HAL Id: jpa-00221789
https://hal.archives-ouvertes.fr/jpa-00221789
Submitted on 1 Jan 1982
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 HAL, est
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.
SEM CHARACTERIZATION OF EFG
POLYCRYSTALLINE SILICON SOLAR CELLS
REALIZED BY ION IMPLANTATION AND LASER
ANNEALING WITH OVERLAPPING PULSES
J. Muller, P. Siffert, C. Ho, J. Hanoka, F. Wald
To cite this version:
CoZZoque supple'ment au nOIO, Tome 43, octobre 1982 page C1-235
SEM CHARACTERIZATION OF EFG POLYCRYSTALLINE S I L I C O N SOLAR C E L L S R E A L I Z E D BY I O N I M P L A N T A T I O N AND LASER ANNEALING W I T H OVERLAPPING PULSES
J . C . Muller, P. Siffert, C.T. Ho*, J . I . Hanoka* a'nd F.V. Wald*
Centre de Recherches NucZBaires,Laboratoire de Physique e t AppZications des Semiconducteurs (PHASE), 67037 Strasbourg Cedex, France
*MobiZ Tyco SoZar Energy Corporation, 16 Hickory Drive, WaZtham, Massachusetts 02254, U. S. A.
R6sm6. On a Btudid par microscopic Blectmnique (SEM) et par EBIC des cellules solaires p&par6es sur du silicium polycristallin en ruban de type
EFG dont la jonction 6tait r8alisi.e par incrustation ( M I ) dtions atomiques et mol6culaires issus dtune source contenant PF
.
La recristallisation du domnage etait effectuke par recuit laser pulse ~ d - Y A ~ .Alors qu'au voisinage de la surface la recombinaison des portews est due essentiellement aux recouvremts des impulsions laser (didtre 100 pm, taux de r4p6tition 10 kHz), on constate q u t a w pmfondeurs plus grandes la recombi- naison aux joints de grains est dominante. La recristallisation par 6pitaxie en phase liquide conserve exactement la structure polycristalline sous-jacente. Abstract. Atomic and molecular ion implantation (AMI) of non mass separated PF5 ions has been used to form the N layer on EM; ribbon.
A
pulsed Nd-YAG laser operating ata
high repetition rate (10 kHz) was then employed to anneal the amorphized layer via liquid phase epitaxy.The properties of the recrystallized layers have been analyzed by SEN w i t h
secondary electron micrography and EBIC. The following results have been observed. At low electron beam energies (below 12 keV), strong EBIC recoinbi- nation currents resulting from the overlapping laser pulses is clearly
visible. At higher energies (above 20 keV), the recombination due to the grain boundaries in the ribbon becomes dominant. The regrowth after annealing repli- cates exactly the underlying structure. Finally, some electrical characteris- tics of the doped layers (doping profile) as well as the photovoltdic
performance are presented. Efficiencies up to 11 %have been obtained under
AMl conditions. 1. Introduction
Ion imlantation, followed by a pulsed annealing treatment in the liquide . .
phase regime induced by either laser or electron be&, is now considered to fulfill the requirements of low-cost silicon solar cell mass production. As
substrate materials, both cast polycristalline bulk and ribbon silicon, especidly FFG ribbon, constitute the most promising candidates. Therefore, it is important to
demonstrate how these large grain sheets, which contain a variety of defects and inhomogeneities, both in the bulk and at the surface, react to the doping perfo-d with the atomic and molecular ion inplantation (AMI) procedure, followed by a high repetition rate (10 kHz) pulsed Nd-YAG laser annealing.
The goal of this study was to investigate by SEM and EBIC the behavior of the doped layer after individual laser pulses (100pm in diameter) have melted the surface. Furthermore, this technique gives information on the surface recombination in the annealed zone, as well as on the quality of regrowth of each grain. For e-ie, an EBIC line scan gives a relative measure of charge collection between
C1-236 JOURNAL DE PHYSIQUE
two adjacent grains. A characteristic property of such polycristalline silicon can be a nonuniform penetration of the dopants during i p l a n t a t i o n due t o channeling conditions, which change with the orientation of each individual grain. The doping profile has been measured by anodic stripping as well a s by SIMS. Finally, the photovoltdic performance of these c e l l s is considered.
2. E3perimental conditions
A. EFG Samples
The ribbon samples were selected fmm material grown i n closed shape form (1).
Some of the samples have been heat treated a t 900°C under pH3 ambient f o r 30 minu- t e s followed by a slow cool down t o 650°C t o improve the minority c a r r i e r lifetime.
B. Junction Formation
A n unanalyzed beam of molecular and atomic ions was extracted from the ion source f i l l e d with PF5 and accelerated t o 10 keV before impinging directly into the sample ( 2 ) . The inplanted ion dose was up to 1017 cm2.
Regrowth of the f i l l y amorphized layer was obtained by using a Nd-YAG laser "Epithermtl-system ( 3 ) , emitting a t 5300
a.
The repetition r a t e of the pulses of t h i s system can be as high a s 10 kHz, the pulse duration being 100 nanoseconds and the beam diameter 0.1 dafter focusing. In the single mode regime, the energy densities up to 3 J/cm can be obtained. Large areas are covered by a micro- processor controlled setup, which allows predetermined pulse overlaps.C. SEN Measurements
A Cambri&e S4-10 SJ94 which was o ~ e r a t e d a t various beam energies from % 5 keV to 30 keV w a s used so that both surfack and bulk effects could be zistinguished. A Keithley 427 current amplifier was employed t o amplify the EBIC signal and feed it back to the SEM video display ( 4 ) . Grain orientation w a s determined using the selected area diffraction mode of the SEM.
D. Solar Cell Processing
After the junction was created, c e l l fabrication was completed with screen printed and alloyed aluminium contacts for the back surface f i e l d (BSF) and vacuum evaporation of Cr-Ti-& for the front contact grid. The metal contact w a s sintered a t 450°C f o r 15 minutes in N2, t h i s heat treatment constituting also the post-laser thermal treatment (5). Finally, the c e l l front surface was covered by an antire- flecting Si3N4 film.
3. Surface analysis S E M MEASUREMENTS
Figure l(a) and (b) represent EBIC
pictures of a periodically pulsed pattern S E M MEASUREMENTS for 10 keV and 30 keV electron energy.
Some well-defined c r a t e r s are visible. which are even enhanced on the second& image of the same zone (figure l ( c ) ) . The usual pulsed pattern has l i t t l e effect on the surface topography, except when craters appear. A t low energies (figure 1 ( a ) ) the periodic pattern dominates the recombination ; at higher energies
(figure l ( b ) ) grain boundaries play the most irqortant role. I1 should be noted that within the craters center, electrical
recombination a c t i v i t i e s can bk seen clearly. a ) Fig. 1 : EBIC and secondary electron
image of the grain boundary structure and pulse pattern for inplanted EFG ribbon sheets annealed with a high repetition r a t e Nd-YAG laser.
a) EBIC lOkeV depth l 5 v m b)EBIC 30keV depth 9 p m c)Secondary ELectron
I m a g e 10 keV
c )
effect of a grain boundary are shown in figure 2. It appears that when a grain boundary is intersected by a laser spot,
the resulting molten area is no longer circular but has a lenticular shape along the grain boundary direction (clearly visible on the secondary image of figure
2 (a). Also, it appears that the EBIC
image shows a lip-like structure with somewhat l e s s recombination along the l i n e of the grain boundary. Finally, note that the centers of the craters show a high degree of recombination.
Fig. 2 : EBIC and secondary electron image given the d e t a i l of the pulsed pattern f o r the same sheet.
grain boundary \
0 ) Secondary Electron Image 10 keV
Some craters
Note the lenticular shape
\
Craters
Lenticular shape wher the pulse intersect a
grain boundary
4. Solar C e l l Evaluation A. Dark I-V Characteristics
-1 -2 -3 -L -5 REVERSE BIAS ( v o l t s ]
Fig. 3 : Reverse dark current of AM1
c e l l s together with a classical diffused c e l l .
Fig.
4 : Evolution of the bulk h f f u s i o n length Ln f o r pH3 trea- ted c e l l s a s compared t o "asgrown" cells. -loo
-
c 1 I 8 0 I- (3 z W A z 60s
cn 3 LL !+ LO 0 Y A 3 m 2 0 0 I I I I - E F G Ribbon - - -pH3 G e t t e r i n g,
0 - A s g r o w n / - +o+ /-
- - 7 - - - - - - Ave -/%
+Q0 /i
- / / Min- * - - -
- / - A M 1 c e l l s ( PFg ,lOkeV) Nd -Yag Laser anneolingI I I I
1olL 1015 1o16 1017
C1-238 JOURNAL DE PHYSIQUE
As shown i n Figure 3, the junction space charge recombination current densities for the ion beam and l a s e r annealed c e l l s are more than one order of magnitude higher than f o r a standard diffused c e l l . However, it should be mentioned
that no chemical handling is necessary f o r junction manufacturing with ion beams.
B. Bulk Diffusion Length L
Just a s for diffbsed c e l l s , the diffusion length Ln i n EFG c e l l s varies with i l l m i n a t i o n (6) f o r AM1 processed structures (figure 4 ) . Gettering treatments im- prove the properties of the material, as is the case for diffused c e l l s .
D. Phosphorus Dopant Profile Both SIMS and anodic stripping
C. Spectral Response
The response of the AM1 and l a s e r $ I I I I I I I
Fig. 5 : Spectral response of a AM1 c e l l and a classical diffused t e s t
.
.
processed c e l l on the short wavelength5
side of the spectrum f a l l s off more 1 0
have shown that the distribution of c e l l .
E F G Ribbon -
phosphorus is nearly exponential, leading t o a junction depth as deep as l p m , both i n EFG and in single-crystal samples. Even if the irrplanted dopants are l e s s than
1OOO h; from the surface a f t e r the AMI process, they are f u l l y redistributed a f t e r laser treatment.
rapidly than that of a diffused con- 2 trol c e l l (figure 5 ) . This may result
"
from several effects, especially f m : -
-
A high surface recombination velo-c i t y due either t o the implantation, -
the laser treatment, o r t o the absence of chemical treatments.
-
-
The very high doping of the surface-
- diffused c e l llayer, due t o the very high solubility
of the active phosphorus, which Nd -Yag Laser annealing -
increases the l i g h t absorption i n the
short wavelength end of the s p e c t m . I I I I I I I
0.4 0.5 0.6 0.7 0.8 0.9 1.0
-
The deep junction, due t o l a s e r WAVELENGTH ( p m )melting (see below)
.
E. Cell Performance
2
The AM1 efficiencies of the M I processed large scale (710 cm ) c e l l s are reported i n Table I. After gettering, efficiencies i n excess of 11 %have been obtained.
T A B L E I
&4 I EFFICImCIES O F E F G P O L S R I S T A L L N E SlLlCaU CELLS O F I I <ma AREA
Several interesting EBIC results appear from this investigation of l a s e r pro- cessed polycystalline silicon :
-
The grain boundary structure a f t e r melting replicates exaclty the underlying grain structure.-
Recombination has disappeared from the grain boundaries from the surface down t o about 2 pm, the depth of phosphorus penetration. This constitutes an intriguing result as it indicates that molecular ion implantation associated with pulse annea- l i n g eliminates grain boundary recombination. There are a t l e a s t two possible explanations f o r t h i s effect : i f inpurities are responsible f o r the recombination a t grain boundaries, they have been f u l l y redistributed a f t e r the short time dura- tion melting. Probably they are now concentrated i n the dark circular areas associated with the periodic pulse patterns shown i n the BIG figures 1 and 7.Another possibilities is that there is a beneficial effect of fluorine on grain boundary dangling bonds o r recombination centers, as it has been suggested t o a c t i n the same wqy as hydrogen.
The evidence supporting the contention of this recombination disappearance comes from the EBIC micrographs s m h as exemplified by figures l ( a ) and l ( b ) . I n ordinary polycrystalline silicon (i .e., not processed by AM1 and recrystallized), grain boundary recombination manifests i t s e l f a s shown i n figure l ( b ) , that is as l i n e s of dark contrast following the trace of the boundary. In figure l ( a ) , where the beam penetration w a s g 2pm, no such lines of dark contrast are seen.
-
Differenceskln grain orientation can have a non-negligible e f f e c t i f the inplan- tation range is'.due t o channeling : In figure 6 we see that a t the edge of two grains A and B, having respectively the orientation <I107 and <124>,
the grain A has a significantly lower current collection than grain B. Figure 7 shows the dif- fraction photograph~~of A, B and the interface. This phenomena can be explained in terms of a more pronounced channeling effect i n t h e < l l O > d i r e c t i o n .S E M MEASUREMENTS
,
grain boundaryEBIC Image ( 5 k e V )
t g r o l n A \ g r a ~ n boundary
,)EBIC LINE SCAN
Z I
W
distance MAGNlFlCATlONx 2000
Fig. 6 : Difference i n current collec- Selected area diffraction on tion between two adjacent grains seen %%&I: given . the orientation of the
JOURNAL DE PHYSIQUE
6. Conclusion
The sharp diffraction images obtained a f t e r laser processing of the AM1 inplanted surfaces indicate that the c r y s t a l l i n i t y of the regrown material is
quite good ; therefore, the presence of multiple regrown zones, corresponding to
the laser used, has not affected the material's quality, a s no amorphous residual zcmes are seen. However, recombination seems t o be high near the center of the annealing spot where the power density of the Gaussian energy distribution is
highest. Substantial microscopic defects still exist after the l a s e r processing, as
indicated both by dark and illwninated performance. However, c e l l aerformance i n excess of 10 % are still possible, even over large areas (
>
10 cm ).7. References
1. A.S. TAYLOR, R. STOWNT, C.C. CHAO a d E.J. HENDEFISON, 15th IEEE Photo- voltdic Specialists Conference Record (New York : IEEE : 1981), p. 589.
2. J.C. MULLER and P. SIFFERT, 1nter.Workshop on Ion Inplantation, Laser Treatment and Ion Beam Analysis of Materials. Bombay (India), Fev. 9-13 (1981),
to be published i n Radiation Effects (1982). 3. QUANTRONIX Fmithtown N.Y. (USA).
4. J.I. HANOKA and R.O. BELL, Ann. Rev. Mat. Sci., Vol. 11 (palo Alto, CA :
Annual Reviews, Inc.
,
1981), p. 353.5. F. 21-1, R. GALUNI, L. PEDULLI, G.G. BENTINI, M. SERVIWRI, F. CEMBALI and A. DESALVO, 2th European Photovoltdc Solar Energy Conf. Berlin (W.Germany)
(1979) p. 213 Ed. D. REIDEL.
