Solar Energy Materials 20 (1990) 405-415 405 North-Holland

X-RAY DIFFRACTION STUDY OF TEXTURE AND GROWTH OF RF-SPU'ITERED CdS FILMS *

A. BENNOUNA, E.L. AMEZIANE, A. HAOUNI, N. GHERMANI, M. AZIZAN

Laboratoire de Physique des Solides et des Couches Minces, Facult~ des Sciences, Universit$ Cadi Ayyad, B.P. S-15, Marrakech, Morrocco

and

M. BRUNEL Laboratoire de Cristallographie, 25 Avenue des Martyrs, 166X, 38042 Grenoble Cedex, France

Received 14 June 1989

The growth and the texture of RF-sputtered cadmium sulfide thin films are investigated using the X-ray diffraction technique. The results show that: (1) the film quality is very sensitive to the argon pressure in the deposition chamber, (2) heating the substrate at 200 o C enhances the degree of preferential orientation, and (3) higher substrate temperatures increase the stacking fault density in the grains. The sizes of crystallites are estimated from the reflection widths, and an analytical model for grain orientation is proposed.

1. Introduction

Cadmium sulfide is a material of which the gap is one of the most interesting for windows in visible-photosensitive systems. Radio frequency (RF) sputtering is a thin-film growing method leading to a satisfactory reproducibility of the electrical properties of CdS films [1-3].

Sprayed [4-6] and evaporated [7-12] CdS thin films have been largely investi- gated over the last decade and some complete structure studies are now available. To our knowledge, only one paper has been devoted to the study of the structural properties of sputtered films [3] and it is essentially based on SEM observations.

Taking into account the diffusion of charge carriers by the grain boundaries, crystallite sizes and orientation can affect the electrical properties of the films. The knowledge of the structural properties and the methods to optimize them can help to improve the characteristics of devices using thin-film technology. For layered materials such as CdS this generally comes down to find a way to obtain a better oriented columnar growth with larger crystallite sizes in the plane as well as in the cross-plane direction.

* This work has been realized in part with the support of the CNR/CNRS scientific cooperation (A.C. MPB3),

0165-1633/90/$03.50 © 1990 - Elsevier Science Publishers B.V. (North-Holland)

406 A. Bennouna et al. / XRD study of RF-sputtered CdS films

2. Experimental details

All the films are prepared in an Alcatel SCM 451 diode RF sputtering system using an undoped 99.999% pure target (7.5 cm in diameter). The pulverization is made on amorphous glass substrates. The target-to-substrate distance is 6.5 cm. Argon is used as sputtering gas (99.9995% pure).

The previous commercial system has been modified to allow multiple deposits without breaking the vacuum. The substrate holder is electrically insulated from ground and can be externally biased, water-cooled or heated by an external source.

The target is always etched for 5 minutes before the pulverization onto the substrate. The deposit time is adjusted to the required thickness according to the deposit rate which is determined for each couple pressure-power. The thickness and rate are monitored during the pulverization by an XTM quartz thickness monitor. The thickness is controlled afterwards by using a visible light interferential micro- scope (for the thinnest films), or interference fringes in the transmission spectra (for films thicker than 5000 ,~).

Temperatures are monitored by two chromel-alumel thermocouples one of which is dipped in a reference temperature.

A specially designed diffractometer [13] is used to carry out grazing incidence X-ray diffraction (GIXD) measurements.

Table 1 shows the deposition parameters related to each sample presented in this paper. From the structural point of view, our experiments are all reproducible.

We used Scherrer's formula and Gaussian reflection shape to compute the sizes of the crystallites. The instrumental resolution has been estimated using the follow- ing formula:

(A~t/X) = AOsimlytan Osi(m ~. (1)

The instrumental widening AO i for each (hk l ) reflection is then computed assuming a constant resolution:

AOi(hkl ) = (AX/X) tan O ( h k l ) . (2)

Table 1 Deposition parameters of the samples presented in this paper

Sample Initial Argon Target Deposition Film External number pressure pressure voltage rate thickness beating

(10 -7 mbar) (10 -3 mbar) (V) (A/min) (A)

# 33/2 8 10 1080 128 7680 No # 37/1 9 10 1080 74 4440 200 o C # 27/2 10 10 1040 - 128 - 1200 No #28/2 10 10 1040 128 1200 No #34/1 10 5 1120 109 7630 No #34/2 10 10 1120 128 8960 No # 34/3 10 50 1080 157 11130 No

For all samples: substrate bias = grounded; RF power = 3.4 W/cm 2.

A. Bennouna et al, / XRD study of RF-sputtered CdS films 407

3. Results and discussion

3.1. X-ray diffraction spectra analysis

All RF-sputtered CdS thin films present evident similarities that will be discussed below.

The lines shown by all our diffraction patterns are attributed to the wurtzite structure [14] and the intense (200), (220) and (311) reflections of the cubic structure [15] are always absent. This is not the case when the films are deposited by spraying [4,6] or thermal evaporation methods [7].

3.1.1. Determination of the lattice parameters Our computations allow us to give the following results:

a = 4.114 + 0.004 ,~, c = 6.724 + 0.006/k.

Comparing these values to the bulk material ones, we find a contraction of the a-axis of about 0.5% and an expansion of the c-axis of about 0.2%. These effects are probably a consequence of the strains inherent to the preparation method. We note at this subject that evaporated films have been prepared and analyzed; the calcu- lated parameters are, as in ref. [9], nearer to the bulk material ones.

3.1.2. Degree of preferred orientation in bulk All spectra show that the preferred growth mode is such that the c-axis of the

crystallites is normal to the surface. So, we calculate the angle ~ between any (hk l ) plane and the (001) one; ~ is defined as follows:

cos (hkl) = 1 + / 12 (3)

The corrected and normalized intensity I c is:

I¢(hkl ) = Im(hk l ) AO(hkl) io(hkl ) , (4)

where 1 m is the maximum intensity of the (hk l ) reflection, A0 the full width at half maximum (FWHM) corrected by the instrumental widening, and I 0 the structure factor (which can be taken as the related powder intensity if there is no texture).

The c-axis of crystallites oriented with (hk l ) direction normal to the substrate make an angle ~ with the normal to the substrate. If the line-shape is rounded to a triangle approximation, I c is proportional to the total number N of crystallites having their c-axis pointing in the direction ~. If a unique preferred orientation exists, N must decrease with increasing ~.

We define a negative orientation energy by

E = - E 0 cos ~(hk l ) , (5)

E decreases with increasing ~, and E 0 increases with increasing degree of preferred

408 A. Bennouna et aL ,/ XRD study of RF-,wuttered CdS film~"

34/1

9(°) 430 36 30 2'7

J i

'>~ ; i

,i

I I I !

23 1~8 _ u ~ , 13

Fig. 1. Classical X-ray diffraction spectrum of sample ~ 34/1 (see text).

orientation and depends on the method of preparation. Applying Boltzmann statis- tics to N, we have:

N = A exp(E 0 cos ~(hkl)/kT). (6) According to this, an estimation of E 0 allows the computation of the mean angle

(~) of disorientation of the crystallites. For sample # 34/1, an X-ray diffraction (XRD) spectrum is shown in fig. 1 and

a fit of ln(lc) versus cos(~) is shown in fig. 2. This latter shows a good agreement between our model and the experience.

~z. ~ ~ 102

o ,I I I I i _ o . 4 0 . 6 o . s ¢os(~)

Fig. 2. Straight-line fit of the logarithm of the corrected normalized intensity versus the cosinus of the orientation angle of crystallites.

A. Bennouna et al. / XRD study of RF-sputtered CdS films 409

Table 2 Orientation data of the films: logarithm of the corrected normalized intensities and results of the orientation analysis

(hkl) ~ Sample number

(deg) # 33/2 # 37/1 # 27/2 # 28/2 # 34/1 # 34/2 # 34/3

101 61.9 0.117 - - - 1.058 0.871 0.116 102 43.1 1.943 - - - 2.971 2.863 1.553 103 31.9 2.755 2.271 1.783 1.983 4.196 3.135 3.123 004 0 4.007 5.075 3.373 - 5.260 4.877 3.071 104 25.1 . . . . 4.991 - - 114 39.1 . . . . 3.473 - - 105 20.5 . . . . 4.465 - -

(~) (deg) 16.9 14.1 a) _ _ 16.0 17.0 16.8

E0(eV ) 0.188 0.262 a) _ _ 0.209 0.187 0.191

a) Values obtained by comparison to sample #33/2.

The found values are l is ted in tab le 2. The found mean d i so r i en ta t ion angles (17 o ) are c o m p a r a b l e to those ob t a ined b y e lec t ron d i f f rac t ion in ref. [7] for the rmal evapora t ed films, bu t higher than the angles (7 o ) o b t a i n e d on l a se r - evapora t ed f i lms depos i t ed on to subs t ra tes hea ted at 350 ° C [10].

I t is now easy to expla in the absence of all the a l lowed (hkO) ref lec t ions even when their s t ruc ture fac tor is high: (hkO) be ing pe rpe nd i c u l a r to (001), the corre- spond ing Ic(hkO ) in tens i ty is abou t exp(9) smal ler t han I t (001) and Im(hkO ) decreases then s t rongly to n o n - m e a s u r a b l e values.

W e th ink that the me thod used here should be app l i ed to o ther l ayered mate r i a l s such as o ther hexagona l I I - V I or o r t h o r h o m b i c I V - V I c ompounds .

The p ropo r t i ona l i t y be tween the n u m b e r of crys ta l l i tes and the m a x i m u m l ine in tens i ty is a lways af fec ted by the crys ta l l i te sizes themselves. W e m a d e an a t t e m p t to conf i rm our mode l wi th rock ing curves b u t it had no success because the sizes a long the c-axis can vary with the o r i en ta t ion 4- F u r t h e r m o r e , the rock ing curve of the (002) l ine al lows analysis in a range l imi ted to [ - 13 o, + 13 o ] and the (006) line, which extends this range to [ - 4 0 o, + 40 o ], has an unsa t i s f ac to ry s igna l - to -noise rat io.

To compare the c ross -p lane tex ture in two d i f fe ren t films, it is somet imes sufficient, as men t ioned in ref. [4], to c o m p a r e the in tens i ty ra t ios of two con t ro l ref lect ions, bu t this requires cons tan t or i m p o r t a n t crys ta l l i te sizes. W e ra ther choose to c o m p a r e the cor rec ted no rma l i zed in tens i ty ( tab le 2).

3.1.3. Apparent grain sizes normally to the substrate Table 3 shows the d o m a i n sizes found for sample # 3 4 / 1 as well as for the o the r

samples p resen ted below. As the (hkl) growth modes o ther than (001) a re no t the p re fe r r ed ones, the

n u m b e r of wel l - s tacked p lanes a long the p reced ing (hkl) di rec t ion is smal le r than a long the (001) one. This is why the d o m a i n size pe rpe nd i c u l a r l y to the subs t r a t e is

410 A. Bennouna et al. / XRD study of RF-sputtered CdS films

smaller for all (hkl) other than (001): 116-273 against 441 A for (001) in sample ¢e34/1. The first data is consistent with that compiled in ref. [11] for evaporated films and brings out the fact that the nucleation kinetics during the growth is not the same on the different crystallographic planes.

It has been shown that the potential (0.035 eV) of stacking-fault creation along the c-axis of h-CdS is very small [7]. As the diffraction lines others than (001) are all related to the wurtzite structure, we think that the reflections at 13.27 °, 27.35 ° and 43.55 ° definitely correspond to an hexagonal (ABABABAB. . . ) stacking with periodic ( l l l)-cubic-l ike stacking faults (ABABCBCBC.. . or ABABCABAB.. . ) that give apparent grain sizes that may be smaller than the real physical size. For # 3 4 / 1 , the related stacking-fault density we find is near 2 × 105 cm --1 which is in the range of values (105-106 ) reported by other authors on evaporated films [7].

3.2. Degree of preferred orientation versus depth

Fig. 3 shows grazing incidence X-ray diffraction spectra obtained for two angles of incidence. The first one (0.15 °) corresponds to an X-ray penetration depth of about 30 A and the second (0.4 ° ) to approximately 500 A. As both intensities of the (002) lines are normalized to have the same amplitude, we can see that the (103) reflection is higher for the 0.15 ° spectrum.

The variation of the degree of preferred orientation with depth can be due to a nucleation process at the end of the growth or to an effect of the surface stoichiome- try. On that subject we note that ESCA measurements on our films exposed to the atmosphere show that the surface contains 24% Cd, 22% S, 8% O and 46% C (excluding the lighter elements), while the bulk has the 50% Cd and 50% S stoichiometry to within about 5% relative precision of the method.

3.3. The effect of the substrate temperature

We studied two cases for substrate temperature T~: (1) the substrate is pre-heated at ~ = 200 °C and the heating power is maintained during all the growth; and (2)

T a b l e 3

C r y s t a l l i t e s izes as d e t e r m i n e d f r o m the (hkl) r e f l e c t i o n s ( r e p e t i t i v e v a l u e s are d u e to m e a s u r e m e n t

F W H M p r e c i s i o n + 0 .025 o )

(hkl) S a m p l e n u m b e r

# 3 3 / 2 # 3 7 / 1 # 2 7 / 2 # 2 8 / 2 # 3 4 / 1 # 3 4 / 2 # 3 4 / 3

100 215 a) 165 a) _ _ _

101 216 - - 168 304 252

102 201 - 178 349 122

110 250 ~) 175 a> _ _ _

103 273 89 99 89 273 273 54

004 441 110 151 441 441 103

104 - - - 116 - -

114 - - 141 - -

105 - - - 162 - -

a) Sizes in the p l a n e o f t he f i lm.

A. Bennouna et al. / XRD study of RF-sputtered CdS films 411

33/2

ol=0.4

o(=0.15 I

3 o 0(') 1 5 2 0 2 5

Fig. 3. Grazing-incidence X-ray diffraction spectra of sample #33/2 at two incidence angles a: 0.15 ° and 0.4 ° , corresponding to penetration depths of about 30 and 500 ,~, respectively.

the substrate is not intentionally heated and a particular experimental procedure allows the thermal effect of the plasma to be brought about.

3.3.1. Intentionally heated substrate Sample # 3 7 / 1 of fig. 4 has been deposited on a substrate pre-heated at

T~ = 200 ° C while # 33 /2 has been grown without external heating (see table 1). By a simple observation of the intensity ratios (004)/(103), we notice that

heating enhances clearly the degree of preferred orientation of the grains in the films. This is also confirmed by the smaller ~ ) and higher E 0 reported in table 2.

#37/1

_A L 33,2 I I I I I I I I I I I I I I I I I

e ( ° ) ~ ' s a o 1 5

Fig. 4. Classical X-ray diffraction spectra of two samples (#33/2 and #37/1) showing the effect of the substrate pre-heating.

412 A. Bennouna et al. / XRD study of RF-sputtered CdS films

/o

O

100

33/2 0 0

I : I , | , , • I 1 1 . 6 1 ~ . 0 1 ~ . 4 2 1 . 2 2 1 . ~ 2 ~ . 0 0 ( ° )

Fig. 5. X-ray diffraction spectra of sample # 33/2 obtained with fixed grazing angle and rotating the detector around the normal to the substrate (Agsi (111) ~ 0 , 0 4 o ) .

We can also see that increasing T~ leads to an increase of the F W H M of the peaks which means that the crystal sizes normally to the substrate are decreased: fig. 4 gives 441 to 110 .~ for (004) and 273 to 89 A for (103) (see table 2).

These remarks seem to show that the adatom with enhanced surface mobility due to higher temperatures prefers the (001) growth mode (compact stacking model) above the others but the frequency of stacking-fault creation is higher so that the number of well-stacked planes is smaUer.

The (200), (220), and (311) reflections of the cubic phase still being absent, heating the substrate up tO 200 °C does not stimulate polytypism in sputtered CdS films unlike the sprayed [4] or evaporated [7] ones.

Figs. 5 and 6 show the (100) and (110) reflections obtained with a fixed grazing incidence angle and rotating the detector around the normal to the substrate. As

I ;. oo 37/1 v ,o.

i ! I | | | I m ea *

~ . B 1 2 . 2 1 2 . 6 2 1 . 2 2 1 . 6 2 2 . O 0{ °)

Fig. 6. Same as fig. 5 with the sample #37/1 (AOs i ( l l l ) = 0.2 ° ).

A. Bennouna et al. / XRD study of RF-sputtered CdS films 413

%

~ # m ' 7 / 2 2 e / 2

• . . .. 1~ - , - i ~ • • • i | . ~ J l i i ! i i i 0(*) a 5 d o 1 5

Fig. 7. Classical X-ray diffraction spectra of two samples (#27/2 and #28/2) showing the thermal effect of the plasma.

these planes are essentially oriented perpendicularly to the substrate, this is the only way to make the related reflections measurable. Attention is attracted on the fact that the two experiments have been conducted with different Soller slits giving different instrumental resolution (see figure captions). The corrected F W H M gives the crystallite size values listed in table 3.

The crystallite sizes in the plane are not strongly affected by the substrate temperature. The grain sizes measured with an electron microscope [3] are larger than the values were report here and we think is because the perfectly periodic domains required to have coherent interferences are smaller than the physical grain sizes seen in SEM.

3.3.2. Thermal effect o f the plasma To bring out this effect we prepared two special samples. Sample # 28 /2 in fig. 7

has been deposited immediately after opening the shutter. On another slide, we first masked the area intended for # 27 /2 and exposed all the rest to the plasma during two hours. After that, a negative mask of the previous allowed us to grow the same thickness as # 28 /2 (1200 A) but on the glass substrate pre-heated by the plasma. Finally the region of # 27 /2 is cut-off by cutting the slide.

Spectra of fig. 7 show the enhancement of the degree of preferred orientation: the intensity ratios (004) or (002) to (103) is better for # 2 7 / 2 than for # 2 8 / 2 . This shows that in the first minutes of the growth at least, the kinetic evolution of the substrate temperature can influence the cross-plane texture of the films. It has been established [2] that approximately 2% of the power (here 3.4 W / c m 2) is lost in substrate heating and that the superficial temperature stabilizes 5 to 10 rain after the exposition to the plasma. From this and from our results, we can conclude that the first thousand hngstriSms of the films always comprise a texture inhomogeneity.

414 A. Bennouna et al. / XR D study qf R F-sputtered UdS ftlms

c3 , i, I

|

0( °} 2 s ~o Is

Fig. 8. Classical X-ray diffraction spectra of three samples showing the effect of the argon pressure.

We also studied seven samples with thickness in the range between 1200 A and 1.5 /~m prepared under the same conditions and we found that the same intensity ratios do not change for films thicker than 3000 A, which confirms our preceding conclusions.

3.4. Effect of the argon pressure and growth rate

We choose to change the deposition rate by varying the argon pressure so that the target voltage and RF power stay nearly constant. By this way, we can free the results from the variation of the thermal effect of the plasma and from the self-bias effect which varies with the target voltage [2,12]. In addition to that, the substrate holder is always externally grounded. Fig. 8 shows the diffraction patterns of samples deposited without breaking the plasma between them: after deposition of each film, a new substrate is placed in front of the target and the new impedance parameters are optimized for the new Argon pressure.

If we take sample # 34 /2 as a reference, we see, besides a lower deposition rate, that a lower argon pressure (sample # 34/3) causes: (i) a decrease of the crystallite sizes (see table 2); and (ii) a change of the lattice parameters stronger along the c-axis than along the a-axis (note the line-shifts in fig. 8):

a = 4.134 _+ 0.012 A, c = 6 . 8 0 8 + 0 . 0 0 8 A . We think that the enhancement of the mean free path of the plasma particles due

to lower pressures causes a stronger bombardment of the films and more damages, which decreases the number of well-stacked planes perpendicularly to the substrate. It is also probable that the observed decrease of the deposition rate is partially due to an enhancement of the resputtering process of the film causing a change in stoichiometry that can enhance the disorder.

A. Bennouna et al. / XRD study of RF-sputtered CdS films 415

Var ia t ions of the op t ica l gap f rom 2.4 to 2.2 eV have been obse rved b y o ther au thors [3] when the argon pressure is dec reased down to 2 x 10 -3 m b a r ind ica t ing some i m p o r t a n t changes in the f u n d a m e n t a l p rope r t i e s of the films.

4. Conclusion

The s t ructura l p rope r t i e s of spu t te red thin f i lms have been ana lyzed as func t ion of some sput te r ing parameters . A n ana ly t ica l mode l of the tex ture of the f i lms is suggested and it indica tes that the p re fe r red o r i en ta t ion co r r e sponds to the c-axis n o r m a l l y to the surface and the mean d i so r i en ta t ion of the crys ta l l i tes is a r o u n d 17 o and somewhat smal ler when the subs t ra te is hea ted to 200 o C.

Increas ing the subs t ra te t empe ra tu r e causes a h igher s tack ing faul t dens i ty a long the no rma l to the subs t ra te p l ane bu t does no t change mean ingfu l ly the sizes of the crysta l l i tes in the p lane of the subst ra te . The results show tha t to get o p t i m u m s t ruc tura l p roper t i e s of the films, it is p re fe rab le to grow the f i lms under h igher a rgon pressures to enhance the size of the crys ta l l i tes in the c ross -p lane d i rec t ion and at lower depos i t ion rates on subs t ra tes hea ted to increase the degree of p re fe r red or ienta t ion .

References

[1] I. Martil, G. Gonzalez-Diaz and F. Sanchez-Quesada, J. Vac. Sci. Technol. A 2 (1984) 1491. [2] I. Martil, G. Gonzalez-Diaz and F. Sanchez-Quesada, Thin Solid Films 90 (1982) 253. [3] I. Martil, G. Gonzalez-Diaz, F. Sanchez-Quesada and M. Rodriguez-Vidal, Thin Solid Films 120

(1984) 31. [4] M.S. Alaee and M.D. Rouhani, J. Electron. Mater. 8 (1979) 289. [5] L. Escosura, E. Garcia-Camarero, F. Arjona and F. Rueda, Sol. Cells 11 (1984) 281. [6] J. Marucchi, M. Perotin, O. Udeacoumar, J. Bugnot and M. Savelli, Proc. 13th IEEE Photovultaic

Specialists Conf., Washington, DC, June 1978. [7] L.L. Kazmerski, W.B. Berry and C.W. Allen, J. Appl. Phys. 43 (1972) 3521. [8] S. Simov, M. Kalitzova, N. Pashov and M. Mikhailov, Thin Solid Films 55 (1979) L5. [9] N.N. Koren, V.F. Gremenok and V.V. Kindiyak, Phys. Status Solidi (a) 90 (1985) K121.

[10] H.S. Kwok, J.P. Zheng, S. Witanachchi, L. Shi, Q.Y. Ying, W. Whang and D.T. Shaw, Appl. Phys. Lett. 52 (1988) 1095.

[11] S. Ray, R. Banerjee and A.K. Barua, Jpn. J. Appl. Phys. 19 (1980) 1889. [12] N. Romeo, G. Sberveglieri and L. Tarricone, Thin Solid Films 55 (1979) 413. [13] M. Brunel, Y. Arnaud and N. Moncoffre, Analysis 16 (1988) 279. [14] Powder Diffraction Data Card #6-0314, Edited by the International Centre for Diffraction Data. [15] Powder Diffraction Data Card #10-0454, Edited by the International Centre for Diffraction Data.