J. TATARKIEWICZ: On Some Proton Radiation Effects in Silicon 423

phys. stat. sol. (a) 63, 423 (1980)

Subject classification: 11 and 20.1 22.1.2

Institute of Nuclear Research, SwierkJOtwockl)

On Some Proton Radiation Effects in Siliconz)

BY J. TATARKIEWICZ

I R optical properties of 9 MeV proton bombarded silicon are studied as a function of implantation dose. Correlations between IR absorption band intensities are observed. To explain the measured effects, the simple “temperature spike” model is discussed. Some numerical calculations con- cerning the dose effects are presented. Comparisons with previously known data are also made.

On a 6tudik les proprietks optiques IR du silicium bombard6 de 9 MeV protons en fonction de la dose implant6e. On a constate une correlation entre les bandes IR d’absorption. Pour expliquer les effets obtenus on a discut6 les ,,temperature spikes“. On a present6 les calculs numbriques concer- nant les effets lies B la dose. On a compare les resultats obtenus actuellement avec les pr6c6dents.

1. Introduction A number of studies have been conducted on hydrogen implanted silicon, and for a brief review the reader is referred to [l]. Prom the pioneer study of Stein [a] the use of increasing implantation energies has been observed. Recently Mukashev et al. [3] have used 7 MeV protons and as a result they measured I R absorption bands quite different from those of Stein (400 keV protons). It is also well known that the ranges and range stragglings for protons in silicon are nearly proportional to ion energies (cf. [4]). It is obvious that the same doses of 7 MeV protons [3] and of 2 MeV protons [5] should give after irradiation different hydrogen concentrations. However, in the present study the dependence of proton radiation effects in silicon on the dose was investigated using IR absorption measurements. The simple model, based on “temper- ature spikes” [GI , predicts that there should be a “critical” dose for which the spikes overlap. Experimental data coincide well with the calculated “critical” dose. This means that the results of earlier papers should be comparable taking some precautions. Much more work is needed to allow a full understanding of the details of proton irradiated silicon I R specbra.

2. Expmimental Dislocation-free, 5 Qcm, n-type (P doped) silicon monocrystals were used. The samples contained about 3 x lo1’ oxygen atoms/cm3 which was checked using the I R transmittance near 9 pm [7]. The samples for irradiations were cut a few degrees off the (111) plane to avoid channeling, and mechano-chemically polished to the thick- ness of about 1.65 mm. The differences in thickness were less than 25 p i , so it was easy to find matching samples for the differential absorption measurements [S]. This technique has been particularly fruitful for studies of radiation defects in silicon. It has been applied primarily to observe all weak hydrogen bands instead of having to observe them on the pure silicon absorption backround.

l ) 05400 Swierk/Otwock, Poland. 2, Partly supported by Polish Academy of Sciences, Institute of Physics, Warsaw, under the

contract MRI-4. 28 physica (a) 63/2

424 J. TATARKIEWICZ

Two double-beam spectrometers were used. In the range of 1.0 to 2.5 pni wave- lengths the Varian Carry 17 and in the range of 2.5 to 52 piii wavelengths the Perkin Elmer 580B spectrometers were used. The relative transmittance Trel of the irradiated sample when a reference non-irradiated sample is used can be expressed as

(1)

where the niultiple-reflection correction [91 is taken into account. The subscript “0” denotes the constants of the non-irradiated material, whereas “1” denotes those of irradiated one; OL being the absorption coefficient and d the thickness. Then aid, is the irradiation induced absorbance; I2 is the dielectric reflectance where R = = ((n - l ) / (n + 1))2, which for pure (non-irradiated) silicon is about 30% because n = 3.446 is the silicon refractive index in the absence of strong absorption bands [lo]. For high-resistivity silicon (no free-carrier absorption) when wavelengths greater than 1.1 p i are involved we always have sod, 5 2 since the sample thicknesses were less than 2 inm. Expanding expression (1) into a Taylor series we get

1 e-&,di ( 1 - R2 e-%& Trel = - _ _ ~ 1 - R2 e-Z((*odo+di) ’

Trel x e-a (1 - R2 e-2(d-aldd + ...) . ( 2 )

Hence the radiation induced absorbance a,d, can be calculated with an error of less than 2%,

x - In (Tre1) (:<)

which can be automatically plotted on both the spectrometers used (ABS function). The absorption units have been calculated from the measured transmittance spectra of irradiated samples (with no reference sample) taken in the region of 2.5 pm wave- length.

The irradiations were made a t the INR linear proton accelerator ANDRZEJ. The sample holder was placed a t RT outside the accelerator vacuum system, hence, the protons reaching the sample had an energy of about 9 MeV. A proton mean current of about 2 pA was used (ANDRZEJ produces 400 pA current pulses with a frequency of several Hz). Because of very intense air ionization taking place during bombard- ment it was impossible to measure exactly the temperature of the samples “on beam”. Instead attempts were made to cool the sample holder with gaseous nitrogen from boiling N2 (more than 30 l/min). The differences in the NIR absorption spectra with

and without cooling are shown in Fig. 1 (cf. also Fig. 4). Cooling is necessary during bombardment because the 9 MeV proton current of 2 pA gives a power of 18 nT

Jhrnj- 5

A to the sample. The implantations described below were I performed with cooling only. 2 It should be emphasized that nearly the whole pro-

ton energy is liberated in the form of heat, although in view of heat conductivity there are no observable teni- perature changes of the sample surfaces. Probably the fullest information about the temperature inside the sam- ples is given neither by the sample surface temperature

Fig. 1. Comparison of the NIR absorbance spectra without (1) and with (2) nitrogen cooling (dose of about 2.5 x 10’’ pro- tons/cm2) ; 9 MeV proton bombarded Si

--wavenunberi70~n ’i

On Some Proton Radiation Effects in Silicon 425

nor by the current and energy used, but by the resulting IR band intensities. Fig. 1 shows the divacancy band [ll] a t about 1.8 p i which is suitable for that purpose because the temperature of annealing for divacancies is about 550 K, hence their lifetime a t slightly lower temperatures is comparable with the times of irradiation.

The times of irradiation have been varied from 10 to 60 h when intensities of about lo1' to 6 x 1017 protons/cmZ were obtained on 3 em2 of the implanted surface. IR optical measurements were performed on the whole implanted areas by means of specal diaphragms only at RT.

3. Results

Fig. 2 and 3 show 1R spectra of proton irradiated silicon samples for various doses. It is obvious that for increasing doses different bands are observed but it is more interesting that the band intensities change also. When band halfwidths (FWHM) are t,he same for various intensities, i t is much sinipler to consider band heights only, which was done in this work. Instead of a table of the band positions (which are not very accurate because of the background of other bands, for more than 30 band positions see [3, 51) here (Fig. 4) we compare band intensities (heights) for varying doses. The bands have been chosen arbitrarily, although all band intensities behave like those shown in Fig. 4.

The vertical scale units in Fig. 4 are different for various bands which means that only the relative intensities are preserved.

Fig. 2 Fig. 3

Fig. 2. NIR spectra of proton bombarded silicon for various doses: (1 ) 0, (2) 6.35 x lo", (3) 5.00 X x lo1?, (4) 3.58 x lo1?, (5) 2.00 x protons/cm2; 9 MeV proton bombarded Si

Fig. 3. I R spectra of proton bombarded silicon for various doses. a-Si: H I R absorbance bands from [13, 141 are also included. (1) Dose 1 x 1017, (2) 2 x 10'7, (3) 3.6 x 1017, (4) 5 x lo1', (5) 6.3 X x

(6) 1.00 x

protons/cm*. Note the change of wave number scale a t 2000 cm-l

28'

426 J. TATARKIEWICZ

7 2 3 4 3 6 dose ilO'~rotom/crn ?--

Fig. 4. The plots of relative intensities of chosen IR bands (- ) observed in proton bombarded silicon versus doses. The results for non-cooled samples (- - - -) are presented for comparison. The band positions are given. (1 t o 11) "Si:H" band with (1) 630, (2) 660, (3) 885, (4) 2100, (5) 2160, (6) 2180, (7) 750, (8) 810, (9) 1830, (10) 1960, (11) 2060 cm-'. (12) 'V-0" band, 829 cm-', (13) Si TO modes band 480 cm-', (14) "V-V" band, 1.8 pm

The decreasing of some band intensities a t about 5 x l0l7 protons/cni2 will be discussed in the next section. Here we shall compare the divacancy band (1.8 pm) and silicon one-phonon vibration modes [12] (the Si TO inode a t about 480 cm-l consists of three bands corresponding to different primitive cell points: 464 cm-l a t X, 491 cn1-l a t L, and 481 cm-1 a t W). These bands were induced by irradiation be- cause the defects break the lattice symmetry. Close correlations of these band intensities with some silane molecule band intensities (Fig. 4) suppose that defects

are not necessarily the cause of the appearance of all the bands observed. For the largest dose implanted sample (6.3 x lo1' protons/cm2) the IR spectrum is similar (Fig. 3, upper part) to the I R spectra of hydrogenated amorphous silicon of various compositions (cf. [13, 141, where from the v.-Si:H band shapes were taken).

A full analysis of the observed silane niolecule bands will be possible with the use of group theory when cross tests of the present data with thermal annealing data will be available. Now only speculations can be done, nevertheless simple models in the next section try to explain some of the dose effects encountered.

4. Models and Calculations

The Monte Carlo simulation program RADDI [15] has been used to compute 9 MeV proton ranges in silicon. The obtained range and damage histograms are shown in Fig. 5. The 9 MeV protons in silicon can travel a total distance of about 600 pni (without including channeling). The projected range and range straggling are 580 and 15 pm, respectively. Notice that the damages (deposited energy) histogram is plotted in a semi-logarithmic scale. Nearly all the defects are produced a t the end parts of the proton tracks. The proton dosesD used in the present study (more than 1017 protons/

(Fig. 5. Monte Carlo simulated range (- ) and damage (deposited energy) (- - - -) histograms for 9 MeV proton bombarded silicon. R,,, = 600 pm; R, x 580 pm and AsRp z 15 pm are simulated values

On Some Proton Radiation Effects in Silicon 427

/cni2) give hydrogen concentrations exceeding N , x D/ A,Rp = lozo hydrogen atoms/cni3 or 0.3 a t % ,which compares with several per cent concentrations in hydro- genated amorphous silicon [13]. As was mentioned before, the silicon samples used contained about 3 x 1O17oxygen atoms/cm3 although, as can be seen in Fig. 3 (lower spectrum for the non-irradiated sample), some samples contained about, 15% oxygen atoms less (compare the negative absorption a t 9 p.m and at about 16 pni to check that the differences in thickness of the samples are minor). After irradiations the band a t 9 pm (about 1100 cm-l) becomes less pronounced (Fig. 3, the negative differential absorbance old being about 0.04) and the “V-0” band a t 829 cm-l ap- pears [16]. Simple calculation reveals that nearly all oxygen atoms in the region of high defect density (cf. Fig. 5) are in the vacancy-oxygen or other similar configura- tions. Negative absorbance of about 0.04 gives for d = A,R, = 15 pm an absorption coefficient of about 2 cin-l, hence [7] there are no oxygen atoms a t the interstitial lattice sites in this region.

Similar calculations for the divacancy band a t 1.8 pni show that for, e.g., a dose of 3.5 x lo1’ protons/cm2 we observe an absorbance old of about 1.3, hence (cf. [17]) the introduced divacancy concentration is Nvv = lo1’ or about 1/3 divacancy per incident proton. For comparison, oxygen ion implantation [18] gives five divacan- cies per incident ion. Assuming that “on beam annealing” is similar in both experi- ments and that only “hard-core” collisions occur one obtains Kinchin-Pease’s [ 191 expression for the number of defects (vacancies) created,

E N d = -, where E 5 E, 2 8 ,

and

is the well-known “hard-core” collision energy limit. The ratio of the resulting numbers of defects for proton and oxygen implantations in silicon is then

which agrees well with the previously described values of divacancy for protons and five divacancies for oxygen ions. These results show also that the irradiations in this work have been done correctly.

Finally let us explain dose effects (Fig. 4). When the simple “temperature spike’’ iiiodel is applied to proton irradiation of silicon, one obtains [20]

where rmelt is the radius of the spike region melted due to the transferred energy (here only spherical spikes are discussed), Tmelt is the melting temperature (for silicon about 1700 K), To the temperature of the sample (here we assumed about 300 K), e the target density (for silicon about 2.33g/cm3), c the specific heat of the target material (for silicon about 0.8 J/gK), and Q the energy transferred to the lattice. Assuming a “hard-core” collision energy limit for protons in silicon (cf. expression (4b)) of about 1 keV and applying the maximum transferred energy expression

428 J. TsTARKIEWICZ

where M denotes the mass, we can easily compute Q = 2.2 x J and hence rmelt = 9 per incident proton; hence the volume of the melted VInelt region is about 3 x loez1 cm3. Because of the 9 MeV proton range straggling in silicon of about 15 pm the “critical” dose when the “temperature spike” overlap is

which agrees well with the observed “critical” dose (Fig. 4). Increasing of defects and some silane band intensities above that dose can be explained by long-range migrations of atoms and defects [OJ during spike cooling. The defects produced by the first spike are moved to the undamaged region by the overlapping spike.

Now it is understood why 2 MeV and lower-energy experiments have been giving results different from those of high-energy implantations: for 2 MeV protons the range straggling is less than 3 pn’ [4], so the dose of 1.5 x 1017 protons/cm2 [5] is slightly above the “critical” dose, whereas the same dose of 7 MeV protons [3 ] is niuch lower than the “critical” one (A,R, = 12 pm). As the “hard-core” limit is very low the whole effect does not depend on the starting proton energy.

5. Summary and Conclusions The results of this investigation support the concept of complex hydrogen defects [21] in silicon, which for higher doses makes possible a simulation of nearly amorphous hydrogenated silicon by proton irradiation. This method seems to be much more precise, especially for IR measurements when high-energy protons are used, because of the wide damaged region. Methods used to produce a-Si:H [13] are not very accurate, because the hydrogen concentration cannot be controlled. Also a better understanding of the complex hydrogen-defect pairs of interest in itself [22J was the purpose of this work. However, much more work is needed to fully understand the influence of hydrogen and defects on the optical properties of silicon for future solar cells.

Acknowledgements

The author wishes to thank Prof. H. Rzewuslri for his kind interest in the work. Many thanks are due to Dr. M. Grynberg and Dr. A. Witowski of the Institute of Experimental Physics, Warsaw University for their helpful teaching and for making their IR instruments available. I owe special thanks to Dr. K. Szczepaniak and Dr. M. Nowak of P.A.S. Institute of Physics for making their far-infrared spectrometer available. Thanks are also due to Dr. Z. Mazur and Dr. A. Stegner of the INR for making the proton irradiations possible. The author wishes to thank Dr. Z. Werner for the very valuable discussions.

Technical assistance of Miss E. Samsel and Mr. T. Lewandowski is acknowledged.

References [l] S. T. PICRAUX, F. L. VOOK, and H. J. STEIN, Inst. Phys. Conf. Ser. 40, 31 (1979). [2] H. J. STEIN, J. Electronic Matter 4, 159 (1975). [3] B. h‘. MUKASHEV, I<. N. Nussu~ov, and M. F. TAMENDEROV, Phys. Letters A 71, 38 (1979). [4] J. TATAHKIEWICZ, Nuclear Instrum. and Methods 146, 446 (1977). [5] N. N. GERASIMENKO, If. ROLL^:, LI-JEN CHENG, Y. H. LEE, J. C. CORRELLI, and J. Vv. COR-

[6] L. T. CHADDERTON, Radiation Damage in Crystals, Methuen Co./J. Wiley & Sons, London/

[7] W. KAISER and H. KECH, 5. appl. Phys. 28, 882 (1957).

BETT, phys. stat. sol. (b) 90, 689 (1978).

New York 1965 (p. 36).

On Some Proton Radiation Effects in Silicon 429

[ 8 ] R. C. NEWMAN and R. S. SMITH, J. Phys. Chem. Solids 30, 493 (1969). [9] F. D. PALIK and J. J. FURDYNA, Rep. Progr. Phys. 30, 1193 (1970).

1101 R. HULTHEN, Physica Scripta l Z , 342 (1975). [ I l l L. J. CHENC, J. C. CORRELLI, J. W. CORBETT, and G. D. WATKINS, Phys. Rev. 152, 761

[ 121 M. BALKANSKI and W. NAZAREWICZ, J. Phys. Chem. Solids 23, 573 (1962). 1131 M. H. BRODSKY and M. CARDONA, J. non-crystall. Solids 31, 81 (1978). 1141 H. J. STEIN and P. S. PEERCY, Appl. Phys. Letters 34, 604 (1979). 1151 J. TATARKIEWICZ, Computer Phys. Commun. 20, 133 (1980) [I61 J. W. CORBETT, G. D. WATKINS, and R. S. MCDONALD, Phys. Rev. 135, A1381 (1964). [I71 H. J. STEIN, F. L. VOOK, D. I<. BRICE, J. A. BORDERS, and S. T. PICRAUX, Radiat. Eff. 6,

[18] H. J. STEIN, F. L. VOOK, and J. A. BORDERS, Appl. Phys. Letters 14, 328 (1969). [191 G. KINCHIN and R. PEASE, Rep. Progr. Phys. 18, 2 (1955). 1201 M. KOPCEWICZ, I. SOSNOWSKA, and 5. TATARKIEWICZ, Radiat. Eff. 30, 207 (1976).

(1966).

17 (1970).

1211 R. L. KLEINHENZ, Y. H. LEE, V. A. SINGH, P. M. MOONEY, A. JAWOROWSKI, L. M. ROTH, J. C. CORRELLI, and J. W. CORBETT. Inst. Phys. Conf. Ser. 48, 200 (1979).

[22] R. C. XEWMAN, Adv. Phys. 18, 545 (1969). (Received August 8, 1980)