720 IEEE 'I RANSACTIONS ON ELECTRON DEVICES, VOL. ED-31, NO. 5 , MAY 1984

Structurally Stable, Thin Silicon Solar Cells

A. MEULENBERG, R. A. ARNDT, AND J. U. TYLER

Abstract-A modification, during the wafer-etching process for thin solar cells, provides the thin structure with silicon reinforcing ribs. 'The result is a lightweight solar cell with significantly increased fabrica h~on yield.

Since 1970, we have pursued a goal of solar cells with increased power and powerlweight ratio at end-of-life in a space radiation environment [ 11. The present work complements prior reslllts (e.g., the COMSAT Violet [ 21 and nonreflective [ 31 cells, which increased end-of-life power) by making more practical the ma nu- facture and use of thin (- 50 pm) solar cells with their inher:nt advantages over thick cells for space applications. Since :!he cell mass is much less and since the power output is only fiac- tionally smaller, the power-to-weight ratio is greater. In a(: di- tion, after moderate (- 1015 cm-') levels of 1-MeV elect1 on irradiation, the maximum power of thin cells is often greater than that of thicker cells that are otherwise identical. Thus {lot only is excess weight eliminated by use of thin cells but the ac- tual power available at end-of-life is greater for satellite missims in most radiation environments.

One problem that seems to persist with thin cells is the 1 )w production yield. Even though 50-pm-thick cells are flexikle, they are fragile and subject to breakage in almost all stages of fabrication. Reported fabrication yields are almost nesw greater than 50 percent. The fact that these thin wafers ;re often bowed increases the handling problems during fabrication and array assembly,

We have developed a structure that overcomes many of 1:h.e handling (and, therefore, breakage) problems and which ] . a s exhibited yields on a limited number of cells (- 100) of at leilst 75 percent. The cells resulting from this structure are quite f a t and structurally reinforced to reduce the losses from handli.ng thin cells after fabrication. A schematic cross section of tlus structure is shown in Fig. 1.

The fabrication process as developed by COMSAT Labs is (briefly) as follows. A 6-mil, circular wafer is oxidizedon both sides. One side is then patterned with a rectangular array 3f holes in the oxide that are nominally 7 5 mils square and sel: a- rated by 2-mil spacings. Wells are then etched into the silic1:m with KOH to a depth of 4 mils, leaving a 2-mil unetched thick- ness. Two areas on the surface are left unetched to provide pa I s for bonding or testing. All oxide is then removed and the rest of the processing is normal; the unetched face is used as the 1- luminated face. When all other processing is complete, a 2 X 2 cm cell is sawed from the starting wafer leaving a border th,r: varies between 2 and 10 mils in width. The effective thickness, determined by weighing an unmetallized sample, of such a ccill with a 2-mil border is -2.4 mil (61 pm).

The structure is not sensitive to variations in processing. Sta1:t- ing wafers that are square, instead of circular, can be used 1 0 avoid the final sawing step. However, the wide borders of tkt: circular wafers do provide support during processing and a:'€; convenient handles. Since most of the strength of the fink1 structure is in the ribs, variations in wafer thickness and in tllt: thinning process are less critical than in the case of planar, thin cell processing. Nonreflective front surfaces, different etcil- mask oxides, and different etchants could be used without d,:.

Q $8 ,." // //" - / N - 6 mils

Fig. 1. Perspective view showing one corner of a rib-reinforced cell.

l ( l ~ l ~ l ~ I ~ I ) l

180- - e -

160- - - -

140- - - -

- a - 120-

- g 100- 1 5 80- - w - -

0 - -

60 - - - -

40 - - - -

20 - - - -

'0 ' I d 0 ' 2 ~ 0 1 3 ~ 0 1 4 ~ 0 1 5 d 0 ' 6 b 0 ' VOLTAGE (mv)

Fig. 2. Typical I-V curve for a 1-s2 9 cm 4-cm2 rib-reinforced cell under AM0 illumination at 25°C.

tracting from the structure. Observation of the breakage modes indicate that larger (e.g., 4 x 4 cm or 2 X 6 cm) rib-reinforced cells might be viable as well. Variations in the structure could also be introduced to optimize the cell mass and production yield.

Several cells were fabricated in each of two low-resistivity, p- type wafers; 0.1 and 1 .O C k m . T h e cells had double antireflec- tive (DAR) layers of Taz O5 and Alz 0 3 . The cells were not opti- mized for beginning-of-life performance by the use of tech- niques such as p+ back contacts, passivated surfaces, or dot contacts [ 41 . A typical I-Vcurve of a 1 -C! cm cell with a fused silica coverslide (0.35-pm filter) obtained under AM0 illumina- t ion at 25OC is shown in Fig. 2. The 0.1-C! * cm cells exhibited much the same maximum power, but V,, was - 6 15 mV and I , was rv 142 mA. Cells of both types exhibited average maxi- mum-power outputs of % 66 mW, with a few as high as 68 mW.

Sample cells from each resistivity were irradiated at the Naval Research Laboratory with 1-MeV electrons to fluences of 4 x 1014 cm-' and 9 X 1014 cm-'. The maximum power (AMO, 25°C) for the 1-!J * cm cells. after these fluences. is shown in Fig. 3 by the X's. For comparison, Fig. 3 also shows the out-

paper is based upon work performed at COMSAT Laboratories undcr a 1-MeV electron fluence. The Of o*l-C! cm did not the sponsorship of the Communications Satellite Corp. degrade much ("5 percent) from the 9 X 1014 cm-' fluence,

The authors are with COMSAT Laboratories, Clarksburg, MD 20871, but Isc dropped a great deal and the output power dropped (to

Manuscript received January 4, 1984; revised January 10, 1984. Th s put power of a 2-C! * cm o-mil-thick as a function Of

0018-9383/84/0500-~~'720$01.00 0 1984 IEEE

IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. ED-31, NO. 5 , MAY 1984 72 I

90- 1 1 I 1 1 1 1 1 1 I I I I I I I I I I I l l l l l l l I I I , , I l , tured surface cells (61 mW) [ 5 1 for the same irradiation

60 - conditions, The rib-reinforced cell described here should be a very good

candidate for space use in tha t it is lightweight and produces

percent, or greater) is such that it should be more economical to fabricate than nonreinforced structures of similar thickness.

-

N- 70- - E, :L--- good power output after irradiation. Further, the yield (- 75

; 60; I

2 50- h -

401 30

o J u I ’ I I I I l I I ’ ‘ “ ‘ I l l I 1 ~~~~~~~

L [21

0 1013 1014 1015 10‘6

Fig. 3. Maximum power output of a violet cell (2 s2 cm, 10 mil) as a [31 1 MeV ELECTRON FLUENCE (crn-2)

function of 1-MeV electron fluence under AM0 illumination at 25°C (solid curve). The X’s are data points for a rib-reinforced cell (1 s2 . cm) under the same conditions (dashed curve).

[41

[51 44 mW) commensurately. The output of the 1 -a cm rib-rein- forced cells compares favorably with that of the 2-fJ-cm 10- mil violet cells as well as with that of 50-Mm 10-s2 * cm tex-

REFERENCES

J. F. Allison, R. A. Arndt, and A. Meulenberg, “A comparison of the COMSAT Violet and non-reflective solar cells,”COMSAT Tech. Rev., vol. 5, no. 2, pp. 211-224, Fall 1975. J. Lindmayer and J. Allison, “An improved silicon solar cell-The Violet cell,” in Proc. 9th IEEE Photovoltaic Specialists Conb (Sil- ver Spring, MD), May 2-4, 1972, pp. 83-84. J. G. Haynos et al., “The COMSAT non-reflective silicon solar cell: A second generation improved cell,” in Proc h t . Con5 on Photo- voltaic Power Generation (Hamburg, Germany), Sept. 25-27,

A. Meulenberg, Jr., “Development toward an 18% efficient silicon solar cell,” NASA Contract NAS-3-22217, Final Rep., Apr. 1983. B. E. Auspaugh, et al., “Electrical characteristics of Spectrolab BSF, BSR, textured, 10 a-cm, 50 micron advanced OAST solar cells as a function of intensity temperature, and irradiation,” JPL Pub. 78-15, vol. VI, June 15, 1979.

1974, pp. 487-500.

Contributors

Thomas C. Anthony, time of publication.

phot( and biography not a vaila rble at the

Allen M. Barnett (S164-M’664M’79) was born in Oklahoma City, OK, on June 20, 1940. He received the B.S.E.E. and M.S,. degrees from the University of Illinois,-Urbana,, in 1962 and 1963, respectively. In 1966 he received the Ph.D. degree from Carnegie Institute of Technology, Pittsburgh, PA.

He has been Professor of Electrical Engi- neering at the University of Delaware, Newark, since 1976. He was also Director of the Institute of Energy Conversion at the University of

Delaware from 1976 to September 1979. He was president of Xciton Corporation from 1972 to 1975 and worked for the General Electric Company at the Electronics Laboratory and the Research and Develop- ment Center from 1966 to 1971. He is continuing his work in the planning and implementing of strategies for the transition of new tech- nologies from the laboratory through product development and produc- tion on to the market place. He is currently working on the development of thin-film crystalline silicon, thin-film gallium arsenide, and single crystal 111-V compound solar cells.

Sharleen K. Beckwith received the B.A. degree in art and the Elementary Teaching Credential from San Jose State College, CA, in 1971, and 1972, respectively.

She has over 13 years experience in semicon- ductor processing arud’has been responsible for creating and implementing unique processing techniques for several research and develop- ment projects. She has been at Stanford Uni- versity since 1977 where she is currently a Principal Physical Science and Engineering Technician.