Presented at the 21st European Photovoltaic Solar Energy Conference, September 4-8, 2006, Dresden (2CV.4.26)
`
`PECVD PSG AS A DOPANT SOURCE FOR INDUSTRIAL SOLAR CELLS
`
`J. Benick, J. Rentsch, Ch. Schetter, C. Voyer, D. Biro, R. Preu
`Fraunhofer Institute for Solar Energy Systems (ISE), Heidenhofstrasse 2, D-79110 Freiburg, Germany
`Phone +49-761-4588-5287; Fax +49-761-4588-9250; email: jan.benick@ise.fraunhofer.de
`
`ABSTRACT: Emitter formation by in-line deposition of a PECVD PSG and subsequent diffusion in an in-line belt
`furnace is suggested as an alternative to quartz tube diffusion. The PSG deposition process reached excellent layer
`homogeneities, both across a single wafer and across the whole carrier, resulting in very uniform sheet resistance
`distributions. Relative standard deviations of 2,5 % have been achieved.
`Solar cells with the developed emitter reached efficiencies of up to 17,5 % and 16,8 % on textured and untextured
`Cz-Si respectively.
`Keywords: Deposition, Doping, c-Si,
`
`1
`
`INTRODUCTION
`
`The standard method for forming the solar cell
`emitter which used by the majority of solar cell
`manufacturers, is the diffusion in a batch-like tube
`diffusion furnace. To reduce mechanical stress during
`processing and handling and to obtain high mechanical
`yields, the use of in-line PECVD systems for phosphorus
`deposition with an horizontal transport of wafers is
`favourable. Due to the low impact of forces on wafer
`edges and surfaces, in-line processing is well suited for
`the production of large and very thin wafers [1].
`Besides the PECVD process presented in this paper
`emitter forming has been done by deposition of an
`APCVD PSG and adjacent rapid thermal annealing
`(RTA) [2,3]. Another alternative is the emitter formation
`by a spray-on process [4,5] which is also promising for
`industrial realisation.
`One advantage of PECVD is its flexibility. Single-
`sided and double-sided deposition are possible in one
`process step. Furthermore, the deposition of PSG can
`directly follow an in-line plasma texturing process, also
`in development at Fraunhofer ISE, further reducing
`necessary wafer handling steps [6].
`This paper presents some features of PECVD-
`deposition of PSG as well as solar cell results.
`
`2 DEPOSITION OF DOPANT SOURCE
`
`2.1 Deposition homogeneity
`A large-scale in-line PECVD reactor from Roth &
`Rau AG [6] has been used to deposit PSG as a dopant
`source for forming the emitter of industrial solar cells.
`The plasma excitation was performed by a microwave
`linear plasma source operating at a frequency of 2,45
`GHz. As process gases for the PSG deposition, TMCTS
`(Tetramethylcyclotetrasiloxane),
`TMPi
`(Trimethyl-
`phosphite) and O2 have been used.
`Ellipsometric measurements have been performed in
`order to characterise layer thickness homogeneity. The
`measurements were done on a single wafer positioned in
`the middle of the carrier as well as on wafers occupying
`the whole carrier width.
`A capping layer is deposited on top of the PSG for
`moisture protection because of the strong hygroscopic
`nature of PSG.
`Various deposition parameters such as pressure,
`temperature, microwave
`power,
`gas
`flow
`and
`composition of gases have been examined. As an
`example, Fig. 1 -3 show the parameters that influence
`
`layer uniformity the most: microwave power and process
`pressure.
`
`Fig. 1 Deposition homogeneity across a 50x50 mm2
`FZ Si wafer varying with the microwave power.
`
`Fig. 2 Deposition homogeneity across a 50x50 mm2
`FZ Si wafer varying with the process pressure.
`
`As can be seen, deposition homogeneity increases
`with increasing microwave power as well as with
`decreasing process pressure. Because of the linear
`decrease of microwave power along the microwave
`antenna, high powers are needed to form a uniform
`plasma along the whole length of the microwave antenna.
`Decreasing process pressure enhances the diffusive
`transport of reactive species, thus improving layer
`homogeneity. Moreover, both an increase in microwave
`power and a decrease in process pressure result in an
`enhanced deposition speed. Other parameters such as
`
`HANWHA 1030
`
`1
`
`

`

`Presented at the 21st European Photovoltaic Solar Energy Conference, September 4-8, 2006, Dresden (2CV.4.26)
`
`deposition temperature or O2 content show an opposite
`effect. Increasing deposition temperature as well as
`increasing the O2 content improve deposition uniformity,
`but reduce deposition speed. Deposition speed for
`processes with highest homogeneities is in the range of
`about 2 nm/s. As can been seen, very homogeneous
`layers have been deposited, not only across a single
`wafer (relative standard deviation below 1 %) but also
`across the whole deposition carrier (relative standard
`deviation of about 3 %) whose width is approximately 90
`cm.
`
`Fig. 3
`Layer thickness across the whole deposition
`carrier at varying microwave power.
`
`2.2 Layer composition
`Layer composition has been determined with Energy
`Dispersive X-ray Analysis
`(EDX)
`and X-ray
`Photoelectron Spectroscopy
`(XPS) measurements.
`Because of the organic nature of deposition precursors
`TMCTS and TMP, a significant carbon content in the
`deposited PSG layer could not be excluded, therefor the
`phosphorus as well as the carbon content have been
`determined.
`Since PSG is an electrical insulator, it has to be
`covered with a
`thin carbon
`layer before EDX
`measurement to avoid surface charging. The carbon
`content can therefore not be determined by EDX. XPS
`measurements have been used instead. By sputtering the
`surface, organic contamination can be removed from the
`sample surface prior to measurement. Measurements
`before and after surface sputtering have been performed.
`
`Table I Composition of PSG layer measured by EDX
`technique.
`layer
`
`Fig. 4 XPS spectrum of a PSG layer before and after
`sputtering of the surface.
`
`3 EMITTER FORMATION
`
`The diffusion step, subsequent to the PSG coating,
`has been carried out in an in-line walking string furnace
`[7]. Two processes, one with high temperature and high
`conveyor speed i.e. a short diffusion time (process 1), the
`other with low temperature and low conveyor speed
`(process 2), both yielding sheet resistances of about
`45 (cid:58)/sq, have been used.
`four point probe
`For emitter characterisation,
`measurements of sheet resistance and SIMS doping
`profile measurements have been carried out.
`Sheet resistance shows a strong decrease with
`increasing layer thickness up to 60 nm, proving that
`beyond this thickness, the deposited PSG acts as an
`inexhaustible dopant source (see Fig. 5). The resulting
`lateral sheet resistance distribution is very homogeneous
`(see Fig. 6). With a relative standard deviation over the
`full wafer (125 x 125 mm²) of 2,5 % for a polished Cz
`wafer, the developed process shows results even superior
`to
`the POCl3 reference process (relative standard
`deviation 3,8 %).
`
`Fig. 5 Dependence of sheet resistance on
`layer
`thickness. The diffusion has been performed at low
`temperature and long diffusion time (process 2).
`
`C
`[at%]
`6,5
`
`O
`[at%]
`66,3
`
`Si
`[at%]
`18,5
`
`P
`[at%]
`8,7
`
`PSG
`
`As expected, EDX measurements show a high carbon
`fraction. The proportion of phosphorus in the PSG is
`about 9 %. As can been seen on the XPS spectrum the
`carbon peak
`(labeled with C_1s) vanishes after
`sputtering, indicating insignificant carbon content of the
`deposited layers.
`
`2
`
`

`

`Presented at the 21st European Photovoltaic Solar Energy Conference, September 4-8, 2006, Dresden (2CV.4.26)
`
`RSH: 43,1 (cid:58)/sq (cid:114) 2,5 %
`
`RSH: 44,9 (cid:58)/sq (cid:114) 3,8 %
`
`b)
`a)
`Fig. 6
`Four point probe measurements of sheet
`resistance topography. Compared are the homogeneities
`of a) a wafer processed with the developed plasma
`process, b) a POCl3 reference wafer.
`
`Although all emitters characterised with SIMS have
`similar sheet resistances in the range of 45 (cid:58)/sq, their
`doping profiles are significantly different. The POCl3
`emitter, which serves as a reference, has the deepest
`emitter profile of about 0,5 μm. The developed in-line
`emitters have shallower diffusion profiles with 0,3 μm
`for process 1 and 0,4 μm for process 2. This could easily
`be changed by varying the diffusion parameters of the
`walking string furnace, e.g. conveyor speed (time) and
`temperature.
`
`Fig. 7
`SIMS profiles of FZ Si material. The line
`displayed with green stars represents the diffusion profile
`of the POCl3 reference emitter, the one depicted in red
`triangles emitter 2 and the last one (black squares)
`emitter 1.
`
`4 SOLAR CELL PROCESSING AND RESULTS
`
`Industrial-type solar cells were fabricated on textured
`and untextured 3-6 (cid:58)cm Cz-Si wafers (125x125 mm2).
`They have been processed using the PSG PECVD
`deposition process, screen printing metallisation and fast
`firing through a passivating SiNx-ACR layer. The solar
`cell process scheme is shown in Fig. 8.
`
`
`
`wet chemical etching and cleaningwet chemical etching and cleaning
`
`
`
`deposition of dopant sourcedeposition of dopant source
`
`
`
`emitter diffusion in an in-line furnaceemitter diffusion in an in-line furnace
`
`
`
`sputtered SiN antireflection coatingsputtered SiN antireflection coating
`
`
`
`screen printing of contacts screen printing of contacts
`
`
`
`in-line fast firingin-line fast firing
`
`
`
`edge isolationedge isolation
`
`
`
`characterisationcharacterisation
`
`Fig. 8 Used industrial solar cell process scheme.
`
`two previously described emitter profiles
`The
`(relatively shallow emitter 1 and the deeper emitter 2),
`both yielding sheet resistances of about 45 (cid:58)/sq, have
`been tested. Firing temperatures have been varied from
`840°C to 900°C.
`For all cells, the best results have been reached at the
`firing temperature of 860°C.
`Very good cell results (Table 1) have been achieved
`with this in-line process, leading to efficiencies up to
`17,5 % for alkaline textured and 16,8 % for untextured
`Cz-Si material respectively.
`
`Table II Solar cell results on alkaline textured and
`planar Cz-Si material (thickness 270 μm). Two diffusion
`processes are compared which differ in process time and
`temperature.
`
`emitter 2
`
`emitter 1
`
`emitter 2
`
`emitter 1
`
`best cell
`average
`
`best cell
`average
`
`best cell
`average
`
`best cell
`average
`
`VOC
`JSC
`[mA/cm2]
`[V]
`textured
`619,1
`35,4
`617,3
`35,4
`± 1,0
`± 0,2
`613,4
`35,2
`611,6
`35,1
`± 1,6
`± 0,1
`untextured
`625,4
`33,3
`624,4
`33,1
`± 0,9
`± 0,2
`624 3
`33,4
`623,1
`33,2
`± 1,4
`± 0,2
`
`FF
`[%]
`
`ETA
`[%]
`
`80,1
`80,0
`± 0,5
`78,9
`78,8
`± 0,1
`
`80,6
`80,0
`± 0,3
`79,8
`79,9
`± 0,2
`
`17,5
`17,3
`± 0,1
`17,0
`16,9
`± 0,1
`
`16,8
`16,6
`± 0,2
`16,6
`16,5
`± 0,1
`
`Uniform FF values of about 80 % have been reached,
`proving the ability of the emitter to form very good
`contact with the used screen printing metallisation
`technique. In the case of untextured cells, both emitters
`show almost identical results, while there is a significant
`difference for textured cells. For the shallower emitter 1,
`a drop in fill factor of about 1 % absolute down to
`approximately 79 % can be observed. The same drop can
`be observed in VOC where the voltage decreases from an
`average value of 617 V for emitter 2 down to 611 V for
`emitter 1. So obviously there are difficulties in contacting
`the relatively shallow emitter on textured surfaces.
`
`3
`
`

`

`Presented at the 21st European Photovoltaic Solar Energy Conference, September 4-8, 2006, Dresden (2CV.4.26)
`
`Photovoltaic Energy Conversion, Osaka, Japan
`(2003)
`[2] S. Sivoththaman, J. Horzel, W. Laureys, F. Duerincx,
`P. De Schepper, J. Szlufcik, J. Nijs, R. Mertens,
`Proceedings of the 14th EU PVSEC, Barcelona,
`Spain (1997)
`[3] S. Sivoththaman, W. Laureys, P. De Schepper, J.
`Nijs, R. Mertens, IEEE Electron Device Letters, Vol.
`21, No 6 (2000)
`[4] K. Nakano, A. Hariharan, K. Matthei et al., 12th
`Workshop on Crystalline Silicon Solar Cells
`Materials and Processes (2002)
`[5] C. Voyer, D. Biro, K. Wagner, J. Benick, R. Preu,
`Proc. 20th EU PVSEC, Barcelona, Spain (2005)
`[6] J. Rentsch, N. Kohn, F. Bamberg, K. Roth, S. Peters,
`R. Lüdemann, R. Preu, IEEE PVSC 31, Orlando,
`USA (2005)
`[7] D. Biro, G. Emanuel, R. Preu, G. Willeke, F.
`Schitthelm, G. Wandel, Proceedings of PV in Europe
`- From PV Technology to Energy Solutions, Rome,
`Italy (2002)
`
`IQE measurements of the best cells for both textured
`and untextured surfaces processed with emitter process 1
`and 2 are displayed in Fig. 9.
`
`Fig. 9
`IQE of best solar cells for both textured and
`untextured surfaces processed with emitter processes 1
`and 2.
`
`4 CONCLUSIONS
`
`The developed process for emitter formation, in-line
`deposition of a PECVD PSG as a dopant source with
`subsequent high-temperature treatment in an in-line
`diffusion furnace, is a promising alternative to diffusion
`in a quartz tube diffusion furnace particularly when
`combined with an inline plasma texturing process.
`Deposition of PSG has proven
`to be very
`homogenous across the single wafer as well as across the
`whole deposition carrier. Relative standard deviations of
`about 1 % for single wafer and about 3 % for the whole
`deposition carrier respectively have been reached.
`Analysis of layer composition showed a phosphorus
`content of approximately 9 % and insignificant carbon
`content.
`Resulting sheet resistances have been shown to be
`independent of layer thickness for a PSG thickness above
`approximately 60 nm. Uniformity of sheet resistance
`across a single wafer shows very low relative standard
`deviation of about 2,5 %.
`First results
`in solar cell production are very
`promising yielding efficiencies up to 17,5 % for textured
`and 16,8 % for untextured Cz-Si material.
`
`5 ACKNOWLEDGEMENTS
`
`The author would like to thank Alexander Pohl for wet
`chemical processing and Elisabeth Schäffer for cell
`measurements. Financial support by the German Federal
`Ministry for the Environment, Nature Conservation and
`Reactor Safety (BMU) under contract No. 329933E as
`well as by the companies Roth&Rau AG and Deutsche
`Cell GmbH is gratefully acknowledged.
`
`6 REFERENCES
`
`[1] R. Preu, D. Biro, G. Emanuel, A. Grohe, M.
`Hofmann, D. Huljic, I. Reis, J. Rentsch, E.
`Schneiderlöchner, W. Sparber, W. Wolke, G.
`Willeke, Proceedings of the 3rd World Conference on
`
`4
`
`