strain rates between 10–5 and 1 s–1.Microstructures of the samples before and afterdeformation were examined using a transmissionelectron microscope.

′γ

Figure 1. TEM dark-field image from the gripregion of a sample tested at 350°C at a strain rateof 10–5 s–1.



revealed that fine (L12-type) Al3Sc precipitates,with particle size ranging from 10 to 100 nm, werefound to be uniformly distributed within alumi-num grains. The spacing of the Al3Sc particlesranged from 50 to 100 nm.

In a manner similar to that in Fig. 1, themicrostructure from the grip region of a sampletested at 475°C at a strain rate of 10–2 s–1 alsoconsists of primarily low-angled grain boundarieswith boundary misorientation angles of less than4–5°. The subgrain size is about 1 µm, which isslightly larger than that observed at 350°C (~1 µm).In comparison, Fig. 2 shows the gage region exhibitsa different structure, illustrating a recrystallizedmicrostructure in which the majority of grainboundaries are high-angled. The average grainsize (~3 µm) is slightly larger than that of thesubgrain size in the grip region. From a super-plasticity point of view, a 5-µm grain size isconsidered to be fine for an aluminum alloy. Thefine microstructure is evidently a result of theextremely fine (10–20 nm) and uniform distribu-tion of the L12 Al3Sc precipitates. In fact, theseprecipitates not only effectively pin high-angledgrain boundaries but also subgrain boundaries, asillustrated in Fig. 3.

Mechanical properties

Figure 4 shows the stress-true strain curves forAl–5Mg–0.3Sc alloy, tested at different tempera-

Figure 2. Microstructure of the gage region ofthe sample tested at a strain rate of 10–2 s–1 and475°C. Stress axis is indicated and cavities atgrain triple junctions are marked.

SASA

Figure 3. Al3Sc precipitates effectively pinsubgrain boundaries.

Figure 4. True stress-true strain curves forAl–5Mg–0.3Sc alloy tested at a true strain rate of10–2 s–1 at different temperatures (350–500°C).

0

20

40

60

80

100

True strain

Tru

e st

ress

(M

Pa)

350°CAl-6%Mg-0.3%Sc

400°C450°C

475°C

500°C

ε = 10-2s-1

•

1.00.80.60.40.20.0

tures (350–500°C) at a true strain rate of 10–2 s–1.At all temperatures there was an immediatehardening upon loading; at temperatures lowerthan 450°C, this hardening was followed by acontinuous softening. In contrast, at temperatureshigher than 475°C, the hardening was followedby an apparent steady-state flow. The elongationvalue is a function of testing temperature andexhibits a maximum at 475°C.



Figure 5 illustrates the true stress-true straincurves for the alloy tested at 475°C and atdifferent true strain rates. Except for strain ratesfaster than 2 × 10–1 s–1, a region of steady-stateflow appears at all strain rates. Within the strainrates from 1.1 × 10–3 to 1.1 s–1, the maximumtensile elongation occurs at strain rates ofabout 10–2 – 10–1 s–1.

Figure 6 summarizes experimental resultsobtained at various temperatures. The strain ratesensitivity value, m, in the equation σ = k ε m isnoted to increase with testing temperature. At350°C, m is about 0.35 and increases to 0.45 at 475°C.Tensile elongation approximately follows theexpected trend, i.e., a higher m value results in alarger elongation, showing elongation always <200%.

1.21.00.80.60.40.20.00

20

40

60

80

100

120

140

Al–6Mg–0.3Sc T = 475°C

True strain

Tru

e st

ress

(M

Pa)

10-1s-1100s-1 2x10-1s-1

5x10-2s-1

5x10-3s-110-3s-1

10-2s-1

Figure 5. True stress-truestrain curves forAl–5Mg–0.3Sc tested at475°C and at differentstrain rates.

Figure 6. (a) The flowstress (at a fixed strain of0.2) as a function ofstrain rate, and (b) theelongation-to-failure as afunction of strain rate.

101

102

101

102

103

10010-110-210-310-4

Strain rate (s-1 )

10010-110-210-310-4

Strain rate (s-1 )

Elo

ngat

ion

(%)

350°C

450°C

475°C

350°C

450°C

475°C

Al−6%Mg–0.3%Sc

Al−6%Mg–0.3%Sc

(b)

(a)

Stre

ss (

MPa

)

m = 0.45

m = 0.35

ε = 0.2 m = 0.40



Discussion

Al–Mg alloys are known to exhibit Class 1solid solution behavior, namely, deformation iscontrolled by solute-drag on gliding disloca-tions.3,4 In the present study, the Al–Mg–Sc alloypossesses a fine-grained structure, thus, grainboundary sliding is also expected to operateunder certain test conditions. Since solute-dragand grain boundary sliding are two independentmechanisms, the resultant deformation in thealloy is the summation of contributions from bothmechanisms. That is, the deformation strain ratecan be described by

ε ε ε σ σtotal gbs drag gbs LD= + = +AD B2 3 , (1)

where εtotal is the total strain rate; εgbs and εdragare the strain rates caused by grain boundarysliding and solute drag, respectively; Dgbs and DLare the grain boundary and lattice diffusioncoefficients, respectively; σ is the flow stress; andA and B are material constants.

According to Eq. (1), depending upon the testconditions, the strain rate sensitivity value m shouldhave an upper bound value of 0.5 and lowerbound value of 0.33. The experimental results(Fig. 6) indeed showed that m ranges from 0.33 to0.5. Specifically, at 350°C the m value is about 0.35.At this temperature, although the grain sizeappears to be fine, the grains are primarilysubgrains (see Fig. 1). Subgrain boundaries aregenerally immobile with respect to grain bound-ary sliding. As a result, the grain boundarysliding process is not expected to prevail at 350°Cand deformation would be mainly controlled bydislocation glide through the lattice (i.e. m ~ 0.33).This is consistent with the previous result that thesolute-drag mechanism in Al–Mg usually takesplace at intermediate temperatures around 300°C.3

At 475°C, the m value is about 0.45. At thishigh temperature, as a result of the pinning effectsof the Al3Sc particles, fine subgrains are stillthermally stable under static conditions. Understresses, however, high-angled grain boundariesrapidly evolve (Fig. 2). These deformation-induced, high-angled, boundaries are readily able

to slide and dominate the overall deformation inthe sample. This results in a high strain ratesensitivity value, in the proximity of 0.5, reflect-ing the grain boundary sliding mechanism. Othermicrostructural evidence for the prevalent grainboundary sliding at high temperature is shown inFig. 2, in which cavities at grain triple junctionsare readily observed. These cavities were formedbecause grain boundary sliding was not properlyaccommodated. In contrast, only a limitedamount of cavity formation was observed at graintriple junctions in samples deformed at 350°C.This is because dislocation glide is primarily anintragranular process which is not expected tolead to cavity formation at triple junctions.

A final comment is noted about the elonga-tion: although the total elongation was <200%, itshould be pointed out that these data were obtainedfrom testing extremely thin samples (90 µm).Tensile elongation is expected to be stronglysensitive to surface defects on thin samples. Infact, our most recent data measured from thicksamples (~2 mm) indicated that elongation can beover 700% at 475°C.

Summary

The microstructure and mechanical proper-ties of an Al–6Mg–0.3Sc alloy were characterized.The presence of Sc results in the uniform distribu-tion of fine L12 precipitates, which stabilize thegrain substructure/structure in the alloy. At anintermediate temperature of 350°C, subgrains areformed and they are stable even under a dynamiccondition (i.e., under stress). The deformation ofthe alloy at this temperature is controlled by solutedrag on gliding dislocations. Thus, the strain ratesensitivity value is about 0.33. At a high tempera-ture of 475°C, fine subgrains are still preferentiallyformed under static conditions (i.e., withoutstress). But, under a dynamic condition, the low-angled subgrain boundaries quickly convert intohigh-angled grain boundaries and lead to exten-sive grain boundary sliding. Therefore, thedominant deformation mechanism of the alloy at475°C is grain boundary sliding, and the alloyexhibits a strain rate sensitivity value close to 0.5.



References

1. T. G. Nieh and J. Wadsworth, JOM 44(11), 46(1992).

2. N. Blake and M. A. Hopkins, J. Mater. Sci. 20,2861(1985).

3. H. Oikawa, K. Sugawara, and S. Karashima,Trans. JIM 19, 611 (1978).

4. O. D. Sherby and P. M. Burke, Prog. Mater.Sci. 13, 325 (1967).

Publications

1. E. M. Taleff, G. A. Henshall, D. R. Lesuer,T. G. Nieh, and J. Wadsworth, “EnhancedTensile Ductility in Al–Mg Alloys by SolidSolution Interactions,” in Aluminum andMagnesium for Automobile Applications,J. D. Bryant and D. R. White, Eds. (TheMinerals, Metals, and Materials Society,Warrendale, PA, 1996) pp. 125–134.

2. J. Huang, A. J. Schwartz, and T. G. Nieh,“High Temperature Deformation in 2036 Aland 0.2 wt% Zr–2036 Al,” in Aluminum andMagnesium for Automotive Applications,J. D. Bryant and D. R. White, Eds. (The Minerals,Metals, and Materials Society, Warrendale,PA, 1996) pp. 149–162.

3. J. Huang, L. M. Hsiung, and T. G. Nieh, “Effectof Strain Rate on the Elevated–TemperatureTensile Properties of an Al–Pb Alloy,” ScriptaMetallurgica et Materialia 35, 919–924 (1996).

4. T. Imai, S. Kojima, G. L’Esperance, B. D. Hong,D. Jiang, and T. G. Nieh, “Effects of Tempera-ture on the Superplastic Characteristics of aPower–Metallurgy Pure Aluminum,” ScriptaMetallurgica et Materialia 35, 1189–1193 (1996).

5. T. C. Schulthess, P. E. A. Turchi, A. Gonis,and T.G. Nieh, “Stacking Fault Energies inAl–Based Alloys,” in Proc. First InternationalAlloy Conference, A. Gonis and A. Meike, Eds.(Plenum Publishing Co., New York, 1996, inpress).

6. T. G. Nieh and J. Wadsworth, “Effect of LiquidPhase on Superplasticity at High Strain Ratesin Metals and Their Composites,” in Proc.First International Alloy Conference, A. Gonisand A. Meike, Eds. (Plenum Publishing Co.,New York, 1996, in press).

7. T. G. Nieh and J. Wadsworth, “The Role ofLiquid Phase on Superplasticity in Metals andCeramics,” in International Symposium TowardInnovation in Superplasticity, (Amanohashidate,Japan, July 23–24, 1996, in press).

8. T. G. Nieh, R. Kaibyshev, L. M. Hsiung,N. Nguyen, and J. Wadsworth, “SubgrainFormation and Evolution during the Defor-mation of an Al–Mg–Sc Alloy at ElevatedTemperatures,” in Scripta Metallurgica etMaterialia (1997, in press).

MOLECULAR SCALE INVESTIGATION OF CRYSTAL GROWTH FROM SOLUTIONS


MOLECULAR SCALE INVESTIGATION OF

CRYSTAL GROWTH FROM SOLUTIONS

T. Land, J. De Yoreo

Introduction

Over the past decade, scanned probe micros-copy has been a vital tool for investigating thephysics of crystal surfaces grown in ultra highvacuum environments, providing a wealth ofinformation on the dynamics of surface evolution,atomic mobility, and step motion. In contrast, ourknowledge of step dynamics and surface morpho-logical evolution on crystal surfaces in solution isbased upon either theoretical considerations,most notably the theory of Burton, Cabrera, andFrank (BCF),1 and its later modifications,2 andindirect experimental techniques such as electronmicroscopy on ex situ surfaces or in situ opticalinterferometry.3 Indeed, while most of our currentunderstanding of growth in this environment is aresult of the latter, because of the limited resolu-tion of this technique, significant questions remainconcerning fundamental aspects of growth, includ-ing: the effects of dislocation structure on growthsource activity and surface morphology, evolutionof growth modes as a function of supersaturation(chemical potential); the effect of strain and defectson nucleation and step motion; and, perhaps mostimportantly, the pathway by which ions leave thesolvated state in solution to become molecularunits in the solid.

Recently, the advent of atomic force micros-copy (AFM) has made it possible to investigate insitu the growth from solution of a wide range ofsystems. These systems range from simple inorganicsalts4–6 such as KH2PO4 (KDP) where moleculardiameters are a few angstroms, to large, complexmacromolecules7–13 with diameters up to 160Å.9,10

The purpose of this project is to use AFM toinvestigate the evolution of surface morphologyand step dynamics during growth of crystalsfrom solution to understand the rate controllingmechanisms. Two primary systems were investi-gated, KH2PO4 (Mr = 135g/mole)—the canonical,inorganic solution grown crystal—and canavalin(Mr = 1.47 × 105g/mole), a typical storage protein.

Effect of Dislocations on KDPSurface Morphology andGrowth Rate

The results of AFM measurements on KDP101 surfaces5,6 show that over the range ofsupersaturations, 3%

<~ σ ≤ 30%, the terrace

widths on spiral growth hillocks formed bydislocations (Fig. 1) are nearly independent ofboth supersaturation and dislocation structure, incontradiction to the predictions of simple BCFmodels. The data also show that, for Burgersvectors in excess of one unit step height, thedislocations generate hollow cores in accordancewith theoretical predictions (Fig. 1). The mea-sured radii, r, of these cores is in good agreementwith predicted rF approximately given by14, 15

r

GF

F= b2

28π α. (1)

Here, G is the modulus of rigidity, b is the Burgersvector and αF is the free energy per unit stepheight of the step edge. Failure to take intoaccount the effect of these cores on the period ofspiral rotation as in the simple BCF models,1results in a prediction of hillock slope whichstrongly disagrees with our measurements (Fig. 2).Both analytical and numerical analyses have beenperformed to consider the effect of these cores.6These calculations show that, even in the case ofanisotropic step kinetics, a simple analyticalexpression can be used to describe the slope whichdepends primarily on the size of the Burgers vectorof the dislocation source and the radius of thehollow core relative to the critical radius of stepcurvature. However, because the size of the coreincreases with increasing Burgers vector [see Eq. (1)],when the core radius is ≥ the critical radius, theslope is nearly independent of Burgers vector.

This analysis predicts a dependence of slopeon supersaturation and Burgers vector which is ingood agreement with the experimental results, as



shown in Fig. 2. Because the core diameter isproportional to both G and α–1, the rate at whicha growth source generates steps is as stronglycontrolled by the elastic properties of the crystal(through the modulus of rigidity), as it is by thefree energy of the step edge, the fundamentalgrowth parameter of the BCF model.

The normal growth rate of a crystal face, R, isgiven by the slope times the tangential step speed,pv. A number of theoretical3 and experimental3,12

studies have shown that

v = ωβ(C – Ce) , (2)

where ω is the volume per molecule in the solid,β is the kinetic coefficient, and C – Ce is thedifference between the actual and equilibriumsolute concentrations. In most theoretical models,β is assumed to show Arrhenius behavior

β = exp(Ea/kT) , (3)

where Ea is the activation energy for step motion,k is Boltzman’s constant and T is the temperature.Our results show that above supersaturations ofabout 5%, growth hillock slope is nearly indepen-dent of supersaturation and dislocation structure.Consequently, for any growth experiment,R/p(C– Ce) should be a universal function whichis exponential in temperature with an exponentgiven by Ea. Figure 3 shows the dependence of

Figure 1. Hillocks formed by simple dislocationsources with Burgers vectors of: (a) one, (b) two,(c) three, and (4) four unit steps.

(a) 5µm (b) 3µm

(c) 3µm (d) 2µm

Figure 2. Measured hillock slope (data points) vssupersaturation along with the curves predictedby the simple BCF model for b⊥ = 1 (dashed lines)and by a model which includes the effect of thehollow cores (solid curves). r0 is radius of hollowcore used in the calculations.

b=3, ro=25

b=1

Shallow Medium Steep

b=2

b=3

b=1, ro=25

5 10Supersaturation (x100)

15 20 250

Hill

cock

Slo

pe

0.014

0.012

0.010

0.008

0.006

0.004

0.002

0.000

b=2, ro=257.5

10mm/day20mm/day30mm/day40mm/day

7.0

6.5

6.0

5.5

5.03.02.9 3.2 3.3 3.43.1

1000/T (1/K)

ln[R

/p(C

-Ce)

]

Figure 3. (a) Dependence of ln[R/p(C – Ce)] ontemperature for growth rates of 10, 20, 30, and40 mm/day.



ln[R/p(C – Ce)] on T for a number of differentmacroscopic growth rates. The data agree withthe prediction of Eqs. (2) and (3) and give anactivation energy for step motion of 0.33 eV.

Canavalin Kinetics

The results of our investigations of canavalincrystallization8,10,13 show that, depending on thesupersaturation, growth occurs on both simple andcomplex screw dislocation sources, two-dimensional(2-D) nucleating islands, or three-dimensional(3-D) nuclei that sediment onto the surface beforespreading laterally as step bunches (see Fig. 4).The step velocity at three different values of thepH (pH = 7.0, 7.7, and 8.0) was found to dependlinearly on canavalin concentration, C, as shownin Fig. 5, in accordance with Eq. (2). The kineticcoefficient, β, determined from these measurementsdepends strongly on pH with β ≅ 2.6 × 10–3 at pH7.3 to β ≅ 5.8 × 10–4 (cm/sec) at pH 8.0. This isapproximately three orders of magnitude less thanfor KDP. Insight into the processes determining βis obtained from the 3-D nuclei that deposit onthe surface. The velocity of single steps was 25 to100% greater than that of the step bunches of the3-D nuclei, demonstrating that the diffusion fieldsof adjacent steps overlap on very narrow terracesand providing a rough estimate for the lengthscale of diffusion. At moderate supersaturations,2-D nucleation was observed to take place on the

Figure 4. A series of 40 × 40 µm images collectedat 30 sec intervals, showing the growth of a 3-Dnucleus that has landed on the surface. Thenucleus grows out radially, forming macrostepsat the edge and a large plateau on top. Note thatthis nucleus merges flawlessly into the existingcrystal.

(a)

(b)

(c)

40x40 µm

Figure 5. A plot of step speed vs concentrationof canavalin as a function of pH. The kineticcoefficient, β, is determined from the slope ofthe line. Error bars are typically ±10%.

800

600

400

200

00 10 20 30 40

p H

7.3

7.7

8.0

Canavalin conc (mg/ml)

Step

spe

ed (

nm/s

)



Figure 6. (a) 44 × 44 µm image, showingdecoupling of steps from a dislocation source;(b) terrace width, W, vs time for step pairs, whichinitially emerge coupled from a dislocationsource; and (c) ln[(W – WF)/(W0 – WF)] for thestep pairs in (a) where WF and W0 are the finalasymptotic and initial terrace widths, respectively.

(a) 44 × 44 µm

7.2

pH

8.08.0

1.7x1016

7.3x1016

7.3x1016

C (cm-3)

0.5

1.0

1.5

0.00

(b)

100 200 300t (s)

Terr

ace

wid

th (

µm)

7.2

pH

8.08.0

1.7x1016

7.3x1016

7.3x1016

C (cm-3)

-3.5

-1.5

-2.5

-0.5

-4.50

(c)

100 200 300t (s)

ln[(

W-W

f)/(

Wo-

Wf)

]

large ( >~10 µm) plateaus generated on the tops of

these 3-D nuclei whereas it did not occur on thenarrower (1–10 µm) terraces generated by thescrew dislocations. This demonstrates that thelength scale for diffusion is of order 1 µm. Mea-surements of the step current combined with thisestimate for the diffusion length strongly indicatethat surface diffusion rather than bulk diffusion isthe controlling mechanism of solute transport tothe steps.

Further experimental evidence for surfacediffusion control of growth is provided by theprocess of step homogenization. We observedthat dislocation sources with components of theBurgers vector normal to the surface which aregreater than one lattice parameter produce steps

that initially emerge coupled and then homogenizewith time as shown in Fig. 6. This process of stephomogenization, as outlined by the Schwoebel–Ehrlich model,16 can result from surface diffusioncontrol of growth coupled with a barrier to down-ward transport at step edges. In accordance withthis model, the inhomogeneity in terrace widthdecays exponentially with time (see Fig. 5). Therelationship between step speed and terrace widthduring step homogenization was investigatedquantitatively, using a model of adsorption,diffusion, and incorporation.2 The best fit to thedata is obtained with a surface diffusion length of0.4–0.9 µm. The analysis also provides estimatesfor values of the activation energy for adsorption ofmolecules to the terrace, Ead, and for incorporation



3. P. Vekilov, Yu. G. Kuznetsov, andA. A. Chernov, J. Cryst. Growth 121, 643–655(1992).

4. A. J. Gratz, S. Manne, and P.K. Hansma,Science 251, 1343 (1991).

5. J. J. De Yoreo, T. A. Land, and B. J. Dair Phys.Rev. Lett. 73(6), 838–841 (1994).

6. J. J. De Yoreo, T. A. Land, L. N. Rashkovich,T. A. Onischenko, J. D. Lee, O. V. Monovskii,and N. P. Zaitseva, “Effect of DislocationCores on Growth Hillock Vicinality andNormal Growth Rates of KDP(101) Surfaces,”Lawrence Livermore National Laboratory,Livermore, CA, UCRL-JC-126010; submittedto J. Cryst. Growth (1996).

7. S. D. Durbin, W. E. Carlson, and W. E. Sarow,J. Phys. D. Appl. Phys. 26, B128 (1993).

8. T. A. Land, A. J. Malkin, Yu. G. Kutznesov,J. J. De Yoreo, and A. McPherson, Phys. Rev.Lett. 75, 2774 (1995).

9. A. J. Malkin, T. A. Land, Yu. G. Kutznesov,A. McPherson, and J. J. De Yoreo, Phys. Rev.Lett. 75, 2778 (1995).

10. T. A. Land, A. J. Malkin, Yu. G. Kuznetsov,A. McPherson, and J. J. De Yoreo, J. Cryst.Growth. 166, 291 (1996).

11. T. A. Malkin, Yu. G. Kutznesov, T. A. Land,J. J. De Yoreo, and A. McPherson, NatureStructural Biol. 2(11), 956–959 (1995).

12. A. J. Malkin, Yu. G. Kuznetsov, W. Glantz,and A. McPherson, J. Phys. Chem. 100, 11736(1996).

13. T. A. Land, J. J. De Yoreo, A. J. Malkin,Yu. G. Kutznesov, and A. McPherson, “An InSitu AFM Investigation of Canavalin Crystal-lization Kinetics,” Lawrence LivermoreNational Laboratory, Livermore, CA, UCRL-JC-126077; submitted to Surf. Sci. (1996).

14. F. C. Frank, Acta Cryst. 4, 497 (1951).15. N. Cabrera and M. M. Levine, Phil. Mag. 1,

450 (1956).16. R. L. Schwoebel and E. J. Shipsey, J. Appl.

Phys. 37(10), 3682 (1966).

at the step, Einc, of 0.27 and <0.1 eV, respectively.The results of this analysis can be compared withthat performed previously on (NH4)H2PO4 (ADP)based on interferometric measurements. Withinthis model, the kinetic coefficient expressed inunits which are independent of molecular size,β/a, can be written as a product of two terms, βadand βinc, which describe the kinetics of adsorptionand incorporation, respectively.

β β βa a

adinc= , (4a)

βad ada

DP

a=

2 , (4b)

β λλinc

inc

ah

aP

=

+

−

2

122

1

, (4c)

where D is the bulk diffusivity, h is the step height,a is the lattice spacing, λ is the surface diffusionlength, and Pad and Pinc are the sticking probabili-ties for adsorption and incorporation, respectively.Both probabilities are assumed to follow anArrhenius relationship with activation energiesEad and Einc, respectively. The ratios of thevalues of βad and βinc of ADP to those of canavalinare 0.9 and 3 × 104 respectively, the major differ-ence coming from the bulk diffusivity. Thiscomparison indicates that the slow kinetics ofcanavalin growth are mainly due to the slowadsorption rate of these large, immobile molecules.

References

1. W. K. Burton, N. Carbrera, and F. C. Frank,Royal Soc. London Philos. Trans. A243, 299–358(1951).

2. G. H. Gilmer, R. Ghez, and N. Cabrera, J. Cryst.Growth 8, 79–93 (1971).

ISOTOPE MEASUREMENTS FOR INNOVATIVE GROUNDWATER MANAGEMENT


ISOTOPE MEASUREMENTS FOR INNOVATIVE

GROUNDWATER MANAGEMENT

G. B. Hudson, M. L. Davisson

Water is easily identified as a fundamentalresource with issues abounding at the global,national, state, and local levels. In proclaimingWorld Water Day in March of 1996, the UnitedNations recognized that managing water resourceswill be a critical element in sustaining worldpeace. Therefore, the goal of this project is todevelop a set of water resource management toolsbased on isotope abundance measurements. Inthe past two years, we have focused primarily ongroundwater investigations, showing that ournew isotope tools can delineate groundwater flowpaths, determine groundwater ages and traveltimes, and estimate percentages of water fromvarious sources.

Concurrently, we have met with a variety ofwater management agencies and research organi-zations [e.g., Orange County Water District(OCWD) and the American Water Works Associa-tion Research Foundation (AWWARF)] to betterunderstand their needs and interests. Accordingto their studies, non-point source pollution appearsto be the most significant threat to water qualityand impacts the majority of drinking watersupplies. Among the contaminants from non-point sources are nutrients such as nitrogen andphosphorus, total dissolved solids, and toxic traceelements such as arsenic, cadmium, and mercury.A focus of many studies has been on dissolvedorganic carbon and the formation of halogenatedhydrocarbons during chlorination in municipaldrinking water systems. In addition to thesechemical contaminants, microbiological contami-nation is of concern where isotope abundances canbe used as tracers of non-point source pollutants.Where simple concentration measurements ofwater constituents may show a non-systematicvariation within a watershed, the isotope abun-dances are indicative of its source. For example,15N is more abundant relative to 14N in animalwastes than in agricultural fertilizer, providing a

distinguishing characteristic of its source. A simulta-neous measurement of 18O/16O can provide evenmore information on its source. Viruses and microbeshave limited lifetimes; therefore, groundwaterage-dating can help identify water sources at riskfor microbiological contamination.

We completed measurements of hydrogen/oxygen (H/O) isotopes and noble gases in theforebay area of the OCWD. These measurementsdemonstrate the applicability of isotope tracingand dating of the natural and artificial recharge ofSanta Anna River water into the exposed forebayaquifer unit. Temporal variation of hydrogen andoxygen isotopes provide a tracer for younggroundwater. For these studies, we match sourceterm with observed variation in groundwaterhydrogen and oxygen isotopes to provide a datingtechnique for young groundwater (1 yr) with ageresolution of a few months. As a result, we canidentify water production wells that receive alarge portion of their water in less than 6 monthsfrom time of recharge—six months is the proposedminimum retention time for recycled water. In asimilar fashion, we are able to use the relativeabundances of neon, argon, krypton, and xenon.Because of temperature dependent solubilities,noble gas abundances record the seasonal tem-perature variations which can be looked for ingroundwater samples. The 3H/3He dating systemhas shown large vertical variation in groundwaterage, demonstrating how critical three-dimensionalmodeling of the system is for accurate results. Ingeneral, water near the surface flows much fasterthan deeper water. Our work on age-dating hasbeen done in conjunction with microbiologicaltesting sponsored by the OCWD. The correlationsobserved between groundwater age and micro-biological activity has caught the attention of theUnited States Environmental Protection Agency,which is now beginning to evaluate whether age-dating might be a useful part of the proposedGroundwater Disinfection Rule.

ISOTOPE MEASUREMENTS FOR INNOVATIVE GROUNDWATER MANAGEMENT


We have developed techniques for isotopicanalysis of carbon in organic compounds usingcombustion to CO2. We have developed tech-niques for isotopic analysis of 15N/14N andsimultaneous 18O/16O in nitrates, a necessaryprerequisite for the proposed work on isotopesignatures for non-point source pollution. We arecurrently planning sampling of surface waters withthe City of St. Louis to study carbon and nitrogenisotope signatures in the Missouri River. We arealso looking to make similar measurements inTexas, working with the Texas River Authorityand in California in the aqueduct systems of thestate water project.

SELENIUM ISOTOPE GEOCHEMISTRY


SELENIUM ISOTOPE GEOCHEMISTRY: A NEW

APPROACH TO CHARACTERIZING THE

ENVIRONMENTAL CHEMISTRY OF SELENIUM

A. Volpe, B. Esser

High levels of selenium in the environmentwill be a prominent water quality issue in thewestern United States for many years. Seleniumaccumulation is linked to increased rates of deathand deformity in migratory birds, blind staggersin domestic livestock, and selenosis in humans. InCalifornia, agricultural drain waters, oil refineryeffluent, and semiconductor industry effluent contrib-ute to high selenium content in the San JoaquinValley and the San Francisco Bay. The importanceof these industries to California’s economy pre-cludes simple abatement, while the complexity ofselenium cycling precludes simple remediation.

The purpose of this project is to measurevariations in the isotopic composition of seleniumin water and soil samples caused by naturalprocesses and to show, for the first time, the valueof isotopic measurements in characterizingselenium pollution. The research seeks to identifysources of selenium pollution, determine processesin the selenium cycle, and support seleniumremediation studies.

During the first two years, we developedchemistry to extract and purify the various speciesof selenium found in waters and soils. We use ananion exchange column chemistry that allows usto pre-concentrate, separate, and determine lowlevels of different selenium species (selenite, selenate,and reduced selenium) in natural waters. Devel-opment of this technique involved the preparationof standards and enriched isotope solutions, which

will be useful in future isotope dilution and isotopiccomposition work. The ability to separate thenaturally occurring species of selenium is criticalto understanding selenium isotope fractionation.

In the second year, we completed conversionof an existing mass spectrometer to negativethermal ionization mass spectrometry (NTIMS).This project successfully provided a new analyti-cal capability for LLNL isotopic measurements.Using NTIMS, we measured selenium isotopiccomposition with high precision and accuracy in70 samples. These samples were selected to studyionization efficiency, instrumental isotopicfractionation, ion-beam stability for differentspecies of selenium measured as standard solu-tions, including standards processed throughchemistry. Precision and accuracy of the isotopemeasurement are excellent, better than 0.3% forthe minor isotope (74Se) and better than 0.05% forthe major isotope (80Se). Repeated measurementof standard solutions of selenite and selenateprocessed through chemistry demonstrates thatreduction steps in our chemical procedures do notresult in isotopic fractionation of selenium. Therefore,we are confident that our chemical procedures andmeasurement techniques are capable of determin-ing whether or not natural processes lead toselenium isotopic fractionation in the environment.We hypothesize that microbially mediated oxida-tion and reduction processes will affect both thespeciation and isotopic composition of seleniumin natural and contaminated systems.

STUDIES IN THE REGION OF ENHANCED NUCLEAR STABILITY


STUDIES IN THE REGION OF ENHANCED NUCLEAR

STABILITY AROUND N = 162 AND Z = 108J. F. Wild, R. W. Lougheed, K. J. Moody, N. J. Stoyer

In FY96, we carefully examined our raw datafrom the experiment to produce element 110 viathe bombardment of 244Pu with 34S ions from theU400 cyclotron at the Joint Institute for NuclearResearch, Dubna, Russia.1,2 We did this to assureourselves that (1) the one Z=110 event we discov-ered in our data was the only Z=110 event that wecould identify with a high degree of surety, and(2) that our claim to the discovery of element 110could not be denied due to a failure on our part toexplain some aspect of our results. With regard topoint (2), we also performed collateral bombard-ments at the Dubna U400 cyclotron to characterizethe response of the time-of-flight detectors at theend of the Dubna gas-filled mass separator tobetter understand an unusual signal we incurredin the discovery event. This signal is believed tohave arisen from the coincident detection of thealpha particle from the decay of one of the daugh-ter nuclei in the 110 chain and a conversionelectron from de-excitation of the subsequentlyproduced nucleus.

During the year we also prepared for the nextexperiment in this collaboration, which is thesearch for an isotope of element 114, produced bybombarding 244Pu with 48Ca ions. Element 114 isbelieved to be at the center of a region of nuclei(superheavy elements) which is very stable due tothe presence of both proton and neutron sphericalclosed shells. Characterization of the decay proper-ties of these nuclei would be of prime importanceto the theoretical understanding of the behaviorof nuclear matter. This experiment is probably anorder of magnitude more difficult than the search

for element 110 and would require a considerablymore efficient detection system than previousexperiments. Therefore, since the actual bombard-ment will probably not begin before mid-FY97,we contributed by purchasing a suite of surface-barrier detectors for use in the experiment. Ourprevious experiments demonstrated the capabilityfor producing high-intensity cyclotron beams tobombard targets capable of withstanding thisintensity over a long period of time. This capabilitycoupled with a very stable detection and dataacquisition system are all necessary to achievepositive results.

References

1. C&MS, “Enhanced Nuclear Stability,” Chemistry& Materials Science Progress Report, 95-ERD-040(1995) p. 74.

2. Yu. A. Lazarev, Yu. V. Lobanov,Yu. Ts. Oganessian, V. K. Utyonkov, et al.,Phys Rev C 54, 620 (1996).

Presentation

1. J. F. Wild, R. W. Lougheed, K. J. Moody,E. K. Hulet, Yu. A. Lazarev, Yu. V. Lobanov,et al., “Discovery of 267108 and 273110: ShellClosure at N = 162,” LLNL, UCRL-JC-120275(ABS), invited paper presented at thePacificchem ‘95 Conference, Honolulu, HI,Dec. 17–22, 1995.

Technical Inform

ation Departm

ent • Lawrence Liverm

ore National Laboratory

University of C

alifornia • Livermore, C

alifornia 94551

Documents

UNT Digital Library/67531/metadc693815/m2/1/high_re… · Lawrence Livermore National Laboratory This is an informal report intended primarily for internal or limited external distribution