1

Structural transformations induced by swift heavy ions

in polysiloxanes and polycarbosilanes

Dedicated to Prof. G. Petzow on the occasion of his 75th birthday

Jean-Claude Pivin 1), Eckhard Pippel 2), Jörg Woltersdorf 2), Devesh Kumar Avasthi 3)

and Sanjev Kumar 3)

1) Centre de Spectrométrie Nucléaire et de Spectrométrie de Masse, Orsay Campus,

France

2) Max-Planck-Institut für Mikrostrukturphysik, Halle, Germany

3) Nuclear Science Center, New Delhi, India

Abstract

High resolution electron microscopy in connection with electron energy filtered

microscopy is used for evidencing the precipitation of free C clusters in polysiloxanes

and polycarbosilanes, responsible for the hardening and luminescence of these classes

of inorganic polymers after ion irradiation. While during irradiation with 3 MeV Au

ions randomly distributed carbon clusters are formed, in the case of irradiation with 100

MeV Au ions, the carbon clusters are aligned along tubular ion tracks without forming

continuous wires. Contrary to observations in some other polymers and semiconductors

no tubular voids or crystallization of amorphous SiC were found. It is concluded that in

both cases the carbon precipitation is due to a solid state transformation induced by

electronic excitations.

2

Strukturelle Transformationen in Polysiloxanen und Polycarbo-

silanen durch Bestrahlung mit schnellen Schwerionen

Kurzfassung

Hochauflösende und energiegefilterte Elektronenmikroskopie wird zum Nachweis der

Bildung freier Kohlenstoff-Cluster in Polysiloxanen und Polycarbosilanen benutzt, die

verantwortlich für die große Härte und die Lumineszenzeigenschaften dieser

anorganischen Polymere nach Ionenbestrahlung sind.

Während bei Bestrahlung mit 3-MeV-Goldionen zufällig verteilte Kohlenstoff-Cluster

auftreten, ordnen sich diese im Falle der Bestrahlung mit Goldionen von 100 MeV

entlang röhrenförmiger Ionenbahnen an, ohne jedoch zusammenhängende "Nanodrähte"

zu bilden. Im Gegensatz zu Beobachtungen an einigen anderen Polymeren und

Halbleitern wurden keine Porenschläuche oder eine Kristallisation von amorphem SiC

gefunden. Aus den Ergebnissen kann geschlossen werden, daß in beiden Fällen die

Kohlenstoff-Ausscheidungen das Ergebnis einer durch elektronische Anregung

induzierten strukturellen Transformation sind.

3

1. Introduction

Inorganic polymers are of great interest in materials science as their thermal conversion

into ceramics permits to fabricate bulk pieces, fibers and films of a large variety of hard

and refractory materials (for instance SiC, SiCxOy glasses, Si-C-B-N materials, ZrO2,

TiO2). These different ceramic products can be formed via the polymer pyrolysis route,

requiring treatments at temperatures around 1000°C which are lower than the

temperatures of the conventional sintering process. On the other hand, it has been

shown recently that ion irradiation at room temperature can enable the transformation

of films of polysiloxanes and polycarbosilanes into amorphous ceramics without

unwanted loss of carbon and silicon as well as without increase of the oxygen content,

nor interdiffusion with their substrate and cracking due to the thermal expansion

mismatch between film and substrate [1-4]. Furthermore, the obtained ceramics are

harder than the pyrolyzed materials, and polymer films which are partially converted

into ceramic films exhibit a strong green photoluminescence (PL). According to spectra

of Raman scattering, this emission and the hardening can be ascribed to the segregation

of diamond like carbon in the amorphized structure when the amount of energy

transferred by ions to target atoms exceeds the value of Ei ~10 eV [1]. It is worth to note

that the yield of transformation depends essentially on the amount of electronic

excitations; displacements scarcely contribute to the conversion, as evidenced by

experiments performed with different ion species. The luminescence yield goes through

a maximum for Ei ~ 20 to 40 eV according to the polymer composition [2, 4]; then it

decreases and vanishes in the case of polymer precursors with high C contents.

Therefore the variation of the PL yield was interpreted as due to the growth of C

4

clusters and/or to their percolation in matrices with stoichiometries Si C1...6 O0...1.5,

containing up to 100% C in excess with respect to the maximum number of Si-C bonds

in SiCx or SiCxOy glasses (x = 1 or 2-y, respectively). However, this hypothesis of

carbon clustering could not be proven on the basis of Raman scattering analyses nor of

grazing incidence X-ray diffraction or conventional transmission electron microscopy

(TEM). The yield of Raman scattering levels off at high ion fluence (Ei > 40 eV) [1]

because of the increase of light absorption. On the other hand, clusters of a low Z

element exhibit no contrast in a heavier matrix for X-ray beams.

According to our knowledge only two techniques can be used for observing directly this

type of clusters, which are the near field microscopy of the PL and the combination of

high resolution electron microscopy (HREM) with electron energy loss spectroscopy

(EELS) at a high spatial resolution in the nanometre range. The latter technique was

used here, and two preliminary results are presented.

The first result concerns the observation of C clusters in a polysiloxane after irradiation

with swift heavy ions (energy >1MeV/nucleon). Incremented fluences of 3 MeV Au ions

were used. (The properties of such polysiloxanes after irradiation with incremented

fluences of 500 keV C and 3 MeV Au ions have been studied in previous publications

[1-4]).

On the other hand, irradiations with ions of energies of ~100 MeV have been

undertaken very recently on polycarbosilanes and polysiloxanes. Still little is known of

structural transformations in organic polymers under such high irradiation energies, and

nothing in the case of inorganic ones. Irradiation with swift heavy ions of energies in the

range of 100 MeV to a few GeV promotes essentially electronic excitations along

tubular tracks during most part of their travel in any type of target. The diameter of

5

these tracks is typically from a few nm to 10 nm in organic polymers [5-7], and the

excitations within the tracks have similar effects as those induced by less energetic ions

(apart for their spatial distribution). But in the case of semiconductors or metals

structural changes are observed when the linear density of deposited energy exceeds a

threshold value, similar to those produced by the shock wave of a laser beam. Atomic

rearrangements during this energy spike lead to the formation of dislocations in

crystalline semiconductors but not to their amorphization while amorphous SiC can be

crystallized [8]. Therefore, the formation of crystalline particles of SiC or of turbostratic

graphite could be expected in the materials derived from polycarbosilanes.

HREM and electron energy filtered microscopy (ETFEM) observations of tracks in

films of polycarbosilanes and polysiloxanes irradiated with such high energies are

interesting in two respects, being the determination of the size and the crystallographic

structure of the tracks as well as the distribution of carbon in these tracks. The second

result presented in this paper concerns such investigations, carried out on

polycarbosilanes irradiated with 100 MeV Au ions.

2. Experimental

An ethanolic solution of methyltriethoxysilane CH3Si(OC2H5)3, (MTES, Aldrich

Chimica, Milan) was hydrolyzed by addition of H2O+HCl for obtaining a polysiloxane

constituted of -[(CH3)(OH)-Si-O]- chains. Allylhydridopolycarbosilane [-(H)2Si-CH2-]n

(HPCS, Starfire Systems Inc., Watervliet, N.Y.) and polycarbosilane [-(CH3)(H)Si-

CH2-(CH3)2Si-CH2-]n (PCS, Dow Corning X9-6348), were dissolved in HPLC hexane

6

for studying films with 2 different C/Si content ratios. The films of MTES, HPCS and

PCS were spun at 1500 to 3000 r.p.m. on clean <100> Si wafers.

Irradiations were performed at room temperature. The ARAMIS accelerator of CSNSM

was used for those with incremented fluences Φ of 3 MeV Au ions, and the 15 MV

Pelletron accelerator of Nuclear Science Centre, New Delhi for the irradiation with 100

MeV Au7+ ions at a fluence of 1013 ions/cm2. Calculations using the TRIM (transport of

ions in matter) code [9] indicate that 3MeV Au ions are implanted in the substrate for

films with a maximum thickness of 1 µm used in present experiments. Values of the

linear density of energy transferred to electronic shells, Se, and to nuclei, Sn, by 3 MeV

Au ions are for instance in PCS: Se = 300 eV / (atom × 1015 ions / cm2), including the

energy loosed by recoils, and Sn = 350 eV / (atom × 1015 ions / cm2). They differ by less

than 20% in other polymers but increase by about 50% during the irradiation due to the

change of stoichiometry and atomic density. The linear density of energy deposited by

100 MeV Au ions in electronic excitations is 3 times as high ( 1000 eV / (atom × 1015

ions / cm2), and these ions have insignificant elastic collisions with target atoms as

revealed by the estimated low value of Sn. Note that TRIM does not permit to estimate

the lateral distribution of electronic excitations around the ion path.

Changes in the chemical structure of the coatings were assessed by Raman

spectroscopy, using a DILOR X-Y micro-spectrometer fitted with a multi channel CCD

detector. The excitation source was the 514.5 nm line of a Spectra Physics 2017 argon-

ion laser, operated at 2mW, and the beam was scanned over 75 µm, in order to avoid

heating of the films. For the PL experiments a double SPEX 1403 monochromator and

an EMI 9863B photomultiplier were used. The PL was excited with the 488 nm line of

an argon-ion laser with power density below 1 W/cm2 on the sample.

7

For electron microscope investigations, thin specimens were prepared by the standard

cross-section techniques, i.e., gluing the coated Si-wafers face to face, cutting thin (<

200 µm) slices, dimple-grinding to about 10 µm, and final Ar-ion milling (Gatan Duo-

Mill) down to electron transparency. This enabled the high resolution and energy

filtered imaging of specimens of only a few nanometres up to ~50 nm in thickness and a

tolerable surface roughness. Microstructure and nanochemical investigations were

carried out using the high resolution Philips CM 20 FEG field emission electron

microscope, run at 200 kV and equipped with a Gatan Imaging Filter (GIF 200),

mounted below the microscope column. Besides electron energy loss spectroscopy at an

energy resolution of 0.8 to 1 eV, this filter enables the imaging with inelastically

scattered electrons of a certain energy range. The combination with an appropriate

computer equipment allows the sensitive mapping of a specific element (energy filtered

electron microscopy, EFTEM) with a high spatial resolution (<1nm) within a few

seconds so that a structure/chemistry relation is easy to obtain. The filtered images were

digitally recorded by a slow-scan CCD camera within the GIF. Further details

concerning the proper operation conditions of the GIF equipment are given in [10, 11].

3. Results and discussion

3.1. MTES, irradiated with 3 MeV Au ions

Raman spectra of the specimens studied are displayed in Fig. 1. All the Stokes Raman

peaks of C in these films are centered at 1510 ± 20 cm-1 like those of amorphous C

coatings with a noticeable degree of sp3 hybridization. Whatever the ion species used

8

for the irradiation and the fluence, they are never splitted in 2 components at 1350 and

1580 cm-1 like in spectra of evaporated C, pyrolyzed polymers and other types of

turbostratic graphite. The Raman spectrum of MTES irradiated with 3 MeV Au ions at a

fluence of 5 ×1013 ions / cm2 is a weak bump on a continuous background of

luminescence, which has been subtracted in the figure. The intensity of the

photoluminescence peak at 2.1 eV is close to its maximum, reached for a fluence of 1014

ions / cm2. The PL emission has vanished for 3 MeV Au fluences as high as 2.5 × 1015

ions / cm2, so that the Raman peak of the C phase is clearly resolved before subtracting

a background while it is only 3 times more intense than for the fluence of 5 ×1013 ions /

cm2.

Two-dimensional distributions of the elements carbon (using the C-K edge at 284 eV)

and silicon (using the Si-L23 edge at 99 eV) are shown on the EFTEM images of Figs.

2a and 2b for MTES irradiated with 3 MeV Au ions at these two fluences of 5 x 1013

ions/cm2 and 2.5 x 1015 ions/cm2.. The images clearly evidence that part of the C atoms

has segregated into clusters for both irradiations and that the degree of segregation

increases with the fluence. The carbon clusters seem to have larger diameters in the film

irradiated with 5x1013 Au ions because of the unresolved contribution of C atoms

bonded to Si. This fact is proven by observing the Si-rich areas appearing more

overlapped by areas containing C in Fig. 2a. According to HREM observations (cf. Fig.

3) the films exhibit a completely amorphous structure, in particular, without local

ordering of graphitic carbon or crystalline SiC . The real cluster sizes should be nearly

identical in both specimens. This size of ~5 nm is consistent with the idea that the PL

emission of these films is due to the confinement of electrons in quantum dots of a

semiconductor. Note that the variation of the energy of this emission with the size of the

9

clusters has been established only in the case of Si clusters, preferred by theoreticians

and easier to fabricate in a controlled manner: It vanishes for clusters smaller than 2 nm

or bigger than 6 nm [12, 13]. The fact remains that C clusters do not appear to percolate

or grow during the polymer to ceramic conversion of the MTES, and one needs to find

an other explanation for the quenching of the PL emission. Two further experimental

results may be invoked to understand this behavior: a) the electronic gap of the matrix

decreases (without closing), indicating that defect related states constituting non-

radiative recombination centers multiply; b) first the matrix, and then the C clusters

become depleted in hydrogen. The remaining 7 at% H in the completely converted

material [1, 4] are probably dissolved within clusters, taking into account the strength of

the C-H bonds and the comparison of radiolytic yields in polysiloxanes with various C

concentrations. However, larger amounts of H could be necessary for keeping to the

clusters a semiconducting character [14]. It was not possible to obtain an unambiguous

hint to the hybridization of C atoms in the clusters from the analysis of C near edge

structure in EEL spectra because of the superimposed contribution of C atoms bonded

to Si.

3.2. HPCS/PCS, irradiated with 100 MeV Au ions

The PL emission of PCS and HPCS films irradiated with 100 MeV Au ions at a fluence

of 1013 ions/cm2 is comparable to that of MTES films irradiated with 3 MeV Au ions at

a fluence of 3x1013 ions/cm2, in agreement with the ratio of electronic energy losses. It

would probably still increase with the ion fluence as also other physical quantities

10

characteristic of the polymer to ceramics conversion (hardness, yield of H radiolysis,

compaction) which were found also relatively low for the films irradiated with 1013

ions/cm2 Au of 100 MeV. This behaviour should be caused by the fact that the C atoms

are aligned along channels with a width of ~5 nm in the HPCS film and ~10 nm in PCS.

The segregation can be clearly seen on the element specific EFTEM images of Figs. 4b

and 5b for PCS and HPCS, respectively. Figs. 4a and 5a show the corresponding TEM

bright field images with a lower absorption contrast in the carbon rich areas and, again,

a totally amorphous structure of the film. Note also that these channels are not

completely decorated with C atoms and that between them the matrix appears almost

depleted in C. One can suppose that the C-enriched channels represent sections of the

tubular tracks typical of ion irradiation at such energies, or on the contrary the parallel

channels containing more Si. Indeed, the tracks appear very close one from the other, in

part because of the projection to the image plane of tracks located within a certain

sample thickness (about 10 times the track diameter). However, considering their

diameter of 5-10 nm one can easily calculate that they cover all the film surface if the

1013 ion tracks per cm2 do not overlap each other. The question of the segregation of C

inside the core of the tracks or at their periphery can be discussed on the basis of the

following arguments: It is generally believed that in very short durations of time (10-13

to 10-11s) spikes of energy or temperature increase occur in the core of the tracks. A

noticeable part of the excited particles are desorbed in this part of the damaged material,

explaining the observation of pipes even in heat-resistant polymers like polyimides [6,

7]. Voids form in polycarbosilanes or in polysiloxanes when heat-treated at moderate

temperatures of 600-1000°C by coalescing of the free volume resulting from both

11

compaction and evolution of gaseous species (H2, CO, SiH4...) but only H evolves

during ion irradiation.

The discontinuous distribution of C clusters in the tracks suggests the occurrence of a

solid state transformation instead of a resolidification at the end of a spike, so that there

should be no reason for C to segregate outside tracks. Moreover, the difference in the

width of C enriched areas in the two polymers can simply be explained if carbon

precipitates inside the tracks due to the twice higher concentration of CH3 groups in

PCS with respect to HPCS.

4. Conclusions

HREM and ETFEM observations permitted to evidence the clustering of C in

polysiloxanes and polycarbosilanes under irradiation with swift heavy ions, sustaining

the previously proposed mechanisms of hardening and photoluminescence of the

formed ceramics by such particles. According to the few observations performed until

now the C clusters do not grow with increasing ion fluence indicating that the decay of

the luminescence at high fluence must be ascribed to a change of the electronic structure

of the clusters or of the matrix. Since clusters consist of C and H only, the reason for

the decrease of luminescence is assumed to be the multiplication of non-radiative

deexcitation centers in the matrix associated to structural defects.

The morphology and size of the observed C clusters are comparable in films irradiated

at energies of 3 and of 100 MeV although 100 MeV Au ions induce a high density of

electronic excitations along tubular tracks while the paths of 3 MeV Au ions are

12

stochastic, and displacements as well as ionizations are produced in much larger

regions. The linear densities of energy transferred to electrons Se and to nuclei Sn are

almost equal in case of 3 MeV Au ions whereas in the case of 100 MeV Au ions the

value of Sn is negligible. It is therefore clear that the electronic excitation is the reason

for the formation of the clusters. Since the results obtained by irradiation with 3 MeV

Au ions at a fluence of 3x1013 ions/cm2 and with 100 MeV Au ions at a fluence of 1013

ions/cm2 are comparable, it can be concluded that the process of cluster formation is

dependent on the product of Se and the fluence.

Acknowledgments

Thanks are due to Prof. Paolo Colombo, University of Bologna, for preparing the

polycarbosilane films.

13

References

[1] J.C. Pivin, P. Colombo, J. Mat. Sci. 32 (1997) 6163 - 6175.

[2] J.C. Pivin, M. Sendova-Vassileva, Solid State Commun 106 (1998) 133 - 138.

[3] J.C. Pivin, M. Sendova-Vassileva, P. Colombo, A. Martucci, Materials Science

and Engineering B69 (2000) 574-577.

[4] J.C. Pivin, P. Colombo, G.D. Soraru, J. American Ceramic Society 83 (2000)

713 -720.

[5] C. Trautmann, K. Schwartz, T. Steckenreiter, Nucl. Instrum. and Methods in

Phys. Research B 156 (1999) 162 - 169.

[6] Y. Eyal, K. Gassan, Nucl. Instrum. and Methods in Phys. Research B 156 (1999)

183-190.

[7] V. K. Mittal, S. Lotha, D. K. Avasthi, Radiation Effects & Defects in solids 147

(1999) 199-208.

[8] M. Levalois, P Marie, in proceedings conference "swift heavy ions in materials

egineering and characterization" SHIMEC-98, NSC, D.K. Avasthi, D. Kanjilal

eds., NIMB 156 (1999), p. 64 .

[9] J.P. Biersack, Nucl. Instrum. and Methods B27 (1987) 21-30.

[10] O.L. Krivanek, A.J. Grubbens, N. Dellby and C.E. Meyer, Micr. Microanal.

Microstruct. 3 (1992) 187-199.

[11] E. Pippel, J. Woltersdorf, G. Pöckel, G. Lichtenegger, Materials Characterization

43 (1999) 41-55.

[12] C. Delerue, M. Lannoo, G. Allan, Phys. Rev. Lett. 76 (1996) 3038-3044.

14

[13] M.A. Laguna, V. Paillard, B. Kohn, M. Ehbrecht, F. Huisken, G. Ledoux, R.

Papoular, H. Hofmeister, J. of Luminescence 80 (1999) 223-228.

[14] J. Robertson, Adv. Phys. 35 (1986) 317 - 327.

15

Captions

Fig. 1: Raman spectra of the 4 studied films, microcrystalline graphite and of a PECVD

films of polycrystalline diamond with stacking faults and dislocations.

Fig.2: EFTEM image of MTES, a) irradiated with 3MeV 5x1013 Au/cm2 and b)

irradiated with 3 MeV 2.5x1015 Au/cm2 (red: carbon, blue: silicon).

Fig.3: HREM image of MTES irradiated with 3 MeV Au ions.

Fig.4: a) TEM, b) EFTEM image (red: carbon, blue: silicon) of PCS, irradiated with

100 MeV Au ions.

Fig.5: a) TEM, b)EFTEM image (red: carbon, blue: silicon) of HPCS, irradiated with

100 MeV Au ions, (lower left: epoxy).

16

Korrespondenzanschriften

Jean-Claude Pivin

Centre de Spectrométrie Nucléaire et de Spectrométrie de Masse, Bâtiment 108,

91405 Orsay Campus, France

Eckhard Pippel, Jörg Woltersdorf

Max-Planck-Institut für Mikrostrukturphysik, Weinberg 2, D-06120 Halle, Germany

Devesh Kumar Avasthi, Sanjev Kumar

Nuclear Science Center, Post Box 10502, Aruna Asaf Ali Marg, New Delhi-110067,

India

Fig.1J. C. Pivin et al.

wavenumber (cm-1)

1000 1200 1400 1600

intensity (a.u)

0

100

200

300

400

500

600

700

800

MTES 5 1013Au

MTES 2.5 1014Au graphite defected diamond

PCS 1013Au

HPCS 1013Au

Fig.2J. C. Pivin et al.

5 nmSi substrate

layer

a) b)

Fig.3J. C. Pivin et al.

layer

Si substrate, d(111) = 0.313 nm

Fig.4J. C. Pivin et al.

20 nma) b)

Fig.5J. C. Pivin et al.

20 nma) b)