Co-Diffused Back-Contact Back-Junction Silicon Solar ... - IEEE Xplore

5 downloads 2715 Views 4MB Size Report
Sep 18, 2013 - Index Terms—Diffusion processes, printing, solar energy. I. INTRODUCTION ... currents in gap-free BC-BJ solar cells, deeply diffused doped.
1236

IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 3, NO. 4, OCTOBER 2013

Co-Diffused Back-Contact Back-Junction Silicon Solar Cells without Gap Regions Roman Keding, David St¨uwe, Mathias Kamp, Christian Reichel, Andreas Wolf, Robert Woehl, Dietmar Borchert, Holger Reinecke, and Daniel Biro

Abstract—In this paper, first generation back-contact backjunction (BC-BJ) silicon solar cells with cell efficiencies well above η = 20% were fabricated. The process sequence is industrially feasible, requires only one high-temperature step (codiffusion), and relies only on industrially available pattering technologies. The silicon-doping is performed from pre-patterned solid diffusion sources, which allow for the precise adjustment of phosphorus- and boron-doping levels. Based on the investigated process technologies, BC-BJ solar cells with gap and without gap between adjacent n+ - and p+ -doped areas were processed. On the one hand, a strong reduction of the process effort is possible by omitting the gap regions. On the other hand, parasitic tunneling currents through the narrow space charge region may occur. Hence, deep doped areas were realized to avoid tunneling currents in gap-free BC-BJ cells. This paper finishes with a detailed characterization of the manufactured cells including important cell measurements like I–V, SunsVOC, quantum efficiency, and an analysis of the cell specific fill factor losses. Index Terms—Diffusion processes, printing, solar energy.

I. INTRODUCTION HE main difference between back-contact back-junction (BC-BJ) and standard solar cells is that the electrical contacting of silicon, and thus, the extraction of excess carriers is carried out on the rear side of the cell. Hence, shading losses due to a metal grid on the front side are completely avoided, which result in an increased photocurrent and efficiency of the cell [1]. A simplified module assembly of BC-BJ solar cells is possible as well, since the cell to cell wiring by cell connectors is only performed on the rear side.

T

Manuscript received July 8, 2013; accepted July 15, 2013. Date of publication August 15, 2013; date of current version September 18, 2013. R. Keding is with the Fraunhofer Institute for Solar Energy Systems, Freiburg 79100, Germany, and also with the Freiburg Materials Research Center, University of Freiburg, Freiburg 79085, Germany (e-mail: roman.keding@ ise.fraunhofer.de). D. St¨uwe, M. Kamp, C. Reichel, A. Wolf, and D. Biro are with the Fraunhofer Institute for Solar Energy Systems, Freiburg 79110, Germany (e-mail: [email protected]; [email protected]; [email protected]; [email protected]; daniel. [email protected]). D. Borchert is with the Fraunhofer Institute for Solar Energy Systems, Gelsenkirchen 45884, Germany (e-mail: [email protected]). R. Woehl is with Total New Energies, Emeryville, CA 94608 USA (e-mail: [email protected]). H. Reinecke is with the Institute for Microsystems Engineering, University of Freiburg, Freiburg 79104, Germany (e-mail: [email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/JPHOTOV.2013.2274382

However, there are also some challenges. First, a highly efficient BC-BJ solar cell usually exhibits three different doped areas: the so-called front surface field (FSF), the back surface field (BSF), and the emitter. Hence, several diffusion steps with high process temperatures have to be carried out, which lead to an increase in process costs. Second, for an efficient collection of excess carriers, locally doped areas have to be integrated on the rear side. The successful integration of locally doped areas in BC-BJ cells strongly depends on the resolution and accuracy of the used patterning technologies. Unfortunately, the implementation of patterning technologies in the process sequence leads to a further increase in process effort. Third, since excess carriers are generated on the front side and collected at the p-n junction on the rear side, the solar cell performance strongly depends on the front surface passivation, bulk carrier life time, and especially the wafer thickness. In conclusion, the higher efficiency potential of BC-BJ solar cells compared with standard silicon solar cells can only be fully exploited, if the additional process effort and costs of this advanced cell-type are kept as low as possible. This study presents BC-BJ solar cells that are manufactured in an industrially feasible manner. One single high-temperature treatment, the so-called co-diffusion, forms all the necessary doped areas. Therefore, pre-patterned solid diffusion sources are used. A highly precise inkjet process enables patterning of both solid diffusion sources and metallization, allowing for minimal feature sizes far below 100 μm. Aiming at even further reducing the process effort, BC-BJ solar cells are manufactured without gap between adjacent n+ - and p+ -doped areas on the rear side of the cell. Compared with cells with gap, one patterning step can be omitted in this case. To prevent tunneling recombination currents in gap-free BC-BJ solar cells, deeply diffused doped areas were integrated. The presented BC-BJ process flow solely relies on processing steps without mechanical contact to the cell. By aiming at further reduction of the overall material costs, also ultra-thin wafers could be introduced as base material. The first part of this study focuses on the BC-BJ solar cell design and the corresponding process sequence. In the second part, the performance of the manufactured solar cells is discussed. A power loss analysis, which focuses on fill factor losses, is given in the last part of this study. II. SOLAR CELL PROCESS TECHNOLOGY A. Solar Cell Structures and Process Sequences The BC-BJ solar cells were manufactured from n-type Czochralski grown monocrystalline silicon wafers (Cz-Si) with

2156-3381 © 2013 IEEE

KEDING et al.: CO-DIFFUSED BACK-CONTACT BACK-JUNCTION SILICON SOLAR CELLS WITHOUT GAP REGIONS

1237

Fig. 1. In this figure, the process effort for manufacturing a BC-BJ doping structure is compared for successive diffusion and co-diffusion. With the numbering, the number of process-steps for every mentioned process-type is depicted. Successive diffusion means the diffusion from gaseous sources with local SiOx -diffusion barriers. Co-diffusion means the simultaneous diffusion from pre-patterned doped silicate glasses as carried out in this study. By manufacturing a co-diffused BC-BJ cell without gap shallowly diffused (00-cell) instead of a successively diffused cell with gap shallowly diffused (10-cell), 12 process steps can be saved. This effort analysis considers the processes needed for just the assembly of the BC-BJ doping structure. The bordered box on the bottom right gives an alternative designation for the several cell-types.

a base resistivity ρbase of 4 Ωcm. One wafer with an edge to edge length of 12.5 cm features 16 BC-BJ solar cells each with a designated area of 4 cm2 . Silicon ablation during saw damage removal, cleaning, and single-side texturing processes led to a final wafer thickness between 140 and 150 μm. After cleaning, the application of solid diffusion sources, and patterning, the co-diffusion step was carried out. The corresponding patterning steps and the formation of the locally doped areas during co-diffusion are described in detail in the next section. In Fig. 1, the different cell and doping structures are depicted. The figure includes the sum of process steps needed, and hence, an analysis of the process effort for every single doping structure realized by co-diffusion. For comparison also, the process effort of a successively diffused doping structure with gap shallowly diffused (10-cell) is given. Successive diffusion describes the independent diffusion of the three doped areas featuring gaseous diffusion sources in combination with locally opened diffusion barriers. In the following, the process of successive diffusion is explained and compared with the one of co-diffusion carried out in this study. Aiming at FSF formation, the rear side of the cleaned wafer is coated with a SiOx -layer acting as diffusion barrier. After n+ doping in a thermal diffusion process featuring reactant gases, the SiOx layer as well as oxide residuals grown during the diffusion process are wet chemically removed. Persistent residuals are removed in a subsequent cleaning process. Aiming at emitter, and respectively, BSF formation, SiOx layers are deposited

on both sides of the wafer followed by the application of local (rear side) and nonlocal (front side) etching barriers. After etching and dissolving the etching barriers, organic residuals are removed in a subsequent cleaning process. Finally, the local doping process is carried out in a thermal diffusion process featuring reactant gases for n+ - or, respectively, p+ -doping; this process is followed by etching and cleaning processes. Compared with successive diffusion of a doping structure with gap shallowly diffused (10-cell), co-diffused doping structures of the same type feature a lower process sum (six processes less) as well as less expensive thermal diffusion treatments (two processes less). By omitting the gap region, the process sum of co-diffused doping structures can be further decreased (process sum of 14 processes; one diffusion process). The cell featuring a co-diffused doping structure without gap shallowly diffused (00-cell) is, therefore, most attractive for industrial applications. Compared with a co-diffused doping structure shallowly diffused (x0-cell), a deeply diffused doping structure (x1-cell) requires an additional cleaning and thermal diffusion process. Nevertheless, due to the low process sum, cells featuring a codiffused doping structure without gap deeply diffused (01-cell) are of high industrial interest as well. After co-diffusion, the cells were passivated by a plasmaenhanced chemical vapor deposition (PECVD) stack on the front side consisting of a silicon rich siliconoxynitride (SiONx ) film [2] and a silicon nitride antireflection coating (AR-SiNx ). The cell rear side was passivated by a stack consisting of an

IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 3, NO. 4, OCTOBER 2013

charge carrier concentration n - p (1/cm )

1238

Fig. 2. Figure shows the process sequences for the fabrication of the BC-BJ doping-structure without gap (00-cell) and with gap (10-cell) shallowly diffused. The simultaneous doping of silicon to n + -Si (FSF and BSF) and p + -Si (Emitter) is performed by pre-patterned solid diffusion sources. For the fabrication of the cell with gap, an additional deposition and patterning step is necessary.

atomic layer deposition aluminum oxide (AlOx ) film [3] and a PECVD-SiNx . Insulation layers for decoupling the metallization pattern from the doping pattern were not integrated yet [4]. To avoid spiking issues, the cells were metalized with an evaporated aluminum–silicon (Al–Si) alloy featuring a silicon content of 1%. The contact separation was carried out wet chemically featuring a selectively printed etching barrier [5]. The cell process sequence was finished by an annealing step for electrical contact formation and activation of the hydrogen passivation. B. Co-diffusion As already mentioned, the cells manufactured within this study were realized by an industrially feasible co-diffusion process. In Fig. 2, one possible process sequence for the realization of a doping structure with and one without gap shallowly diffused is depicted. In the case of the gap-free cell (00-cell), first the front side of the cell is coated with a phosphorus-doped silicate glass (PSG) as solid diffusion source. Afterwards, the rear side is PSG coated as well. Since the n+ -area (BSF) on the cell rear side usually requires a higher phosphorus-doping level, the phosphorus concentration in the PSG needs to be adapted [6]. To avoid reactions of the PSG with the environment, PSG layers are capped with an undoped silicon oxide (SiOx ) layer. After PSG and SiOx deposition, the PSG–SiOx stack on the rear side is patterned by the precise application of inkjet-printed etching barriers followed by wet chemical etching and cleaning processes [7]. Finally, a boron-doped silicate glass (BSG) as solid boron diffusion source is deposited on top of the patterned PSG–SiOx stack. The sheet resistance of the p+ -doped areas diffused from the BSG can be precisely adjusted by varying, e.g., the gas flows during the deposition process [8]. The SiOx , PSG, and BSG are deposited by means of PECVD processes. For the fabrication of a doping structure with gap shallowly diffused (10-cell), an additional SiOx layer is deposited on the

n -Si (BSF) n -Si (FSF) p -Si (Emitter)

shallowly diffused

10

10

10

61 Ω/sq

556 Ω/sq

48 Ω/sq

10

0.0

0.4

0.8

1.2

148.8 149.2 149.6 150.0

Depth x (µm)

Fig. 3. ECV-measured charge carrier concentration n(x) and p(x) of the BSF, FSF, and emitter shallowly diffused (x0-cell). The silicon surface of the cell rear side corresponds to the position x = 0 μm.

patterned PSG–SiOx stack. Afterwards, the SiOx layer is patterned as well. The additional SiOx layer acts as a diffusion barrier against boron, and hence, defines the gap region. Finally, the BSG layer is deposited on the top of the patterned SiOx layer. Providing that the diffusion sources are in direct contact with silicon, diffusion from dopants in silicon takes place. After diffusion, PSG, BSG, and SiOx layers are removed in a wet chemical etching step. For creating a deeply diffused doping structure (x1-cell), the doped silicate glasses are etched back as well after co-diffusion but followed by an additional high temperature step, which is called the drive-in process. C. Characterization of Locally Doped Areas Diffused by Co-diffusion The dopant concentration profiles of the co-diffused doped areas were determined by electrochemical capacitance voltage (ECV) measurements. Fig. 3 depicts the electron and hole concentration n(x) and p(x) of the doping structures shallowly (x0-cells) and deeply diffused (x1-cells) in dependence of the wafer depth. The measurements were carried out with test samples on the cell wafer. Since no significant amount of inactive dopants is expected, the measured carrier concentration represents the dopant concentration. The boron-doped p+ -Si (emitter) exhibits a maximum dopant concentration of 1 × 1020 L/cm3 . The significant decrease of the dopant concentration near the silicon surface is probably an artifact that results from the measurement. The p-n junction is located at a depth of approximately 700 nm. Since the n+ -doped areas are diffused from exhaustible diffusion sources (thin PSG layers), the phosphorus surface concentration of both, FSF and BSF, is quite low leading to concentration profiles without the typical so-called kink-and-tail shape. Suggesting ideal silicon surface conditions, the dopant profiles provide both a high passivation ability, which leads to a low dark saturation current density (J0,FSF < 65 fA/cm2 ), as well as a high contacting ability of silicon by evaporated Al–Si alloys, which leads to a low specific contact resistivity.

charge carrier concentration n - p (1/cm )

KEDING et al.: CO-DIFFUSED BACK-CONTACT BACK-JUNCTION SILICON SOLAR CELLS WITHOUT GAP REGIONS

n -Si (BSF) n -Si (FSF) p -Si (Emitter)

deeply diffused

10

10 619 Ω/sq 10

44 Ω/sq 66 Ω/sq

10

0.0

0.4

0.8

1.2

148.8 149.2 149.6 150.0

Depth x (µm)

Fig. 4. ECV-measured charge carrier concentration n(x) and p(x) of the BSF, FSF, and emitter deeply diffused (x1-cell).

Fig. 4 shows the carrier concentration profiles of the doping structures deeply diffused (x1-cells). The influence of the drive-in process is clearly visible. For each profile, the surface concentration decreases, whereas the depth of the doped areas increases. Depending on the thermal budget during the drive-in process, the profiles can be adjusted to be even deeper. Deep profiles can be advantageous for avoiding metal spiking issues through the emitter as well as for decreasing the surface recombination velocity. The difference between the sheet resistance of the doped areas shallowly diffused and deeply diffused is probably caused by the homogeneity limit of the PECVD process for PSG and BSG deposition. The dependence of the carrier mobility on the carrier and dopant concentration might be considered as well [9]. III. SOLAR CELL RESULTS Table I lists the cell parameters extracted from the I–V curve (one sun conditions) and the mean value and the standard deviation for each cell parameter and cell-type (four cells measured per cell-type). The cell parameters of the best cell per cell-type are highlighted as well. Comparing the cells with (10-cell) and without (00-cell) gap shallowly diffused, no significant change in the cell performance is detectable. Since the pseudo fill factor (pFF, SunsVOC , [10]) as well as the open-circuit voltage (VOC ) of the cells with or without gap are in the same range, enhanced shunting of the cells without gap due to enhanced recombination through the space charge region is considered to be negligible. Because of the higher emitter coverage (EC) of the cells without gap, a higher minority carrier collection ability, and hence, an increased shortcircuit density (JSC ) can be expected. Unfortunately, this effect can only be observed by comparing the best cells of the respective cell-types. Comparing the cells shallowly diffused (x0-cell) with the cells deeply diffused (x1-cell) independent of the availability of a gap, a strong increase in the cell efficiency is detectable for the cells deeply diffused. To a great extent, the increase in cell efficiency can be attributed to an increase in JSC but also to a slight increase in VOC . This improvement in the cell performance is caused by

1239

the decreased dopant concentration within the doped areas that leads to a lower surface recombination velocity especially on the front side. The pFF values indicate a less pronounced cell shunting due to Al spiking, as well as a lower recombination in the space charge region. Since the space charge region is wider for decreased dopant concentrations as realized by the deeply driven-in doped areas, a lower shunting due to tunneling currents can be expected. Comparing the cells with gap (11-cell) and without gap deeply diffused (01-cell), a significant increase in JSC up to 40.7 mA/cm2 is detectable for the cells without gap. Since the limitation of JSC , caused by the front side recombination, is reduced for the cells deeply diffused, the geometry of the rear side has a relevant impact on JSC . Hence, the higher EC, and respectively, the shortened minority carrier diffusion path of the cells without gap lead to the gain in JSC . The cells without gap deeply diffused (01-cell) include the best cell manufactured within this study featuring an efficiency of 20.2% independently measured by the Fraunhofer ISE calibration laboratory (CalLab). Remarkable are also the cells with gap shallowly diffused (10-cell) that were manufactured with an alternative BSG deposition process. This process uses an rf-source instead of a MW-source for plasma activation as well as shorter deposition times. The corresponding cells feature an increase in VOC , JSC , and FF that results in an absolute efficiency gain Δηabs of 1.5%. The increased VOC as well as JSC are probably due to a higher bulk life time. The increased pFF might be attributed to fewer impurities in the space charge region. In general, the cells feature special characteristics: high JSC values as well as low FF-values. Next, a more detailed cell analysis is carried out, focusing on both parameters. IV. DETAILED CELL ANALYSIS A. Quantum Efficiency Measurements The internal quantum efficiency (IQE) as a function of the wavelength (λ) for a solar cell with (11-cell) and without gap (01-cell) deeply diffused is depicted in Fig. 5. The reflection curve corresponding to both cell-types is shown as well. Because of the absence of a front side metallization grid, the reflection is almost zero in the wavelength range of 530– 550 nm. Around the wavelength λ = 1000 nm, the reflectance is still less than 5%. Regarding the IQE of both cell-types in the wavelength range 500 to 900 nm, the IQE is almost independent on the wavelength. In this range, the cell without gap reaches an IQE mean value of 98.0% compared with 96.6% for the cell with gap. This increase is attributed to the increased emitter fraction of the cell without gap that leads to an increased spectral ability for collecting minority charge carriers [11]. Because of the bulk recombination above the BSF regions (electrical shading), the IQE of the cell without gap does not reach the maximum value of 100.0%. In the short wavelength range (λ < 500 nm), the IQE decreases for both cell-types. Since the surface recombination velocity of the front side is relatively high for both cells, this decrease of the IQE can mainly be attributed to the front side recombination. In the long wavelength range

1240

IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 3, NO. 4, OCTOBER 2013

TABLE I CELL CHARACTERISTIC PARAMETERS EXTRACTED FROM I–V MEASUREMENTS UNDER ONE SUN CONDITIONS (DESIGNATED AREA A = 4cm2 WITHOUT BUSBARS)

B. Fill Factor Analysis and Further Cell Improvement It has already been discovered that, within the I–V measurements, all the processed cells show moderate FF-values. A more detailed analysis of these FF-losses can be carried out, by comparing FF-values with pFF-values [12]. Since the pFF is independent on the series resistance RS , the difference of the FF and pFF indicates the absolute FF-loss of a cell due to the RS . The pFF itself is mainly limited by recombination effects especially in the space charge region as well as by shunting. Fig. 6 depicts the pFF and the FF of cells with (11-cell) and without gap (01-cell) deeply diffused for varying pitch distances xpitch . Independent of the pitch distance and the cell structure, the pFF is constant (pFF ≈ 80%). Additional shunts or enhanced recombination in the space charge region for cells without gap can be neglected, since the latter defects would probably depend on the pitch distance. In contrast with the pFF, the FF strongly decreases for increasing xpitch , and thus, the series resistance RS increases [13]. The increase in RS for an increasing xpitch is mainly caused by the prolonged electrical path length of majority charge carriers within the bulk. Nevertheless, since the FF is not in the range of the pFF even for low pitch distances, an additional limitation of the FF by a nonideal metallization is possible. Thus, aiming at decreasing the resistance of the metal fingers, the Al–Si metallization was thickened by a plating process. Therefore, the surface of the Al–Si metallization was modified by a zincate solution and afterwards thickened by Ag-plating [14]. Table II exhibits the cell results of a cell without gap deeply diffused (01-cell) with an EC of 85% before and after plating. Because of the enlarged cross section of the metal fingers after

1.0 0.9 0.8 0.7

IQE, R (-)

(λ > 900 nm), both cell-types show similar IQE-values. The antireflecting surface and the sufficient bulk carrier lifetime enable JSC -values well above 40.5 mA/cm2 . Higher JSC -values can probably be reached by a careful reduction of the front surface recombination and an adjusted BC-BJ design of the rear with reduced electrical shading.

0.6

no gap cell gap cell Reflection

0.5 0.4 0.3 0.2 0.1 0.0 300 400 500 600 700 800 900 1000 1100 1200

Wavelength λ (nm)

Fig. 5. Internal quantum efficiency (IQE) and reflectance (R) as a function of the wavelength (λ) for a BC-BJ solar cell with (11-cell) and without gap (01-cell) deeply diffused. Within this comparison, cells with constant cell dimensions were selected. Each cell has a pitch of xp itch = 1.5 mm and a BSF finger width of xB S F = 300 μm. The IQE of both cells is mainly limited by front side and bulk recombination above the BSF regions on the cell rear side. JS C values as measured by I–V measurements under one-sun conditions are depicted as well. TABLE II CELL CHARACTERISTIC PARAMETERS EXTRACTED FROM I–V MEASUREMENTS UNDER ONE SUN CONDITIONS (DESIGNATED AREA A = 4 cm2 WITHOUT BUSBARS)

plating, the FF increases. Simultaneously, the pFF is almost constant before and after plating. Therefore, additional shunting problems due to the plating step can be neglected. The increased FF induces an increase of the overall cell efficiency up to 20.5%. This efficiency is independently confirmed by the Fraunhofer

KEDING et al.: CO-DIFFUSED BACK-CONTACT BACK-JUNCTION SILICON SOLAR CELLS WITHOUT GAP REGIONS

metal spikes penetrating the silicon bulk until a depth of approximately 600 nm. This penetration depth is almost sufficient to shunt, e.g., the space charge region of cells shallowly diffused (x0-cells). The latter is a possible explanation for the increase in pFF of the cells deeply diffused (x1-cells) compared with the cells shallowly diffused (x0-cells). A complete prevention of spikes short circuiting the emitter, and hence, the realization of higher pFF-values might be reached by further increasing the junction depth. This can be accomplished by increasing the thermal budget (temperature and time) during the co-diffusion or the drive-in process.

85.0 82.5

Fill factor FF (%)

80.0 77.5

R

75.0 72.5 no no gap gap no no gap gap // pFF pFF gap gap gap gap // pFF pFF

70.0 67.5 65.0

1000

1500

2000

2500

Pitch xpitch (nm) Fig. 6. Fill factor (FF) and pseudo fill factor (pFF) of cells with (10-cell) and without gap (01-cell) deeply diffused as a function of the pitch distance xp itch . For every cell-type and pitch distance four cells were measured. Depicted are the mean values as well as the standard deviation of the single measurements. The difference of the FF and the pFF indicates the FF-losses due to the series resistance R S . p+-Si (Emitter) n-Si

Al-Si

n+-Si (BSF)

V. CONCLUSION The BC-BJ solar cells presented within this study are assembled in an industrially feasible manner and reached solar cell efficiencies up to 20.5%. The cells feature high short-circuit current densities up to 40.9 mA/cm2 on the one hand, and yet, moderate fill factors and VOC s , on the other hand. Especially the fill factor, mainly limited by RS losses and parasitic shunts, indicates the high potential of this very promising cell-type. Further studies certainly will focus on the improvement of the fill factor and the VOC . The integration of thin defect free insulation layers on the cell rear side, design studies, as well as up scaling the cell format are key issues for future investigations. ACKNOWLEDGMENT

Al-Si

n-Si

1241

Al-Si spiking

The authors would like to thank all their PV-TEC co-workers, especially R. Efinger, M. Jahn, R. Neubauer, and D. Trogus for sample pre-paration, and E. Sch¨affer for I–V measurements. REFERENCES

Fig. 7. Cross-sectional view (SEM) on the metal silicon contact of a BC-BJ solar cell. An evaporated Al–Si alloy is used as the metallization layer. The formation of metal spikes after annealing is clearly visible.

ISE CalLab and the highest efficiency value reached within this study so far. C. Shunt Detection The relatively low pFF of the best cell (pFF = 80.1%) indicates a possible limitation of the cells due to shunting problems. Shunting problems can occur due to several reasons, e.g., design faults and edge effects, incomplete contact separation [5], defect-rich rear side passivation layers [15], or due to metal spikes short circuiting the emitter [16]. Especially by using Al or Al–Si alloys as metallization, spiking issues can occur. Aiming at the detection of metal spikes with scanning electron microscope (SEM) imaging on adequate cross-sectional samples, some BC-BJ cells were annealed, cut into pieces, and polished. Fig. 7 presents a magnified cross section of a BC-BJ solar cell, depicting the electrical contact between the Al–Si alloy and the n+ -Si (BSF). Clearly visible is the formation of

[1] P. J. Cousins, D. D. Smith, H.-C. Luan, J. Manning, T. D. Dennis, A. Waldhauer, K. E. Wilson, G. Harley, and W. P. Mulligan, “Generation 3: Improved performance at lower cost,” presented at the IEEE 35th Photovoltaic Specialists Conf., Honolulu, HI, USA, 2010. [2] J. Seiffe, L. Weiss, M. Hofmann, L. Gautero, and J. Rentsch, “Alternative rear surface passivation for industrial cell production,” presented at the 23rd Eur. Photovoltaic Solar Energy Conf., Valencia, Spain, 2008. [3] B. Hoex, J. Schmidt, P. Pohl, M. C. M. van de Sanden, and W. M. M. Kessels, “Silicon surface passivation by atomic layer deposited Al2 O3 ,” J. Appl. Phys., vol. 104, pp. 044903-1–044903-12, 2008. [4] J. Robbelein, B. Vermang, T. Janssens, M. Alem´an, N. E. Posthuma, and J. Poortmans, “Passivation layers for large area interdigitated back junction cells and their electrical isolation properties,” presented at the 26th Eur. Photovoltaic Solar Energy Conf. Exhib., Hamburg, Germany, 2011. [5] N. Mingirulli, R. Keding, J. Specht, A. Fallisch, D. St¨uwe, and D. Biro, “Hot-melt inkjet as masking technology for back-contacted cells,” presented at the 34th IEEE Photovoltaic Spec. Conf., Philadelphia, PA, USA, 2009. [6] A. Fallisch, D. Wagenmann, R. Keding, D. Trogus, M. Hofmann, J. Rentsch, H. Reinecke, and D. Biro, “Analysis of phosphorus-doped silicon oxide layers deposited by means of PECVD as a dopant source in diffusion processes,” IEEE J. Photovoltaics, vol. 2, no. 4, pp. 450–456, 2012. [7] R. Keding, R. Woehl, D. St¨uwe, A. Fallisch, A. Hofmann, J. Rentsch, and D. Biro, “Diffusion and characterization of doped patterns in silicon from prepatterned boron- and phosphorus-doped silicate glasses,” presented at the Proc. 26th Eur. Photovoltaic Solar Energy Conf. Exhib., Hamburg, Germany, 2011. [8] R. Keding, P. Rothhardt, C. Roters, A. Fallisch, S. Hohage, M. Hofmann, R. Woehl, D. Borchert, and D. Biro, “Silicon doping performed by different diffusion sources aiming co-diffusion,” presented at the 27th Eur. Photovoltaic Solar Energy Conf. Exhib., Frankfurt, Germany, 2012.

1242

[9] D. B. M. Klaassen, “A unified mobility model for device simulation–II. Temperature dependence of carrier mobility and lifetime,” Solid State Electron., vol. 35, pp. 961–967, 1992. [10] R. A. Sinton and A. Cuevas, “A quasi-steady-state open-circuit voltage method for solar cell characterization,” presented at the 16th Eur. Photovoltaic Solar Energy Conf., Glasgow, U.K., 2000. [11] C. Reichel, F. Granek, M. hermle, and S. W. Glunz, “Investigation of electrical shading effects in back-contact back-junction silicon solar cells using two-dimensional charge collection probability and the reciprocity theorem,” J. Appl. Phys., vol. 109, pp. 024507-1–024507-12, 2011. [12] D. Pysch, A. Mette, and S. W. Glunz, “A review and comparison of different methods to determine the series resistance of solar cells,” Solar Energy Mater. Solar Cells, vol. 91, pp. 1698–1706, 2007. [13] F. Granek, M. Hermle, D. Huljic, O. Schultz-Wittmann, and S. W. Glunz, “Enhanced lateral current transport via the front n+ diffused layer of n-type high-efficiency back-junction back-contact silicon solar cells,” Progress Photovoltaics: Res. Appl., vol. 17, pp. 47–56, 2008. [14] M. Kamp, J. Bartsch, G. Cimiotti, R. Keding, A. Zogaj, C. Reichel, A. Kalio, M. Glatthaar, and S. W. Glunz, “Zincate processes for silicon solar cell metallization,” Solar Energy Mater. Solar Cells, 2013. Available: http://dx.doi.org/10.1016/j.solmat.2013.05.035i. [15] P. Saint-Cast, F. Tanay, M. Aleman, C. Reichel, J. Bartsch, M. Hofmann, J. Rentsch, and R. Preu, “Relevant pinhole characterisation methods for dielectric layers for silicon solar cells,” presented at the 24th Eur. Photovoltaic Solar Energy Conf., Hamburg, Germany, 2009. [16] A. Fallisch, S. Werner, M. Retzlaff, R. Neubauer, F. Lottspeich, and D. Biro, “18,7% emitter wrap-through silicon solar cells with screenprinted silver contacts acting as a barrier for evaporated aluminium metallization,” presented at the 26th Eur. Photovoltaic Solar Energy Conf. Exhib., Hamburg, Germany, 2011.

Roman Keding was born in 1984. He studied microsystems engineering with the Institute for Microsystems Engineering, Albert-Ludwigs-University, Freiburg, Germany, where he received the B.Sc. degree in 2008 and the M.Sc. degree in 2010. Since 2010, he has been working toward the Ph.D. degree, developing highly efficient back-contact back-junction silicon solar cells with industrially feasible technologies at Fraunhofer ISE. He joined the Division of PV Production Technology and Quality Assurance, Fraunhofer Institute for Solar Energy Systems, Freiburg, in 2008.

¨ was born in 1980. He studied print and media technology with the David Stuwe Media University, Stuttgart, Germany and with EVTEK University of Applied Sciences, Espoo, Finland. He received the Diploma degree from the Media University, in 2007 for his work on inkjet printing of hotmelt etching masks for the structuring of antireflective coatings. Since 2011, he has been working toward the Ph.D. degree with the Fraunhofer Institute for Solar Energy Systems (ISE), Freiburg, Germany. He joined the Division PV Production Technology and Quality Assurance, Fraunhofer ISE, in 2007. His research interests include the investigation of novel inkjet applications for the production of crystalline silicon solar cells.

Mathias Kamp was born in Ahaus, Germany, in 1985. He received the Diploma degree in chemistry from the Westphalian University of Applied Sciences, Gelsenkirchen, Germany. He finished the Diploma thesis on light-induced silver plating processes for silicon solar cells in 2010. Since 2012, he has been working toward the Ph.D. degree with the Fraunhofer Institute for Solar Energy Systems, Freiburg, Germany, focusing on plating processes for back-contact solar cells. Mr. Kamp received the SiliconPV Award 2013 for the contribution entitled “Zincate Processes for Silicon Solar Cell Metallization.” The article is published in the Journal Solar Energy Materials and Solar Cells.

Christian Reichel, photograph and biography not available at the time of publication.

IEEE JOURNAL OF PHOTOVOLTAICS, VOL. 3, NO. 4, OCTOBER 2013

Andreas Wolf studied physics with the Technical University of Darmstadt, Darmstadt, Germany, and the KTH Royal Institute of Technology, Stockholm, Sweden. In 2007, he received the Ph.D. degree from the Leibniz University of Hannover, Hanover, Germany, for his work on sintered porous silicon and layer transfer silicon thin-film solar cells, which he carried out with the Institute for Solar Energy Research Hamelin, Emmerthal, Germany. He is currently the Head of the Thermal Processes/Passivated Solar Cells Group, Fraunhofer Institute for Solar Energy Systems, Freiburg, Germany.

Robert Woehl completed the Undergraduate studies in physics with the University of Hanover, Hanover, Germany. He continued with the University of Manchester, U.K., and received the Graduate degree from the University of Freiburg, Freiburg, Germany, in 2008. At Fraunhofer Institute for Solar Energy Systems (ISE), Freiburg, he completed the Diploma thesis and the Ph.D. thesis, developing different concepts for back-contact back-junction silicon solar cells with industrially feasible technologies, and afterwards was leading a team which worked on back-contact back-junction cells. Since 2013, he has been with the R&D Solar Cell Group, Total New Energies, Emeryville, CA, USA.

Dietmar Borchert was born in 1963. He studied physics with the University of Dortmund, Dortmund, Germany. He received the Ph.D. degree in electrical engineering from the University of Hagen, Hagen, Germany, in 1998 in the field of silicon heterojunction solar cell technology. In 1999, he joined the Fraunhofer Institute for Solar Energy Systems, Fraunhofer, Germany. Since then, he has been the Head of the Laboratory and Service Center, Gelsenkirchen, Germany. He is currently working on silicon thin-film and heterojunction solar cells.

Holger Reinecke was born in Bad Harzburg, Germany, in 1964. From 1983 to 1988, he studied chemistry with the Technical University, Clausthal-Zellerfeld, Germany and received the Graduate degree in electrochemical analytics from the Institute for Inorganic and Analytical Chemistry, Jena, Germany, in 1990. From 1988 to 1990, he was a Scientific Assistant with the Institute for Inorganic and Analytical Chemistry. In August 1990, he joined the Electroplating Group of microParts GmbH as a Scientific Assistant and became Head of the group in 1991. In 1993, he took over the Department of Chemical Process Technology and then the Department of Process Technology in 1995. In 1999, as an Area Manager, he additionally became the Head of the microfluidics and microoptics product branches. Since November 2004, he is the Head of the Chair of Process Technology with the Department of Micro Systems Engineering, University of Freiburg, Freiburg, Germany. Mr. Reinecke has been the Speaker of the Board of Directors of the HSGIMIT in Villingen-Schwenningen (www.hsg-imit.de), since May 2005.

Daniel Biro was born in 1972 and studied physics at the University of Karlsruhe, Karlsruhe, Germany, and the University of Massachusetts Amherst, Amherst, MA, USA. He received the Ph.D. degree from the University of Freiburg, Freiburg, Germany, in 2003 in the field of silicon solar cell diffusion technologies. In 1995, he joined the Silicon Cell Characterization Group, Fraunhofer Institute for Solar Energy Systems (ISE), Fraunhofer, Germany, and started working in the Silicon Solar Cell Production Technology Group, Fraunhofer ISE. In 2004–2005, he coordinated the design and ramp-up of the production oriented research platform PV-TEC and is currently the Head of the “Thermal, PVD and Printing Technology/Ind. Cell Structures” Department with Fraunhofer ISE.