INCOHERENT-LIGHT ANNEALING OF PHOSPHORUS IMPLANTED SILICON, WITH SOLAR CELL PRODUCTION IN VIEW

HAL Id: jpa-00223142

https://hal.archives-ouvertes.fr/jpa-00223142

Submitted on 1 Jan 1983

HAL

is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire

destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

INCOHERENT-LIGHT ANNEALING OF PHOSPHORUS IMPLANTED SILICON, WITH

SOLAR CELL PRODUCTION IN VIEW

L. Nielsen, A. Larsen, V. Borisenko

To cite this version:

L. Nielsen, A. Larsen, V. Borisenko. INCOHERENT-LIGHT ANNEALING OF PHOSPHORUS

IMPLANTED SILICON, WITH SOLAR CELL PRODUCTION IN VIEW. Journal de Physique Col-

loques, 1983, 44 (C5), pp.C5-381-C5-385. �10.1051/jphyscol:1983556�. �jpa-00223142�

JOURNAL DE PHYSIQUE

Colloque C5, supplement au nDIO, Tome 44, octobre 1983 page C5-381

INCOHERENT-LIGHT ANNEALING OF PHOSPHORUS IMPLANTED SILICON, WITH SOLAR CELL PRODUCTION IN VIEW

L.D. Nielsen, A.N. ~arsen' and V.E. Borisenko

+*

Physics Laboratory 111, TeehnicaZ University o f Denmark, 9K-2800 Lyngby, Denmark

I n s t i t u t e o f Physics, University of Aarhus, DK-8000 ark us C, Denmark

~ e s u m 6 - Des illuminations de courte dure'e d'une lampe 2 xenon ont Bte' utilisees pour supprimer le domage dG 2 l'implantation ionique dans du silicium monocri- stallin d'orientation <loo> et polycristallin implante avec du phosphore. L'in- fluence des paramittres d'implantation (18-39 kev, 10'~-10'~ ions/cm2) et de re- cuit (950-1100~~ pour 10 sec) sur la qualit6 du silicium dope a 6t6 BtudiQe par des mesures de resistance superficielle et par canalisation. Cellules so- laires realisees 2 partir des echantillons implantees et recuits des deux ma- teriaux ont Qt6 caract&risGes par leurs rsponses spectrales et leurs caract6- ristiques courant-tension en lumisre AM1. Les param&zres.optimaux de realisation des cellules sont discutees.

Abstract - Short-duration incoherent xenon light exposure has been used to re- move the ion implantation damage in phosphorus implanted <loo> single crystal and polycrystalline silicon. The influence of implantation parameters 118-39 keV, 10'~-10'~ ions/cm2) and annealing parameters (950-1000°~ for 10 sec) on the qua- lity of the doped layers has been investigated by measurements of sheet resisti- vity and by ion channeling experiments. Solar cells made from implanted and an- nealed samples of both materials have been characterized by spectral response measurements and recording of AM1 current-voltage characteristics. Optimal pro- cess parameters are discussed.

I - INTRODUCTION

For several reasons, ion implantation is considered to be an attractive technique for production of pn-junction silicon solar cells. Implanted doses and depth pro- files can be a accurately controlled, and very shallow junctions can be obtained, leading to cells with an enhanced violet response. Dopant concentrations are not necessarily limited by solid solubility considerations, and low sheet resistivities may thus be obtained in very thin top layers. Finally, advantages are obtained in connection with polycrystalline materials, where standard diffusion techniques may be susceptible to the influence of grain boundaries. Large scale production of ion implanted solar cells may be envisaged with efficient high-current implantation systems, possibly without mass separation of the ion beamlt2).

Removal of the implantation damage may be accomplished by traditional furnace anneal- ing or by one of the various beam annealing techniques (electron beam, CW or pulsed laser) proposed during the recent years3). The present gaper is concerned with the application of short-duration incoherent-light exposure ) as a means of annealing phosphorus implanted single- or poly-crystalline silicon solar cells. The described method is believed to be a fast, cheap, and efficient way of annealing, and further- more well-suited for automization and high-volume production.

This work was supported by the Commission of the European Communities under contract ESC-R-020-DK (G)

.

* ) On leave of absence from Minsk Radioengineering Institute,

P. Browka 6, 220069 Minsk, USSR.

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1983556

JOURNAL DE PHYSIQUE

I1

-

The present investigation has comprised two different silicon materials, both p-type with a resistivity around 5 k m . As a single crystal material was chosen standard one-side polished <loo> wafers, and as a potentially low-cost polycrystalline mate- rial was taken the Wacker-Chemitronic SILSO material (square slices of cast ingot).

Cutting damage was removed from the latter material by etching prior to implantation.

Samples were cut as 15x15 m m 2 squares and implanted either within a 10x10 mm2 area (earlier batches) or over the entire surface (later batches). It should be emphasized that the sample size was limited only by the capability of our present implantation and annealing equipment and that scaling-up to large areas is technically feasible.

Phosphorus implantations (p3') were carried out at a typical ion current density of 5 !&'cm2. Single crystal samples were tilted with the <loo> axis 7O away from the beam direction to avoid channeling. Table I summarizes the applied implantation para- meters, including mean range, Rp, and straggling, A R ~ , and shows the calculated peak concentrations, N 5) For comparison, the solid solubility of P in Si is about 3 . ~ 1 0 ~ ~ c m - ~ at 9?8% 6j.

Table I

-

Incoherent-light annealing was carried out by means of a short-arc xenon lamp placed in one focal point of an ellipsoidal reflector. Samples were mounted in free air on a thermally isolating support close to the opposite focal point, where the tempera- ture could be monitored by a thermocouple. A further description of the apparatus is given in ref. 7. A rather short pulse duration of 10 sec (controlled by a mecha- nical shutter) was chosen as a standard for all the present annealings, in order to minimize redistribution effects. Equilibrium temperatures of 950, 1000, 1050, or llOOOC were attained within the first 5 sec of the pulse, and the cooling after the end of the pulse had a very fast initial rate.

10' 6cm-2

2 . 0 ~ 1 0 ~ ~ c m - ~

Solar cell samples had thelr back surfaces contacted by evaporated aluminium, fur- nace sintered at 500°c. A few samples were selected for experiments with built-in back surface fields (BSF) and were heated to 800°c (eutectic alloying) to form a pfp-structure. Front surface contact grids on the implanted nf-layers were establish- ed by evaporation of aluminium through metal masks without subsequent sintering, in order to avoid burn-through of the shallow pn-junctions. No antireflective coat- ings were applied. The described procedures have made possible a comparative study of implanted and annealed test cells but do not pretend to be optimized for real solar cell production.

5 ~ 1 0 ' ~ c m - ~ 1.8x10~'cm-~

1 . 0 ~ 1 0 ~ ~ c m - ~ Implanted dose:

18keV: R =250I%, AR =llOI%

P P

39keV: R =500A, AR ^=200A

P P

111

-

Figure 1 displays sheet resistivities in single crystals as a function of induced temperature for the different implantation conditions investigated. At the highest temperature ( 1 1 0 0 ~ ~ ) the results fall into three groups, independent of implantation energy: -90, 25, and 15 Q / o , respectively, for the three doses applied. The values compare very well to values reported by Downey et a ~ . ~ ) after infrared rapid isother- mal annealing (12500C, 10 sec), or after furnace annealing (9000~, 30 min) of 60 keV phosphorus implanted into <loo> silicon. It appears from Fig. 1 that the 18 keV im- plantations need high temperature annealing to attain the same sheet resistivities as the 39 keV implantations. This behaviour cannot be due to lack of electrical ac- tivity as an effect of concentrations exceeding solid solubility, since the as-im- planted peak concentration of the 1 0 l ~ c m - ~ implantation at 18 keV only slightly ex- ceeds the solid solubility limit of P in Si at 9 5 0 ~ ~ . It could be due to phosphorus segregation at the moving crystal/amorphous interface during expitaxial regrowth,

resulting in a phosphorus accumulation at the surface. Thls effect would be expected to be most pronounced for the shallow implants, and high temperature would be needed for a redistribution by solid state diffusion. Further experiments are needed to clarify these points. The sheet resistivities of the polycrystalline samples were, within 5%, identical to those o f --- the single crystal samples.

Some of the implanted samples were analyzed by 2.0-MeV ~ e i channeling to study re- crystallization as a function of implantation and anneal conditions. Examples of backscattering spectra are given in Fig. 2. The degree of recrystallization has been measured by Xmin(Si), defined as the ratio between backscattering yields in aligned and random directions, measured immediately below the surface peak. The &in(Si) value of the annealed crystal of Fig. 2 is identical to the value for a virgin cry- stal (Xmin(Si)=0.030), demonstrating perfect recrystallization. Results for 39 keV implantations are collected in Table 11. It appears that for the two lowest doses annealing at llOOOc is necessary to achieve a xmin(Si) -value identical to the virgin value; for the highest dose the virgin value has not been achieved, and higher tem- peratures and/or longer exposure times are probably needed. Whether these slightly high xmin(Si) values observed at low anneal temperature and at high dose are due to dechanneling by residual damage or non-substitutional phosporus atoms cannot be de- duced from these experiments.

0 -

TEMPERATURE PC)

150 200 250

CHANNEL NUMBER

F i g . ¹ - Sheet r e s i s t i v i t i e s a f t e r Fig. 2

-

10 see incoherent-light annealing of spectra of a phosphorus implanted phosphorus implanted <loo> s i l i c o n <ZOO> s i l i c o n single crystal. Implant:

single crystals. 39 ^ke~/5xl0~ 5cm-2. Anneal: ^2100°C/10 see.

IV

-

Spectral response measurements were carried out by means of a microcomputer control- led set-up similar to that described by Shay et al.'). External quantum efficiency, QEXT: and surface reflectivity, REFL, were measured as functions of wavelength, and the lnternal quantum efficiency was calculated as Q ~ ~ ~ = Q ~ ~ ~ / ( ~ - R E F L ) . Due to the non- specular structure of the etched polycrystalline surfaces, only Q E x ~ could be re- corded for the SILSO cells. By integration of QEXT with a tabulated AM1 solar spec-

Implanted dose (39keV) : 950°C, 10 sec

100O0c, 10 sec 1100°~, 10 sec

Table 11

-

lo1

'ern-'

0.034

0.030

5 ~ 1 0 ' ~ c m - ~ 0.035 0.034 0.030

10' 6cm-2 0.036

0.034

C5-384 JOURNAL DE PHYSIQUE

trum over all wavelengths a measure was obtained of the short-circuit current den- sity, ISC, for an incoming power density of PIN=88.92 mw/cm2 lo)

.

V - RESULTS AND DISCUSSION

Figure 3 shows spectral response curves for representative single- and poly-crystal- line solar cells. The ISC-values for the two cells were 19.6 and 16.1 rnZi/cm2, re- spectively, and were typlcal for the two base materials investigated. The difference was almost exclusively arising from different bulk minority-carrier lifetimes, lead- ing to a lower infrared response of the SILSO cells. Spreading on lifetime values within each of the two groups of cells made it difficult to analyze the ISC-values for systematic influence of process parameters. BSF-processed cells appeared to have their ISC increased by 1-2 m~/crn', due to improved infrared response, but the me- chanism might as well be a simple restoring of bulk lifetimes, degraded during the abrupt incoherent-light annealing. ISC-variations from one batch to another were of the same order of magnitude.

WAVELENGTH [ nml

CURRENT DENSITY [ r n ~ lcm21

30 ' ' ' 1 ~ 1 ' 6 . 2 3

0 0 100 200 300 LOO 500 600 VOLTAGE [mVI

- -- -

Fig. 3

-

-

<loo>

Implant: 3 9 k e ~ / 5 ~ 1 0 ' 5ern-2. Implant: 1 8 k e ~ / 5 x 1 0 ~

Anneal: 95U0C/1O see. Anneal: 105U0C/1O see.

The violet response, however, is less sensitive to uncontrollable variations in ma- terial properties, and the dependences on process parameters can be studied in de- tail. For instance, at a wavelength of 400 nm all light is absorbed within less than 1 pm, and the detected response will mainly reflect the properties around the pn- junction. Table I11 summarizes the measured values of QEXT(400nm) for single- as well as poly-crystalline cells. It is clearly seen from Table III(a) that the violet response is favoured by low implantation energy (shallow junction) and low dose

(less damage). Table III(b) indicates that low-temperature annealing is essential to preserve the excellent violet response of the 18 keV implantations, whereas the cells implanted at 39 keV remain with a lower, but almost constant value of Q E x ~

(400nm) up to the highest temperatures.

Table III - V i o l e t response: Values of QmT(400nm) for singZe- and poly-crystalline samples. (a) : 1000'~ anneals. ( b ) : 5x10' 'cm-' i m p Zants

.

10' cm-2

12-19%

Implanted dose:

18keV implantations 39keV implantations

Open-circuit voltages were, with a few exceptions, in the range 510-535 mV for <LOO>

samples. A wider scattering of results occurred for the SILSO cells, but typical re- sults were 470-490 mV (up to 515 mV for BSF cells). Fill factors were generally not very good, and no systematic dependences on material type or process parameters could be observed. Earlier batches, implanted within a 10x10 mm2 square, were affect- ed by a certain photosensitivity outside the implanted area, due to a thin n-type surface inversion layer. Artificially high short-circuit currents, but with large series resistances, were therefore observed under the ELH-lamp illumination, and fill factors were typically 60-65%.

r

Anneal temperature:

18keV implantations 39keV implantations

Improvements were observed after removal of the peripherical areas for some of the

<loo> samples. An example of a resultxng current-voltage characteristic is shown in Fig. 4, leading to FF=70% and r)=9%. A logarithmic plot of the dark-diode character- istic revealed a dominating recombination-current contribution, which seems respons- ible for the limitation of both VOC and FF. It should be noted that the incorporation of a perfect AR-coating and an increase of PIN to 100 mw/cm2 would add about 15 mV to VOC and slightly improve FF.

10' 39-40%

32-33%

1 0 0 0 ~ ~ 30-34%

21-258 950°c

33-39%

24%

Under these assumptions, the power conversion efficiency would be around 14% for the single crystalline cell. Best recorded efficiencies for the uncoated polycrystalline cells were around 7% and would similarly be improved to well beyond 10%. Incoherent- light annealing is therefore believed to be a promising technique in conneTtion with large-area ion implanted pn-junctions for solar cell purposes.

5 ~ 1 0 ' ~ c m - ~ 30-34%

21-25%

REFERENCES

1) MULLER J.C., PONPON J.P., GROB J.J., GROB A., STUCK R., SIFFERT P., Proc. 1st EC Photovoltaic Solar Energy Conf., Luxembourg, 1977, p. 897.

2) NIELsEN L.D., BALSLEV P., Proc. 3rd EC Photovoltaic Solar Energy Conf., Cannes, France, 1980, p. 698.

3) HILL C., in: Laser and Electron-Beam Solid Interactions and Materials PrOCeS- sing, eds. J.F. Gibbons, L.D. Hess, T.W. Sigmon (North-Holland, Amsterdam, 1981) , p. 361.

4) CORRERA L., PEDULLI L., Radiation Effects

63,

5 ) DEARNALEY G., FREEMAN J.H., NELSON R.S., STEPHEN J.: Ion 1mplantation

(North-Holland, Amsterdam, 1973), p. 756.

6) NOBILI D., Proc. 4th EC Photovoltaic Solar Energy Conf., Stresa, Italy, 1982, p. 410.

7) NYLANDSTED LARSEN A., BORISENKO V.E., submitted for publication in Appl. Phys.

8) DOWNEY D.F., RUSSO C.J., WHITE J.T., Solid State Technology, Sept. 1982, p. 87.

9) SHAY J.L., WAGNER S., EPWORTH R.W., BAKCMANN K.J., BUEHLER E., J. Appl. Phys. 48, 4853 (1977).

10) THEKAEKARA M.P., "Data on incident solar energy", Supplement to P ~ o c . 20th Annual Meeting of Inst. for Environmental Science, p. 21 (1974).

1O5O0c 20-31%

25-29%

l l o o O ~ 21-252 21-27%