IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. ED-24, NO. 4, APRIL 1977 463

Intensity Effects in SnOZ-Si Heterojunction Solar Cells

WILLIAM G. THOMPSON, STEPHEN L. FRANZ, RICHARD L. ANDERSON, FELLOW, IEEE, AND OLIVER H. WINN, FELLOW, IEEE

Abstract-The saturated photocurrent of SnOrn-Si hetero- junction solar cells is found to be linear with illumination intensity up to 30 suns of simulated AM1 irradiation. An interfacial layer at the SnOrSi junction causes the open-circuit voltage to saturate at a v,alue of light intensity determined by the thickness of this layer. The resistance of this interfacial layer limits the fill factor of these devices. It is shown that the open-circuit voltage depends on an internal diode quality factor rather than on that measured from t.he terminal I-V characteristics. The internal diode quality factor can be determined from measurements of open-circuit voltage and saturated photocurrent, with illumination intensity as a pnrameter.

H I. INTRODUCTION

ETEROJUNCTION solar cells (HJSC’s) of SnO2-Si have been reported to have AM1 conver-

sion efficiencies on the order of 10 percent [l]. A high re- sistarlce at the SnOYSi interface has been attributed to the presence of a semi-insulating interface (I) layer thought to be, at least in part, Si02 resulting from a reduction of the tin oxide by the Si. The increasing series resistance asso- ciated with the slow degradation of the SnO2-Si HJSC’s at elevated temperatures is believed to result from growth of this I layer with time 121, [3]. The larger series resistance of HJSC’s using polycrystalline Si substrates, compared to those using monocrystalline substrates made with similar processing, is thought to result from a thicker I layer in the polycrystalline-based cells [4]. The presence of this I layer is also thought to result in a trapping and recombination of photogenerated carriers in the Si via interface states and a reduction in the collection efficiency [5]. Metal-insulator-semiconductor (MIS) Schottky barrier devices have recently been studied both theoreti- cally (and experimentally [6]-[9]. The purpose of the thin insulating layer is to increase the open-circuit voltage V,, by increasing the diode quality factor, and by reducing the (extrapolated) dark saturation current IO.

Since an insulating layer in a heterojunction device is expected to behave much like one in an MIS structure, SnOZ-n-Si HJSC’s with insulating interface layers were investigated to determine the effects of this layer on con- version efficiency. Because it is thought that the I layer may preclude these cells from use with concentration, the characteristics of the cells with intensity were investi- gated.

This work was supported in part by the National Science Foundation/ Manuscript received September 20,1976; revised November 23,1976.

Research Applied to National Needs. W. C:. Thompson, S. L. Franz, and R. L. Anderson are with the De-

partmmt of Electrical Engineering, Syracuse University, Syracuse, NY . A

I J Z l U . 0. H. Winn is with Atlantic Community College, Mays Landing, NJ

08330.

11. EXPERIMENT The SnOrn-Si HJSC’s discussed here were made by

vacuum deposition of SnO2 onto Si substrates followed by a short heat-treatment [l].

Quartz-halogen (ELH) lamps were used for simulation of the AM1 spectrum, and illumination intensity S (sun) was varied by adjusting the distance (on an optical bench) between lamps and cells.

The illumination intensity at any position was deter- mined by measuring the saturated photocurrent1 Ip sat of a calibrated EG&G UV 100 Si photodiode, hereafter re- ferred to as the reference cell. The reference cell was cali- brated for the ELH illumination spectrum using an HP Model 8330A Radiant Flux meter with an HP Model 8334A thermopile detector having a fused quartz window. This system had a flat (f3-percent) response from 0.3 to 3.0 pm and was calibrated up to an illumination intensity of 100 mW/cm2.

Below 1 sun (100 mW/cm2) the saturated photocurrent of the reference cell was calibrated directly against the reading of the Flux Meter. In this range Ip sat was found to be linear with S.

For S > 1 the reference cell was calibrated in the fol- lowing manner:

1) The reference cell was located at a distance from the optical source such that S = 1 giving a saturated photo- current Ih+Lb2

2) An iris was located between source and cell and ad- justed such that the saturated photocurrent was reduced by a factor B or

l(/iat = -&.E?! 1 ( 4

P corresponding to S ( B ) = UP.

3) The cell source distance was then reduced until

IL‘iat = ILA2at sat. ( B )

4) The iris was then removed from the optical path re- sulting in

S ( D ) = P and Ip sat was measured for this value of S. The above process was then repeated.

At S > 1, to minimize heating effects of .the irradiation, the reference cell was forced-air cooled, and the irradiation

the region (usually a t zero or reverse bias) where photocurrent is inde- Saturated photocurrent is defined as the photocurrent measured in

pendent of ‘unction voltage. Here I$kt, Z(Bkt, I$&, and S(D) refer to the quantities associated with

steps I, 2,3, and 4, respectwely.

464 IEEE TRANSACTIONS ON ELECTRON DEVICES, APRIL 1977

2- r c -I

-1.5

-0.2 -0. I 0 0.1 0.2 0.3 0.4 V, ( V O L T S )

Fig. 1. Dark and illuminated (simulated AM1) terminal I-V charac- teristics of three representative SnOr-Si HJSC's.

was blocked by a shutter which was opened (-1 s) during measurement.

It was found that for the reference cell, Ip sat is propcjlr- tional to S up to an irradiation intensity of 50 suns. Abwe this value the uniformity of illumination was inadequthe to make reliable readings.'

Using the saturated photocurrent of the reference c:~hll as a measure of the illumination intensity, the electri1:;sl characteristics of the test cells were measured with S 8.83 a parameter. The illumination intensity was varied by d - justing the distance between lamps and cells. To ins1.1!11-e uniform illumination, particularly at higher intensiti~.s, only small cells (5 mm2) were used in this study. These ct'lls were mounted on TO-5 headers and encapsulated in ei I-y nitrogen with a hermetically sealed cap with a clear glmss window.

A t low values of S, the illuminated I-V characterist1,c:s were obtained by applying a ramp voltage to the test ~;~:ll in series with a resistor. The voltage across the cell was applied to the X-input of an X-Y recorder while the voltage across the resistor (a measure of the cell currerrlt) was applied to the Y input terminals. To minimize heatirlig at large S, as for calibration of the reference cell, the trt33t cells were forced-air cooled and the incident light u+;m pulsed with a short duty cycle. Data was obtained poiruit- by-point in this case.

111. EXPERIMENTAL RESULTS As indicated in Section 11, the devices in this study ti 1.e

all encapsulated and of 5-mm2 area. Details of three cells are presented. These devices represent a range of device characteristics.

The room temperature dark and AM1 I-V charactlf:.r- istics of three representative SnO2-n-Si cells are shown in Fig. 1. These cells have comparable values of saturat'l?d

".-. 0.0 I 0. I I I O too

8 (sons)

Fig. 2. Saturated photocurrent as a function of the number of suns of simulated AMI irradiation for a SnOz-Si HJSC.

-0.4 -0.2 0 0.2 0.4 V (VOLTS)

Fig. 3. Illuminated I-V, characteristics with illumination intensity as a parameter for unit Z88.

photocurrent Ip sat. Cell 288 has a conversion efficiency of 5.3 percent, an open-circuit voltage Vo, of 0.42 V, and a fill factor Ff of 0.60. Devices Z18 and ZOO are of markedly poorer quality as can be seen from Fig. 1. An inflection in the 4th quadrant of the illuminated I-V characteristics for unit ZOO is evident.

The saturated photocurrent for all cells tested was found to be linear with light intensity up to 30 suns of simulated AM1 irradiation. This is illustrated in Fig. 2 for unit 218.

Fig. 3 shows the I-V characteristics of cell 288 with il- lumination intensity as a parameter.. Normalized short-

et al.: INTENSITY EFFECTS IN SOLAR CELLS 465

\ zoo ‘.,

O.?

-0.7 - - 0.6 c

: - 0

-0.5

-0.4

- 0.3

- 0.2

o L + - - J 1 0 S(suns) 2 3 4

Fig. 4. Normalized short-circuit current and fill factor as functions of illumination intensity for three representative SnOz-Si HJSC’s.

10 - 4 E

- I - U

0 z a G . I D

H

.01 0 Id vs. v,

.I .2 .3 .4 .5 V,, A N D V, (VOLTS)

Fig. 5. Terminal dark k v t characteristics and I p sat-Voc character- istics of two SnOz-Si HJSC’s. Values for terminal and junction quality factors mt and mi, respectively, are indicated.

circuit current Isc/Ip sat, and fill factor are shown in Fig. 4 as functions of S for the three representative units. The short-circuit current is linear with S at the lower light levels but becomes sublinear at higher illumination in- tensity. The fill factor decreases with s. It is noted that at higher illumination, Ff of cell ZOO is less than 0.25-that obtained from a straight line between V,, and I,,, i.e. the limiting case imposed by linear series resistance.

In Fig. 5 are shown semilog plots of the experimental dark current-terminal voltage (Id-Vt) characteristics for two SnOz-n-Si HJSC’s. Also shown in Fig. 5 are plots of the I p sat-Voc characteristics of these cells using S as a parameter. Since Ip sat is proportioned to S , these latter plots also represent the V,,-S relations.

Also indicated in Fig. 5 are values for the terminal quality factor mt and the internal, or junction quality factor mj as discussed in the following section.

It is seen that at low voltages for any device the two plots are indlistinguishable with mt = mi. A t higher voltages mt increases while mj approaches unity. Above about 1.5 suns

Fig. 6. Equivalent circuit of HJSC.

in unit ZOO, mi becomes less than unity, or V,, tends toward saturation. No saturation of V,, is observed for device 288 up to 30 suns.

Iv. DISCUSSION OF RESULTS The equivalent circuit of Fig. 6 can be used to represent

a solar cell. The series resistance is represented by Rs, j represents the intrinsic junction, and the photocurrent, which depends on Vj and S, is represented by Ip.

The applied or terminal voltage is given by

The dark current in such a cell can be written in terms of the terminal voltage [lo].

or in terms of junction voltage

where mt and mj represent the terminal and junction quality factors, respectively. These quality factors can be obtained from (2a) and (2b) at any point:

Vj is not directly measurable, but mj can also be obtained from the Voc-Ip sat characteristic as shown in the appen- dix:

For small IRs products, (2a) and (2%) approximate each other, and mt approximates mj. At larger .IRs products, mt > mj. If Rs is known, Vi and thus mj can be calculated from the measured terminal characteristics. If Rs is linear (independent of current) it can be approximated from the slope of the Id-Vt characteristics at high currents. If it is nonlinear, however, as is the case for solar cells [ll], this procedure is not valid.

466 IEEE TRANSACTIONS ON ELECTRON DEVICES, APRIL 1977

I

a E H -

0

288 _.^ I

I f

zoo

0.011 ' I I 0.01 0. I I

VI (VOLTS) Fig. 7. Z-V characteristics for the series resistance of three cells. I'he

current can be expressed as I a V n with n increasing with decrea ' J IIng cell quality.

We interpret the tendency toward saturation of V,, 'ly9nj < 1) of unit ZOO in Fig. 5 as resulting from photocuramt suppression [5]-i.e. Ip(V,,) < I p sat (see Appendix). #:In unit 288 photocurrent suppression is not observed UI: I to 30 suns.

The value of mj = 1 is thought to result from Schot 1; ky emission over the potential barrier. A t lower voltages,, jnj varies from 1.2 to 1.4 and increases as V,, increases. This is as expected for recombination of carriers within ];he transition region via traps [l], [12]. The value of mj =: 2.8 in unit 288 is thought to result from a multiple step tun- neling process.

The series resistance of these SnOpSi HJSC's is 1!9ri- marily attributed to a semi-insulating interfacial (I) layer (presumably SiO2) at the SnOFSi interface [2], [3]. I:t is noted that the resistance of the interfacial layer is effeci ive in increasing the external diode quality factor mt but :1:1ot the internal or junction diode quality factor mj in t h s e devices. Since it is mj which determines V,, increasing: Int with the corresponding decrease in fill factor can only re- duce the conversion efficiency.

The voltage drop VR (in the dark) across the series ]'e- sistance at any given current is equal to the difference between terminal voltage and junction voltage at t lhat current, as illustrated in Fig. 5 for device ZOO at I = 0.3 al81A. Log-log plots of the Id-vR relations of three cells ijre shown in Fig. 7.

Except at small currents where small differences it:le- tween Vt and Vj result in unreliable data, the current 1 1 : m

be approximated as

I = const V" 4)

where n increases with decreasing cell quality. Such n ~ m - linear resistances have been reported for conductiolrl in amorphous materials [ 131, [ 141.

The exponent n in (4) has been observed to increase on SnO2-Si HJSC's during degradation. Increased values of n are attributed to increases in the thickness of the inter- face layer [2], [3]. We attribute the poor photovoltaic characteristics of cells ZOO and 218 to an excessively thick interfacial layer. The resistance of this layer decreases Ff and I,, with increasing illumination intensity (Fig. 4). It is thought that the presence of fixed positive charge at the Si surface (as is repoFted at SiOfii interfaces) reduces the surface potential, or band-bending in the Si with increasing interfacial layer thickness.

The inflection in the illuminated I-V characteristics for unit ZOO of Fig. 1 results from the magnitude and high degree of nonlinearity of Rs (n = 1.9) of this device.

Although only data of SnOYn-Si HJSC's are presented here, similar results are obtained for InaOs-p-Si cells.

V. CONCLUSIONS 1) The saturated photocurrent in SnO2-n-Si HJSC's

is linear with illumination intensity up to 30 suns. 2) The presence of a semi-insulating (I) layer does not

increase V,, by increasing the diode quality factor. Al- though the external diode quality factor mt does increase, the internal diode quality factor mj, which is related to V,, is not affected by the I layer. The I layer along with fixed interface charge may, however, influence V,, by variation of the barrier height in the Si.

3) Both fill factor and open-circuit voltage are reduced in SnO2-n-Si HJSC's with increasing thickness of the in- terfacial layer; the former from increased resistance and the latter from the presence of fixed positive charge in the I layer, resulting in a smaller surface potential in the Si. 4) The I-V characteristics of the interfacial layer can

be obtained from a comparison of the dark I-V charac- teristics with the Ip sat-Voc characteristics. The resistance of this I layer is current-dependent.

5) The open-circuit voltage is proportional to the log- arithm of illumination intensity at low intensities. Above a critical intensity, V,, tends toward saturation.

6) The active device characteristics and the internal diode quality factor mi can be determined, in the absence of photocurrent suppression, from a plot of saturated photocurrent as a function of open-circuit voltage with illumination intensity as a parameter. Since it is mj rather than the external diode quality factor which determines V,,, a plot of Voc-Zp sat can easily determine the effec- tiveness of I layers in increasing V , by increasing mi. This is expected to apply also to MIS devices.

7) The decrease in conversion efficiency with concen- tration of these cells is believed to result from the high resistance associated with the interfacial layer. From the Ip ,,t-V, characteristics, it appears that if this resistance can be reduced sufficiently, these cells will be capable of operating at concentration up to 30 suns.

APPENDIX Provided the illumination does not affect the Z-Vj

characteristics of those carriers responsible for the dark

THOMPSON et al.: INTENSITY EFFECTS IN SOLAR CELLS 467

current across the junction region, the cell current can be expressed as a linear combination of dark current and photocurrent:

I := I , [ exp -- ‘[” -IRs1 - 1 - Ip(Vt - I&) (Al) mjk T 1

for

and

or

In the case of open circuit, Id = I p and Vi = V,,, and (A2) lbecomes equivalent to (2b). In the absence of pho- tocurrent suppression, I p ( Voc) = I p sat and the I p sat-Voc and the Id-vj characteristics are identical. The true dark juncti.on Id-vj characteristics, including mi, can then be determined, in the absence of photocurrent suppression, from the I p sat-Voc relation with illumination intensity as a parameter.

ACKNOWLEDGMENT

The authors wish to thank G. Kent for many helpful discussions and T. R. Nash for help with measurements.

R:EFERENCES [l] Stephen Franz, Gorden Kent, and Richard L. Anderson, “Hetero-

ference, Salt Lake City, UT, June 23-26,1976, to be published in Journal of Electronic Material.

[2] T. R. Nash and R. L. Anderson, “Degradation of SnOz/Si hetero- junction solar cells,” presented at the IEEE 12th Photovoltaic Specialists Conference, Baton Rouge, LA, Nov. 15-18,1976, to be published in Conference Proceedings.

[3] --, “Accelerated life tests on SnOz/Si heterojunction solar cells,” IEEE Trans. Electron Devices, this issue, pp. 000-000.

[4] S. L. Franz et al., “A comparison of InzOa/Si and SnOz/Si hetero- junction solar cells with polycrystalline and monocrystalline sub- strates,” presented at the National Workshop on Low Cost Poly- crystalline Silicon Solar Cells, Dallas, TX, May 18-19, 1976, to be published.

[5] R. L. Anderson, “Photocurrent suppression in heterojunction solar cell$,” Appl. Phys. Lett., vol. 27, pp. 691-693, 1975.

[6] R. J. Stirn and Yea-Chuan M. Yeh, “A 15% efficient antireflec- tion-coated metal-oxide-semiconductor solar cell,” Appl. Phys. Lett., vol. 27, pp. 95-98, 1975.

[7] S. J. Fonash, “The role of the interfacial layer in metal-semicon- ductor solar cells,” J. Appl. Phys., vol. 46, pp. 1286-1289, 1975.

(81 E. J. Charlson and J. C. Lien, “An A1 p-silicon MOS photovoltaic cell,” J. Appl. Phys., vol. 46, pp. 3982-3987,1975.

[9] D. L. Pulfrey, “Barrier height enhancement in p-silicon MIS solar cells,” IEEE Trans. Electron Devices., vol. ED-23, pp. 587-589, 1 Q7G

[lo] See, for example, R. J. Stirn, “P/N junctions, Schottky barriers and solar cell characterization,” presented at the National Workshop on Low Cost Polycrystalline Silicon Solar cellsi, Dallas, TX, May 18-19,1976, to be published.

[ l l ] R. L. Anderson, “On the determination off series resistance and diode quality factor of solar cells,” presented at the Second ERDA/NASA Workshop on Terrestrial Photovoltaic Measurement Procedures, Baton Rouge, LA, Nov.~lO-12, 1976, to be published.

[12] Richard Williams, “The effect of barrier recombination on pro- duction of hot electrons in a metal by forward bias injection in a Schottky diode,” RCA Review, vol. 30, pp. 306-313, 1969.

[13] B. Dunn and J. D. MacKenzie, “Transport properties of glass-silicon heterojunctions,” J. Appl. Phys., vol. 4’7, pp. 1010-1014, 1976.

[14] 0. H. Winn, R. L. Anderson, and A. Lopez-Otero, “SnOZ/Vz05: P205/Si heterojunctions,” XI11 International Conference on the Physics of Semiconductors, Rome, Aug-Sept. 1976, to be pub-

-.

junction solar cells of SnOz/Si,” presented at the Symposium on Solar Energy Conversion of the 1976 Electronics Materials Con- lished.