scattering with a slight lattice scattering component for doped coatings, while the low mobility observed in undoped coatings is due to grain-boundary scattering.
Solar Energy Materials 18 (1989) 283-297 North-Holland, Amsterdam
283
THE OPTICAL,ELECTRICAL AND STRUCTURAL PROPERTIES OF FLUORINE-DOPED, PYROLYTICALLY SPRAYED TINDIOXIDE COATINGS H. HAITJEMA, J.J.Ph. ELICH and C.J. HOOGENDOORN Delft University of Technology, Applied Physics Department, P.O. Box 5046, Delft, The Netherlands Received 10 October 1988; in revised form 23 December 1988 Tindioxide coatings with different fluorine-doping have been produced by spray pyrolysis. The coatings have been analysed by spectrophotometry, Hall-effect measurements, X-ray diffraction and electron probe microanalysis. It is shown that the the electron mobility is limited by impurity scattering with a slight lattice scattering component for doped coatings, while the low mobility observed in undoped coatings is due to grain-boundary scattering. The diffraction measurements show a strong preferential orientation which is independent of the doping level. Electron probe microanalysis shows that it is a plausible assumption that the fluorine occupies the oxygen positions in the lattice, while the free electron density is about one third of the fluorine concentration.
1. Introduction Tindioxide layers are electrically conducting and infrared reflecting coatings with a wide range of applications. In the field of solar energy conversion they can be used as a conducting anti-reflection electrode for photovoltaic cells or as a spectrally selective layer for photothermal conversion. Tindioxide acts as a spectral window. Light is transmitted in the wavelength region confined by ~gap and ~plasma"Below ~g band gap absorption occurs, above ~p reflecti~m due to free electron plasma behaviour takes place. As most of the solar spectrum is in the 0.3-2.0/~m region and thermal radiation is in the infrared above 2 /~m, ~p should preferably have a value of - 2 /~m. Th~s can be achieved by suitably doping the tindioxide with antimony or fluorine, which enhances the metal character of the layer. Tindioxide layers are mechanically and chemically highly stable, also at high temperatures, and can be produced in a relatively cheap and simple way by spray pyrolysis [1-4]. The compositional and structural properties of these films were J-vestigated by various authors [5-7]. The major problem when producing tinoxide coatings is a relatively high emissivity, which is related to a limited electrical conduction in the layer. In this paper the cause of this limited conductivity is studied in connection with the role of the fluorine doping in the layer. 0165-1633/89/$03.50 © Elsevier Science Publishers B.V. (North-Hol!and Physics Publishing Division)
284
H. Haitjema et al. / F-doped pyrolytically sprayed SnO 2 coatings
2. Preparation and measuring techniques The coatings are produced by a spray pyrolysis method as described in an earlier paper [1]. For this study we used a solution of SnCI4 in water and alcohol with NH4F as a dopant. The pyrex substrate is heated to 600 °C and sprayed 4 times 2 seconds to minimize a large temperature fall during spraying. Reflectance and transmittance in the 0.3-2.5 pm region are measured with a Perkin-Elmer Lambda 9 Spectrophotometer, which is equipped with an integrating sphere. The reflectance in the infrared region (2.5-50 pln) is measured with a Perkin-Elmer 883 IR-Spectrophotometer with a specular reflectance attachment. 3~ne electrical properties, resistivity and Hall coefficient, are measured according to the van der Pauw method [8] using circular-shaped samples in a 0.6 T magnetic field. A sample holder is used which enables the sample temperature to be varied between 77 and 600 K in vacuum. Layer thickness is usually calculated from the optical extrema using' a method described in an earlier paper [1]. When the layer thickness is less than about 100 nm a Tencor alphastep 200 stylus apparatus is used. In that case a step is made by etching the coating with Zn powder and HCI. X-ray difraction measurements are made using an Earaf-Nonius Diffractis 583 stabilized generator and a Philips PW 105~ powder diffr~ctometer. Cu K a radiation is used and diffractograms are taken using continuous s~:~Lning. The electron probe microanalysis is can~/ed out with a Jeol 733 apparatus. The Sn and O concentrations are measured re~atiw~ to a SnO2 s~tandard. The fluorine concentration is measured relative to a CaF~ ~tandard. In the computation matrix effects are taken into account.
3. Determination of optical constants The determination of optical constants as a function of wavelength is essential for comparing the optical and electric properties. In the 0.3-2.5 pm wavelength region the complex refractive index is determined from reflection and transmission measurements using the R - T method [9]. In the 2.5-25 pm region, where the substrate is absorbing, we use an iterative method to determine the optical constants from the reflection spectrum only. A first approximation is obtained by applying a Kramers-Kronig relation in the 2.5-25/tm region: ¢ ( 2 % ) - ~-2°f °° In R ( ; ~ ) - In R(~,o) d~,.
(1)
The spectrum is extended to 0.3 pm by calculating R for a semi-infinite medium from the refractive index obtained by the R - T method using: R = ( n - 1) 2 + k 2 (n+l)2+k
2
•
(2)
H. Haitjema et aL / F-dopedpyrolytically sprayed SnO2 coatings
285
1.0
/
/"
/
mR
//
0.5
(corrected)
m-R
// .
.
.
.
(measured) ( P y r e x .~
~
../. ~" ",.~ . J JD
•
1
"~'~,"
"" . . . . ~ ' "
2
------>
tt
t
i ~'1"--
5
Idlve I engCh (~m)
10
20
Fig. 1. Measured and corrected spectr'am of R(A) of sample 310. For comparison the reflectance of the bare substrate is also shown.
Above 25 gtm the spectrum is extrapolated using the Hagen-Rubens relation R'-l-c~
-1,
(3)
with c a constant which is derived from the reflectance at A = 25/tm. When the reflecting surface is semi-infinite the complex refractive index is given by: n(X)fn-ik=
1 - R - 2 i ~ s i n q) . 1 + R - 2qrR"cos q,
(4)
Now eq. (4) is used as a first approximation. Using n and k from (4) together with the film thickness and the substrate optical constants we can calculate the reflectaz,ce from the film-substrate system R1(X). When this calculated reflectance differs from the measured spectrum R(~) we can calculate a "synthetic" spectrum R2(?0 wl~ch can be expected to give the correct optical constants when using (1) and (4):
) R(x)[ R(x )/R, (x)]. =
(5)
This procedure is repeated until a spectrum R2()t) is obtained which gives optical constants by (1) and (4) which give the measured reflectance R(~) when the reflectance of the film-substrate system is calculated. As an example fig. 1 gives a measured spectrum of R(),), together with the corrected spectrum R2(),) from (5) which is extrapolated according to (2) end (3). For comparison the substrate reflection spectrum is shown. In fig. 1 we see that the substrate reflection peak at A---9 ttm, which influences the measured reflection spectrum of the coating, disappears due to the correction made. The small peaks at 16 and 21 tim which we present in the measured and in the corrected spectrum correspond to molecular resonances in the SnO2 coating.
286
H. Haitjema et ~L / F-doped pyrolytically sprayed Sn02 coatings
From the optical constants, as calculated from (1) and (4), the complex resistivity p(t0) is deduced using the definition: Or(to) = ( n -
_
1
ik)2 = Croo
p(to)Co¢ ° .
(6)
Here Eroo is the relative permittivity at to - oo, which is about 4.1 for SnO 2.
4. Comparison of optical and electrical measurements
Fig. 2 gives the electrical properties of the samples used. The coating thickness is about 0.35 ttm in all cases. In this figure it can be seen that the electron density n_ and the mobility/t both increase with increasing fluorine concentration of the spray solution. The real and complex part of the complex resistance, as defined in (6) are given for the undoped and the 100% F-doped case in fig. 3. From the results as given in figs. 2 and 3 we can derive some useful quantities using the Drude theory, which can be written as: p(co) -_ "1'- ico (0(r~cOp
(7)
where top is the plasma frequency which can be defined as:
n-e2 2= COp EOCrov/Heff,
(8)
with e the electron charge and reeff the effective electron mass. y is the relaxation frequency given by (9)
y = e,/mefftt.
Comparison between the measured electron density and the complex part of p(co)
1,0-
-3
3020
/•/
/a~*
,
• n_
1
I0 I
0
50
~
d
I
100
150
%F..Sr in solution
Fig. 2. Mobiliq/t and electron density n_ at room temperature of the samples used.
H. Haitjema et al. / F-dopedpyrolytically sprayed Sn02 coatings
,~
(a) Sample
5 ." o ~ •~
, , , , , measured ,, calculated
310
(b) Sample 312 $
¢x.
..,....
-
•
,:
. . . .
- - * - measured
F
2 I
287
pfocl
0t'o:/J ..._--;-
O5
0.2
i
i
I
i
2
5
10
20
2
S
IO
Atr~mJ
Fig. 3. Real and complex part of resistivity p(~,) of a doped (a) and of an undoped sample (b). In (a) the curves according the Gerlach-Grosse theory are also drawn.
gives a value for meff. The real part of p(to) can be directly compared to the DC resistivity p(DC) =
(lo)
(n_ep)-'.
In fig. 3 it is shown that for the N'ped sample the Pr(DC) value is in accordance with the AC value, while for the undoped sample the AC value is much lower. In both figs. 3a and 3b we see that Pr('O), the real part of p(,o), is not frequency-independent, as would be expected from the Drude theory. A theory which gives a frequency-dependent pr(o~) is given by Gerlach and Grosse [10]. For scattering due to ionized impurities, which will always be present in doped coatings, as ions are necessary in order to have free electrons, the dynamical resistivity can be written as: p(to) ffi
iZ2Ni
6~'2con~to
fo°°k 2 d k [ ¢ e ' ( k , t o ) - ¢ e ' ( k , 0 ) ] - i
t~ 2"
CoC~=%
(11)
For undoped coatings it can be assumed that the free electrons origin from either oxygen vacancies, which g~ves n_ = 2Ni and Z = 2, or chlorine atoms, which can be treated as fluorine atoms. As a small amount of chlorine is present in our undoped coatings we have taken n _ - Ni and Z - 1 in all cases. For a theory which gives Ce, the complex dielectric function of the free-electron gas, we follow the treatment as given by Hamberg and Granqvist [11]. For •e the longitudinal part of the Linhard dielectric function is taken, which is corrected
H. Haitjema et al. / F-dopedpyrolyticaily sprayed Sn02 coatings
288
according to the Singwi-Sj~lander theory. The lengthy formulas are given in ref. [11] and will not be repeated here. In fig. 3a the real and complex part of p(to) calculated from (11) are sketched. As input parameters we have used Cr~o - 4 . 1 , n _ - 2 . 3 3 × 1 0 2 6 / m 3 and meff-0.2$6 electron mass. The latter has been chosen to fit the pc(w) as measured. When comparing Pr(to) as measured and calculated in fig. 3a we observe some differences, though the essential features are the same. The wavelength dependence at k < ?~p follows about the same power-law of Or " ?l.s for both the calculated curve and the measured points. A similar power-law dependence has also been observed by Frank et al. [12] for SnO2 and In203 coatings and by Hamberg and Granqvist [11] for In203 coatings. For ~ > ~,p the calculated curve of Pr(to) is shifted downward relative to the measured points, which indicates that the measured mobility is -25% lower than the theoretical upper limit when only the effect of ionized impurity scattering is taken into account. These results strongly indicate that ionized impurity scattering is the main damping mechanism in our doped tindioxide coatings. The molecular resonances at 16 and 21 /zm can be observed in the measured spectrum of 0r(to). In fig. 3b only the measured curves are shown as the measured curve of Or deviates about a factor of four from the curve calculated from (11). This means that for undoped samples the effect of scattering by ionized impurities cannot explain the high Or(t°) measured. We can account for the differen~ between p(DC) and Pr by defining an "optical" mobility by: ~(opt) = (n_ep~(X = 5/tin)) -1.
(12)
Analogously we can calculate the maximum DC mobility which can be obtained according to the theory of ionized impurity scattering by taking the limit to-, 0 in
(11). The results are summarized in :able 1. In table 1 it can be observed that m,ff increases with increasing electron density. This indicates that the conduction band will not be parabolic. At high electron density tt(op0 about equals p(DC), while at low electron density/t(DC) is much lower. An explanation for this could be the occurence of grain-boundary scattering which will affect the D e mobility rather
Table 1 Electrical and optical properties Sample number
n_ (x1026/m 3)
~p Otm)
meu/mes
/z(DC) ( × 1 0 -4 m2/V.s)
/t(opt)
~(theory)
312 318 316 314 310 320
0.252 1.04 1.53 2.12 2.33 2.38
4.67 3.14 2.70 2.27 2.37 2,20
0.13 0.23 0.24 0.24 0.28 0.25
6.3 26.4 28.4 35.2 37.3 36,2
-- 27 28.3 28.9 35.6 36.g 36.8
228 81 67 65 49 58
H. Haitjema et al. / F-dopedpyrolytically sprayed SnOz coatings
289
than the optical properties. The occurrence of grain-boundary scattering should also be found from the temperature dependence of the mobility, which will be discussed in the next paragraph.
5. Temperature dependence of electrical properties The electrical properties, electron density n_ and mobility/t, have been measured from 80 to ~ 450 K. For the Hall scattering factor r - 1 has been taken, which is correct for degenerate electron statistics [13].
5.1. Electron density The electron density varied about 7% for the undoped and 1% for the doped sample over the temperature range mentioned above. For describing this small temperature dependence we adopt a simplified model as given by Blakemore [14]. We assume that all free electrons originate from Nd donors which are at one donorlevel E d from the bottom of the conduction band. Then the electron concentration is given by: F1/2 (~) Nd n_ffiNc F ( 3 / 2 ) = 1 + f l - ' exp(~l--ed)'
(]3)
the dimensionless donor energy, ~ ffi qJ/knT the dimensionless
with e d - - E d / k e T
2.6-
2.5.
: staple ,,
310 i 320 t
6
(J
% S C
I
?.3. v
22.
'
O
'''
! I ' ' ' : ....
3 6 LO00/T (K-i )
o
n
I.*
9
•
I ' ' ~ ....
12
*.
15
Fig. 4. Measured values of n_ (T) of samples 310 and 320, with fits made through the measured points as discussed in the text.
290
H. Haitjema et al. / F-dopedpyrolytically sprayed SnO: coatings
Table 2 Donor density and donor energy level Sample number
r.~onor density ( × 102e/m3)
Donor energy level (eV)
312 318 316 314 310 320
0.608 2.72 4.43 6.16 7.16 7.60
0.122 0.268 0.349 0.429 0.454 0.473
Fermi level, p-1 degeneracy factor, we have taken fl-1 _ 2, F~/201) the Fermi-Dirac integral of order 1 / 2 , and Nc is the quantum concentration given by:
Nc = 2(2~rmafkeT/h ~)3/2.
(14)
For a given Nd, E d, T and men, ~ is found iteratively by searching a value for which equals the middle and the right-hand part of (13). For each sample we made a least-squares fit of Nd and E d to the measured values of n_(T). As an example we give the measured values and the resulting fit of samples 310 and 320 in fig. 4. Table 2 gives the values of Nd and E d for each sample. Because of the crude assumptions made in this model (e.g. only one doping level) these values will have rather a practical than a physical meaning. The practical meaning is that we can calculate n_ at any temperature, so we only need to measure the specific resistance p to obtain the mobility/~ from: -1
/~ = (pn_e) . 05) As measuring the specific resistance is much less elaborate we can measure p at many different temperatures and calculate p using the fit made through the n_(T) values measured at much less different temperatures.
5.2. Mobility Fig. 5 gives a plot of the mobility as a function of temperature for the different samples. In this figure we see that the mobility of the samples having a high mobility decreases with temperature. The mobility of sample 312, having a low electrical mobility, however, increases with temperature. The temperature dependence of the samples having a high mobility cannot be explained by ionized impurity scattering as, according to (11), the only temperature-dependent factor is n_ which is as well const~t in our samples. A scatter mezhanism which can account for the temperature dependence is the scattering by acoustic phonons. For many materials the temperature dependence o~ the mobility due to scattering by acoustic phonons is given by the Grfineisen-Bloch relation [15]
OR ~
"o
(e"- 1)(1-e-")dz'
(16)
H. Haitjema et al. / F-dopedpyrolytically sprayed SnOz coatings
I
i-o-sanple 310
t-o-staple 316 I'*" " 318
32
40
291
30
%
% ,.e
26
n,.
I-o-sample312 I 34
I
50 150
I
I
I
350
450
24
.-
5O
T (K)
I
I
I
I
150
250
350
450
3
5O
T (K)
I
I
150
250
I
I
35O 45O
T (K)
Fig. 5. Mobility as a function of temperature for the different samples.
where c is a constant and OR is a characteristic temperature of the lattice scattering. When using Matthiesen's rule for combining (16) with a constant mobility ~o as is found from neutral or ionized impurity scattering we find for the effective mobility/~e:
Pe] : P o ] + N ~
.'o
(e'- 1)(1-e-')dz"
(17)
Fig. 6 gives a plot of the temperature-dependent mobility of samples 310, 314 and 320 with fits made according to (17). In this figure it is shown that the fit follows the
42
. "
., ^ _
..
o sample 310 320 o ,., 314
-"~,,
~0
3B
~
3s
32
0
I
I
I
I
i00
200
300
400
~ >
T
500
(K)
Fig. 6. Measured values of/~(T) of samples 310, 314 and 320, with fits made using eq. (17) (see text).
H. Haitjema et al. / F.doped pyrolytically sprayed SnOz coatings
292
Table 3 Constants describing the temperature-dependence of the mobility Sample number
tt 0 ( x l 0 -4 m2/V-s)
0R (K)
Co ( x l 0 4 V s K / m 2)
310 314 316 320
42.10 + 0.08 38.65+0.09 29.91+0.07 40.40 + 0.07
1096 + 9
68.3 + 1.7
1106+17 1155+69 1084 + 6
62:1:6 60 +12 65.9 + 1.3
measurements perfectly. Table 3 gives the values for ~t0, c and OR for the samples with a decreasing mobility with temperature. The constants OR and c are within the errors the same for the four samples. The decrease of the mobility with temperature has about the same magnitude as has been found by Stapinski [16] for CdSn204 coatings. At 400 K the lattice scattering can be treated as a mechanism limiting the mobility to 250 cm2/V • s. The values of/t o in table 3 should be compared with the theoretical value given in table 1, rather than the mobility measured at room temperature. This gives a difference in the sence that /t o from table 3 closer approaches the theoretical limit. The increasing mobility of sample 312 with temperature can be explained by grain-boundary scattering, which accords with the observed difference in el~trical and optical prope~es: As the mobility of sample 312 will also be determined by impurity and lattice scattering the mobility should be corrected for the impurity and lattice scattering mobility/ti+~ in the following way: /tgb1 ~ ~ -measured(312) 1 ~711,
(18)
where/~gb is the mobility due to grain-boundary scattering only. For/ti+ 1 we have taken the mobility of sample 316 as this mobility about equals the optical mobility of sample 312. However~ this correction hardly influences the values of ~tgb. In the theory of grain-boundary scattering we can distinguish between thermal field emission and electron tunnelling [17]. The criterion, as given by Roth and Williams [17], for a current dominated by tunnelling, is that the parameter E00 is much larger than kT. Eoo is given by: E00- 18.5 × lO-'2(n_/m*e) '/2 eV,
(19)
where e is the static electric consta~lt of SnO2. For sample 312 with m* --0.13m e, n_ "- 0.25 × 1 0 2 ° / c m 3 and e ~ 10 [18] we find that E~ - 0.081 eV, which is always larger than kT in the temperature range in which the mobility has been measured. Though no satisfacto~ analytical model exists for grain-boundary scattering dominated by tunnelling, often an expression is used which has the following form [7,17,19]: ~tgb ~ ~ff--1/2
exp(-EJkT),
(20)
where c is a constant and E a is the pseudoactivation energy which decreases with
H. Haitjemaeta!. / F-dopedpyrolytically sprayedSnOz coatings
I
'4.6
293
I
q,4-,.-,.
q.2-
w
'q.G-
I--
3.8-
3.S
~D
I
I
2
4
~ >
I ooofr
I
I
6
8
..
I
'.
10
12
~,
14
(K -1 )
Fig. 7. Mobility times the square root of temperature versus inverse temperature for sample 312.
decreasing temperature. In fig. 7 ln(ItsbV~) is plotted against 1/T. In case of a constan~ Ea fig. 7 should give a straight line, wl~¢h is clearly not the case. We observe a aecrease of Ea with decreasing temperature, whicl~:.is to be expected as is stated above. As the ~ term influence the product Itv~ as much as the change in It over the temperature interval, we think that (20) is a too weak base to derive quantities such as the barrier gap height, the grain size and the electron density in the grain boundary, as is done by some authors [17,19]. However, the increasing of/~ with temperature together with the difference, between "optical" and electrical mobility are sufficient arguments for the occurence of grain-boundary scattering as the main damping mechanism in the undoped coating no. 312.
6. X-ray diffraction measurements In table 4 the results of X-ray diffraction measurements are given. The intensity measurements indicate a strong orientation of the crystallites in the (200) direction for all samples. The grain sizes in the different orientation directions have been determined from the diffraction peak widths. The grain size is largest in the (200) direction, which, together with the preferential orientation, indicates that the crystals are colunm-like oriented perpendicular to the substrate. A striking result is that the microstructure is about the same for all samples, independent of fluorine doping. This is not in agreement with results ob ~t~ined by Fantini and Torriani [5] who found an increase of (110) and (211) reflections with measuring fluorine concentration, and a decrease of (200) reflections. The difference may be due to the lower substrate temperature ( - 350 °C) at which their coatings are produced. Our results mean that the differences in electrical and optical
294
H. Haitjema et al. / F-doped pyrolytically sprayed SnO z coatings
Table 4 Results from X-ray diffraction measurements
Crystal plane
F : Sn in solution (~) 0 25
50
80
100
130
Randomly oriented SnO 2
21 < 6 100 9
20 I'F"I ( 102G/'rn 3 ) Fig. 9. Oxygen/tin ratio and (oxygen + fluorine + chlorine)/fin ratio versus fluorine concentration. 0.
I ;~.
I
the free electron density is plotted against the fluorine concentration plus the chlorine concentration. In this figure we see that the electron density is only about one third of the value expected from the impurity concentration. This can be due to the fact that part of the fluorine atoms occupy interstitial positions in the lattice. An other possibility is that not all fluorine atoms are ~onized, though they occupy an oxygen position in the lattice. The oxygen/th~ ratio is plotted ill fig. ~. We see that this ratio decreases with increasing doping concentration. However, when we add the fluorine and chlorine concentration to the oxygen we find that the ratio [F + O + CI]/[Sn] ~emains 2 within the error of 0.5~, independent of the doping concentration. This means that the fluorine replaces the oxygen in the coating so it is reasonable to presume that the fluorine atoms occupy oxygen positions in the lattice.
8. Discussion
The frequency dependence of the complex resistivity indicates that ionized impurity scattering is the main damping mechanism in doped tindioxide coatings. However, the scattering by ionized impurities predicts an increasing mobility with decreasing cartier concentration, while we find the opposite (see table 1). An explanation for this might be the hypothesis that the fluorine occupies oxygen vacancies in the lattice, thus increasing the mobility. This should happen on a smaller scale than can be detected by electron probe microanalysis. From the electron probe microanalysis we find that the fluorine concentration is about three times the quantity which should be expected from the electron density. Together with the measured tin/oxygen ratio we can conclude that the fluorine atoms occupy the oxygen positions in the coating, but these fluorine atoms are aot
296
H. Haiqema et al. / F-doped pyrolytical!y sprayed $n02 coatings
all ionized. This is in agreement with the simplified theoretical model which is given for the temperature dependence of the electron density. When fitting this model to the measured electron density we also find a donor density which is about three times the measured electron density (see table 2). The low electrical mobility of the undoped sample can be explained by grainboundary scattering. The fact that grain-boundary scattering does not occur in the doped samples, though the microstructure is the same for all samples, can be explained when we consider the fact that the Fermi energy, which is proportional to n2_/3, is much higher for the doped samples. So the free electrons in the doped samples have a much higher energy and will more easily skip the inter-grain energy barriers. An other possibility is that the grain-boundaries are passivated by the fluorine when the coating is doped.
9. Conclmions The combined determination of optical constants, electrical properties, structural properties and elemental composition gives a lot of information on the physics of doped tindioxide coatings. For lightly doped coatings the electron mobility is mainly limited by ionized impurity scattering with a slight lattice scattering component. The fluorine atoms occupy oxygen positions in the coating, but the resulting electron density is about one third of the fluorine concentration. The preferential orientation in the (200) direction and the grain size of - 25 nm are independent of the doping level.
Acknowledgements The authors wish to thank the TPD (Institute of Applied Physics TNO-TU) for the use of their spectrophotometers. E. Sonneveld of the TPD and the Solid State Group are acknowledged for carrying out the X-ray diffraction measurements. D. Schalkoord of the metallurgy department is acknowledged for carrying out the electron microprobe analysis. E.C. Boslooper, G. de Jong and J. Simons are acknowledged for carrying out various measurements and calculations. This investigation in the program of the Foundation for Fundamental Research on Matter (FOM) has been supported (in part) by the Netherlands Technology Foundation (STW). References [1] [2] [3] [4] [5]
H. Haitjema and J. Erich, Solar Energy Mater. 16 (1987) 79. F. Simonis, A.J. Faber and C.J. Hoogendoorn, J. Solar Energy Eng. 109 (1987) 22. F. S!,u~c~s, M. van der Ley and C.J. Hoogendoorn, Solar Energy Mater. 1 (1978) 221. T. Karlsson, A. Roos and C.J. Ribbing. Solar Energy Mater. 11 (1985) 469. M. Fantini and I. Torfiani, Thin Solid Films 138 (1986) 255.
H. Haitjema et al. / F-doped pyrolytically sprayed SnOz coatings -."
297
[6] J.-C. Manifacier, L. Szepessy, J.F. Bresse, M. Perotin and R. Stuck, Mater. Res. Bull. 14 (1979) 109. [7] E. Shanthi, A. Banerjee and K.L. Chopra, Thin Solid Films 88 (1982) 93. [8] L.J. van der Pauw, Philips Res. Rept. 13 (1958) 1. [9] P.O. Nilsson, Appl. Opt. 7 (1968) 435. [I0] E. Gerlach and P. Grosse, Festk6rperprobleme 17 (1977) 157. [11] I. Hamberg and C.G. Granqvist, J. AppL Phys. 60 (1986) R123. [12] G. Frank, E. Kauer, H. K6stlin and F.J. Schmitte, Solar Energy Mater. 8 (19183) 387. [13] E.H. Putley, The Hall-effect and Semiconductor Physics (Dover, New York, 1968). [14] J.S. Blakemore, Semiconductor Statistics (Pergamon Press, Oxford, 1962). [15] G.T. Maeden, Electrical Resistance of Metals (Plenum Press, New York, 1965). [16] T. Stapinski, E. Leja and T. Pisarkiewicz, J. Phys. D 17 (1984) 407. [17] A.P. Roth and D.F. Williams, J. Appl. Phys. 52 (1981) 6685. [18] H.J. van Daal, $. Appl. Phys. 39 (1968) 4467. [191 M.N. klam and M.O. Hakim, J. Phys. D 9 (1986) 615.
