Study of low energy high dose nitrogen implantation in aluminium, iron, copper and gold

SURFACE AND INTERFACE ANALYSISSurf. Interface Anal. 27, 678È690 (1999)

Study of Low Energy High Dose NitrogenImplantation in Aluminium, Iron, Copper and Gold

H. K. Sanghera* and J. L. SullivanSurface Science Research Group, Aston University, Birmingham B4 7ET, UK

To advance the understanding of fundamental physical and chemical processes occurring in ion bombardment ofmetals, ion beam nitridation of aluminium, iron, copper and gold is studied in the energy range 2–5 keV, at currentdensities of 1 and 5 lA cm—2, for each ion energy. The concentration proÐles of nitrogen-ion-implanted metalswere measured by x-ray photoelectron spectroscopy (XPS). The chemical composition and chemical structure ofthe implanted metals were also investigated. Copyright 1999 John Wiley & Sons, Ltd.(

KEYWORDS: ion implantation ; metal nitrides ; XPS; depth proÐling

INTRODUCTION

The implantation of oxygen and nitrogen has beenwidely used in metals to form wear-resistant coatings.At room temperatures, nitrogen implantation intometals can lead to the formation of metal nitrides,1which are hard, durable, refractory materials with awide number of applications including wear resistance,reÑective coatings and catalysts. All of the observedproperty changes in the target are due to compositional,microstructural or topographical changes ; in this work,mainly the compositional changes are studied.

In recent years there has been particular interest inthe use of this technique to form aluminium nitrideÐlms2h11 for applications as diverse as microelectronicand optoelectronic devices and wear- and corrosion-resistant layers. Aluminium nitride is a very interestingceramic because of its unusual combination of highthermal (up to 320 W m~1 K~1) and low electrical con-ductivity. Further to this, AlN is a hard material withhigh thermal and chemical stability (decompositiontemperature 2400 ¡C). It has a low thermal expansioncoefficient and is transparent in the visible and infraredregion. It is IIIÈV wide bandgap (6.2 eV) semiconductorthat crystallizes in a wurtzite hexagonal close-packedstructure. There are other processes for the formation ofAlN layers on aluminium substrates, e.g. reactive sput-tering deposition and chemical vapour deposition, butthe surface-deposited Ðlms formed by these techniquesare in many ways inferior to those produced by implan-tation.

High energy nitrogen implantation into iron leads tovarious nitride phases and is an e†ective method toimprove the mechanical and chemical properties of iron,such as microhardness, wear resistance and corrosionresistance of the metal surface. Not enough work hasbeen done on low-energy implantation for a full evalu-

* Correspondence to : K. Surface Science ResearchH. Sanghera,Group, Aston University, Birmingham B4 7ET, UK. E-mail :sanghehk=aston.ac.uk

ation of the beneÐts. A large number of papers havebeen published on nitrogen ion implantation iniron5,12h17 but most of the work has been done at highenergies ([40 keV) and/or at high temperature([200 ¡C). The use of low-energy high-dose conditionsis potentially more desirable economically but mayresult in signiÐcant e†ects that are negligible underhigh-energy low-dose nitrogen implantation conditions.These include sputtering of the target, back di†usionand loss of implanted nitrogen due to shallow proÐles.Published results of low-energy nitrogen ion bombard-ment at room temperature are rare.18,19

It is well known that nitrogen is insoluble in copperusing conventional techniques, but it forms metastablecompounds with copper by ion implantation ;20,21however, it has not been shown to form any bonds withgold.

It is well recognized that the kinetic conditions avail-able for structure transformation during implantationare very limited.22 Consequently, the phases formed byion irradiation cannot be predicted deÐnitely by theconventional principles of equilibrium thermodynamicsin all cases.

The aim of this paper is to investigate the e†ects oflow energy ion bombardment on metals. The sampleswere selected very carefully to gain an insight into thephysics and chemistry of the ion implantation processthrough a study of the e†ect of the mass and bindingenergy of the target and the chemical affinity betweenthe ion and the target. In terms of enthalpy of formationof nitrides ([318.0, [10.5, ]74.5 kJ mol~1 for AlN,

and respectively,23 no data for AuN), alu-Fe4N Cu3N,minium has a strong chemical interaction and nitrogen,iron has an intermediate chemical interaction, copperhas a small chemical interaction and gold has no chemi-cal interaction with nitrogen. One thus expects to beable to understand the chemistry of implantation bystudying this range of metals.

EXPERIMENTAL

For this work, aluminium bulk samples were D3 mmthick, cut from a 5 mm diameter rod of 99.99% purity

CCC 0142È2421/99/070678È13 $17.50 Received 16 November 1998; Revised 23 January 1999Copyright ( 1999 John Wiley & Sons, Ltd. Accepted 23 January 1999

LOW-ENERGY HIGH-DOSE NITROGEN IMPLANTATION IN METALS 679

Table 1. Tabulation of measured N/Al, N/Fe and N/Cu satu-ration ratio at various ion energies and current den-sities of implantation

Current N/Al N/Fe N/Cu

Energy density saturation saturation saturation

(keV) (lA cmÉ2) ratio ratio ratio

5 5 0.45 À0.07 0.10 À0.02 0.05 À0.02

4 0.47 À0.07 0.10 À0.02 0.05 À0.02

3 0.46 À0.07 0.09 À0.02 0.05 À0.02

2 0.49 À0.07 0.10 À0.02 0.05 À0.02

5 1 0.47 À0.07 0.12 À0.02 0.05 À0.02

4 0.54 À0.07 0.10 À0.02 0.05 À0.02

3 0.50 À0.07 0.08 À0.02 0.05 À0.02

2 0.43 À0.07 0.08 À0.02 0.05 À0.02

Al supplied by Goodfellow Cambridge Ltd. The ironsamples consisted of 0.125 mm thick iron foils of99.95% purity, supplied by Advert Research MaterialLtd. The copper samples used in this work consisted ofcopper bulk samples 3 mm thick and 8 mm in diameter,with 99.95% purity. These were supplied by GoodfellowCambridge Ltd. Gold samples of 0.025 mm thick foils,with 99.99% purity, were supplied by Advert ResearchMaterial Ltd. A standard copper nitride sample(Cu3N)and a standard iron nitride (mixture of sampleFe2,3,4N)were received from Alfa, Johnsons Matthey Plc., both inpowdered form.

The bulk samples of aluminium and copper were pol-ished with emery paper and then with successively Ðnergrades of alumina powder and then rinsed in distilledwater. However, the gold and iron samples were in theform of foils and hence polishing was not necessary. Inorder to remove hydrocarbon contamination, thesamples were rinsed in hexane and then cleaned ultra-sonically in hexane solution for D15 min.

When the samples (with the exception of gold) areexposed to air, native oxide layers are formed on thesample surfaces. To avoid surface charging, thesesamples were placed on a sample-mounting stub usingdouble-sided tape, with aluminium foil contactsbetween the sample and the stub. A CCD camera onthe spectrometer is aligned to the x-ray source using astandard sample for selecting the analysed area in theXPS experiments. A secondary electron detector, con-sisting of a scintillator and a photomultiplier, is used forcollecting secondary electrons in order to obtain aphysical image, which is displayed on a TV monitor.This assembly is very useful for aligning the ion beamand choosing the analysed area, hence ensuring that theanalysed area is the same as the bombarded area.

The implantation and XPS analysis were carried outin the same equipment without breaking the vacuum.The samples prepared by the procedure described abovewere analysed by XPS by taking a wide-scan spectrumto Ðnd out all the elements at the initial surfaces. Afterthe initial analysis, all the samples, with the exception ofgold, were found to be oxidized and these samples werecleaned using an Ar` beam to remove surface contami-nants. This was monitored by narrow-scan XPS mea-surements between successive periods of ionbombardment until the amount of contaminantsdetected was \5%. Owing to strong chemical inter-action between oxygen and aluminium, it was very diffi-cult to remove the oxide and the samples were generally

Ar` etched for [6 h to remove the oxide layer, whichseemed to be reformed continuously even at base pres-sures better that 7 ] 10~10. For the other samples, theAr` etch for D1È2 h was sufficient to remove surfacecontaminants to \5%. After cleaning, a wide-scan XPSspectrum of the Ar`-bombarded samples were taken.

The samples were then bombarded with an N2`beam of the required energy and current density and thesurface compositional changes were monitored bynarrow-scan (high energy resolution) XPS measure-ments. Eight sets of experiments were performed oneach sample, using a rastered beam of energy 2, 3,N2`4 and 5 keV with current densities of 1 and 5 lA cm~2for each energy. These current densities correspond to

Ñuxes of 1.25] 1013 and 6.25] 1013 atoms cm~2N2`s~1, respectively. In all experiments, the ions were inci-dent on the sample at an angle of 51.6¡. For a typicalexperiment, the spectra were collected after 0, 1, 3, 5, 10,15, 25, 35, 50, 65 and 80 min of ion bombardment. Onaverage, the implantation time was 80 min. This corre-sponds to maximum implanted doses of 6] 1016 and3 ] 1017 atoms cm~2 for current densities of 1 and 5lA cm~2, respectively. At the end of nitrogen ion bom-bardment, a wide-scan (low energy resolution) XPSspectrum of samples was taken.N2`-bombardedDuring these experiments, the base pressure was alwaysbetter than 7] 10~10 mbar and during ion bombard-ment experiments the working vacuum was alwaysbetter than 5 ] 10~7 mbar.

After each complete implantation, XPS depth proÐleswere measured using Ar` beam sputtering under 51.6¡at an energy of 2 keV and a current density of 2 lAcm~2 and narrow-scan XPS measurements were used tomonitor the surface compositional changes ; using thisinformation, depth proÐles were created. The total timeof Ar` depth proÐling was slightly di†erent in the dif-ferent experiments because the time taken to remove theimplanted nitrogen was found to be slightly di†erent fordi†erent bombardment conditions. RelativeN2`atomic concentrations in these proÐles were actuallymeasured as a function of time. The time axis is con-verted into a depth scale using the etch rate values(0.0055, 0.0043 and 0.0062 nm s~1 for nitrogen-ion-implanted aluminium, iron and copper, respectively)calculated from SUSPRE.24 It should be noted,however, that the absolute depth values will be di†erentfrom the depth values calculated in this manner,because the sputter yield is matrix dependent. On com-pletion of the experiment, a wide-scan XPS spectrum ofthe Ar` beam bombarded samples was taken.

X-ray photoelectron spectroscopy measurements

The XPS analysis was carried out using both Mg Ka(hl\ 1253.6 eV, FWHM\ 0.7 eV) and Al Ka(hl\ 1486.6 eV, FWHM\ 0.8 eV) x-rays (anodevoltage 15 kV, Ðlament current 20 mA). The choice ofthe x-ray source was dependent on the requirements ofthe experiment. The x-ray-analysed area on the samplesurface was D500 lm in diameter. The sample waspositioned at 0¡ electron take-o† angle relative to thesurface normal.

For the wide-scan spectrum, a pass energy of 50 eV, astep size of 0.2 eV and a dwell time of 50 ms were selec-

Copyright ( 1999 John Wiley & Sons, Ltd. Surf. Interface Anal. 27, 678È690 (1999)

680 H. K. SANGHERA AND J. L. SULLIVAN

Table 2. The values of Al 2p binding energies (eV) and KLL kinetic energies (eV)for clean, and then Ar‘-bombarded aluminium comparedN

2‘-bombarded

with pure aluminium and standard AlN

Al 2p Al KLL Auger parameter

Al metal 72.9 À0.1 1393.3 À0.1 1466.2 À0.2

N2

½-bombarded Al 73.8 À0.1 1389.4 À0.1 1463.2 À0.2

First N2

½ and then Ar½ bombardment 73.0 À0.1 1393.0 À0.1 1466.0 À0.2

Al26 72.9 1393.3 1466.2

AlN26 74.0 1388.9 1462.9

ted, whereas for the narrow-scan spectrum these param-eters were chosen to be 20 eV, 0.1 eV and 200 ms forphotoelectron peaks and 100 eV, 0.2 eV and 100 ms forAuger and valence band peaks. An aperture of 1000 lmwas selected.

Spectra collection and calibration

Aluminium. The XPS spectra of Al 2p, Al KLL, Alvalence, C 1s, N 1s and O 1s lines were collected aftersuccessive periods of bombardment during theN2`implantation, using Al Ka x-radiation. The Al Ka x-radiation was chosen in order to collect low kineticenergy Auger lines of aluminium. Binding energy mea-surements of all the pre-implanted samples werereferred to the C 1s line of the residual carbon set at284.6 eV. After relatively short periods of ion bombard-ment the adventitious carbon signals fell to levels thatrendered their use as calibrants unreliable. In this casesecondary standards were employed, notably the N 1ssignal, taken from very many measurements of photo-electron emission from the stoichiometric AlN standardto be 396.7 ^ 0.1 eV (FWHM\ 1.6 eV). For thenitrogen-implanted samples the energy of the well-deÐned Al 2p signal was 73.6^ 0.1 eV and the separa-tion between Al 2p and N 1s is 323.1^ 0.2 eV.

Iron and iron nitride. Using Mg Ka x-radiation, XPSspectra of Fe 2p, Fe LLM, Fe valence, C 1s, N 1s, NKLL and O 1s lines were collected after successiveperiods of bombardment. In this case, it was pos-N2`sible also to use the Al Ka x-ray source, but Mg Kax-radiation was the choice made. After Ar` cleaningmost of the carbon was removed, but the peak positionof C 1s from the small amount of carbon left in thesurface shifted from 284.6 eV to 283.4 eV for Fe 2p3@2positioned at 707.0 eV. This indicates the formation ofiron carbide. Because the Ar` bombardment changesthe carbon to iron carbide, the C 1s line could not beused for charge referencing. The binding energy mea-surements of all the ion-bombarded samples were refer-enced to the di†erential Fe valence band spectra set at 0eV. The XPS analysis of the standard iron nitridesample was performed for the sample as received andthen after being Ar` etched for half an hour. The spec-tral peaks collected for this sample were the same asabove, but the binding energy measurements were refer-enced to the C 1s peak set at 284.6 eV.

Copper and copper nitride. During implantation,N2`XPS spectra of Cu 2p, Cu LLM, Cu valence, C 1s, N 1s,N KLL and O 1s lines were collected after successive

periods of ion bombardment until steady-state satura-tion of nitrogen had been reached. These measurementswere made using Mg Ka x-radiation. Again, in this case,it was possible to use Al Ka x-radiation. Because theAr` bombardment changes the carbon concentrationsto a low value, the C 1s line could not be used forcharge referencing. The binding energy measurements ofall the ion-bombarded samples were referenced to thedi†erential Cu valence band spectra set at 0 eV. TheXPS analysis of the standard copper nitride sample wasperformed as received and then after being Ar` etchedfor half an hour. For this sample, the spectral peaks col-lected were the same above, but the binding energymeasurements were referenced to the C 1s peak set at284.6 eV.

Gold. After successive periods of bombardment,N2`XPS spectra of Au 4d, Au 4f, Au LLM, Au valence, C1s, N 1s, N KLL and O 1s lines were collected usingMg Ka x-radiation. However, the use of Al Ka-radiation could have been equally informative. The Au

peak positioned at 85.0 eV was used as a calibrant.4f7@2The binding energy measurements were also referencedto the di†erential Au valence band spectra set at 0 eV.

All the samples except aluminium were analysed withMg Ka x-radiation, and the aluminium samples wereanalysed with Al Ka x-radiation. As mentioned above,the use of Al Ka x-radiation for aluminium was tocollect low kinetic energy Auger lines, but the use of MgKa x-radiation for iron, copper and gold was merely bychoice. For a given type of source, say Au, only onetype of x-ray was used. Corresponding to the chosenx-ray sources, the escape depths of photoelectrons forAl 2p, Fe Cu and Au 4f peaks will be 2.5,2p3@2 , [email protected], 0.8 and 1.4 nm, respectively.25

RESULTS

Retained nitrogen

The changes in the relative atomic concentration oftarget material and nitrogen were calculated at suc-cessive intervals of bombardment. In the case of alu-minium, iron and copper, the development of therelative atomic concentration of nitrogen, at all the

Surf. Interface Anal. 27, 678È690 (1999) Copyright ( 1999 John Wiley & Sons, Ltd.


Figure 1. The XPS time profiles of nitrogen ion implantation at current densities of 1 and 5 lA cmÉ2 and ion energies of (a) 2 keV, (b) 3keV; (c) 4 keV and (d) 5 keV for aluminium and (e) 2 keV, (f) 3 keV, (g) 4 keV and (h) 5 keV for iron.

energies and current densities, shows that as the implan-tation proceeds the nitrogen concentration increasesand eventually reaches some saturation value after acertain implantation time. This saturation is achieved

after implantation of D1017 atoms cm~2. The satura-tion values at various energies and current densities areshown in Table 1. From this table, considering possibleexperimental uncertainties, there is no evidence of sig-



Figure 2. The XPS Ar½ depth and time profiles of nitrogen implantation at ion energies and 5 keV in aluminium at (a) l ¼1 lA cmÉ2 and(b) l ¼5 lA cmÉ2, in iron at (c) l ¼1 lA cmÉ2 and (d) l ¼5 lA cmÉ2 and in copper at (e) l ¼1 lA cmÉ2 and (f) l ¼5 lA cmÉ2.

niÐcant variation in N/Al, N/Fe or N/Cu saturationratios with energy, but perhaps the mean value of N/Alsaturation is a little higher at the lower current density.There may be such a trend but, considering the sta-tistical Ñuctuations, it is difficult to make any statementwith full conÐdence. In the case of gold, there washardly any nitrogen retained after implantation. Thesaturation ratios (nitrogen to metal) follow the trend N/Al[ N/Fe[ N/Cu and show little or no e†ect due tocurrent density.

To investigate the e†ect of ion dose in detail, N/Aland N/Fe relative atomic concentration ratios wereplotted as a function of ion dose for 2, 3, 4 and 5 keVions for implantation at 1 and 5 lA cm~2 (Fig. 1). No

inÑuence of current density on the concentration wasfound for the 2 keV experiment [Figs 1(a) and 1(e)] butfrom Figs 1(b), 1(c), 1(d), 1(f ), 1(g) and 1(h) for ionimplantation at ion energies of 3, 4 and 5 keV, thenitrogen concentration is higher for a given dose for theimplantation carried out at a current density of 1 lAcm~2 compared to that for 5 lA cm~2. Figure 1 alsoindicates that the nitrogen saturation value depends onthe current density, with the exception of 2 keV implan-tation. It was not possible to carry out such analysis inthe case of copper or gold because the amount ofimplanted nitrogen was small. Regarding the discrep-ancy between the results of Fig. 1 and Table 1, in Fig. 1the nitrogen concentration values are shown only up to



Figure 3. The XPS-excited spectra of (I) Al 2p, (II) Al Auger, (III) Fe 2p, (IV) Fe Auger, (V) Cu 2p, (VI) Cu Auger, (VII) Au 4d and (VIII)Au 4f : (a) before nitrogen ion bombardment ; (b) after nitrogen ion bombardment.

a dose of 6 ] 1016 atoms cm~2, whereas the implanta-tion was carried out for a dose of 3 ] 1016 atoms cm~2.The values listed in Table 1 correspond to a higher doseof implantation.

Nitrogen depth distribution

After implantation, Ar` bombardment was used to gaindepth proÐle information on the relative atomic concen-

tration of nitrogen and metal in the sample surfaces.These proÐles are shown in Fig. 2 for nitrogen ionimplantation densities of 1 and 5 lA cm~2, respectively.

The proÐles consist of a high concentration of nitro-gen from the surface to a few nanometres deep, followedby a region of much lower concentration of nitrogen.They then follow the expected trend and show that thenitrogen implantation is shallower at lower ion energies.From the proÐles it may be seen that it is not possibleto remove all the nitrogen atoms, even after a long



Table 3. Values of Fe binding energies (eV) and kinetic energies (eV) for2p3¿2

L3VV

as-received, Ar‘-cleaned, bombarded and then Ar‘-bombarded iron com-N2‘

pared with pure iron and a standard mixture of Fe2,3,4N

Fe 2p3@2 Fe L

3VV Auger parameter

Fe metal (as received) 709.0 À0.1 (3.3)a 704.1 À0.2 1413.1 À0.3

Fe metal (after Ar½ cleaning) 707.1 À0.1 (2.5) 703.2 À0.2 1410.3 À0.3

N2

½-bombarded iron 707.5 À0.1 (2.6) 702.9 À0.2 1410.4 À0.3

First N2


Fe26 706.95 702.4 1409.35

Mixture of Fe(2,3,4)N27 706.7 No data No data

a Values in parentheses are FWHMs of the peak.

period of Ar` bombardment. This is especially evidentfor nitrogen ion implantation at higher ion energies. Anexponential decrease of the nitrogen concentration withdepth, i.e. detectable nitrogen also at relatively large

depth, can be explained by Ar` beam mixing of nitro-gen with the matrix during sputter etching. This couldalso be due to recoil implantation or to di†usion ofweakly bound nitrogen via grain boundaries. Di†usion,

Table 4. Values of Cu binding energies (eV) and kinetic energies (eV) for as-2p3¿2

L3VV

received, Ar‘-cleaned, and then Ar‘-bombarded copper comparedN2‘-bombarded

with pure copper and standard Cu3N

Cu 2p3@2 Cu L

3VV Auger parameter

Cu metal (as received) 932.0 À0.1 (1.88)a 931.7 À0.2 1863.7 À0.3

Cu metal (after Ar½ cleaning) 932.3 À0.1 (1.96) 919.0 À0.2 1851.3 À0.3

N2

½-bombarded copper 932.4 À0.1 (1.97) 919.1 À0.2 1851.5 À0.3

First N2


Cu26 932.67 918.65 1851.32

Standard Cu3N 933.0 À0.1 (2.4) No auger lines –

a Values in parentheses are FWHMs of the peak.

Figure 4. Valence band XPS spectra of (I) aluminium (II) Iron (III) copper and (IV) gold : (a) before nitrogen ion bombardment ; (b) afternitrogen ion bombardment.



Figure 5. Deconvolution of various chemical states in Al 2p spectra after (a) 1 min and (b) 80 min of nitrogen ion bombardment at 4 keVand 1 lA cmÉ2 (uncalibrated spectra).

however, is unlikely unless surface heating is substan-tial, and thermocouple measurements of surface tem-peratures during bombardment indicate temperaturerises of no more than a few degrees. The unusual behav-iour of 5 keV curves in Figs 2(b) and 2(d) in comparisonto the more regular trend in the other cases could notbe explained. However, this behaviour was observedrepeatedly in several experimental measurements.

Compound formation by the implanted nitrogen

Figure 3 shows a comparison of 2p and Auger peaks ofaluminium, iron and copper, and 4d and 4f peaks ofgold, before and after nitrogen ion bombardment. Theelectronegativity values of metals and nitrogen (eAl\and eV)1.5, eFe \ 1.8, eCu \ 1.9, eAu \ 2.4 eN \ 3.0suggest that the nitrogen atoms (ion implanted) willattract valence electrons from the metals and thus the2p peak will shift to a higher binding energy. Becausethe Auger parameter is sensitive to the atomic environ-ment and independent of charge referencing, it is anexcellent method of compound identiÐcation. Thevalues of the binding energy of the photoelectron peak,the kinetic energy of the Auger line and the Auger

parameter (equal to the sum of the binding energy ofthe photoelectron peak and the kinetic energy of theAuger line) are shown in Tables 2, 3 and 4 for alu-minium, iron and copper, for pure target material, after

bombardment and after bombard-N2` N2`] Ar`ment. The values for the pure metals and standardnitrides are also reported. The valence band spectrum isvery closely related to the density of occupied states.Figure 4 shows the valence band x-ray photoelectronspectra of aluminium, iron, copper and gold before andafter nitrogen ion bombardment.

The shift in Al 2p peak position and change in Augerspectra conÐrms the bonding between aluminium andnitrogen. This was further conÐrmed by peak synthesisof the Al 2p curve (Fig. 5), which is described in detail ina later section. Table 4 also conÐrms the production ofAlN by nitrogen ion implantation and conversion fromAlN to Al on Ar` bombardment for D80 min. Figure4 (I) is also a clear indication of how nitrogen implanta-tion changes metallic aluminium to insulating AlN.

In the case of aluminium, the N 1sN2`-bombardedpeak [Fig. 6(a)] is symmetric and its position indicatesthe presence of nitrogen atoms in a strongly boundstate. The energy separation between Al 2p and N 1s is323.1^ 0.2 eV, which corresponds to that for AlN. The



Figure 5. Continued

synthesized Al 2p spectrum [Fig. 5(a)] show that not allof the aluminium binds to nitrogen. A small proportionis present at the outer surface as but the majorityAl2O3of the remaining aluminium appears to be in the metal-lic state. The is present due to oxidation of theAl2O3aluminium sample on exposure to the atmospherebefore transferring to the vacuum chamber, but thispeak was rarely completely removed even after prolong-ed periods of bombardment at base pressures of betterthat 7 ] 10~10. Comparison of Figs 5(a) and 5(b) showshow the amount of aluminium in the form of alu-minium nitride increases with an increase in nitrogenion bombardment time. For nitrogen ion implantationinto an aluminium substrate, one might expect stoichio-metric AlN to be formed with an N/Al ratio of unity.However, our experimental results indicate a measuredequilibrium concentration of The fact thatAlN0.7h0.8 .stoichiometric AlN is not formed within the alteredlayer near to the surface is not too surprising because adegree of reconstruction of surface atoms may occurand, during implantation, many vacancies, interstitialsand other defects are formed due to radiation damage.All these e†ects are likely to inÑuence the structure andstoichiometry of the near surface-modiÐed region,giving rise to a non-ideal phase. A further reason for thelower than unity N/Al ratios measured with XPS after

ion implantation may be because of the fact thatN2`

the nitrogen concentration is not constant over thesampling depth of the measurement technique, but itsmaximum is just beneath the oxide layer and then itdecreases rapidly. This concentration gradient coversmuch of the analysed depth (three times the inelasticmean free path or D9.0 nm). Thus the measured totalconcentrations are e†ectively the integrals of the nitro-gen concentration curves to a depth of D9.0 nmweighted by an photoelectron exponential decay term.

The peak position of Fe 2p [Fig. 3 (III)] remainsunchanged, although peak broadening occurs afternitrogen implantation. The XPS-excited Auger spectraand valence band XPS of iron similarly did not changeon nitrogen ion bombardment. From Table 3, it is diffi-cult to say whether nitrogen ion bombardment resultsin any chemical change in iron, because the values fornitrogen-bombarded iron are very close to that of thepure iron. This again, however, is to be expectedbecause the amount of implanted nitrogen is quitesmall, only 10%, and the FeÈN bond photoelectronsignal is swamped by the large metallic FeÈFe peak.Hence, even if the chemical interaction is taking place, itwill be difficult to observe it from these spectra and theAuger parameter values. All of these results suggest thatnitrogen bombardment may not have resulted in chemi-cal change of the metallic iron, although thermodyna-mic considerations show that this is a very unlikely



Figure 6. Photoelectron spectra of N 1s for nitrogen-ion-bombarded aluminium (a), iron (b), copper (c) and gold (d).

event. The enthalpy of formation of iron nitride is nega-tive ([10.5 kJ mol~1 at 25 ¡C23), thus favouringnitride formation. Broadening of the Fe 2p peaks,observed during nitrogen ion bombardment, indicatessome sub-surface modiÐcation, possibly lattice distor-tion or the introduction of lattice defects. But it mayalso be due to a small Fe nitride component peak athigher binding energy.

Hence, in the case of iron, the Fe 2p, the Auger andthe valence band XPS were not very conclusive indeciding whether nitride is formed or not. Curve synthe-sis of Fe is quite unreliable for low concentration/2p3@2high concentration comparisons due to the asymmetricpeak shape and large FWHM. However, the peak posi-tion of N 1s for nitrogen-bombarded samples(397.4^ 0.1 eV) [Fig. 6(b)] was found to be similar tothat for a mixture of (397.3 eV27,29) and thisFe2,3,4strongly suggests that nitrogen bombardment results inthe formation of nitride. Further, the binding energyseparation between the Fe and the N 1s peak was2p3@2found to be 309.6 ^ 0.2 eV, which is similar to that forthe standard iron nitride sample (309.4 eV27). Thisenergy separation together with the low binding energyof the N 1s peak suggests the formation of iron nitride.A complication here, however, is that the separationbetween the Fe and the N 1s peak is also similar2p3@2to that reported for nitrogen segregated on iron.29 Acombination of the two states might be possible. Fromthe implanted nitrogen concentration, in the ion energyrange 2È5 keV, it seems that iron nitride is in the form

with x \ 0.22. Rauschenbach et al.13 haveFe16N2~x,

reported that the phase transformation from toFe16N2is determined by the di†usion process, resulting inFe4N

nitrogen enrichment. In our experiments, consideringthe temperature rise, di†usion would be insigniÐcantand hence this transformation will not take place.Ideally then, bonding of Fe and N should result in 8 : 1stoichiometry, but the fact that the equilibrium valuewas 9 : 1, rather than 8 : 1, may be due to concentrationgradients within the depth integrated by the XPS tech-nique.

Like iron, the Cu 2p, the Auger and the valence bandXPS spectra of ion-bombarded copper did not showany change, but the broadening of the Cu 2p peak sug-gests some sub-surface modiÐcation. Table 4 shows thatas-received copper was oxidized, but no change inAuger parameter is observed on nitrogen ion bombard-ment, which is again expected because the amount ofimplanted nitrogen is very small (only 5%) and, likeiron, the CuÈN bond photoelectron signal is swampedby the large metallic CuÈCu peak.

Curve synthesis of Cu is difficult due to asym-2p3@2metric peak shape and a small amount of implantednitrogen. However, the N 1s spectrum of nitrogen for

copper [Fig. 6(c)] shows two peaks ofN2`-bombardednitrogen at peak positions 396.8 ^ 0.1 eV(FWHM\ 2.12 eV, D50% of the total nitrogen) and403.3^ 0.1 eV (FWHM\ 2.41 eV). The binding energygap between Cu and N 1s is found to [email protected]^ 0.2 and 529.2^ 0.2 eV for both peaks of nitro-gen. For standard copper nitride, the N 1s peak wasfound at 397.6 eV, and the energy gap between Cu 2p3@2and N 1s was found to be 535.4^ 0.2. Thus, analysis ofthe standard nitride sample suggests that the nitrogenpeak at 396.8^ 0.1 eV for copper isN2`-bombardeddue to copper nitride and the peak at 403.3 ^ 0.1 eV



could be attributed to weakly bound nitrogen, nitrogenbound to oxygen or molecular nitrogen adsorbed at thesurface or trapped within the outermost surface.

In the case of gold, Au 4d, Auger and valence bandXPS spectra of the ion-bombarded sample did not showany change. Even after long periods of bombard-N2`ment, nitrogen was very rarely observed and when itwas observed the amount was insigniÐcantly small. Thehighest available nitrogen is shown in Fig. 6(d). Becausethe nitrogen was present at only one or two levels andmissing at most of the levels, it seems that nitrogen isoccasionally present at interstitial sites.

DISCUSSION

During ion bombardment, the ions trapped in theimmediate subsurface region (D9 nm depth) change thesurface composition. Because this process of nitridationis associated with reaction between implanted nitrogenatoms and metal, the thickness of the metal nitride layerwill be of the same order as the penetration depth of theincident ions. The values of this quantity for projectedions (shown in Table 5), suggest that nitridation is con-Ðned to a few nanometres depth from the surface. Herethe penetration depth is the sum of the range and strag-gle. The ion bombardment is a very complex process,involving some processes that induce compositionalchanges, leading to mass correlation, and some pro-cesses based on other e†ects like chemical bonding, elec-tronic processes and di†usion. The surfaceconcentration of the implantant may depend on severalparameters, such as ion energy, ion Ñuence and targetproperties (the mass, binding energy and chemical struc-ture, etc.), sputter yield, ion range distribution andchemical reactivity of the ion/target combination.

From the experimental results, it seems that in therange 2È5 keV the e†ect of ion energy and currentdensity on the nitrogen saturation value is not impor-tant. However, for a given dose, the value of the nitro-gen concentration is higher for implantation performedat 1 lA cm~2 than at 5 lA cm~2, which is probablydue to thermal processes. This can be explained on thebasis of Gibbsian segregation and radiation-enhancedsegregation. Gibbsian segregation results in the segrega-tion of implanted nitrogen towards the sample surface,which is then removed by bombardment. For implanta-tion at 5 lA cm~2 the nitrogen segregates towards thesample surface more rapidly and hence is etched away,giving smaller initial values of relative atomic concen-

tration of nitrogen than for implantation at 1 lA cm~2.The fact that we do not see the e†ect of this segregationfor implantation carried out at 2 keV can be due to thelow sputter rate at that particular ion energy or the factthat the amount of nitrogen available for segregation issmaller. The latter is due to a smaller amount of thetotal implanted dose at lower ion energies.

Figure 7 shows the sputter yield values for varioustargets for all ion energies. Clearly, the trends in sputteryield and the amount of[(Y )Au [ (Y )Cu [ (Y )Fe[ (Y )Al]nitrogen retained (Table 1 ; (N)Al[ (N)Fe [ (N)Cu[

show that the higher the sputter yield, the smaller(N)Au)is the amount of nitrogen retained. However, this is notnecessarily true because there are more importantchemical factors present. Figure 7 also shows that in theenergy range 2È5 keV the sputter yield value increaseswith ion energy and hence the amount of nitrogenretained should be smaller at higher implantation ener-gies. However, the experimental data did not show sucha trend. This may be due to the higher ion range athigher ion energies (Fig. 8) or the fact that the totalimplanted dose is higher at higher implantation ener-gies.

The other aspect to look at is the mass and size di†er-ence between the nitrogen and target atoms, becausethis can cause the energy transfer processes, and hencesurface composition, to be signiÐcantly di†erent in animplanted substrate. All of these factors may a†ect

Figure 7. Variation of sputter yield with ion beam energies fornitrogen-ion-bombardment of aluminium, copper, iron and goldusing SUSPRE.24

Table 5. Longitudinal ion range and straggle values (innm) calculated from SUSPRE24

Ion beam energy

(keV) Target Al Fe Cu Au

2 Range 3.3 2.1 2.0 2.8

Straggle 2.0 1.2 1.2 1.6

3 Range 4.3 2.7 2.8 3.6

Straggle 2.6 1.6 1.7 2.0

4 Range 5.3 3.2 3.3 4.2

Straggle 3.2 1.9 2.0 2.4

5 Range 6.3 3.8 4.0 4.8

Straggle 3.7 2.3 2.4 2.8



Figure 8. Variation of ion range with ion beam energies for nitro-gen ion bombardment of aluminium, copper, iron and gold, usingSUSPRE.24

nitrogen retention at interstitial sites. The target massand density act by their inÑuence on the sputter yieldand projected range of the ions. This has been discussedin the previous paragraph. Because the radius of nitro-gen atoms is very small compared to the radii of thetarget atoms (rAl\ 143, rFe \ 126, rCu \ 128, rAu \ 144and pm), there is a good possibility of nitrogenrN \ 70trapping at interstitial sites. Looking at the radii, thoseof aluminium and gold atoms are larger than those ofiron and copper. The magnitudes of the radii indicate ahigher possibility of nitrogen retention in the case of Aland Au compared to Fe and Cu. However, the mass,density, radius and ion range are not the only param-eters. For example, in the case of Fe and Cu, they havealmost the same density, radius and ion range but thenitrogen saturation value is higher in the case of iron.This could be due partly to the higher chemical affinityof Fe towards N compared to that for a Cu target.

It is clear that the target mass, density, sputter yieldand ion range distribution may a†ect nitrogen retentionat interstitial sites, however the XPS results indicate theformation of nitride. Hence, by far the predominantfactor is the ionÈtarget chemical combination. Theretention characteristics of implantants at high doseseem to be strongly dependent on the chemical reacti-vity between ion species and target material. The elec-tronegativity values show that the chemical interactionof metals towards nitrogen is in the order :Al[ Fe[ Cu[ Au. Here, the interaction is measuredas : the higher the electronegativity di†erence betweenthe metal and nitrogen, the greater the chemical inter-action. Thus, based on the chemical reactivity of targetwith nitrogen, we expect the nitrogen saturation valueto follow the trend (Nsaturation)Al[ (Nsaturation)Fe [

and this is proved by our(Nsaturation)Cu [ (Nsaturation)Au ,experimental results. The saturated relative atomic con-centration ratios of nitrogen and target atoms,

are plotted as a function of electronega-(N/T )saturation ,tivity di†erence (the di†erence between the electronega-tivity values of nitrogen and target atoms) in Fig. 9. Thegraph shows a parabolic relationship between the ion/target saturation atomic concentration ratio and thedi†erence in electronegativity between nitrogen andtarget atoms.

Figure 9. Saturated relative atomic concentration ratios of nitro-gen and target atoms (N /T) as a function of electronegativity dif-ference between the electronegativity values of nitrogen and targetatoms.

Because the injected ions stop in the surface layer andare deposited into the matrix lattice in the form of pre-cipitates in D10~10 s,28 the kinetic conditions forstructural transformation are very limited22 and themechanism of nitride formation by direct implantationis not the same as in conventional treatment. Formationof nitrides is favoured thermodynamically andkinetically if the heat of formation of nitride is negative.The more negative the heat of formation for a nitride,the greater is the driving force for its nucleation andgrowth, and the easier is the formation of nitride. FromTable 1, the formation of aluminium nitride and ironnitride is favoured but the formation of copper nitride isnot. Because the value of the heat of formation is themost negative for aluminium compared to the othermetals chosen for this work, it will be the easiest toform. However, in the case of iron, data were not avail-able for and hence the value is reported forFe16N2The formation of copper nitride shows that,Fe4N.under the conditions of ion implantation, the formationof nitrides that are thermodynamically and kineticallyunfavourable is also possible. The failure to detect anygold nitride under implantation conditions suggests thatthe behaviour of gold upon irradiation is not the sameas that of copper, despite the fact that they are in thesame group in the Periodic Table. This is probably dueto the chemical inertness of gold, but one would stillexpect some retention of nitrogen, which was absent inmost of the experiments performed on gold. Becausenitrogen does not form any compounds with gold, themobility of a single nitrogen atom in a defect-laden goldlattice may result in the segregation of any implantednitrogen and subsequent sputtering due to the smallermass. Liu et al.21 have also reported that gold nitridesare not formed even with ions of energy of several tensto 100 keV, probably due to the electronic structure ofgold. Prabhawalkar et al.20 have reported that N2`bombardment of the AuÈCu system at 150 keV and anion dose of 2 ] 1016 ions cm~2 results in the formationof metastable bonds with copper and the nitrogenremains only as an interstitial in the gold matrix. Thereason for some retention of nitrogen in the bulk athigher ion energies could be due to the greater range at



higher energies compared to the range at lower ionenergies.

The above analysis indicates that bombardment of atarget with low-energy high-dose nitrogen ions resultsin retention of part of the implanted nitrogen, and theamount of retained nitrogen seems to have a relation-ship with the reactivity of the target atoms to nitrogen.Because the heat of formation is a good indication ofthis chemical reactivity,29 retention of nitrogen and theformation of nitride seem to be a†ected strongly by thereactivity of the metal towards nitrogen. Even for lowdoses of implanted nitrogen, nitride is formed, so theaverage concentration of implanted nitrogen does notseen to be a critical factor in determining the formationof nitrides.

CONCLUSION

In the case of aluminium, iron and copper, as implanta-tion proceeds the nitrogen concentration increases andeventually reaches a saturation value, following thetrend : (N/Al) [ (N/Fe)[ (N/Cu). However, in the caseof gold, hardly any nitrogen was retained. Consideringpossible experimental uncertainties, there is no evidenceof signiÐcant variation in nitrogen saturation concentra-

tion with ion energy. The experimental data indicate ahigher value of nitrogen saturation for implantationperformed at lower ion current density ; however, con-sidering the statistical Ñuctuations in the data, it is hardto say if the trend is real. At a given dose, the nitrogenconcentration is higher for implantation carried out at acurrent density of 1 lA cm~2 compared to 5 lA cm~2,which suggests the presence of radiation-enhanced seg-regation and Gibbsian segregation. The argon proÐlesconsist of a high concentration of nitrogen from thesurface to a few nanometres deep, followed by a regionof much lower concentration of nitrogen.

In the case of aluminium, iron and copper there is anindication of the formation of nitrides but stoichiomet-ric nitrides are not observed, which could be due toconcentration gradients within the depth integrated bythe XPS, or due to the inÑuence of reconstruction ofsurface atoms or defect formation during implantation,resulting in a non-ideal phase.

The nitrogen saturation concentration value increaseswith an increase in the di†erence of electronegativityvalues between the target and nitrogen atoms. Theamount of retained nitrogen increases with a decrease inheat of formation of the nitride, showing that retentionof nitrogen is a†ected strongly by the chemical reacti-vity of metal to nitrogen.

REFERENCES

1. R. Kelly, J . Vac. Sci . Technol . 21, 778 (1982).2. S. Ohira and M. Iwaki,Mater . Sci . Eng.A 116, 153 (1989).3. N. Lieske and R. Hezel, J . Appl . Phys. 52, 5806 (1981).4. P. M. Raole, P. D. Prabhawalker, D. C. Kothari, P. S. Pawar

and S. V. Gogawale, Nucl . Instrum. Methods Phys. Res. B 23,329 (1987).

5. J. A. Taylor and J. W. Rabalais, J . Chem. Phys. 75, 1735(1981).

6. B. Rauschenbach and H. Somer, Thin Solid Films 170, 81(1989).

7. J. L. Sullivan, Z. Wronski, S. O. Saied and J. Sielanko,Vaccum 46, 1333 (1995).

8. B. Rauschenbach, K. Breuer and G. Leonhardt, Nucl . Instrum.Methods and Phys.Res.B 47, 396 (1990).

9. L. Carlo, A. Perez-Rodriguez, A. Romano-Rodriguez, J. R.Morante and J. Montserrat, Nucl . Instrum Methods Phys. Res.B 84, 214 (1994).

10. G. Terwagne, S. Lucas and F. Bodart, Nucl Instrum. MethodsPhys.Res.B 59/60, 93 (1991).

11. S. Simson, T. Reier, J. W. Schultz and C. Buchal, Surf . Coat .Tech. 83, 49 (1998).

12. T. Fujihana, A. Sekiguchi, Y. Okabe, K. Takahashi and M.Iwaki, Surf . Coat . Tech. 51, 19 (1992).

13. B. Rauschenbach, A. Kolitsch and K. Hohmuth, Phys. Status .Solidi 80, 471 (1983).

14. S. Ohtani, M. Watanabe and N. Iwamoto, J. Jpn. Inst . Met .59, 851 (1995).

15. Y. Fukui, O. Nishimura, K. Yabe and M. Iwaki, Nucl . Instrum.Methods Phys.Res.B 59/60, 812 (1981).

16. B. Rauschenbach, Nucl . Instrum. Methods, Phys. Res. B80/81, 303 (1993).

17. A. V. Belyi, V. A. Kukareko, O. V. Lobodaeva and S. K. Shikh,Phys.Met .Metall . 80, 657 (1995).

18. J. Jagielski, N. Moncoffre, G. Marest, L. Thome� A. J. Barcz, G.Gawlik and W. Rosinski, J . Appl . Phys. 75, 153 (1994).

19. H. K. Pu and J. W. Rabalais, J . Chem.Phys. 85, 2459 (1981).20. P. D. Prabhawalkar, D. S. Kothari, M. R. Nair and P. M. Raole,

Nucl . Instrum.Methods B 7/8, 147 (1985).21. B. X. Liu, X. Zhou and H. D. Li, Phys. Status Solidi A 113, 11

(1989).22. E. Ma, B. X. Liu, Chen X. Chen and H. D. Li, Thin Solid Films

147, 49 (1987).23. R. D. Lide (ed), CRC Handbook of Chemistry and Physics ,

72nd Edn. CRC Press Boca Raton, FL (1991).24. SUSPRE, Surrey University Sputter Profile Resolution from

Energy deposition program V (1.4) Surrey University (1987).25. S. Tanuma, C. J. Powell and D. R. Penn, Surf . Interface And.

17, 911 (1991).26. C. D. Wagner, in Practical Surface Analysis , 2nd Edn, Vol. 1,

edited by D. Briggs and M. P. Seah, Appendix S. Wiley, Chi-chester (1992).

27. B. M. Biwer and S. L. Bernasek, J. Electron Spectrosc . Relat .Phenom. 40, 339 (1986).

28. W. L. Johnson, Y. T. Cheng, M. Van Rossum and M. A.Nicolet,Nucl . Instrum.Methods B 7/8, 147 (1985).

29. K. Yabe, O. Nishimura, T. Fujihana and M. Iwaki, Surf . Coat .Tech. 66, 250 (1994).


Documents

Study of low energy high dose nitrogen implantation in aluminium, iron, copper and gold