Defect passivation on cast-mono crystalline screen-printed cells

Abstract

Cast-mono crystalline silicon wafers contain crystallographic defects, which can severely impact the electrical performance of solar cells. This paper demonstrates that applying hydrogenation processes at moderate temperatures to finished screen print cells can passivate dislocation clusters within the cast-mono crystalline silicon wafers far better than the hydrogenation received during standard commercial firing conditions. Efficiency enhancements of up to 2% absolute are demonstrated on wafers with high dislocation densities. The impact of illumination to manipulate the charge state of hydrogen during annealing is investigated and found to not be significant on the wafers used in this study. This finding is contrary to a previous study on similar wafers that concluded increased H⁻ or H⁰ from laser illumination was responsible for the further passivation of positively charged dangling bonds within the dislocation clusters.

Keywords

silicon solar cell / dislocation / cast-mono / laser / hydrogen passivation

Introduction

It is well known that localized regions with high dislocation densities are detrimental to solar cell efficiency, with dislocations being considered the worst of all crystal defects [ 1]. Dislocations on their own have been shown to reduce the minority carrier lifetime [ 2, 3], however this recombination is often made even worse when these regions become decorated by metallic impurities and precipitates that tend to aggregate at such structural defects [ 4– 7]. These decorated structural defects are not only detrimental for minority carrier lifetime but can also form shunt paths across the p-n junction for the majority carriers [ 1, 4]. Furthermore they cannot be gettered by conventional thermal processes and as such cannot be fixed during the emitter formation process [ 8]. With dislocations almost unavoidable in crystalline silicon, and particularly prominent in multi and cast material, the ability to passivate dislocation clusters with hydrogen becomes important for improving the cell’s open circuit voltage, fill factor and efficiency.

Hydrogenation is a process known for its ability to improve the electrical properties of heavily defected silicon, although the exact interactions are not fully understood [ 8– 20]. Certainly the situation appears very complex as evidenced by various studies of cast multi-crystalline material (like the cast-mono wafers used in this study). Tan et al. [ 21] reported that dislocation clusters could be passivated, resulting in a significant enhancement in lifetime. However, Martinuzzi et al. [ 22] reported that only individual/intragrain dislocations could be fully passivated, while the deep levels of dislocation tangles could not be fully neutralized, merely transformed to shallow levels. Sheoran et al. [ 12] found that the bottom parts of ingots were better passivated than the top parts, but this also varied between ingot manufacturers. Despite volumes of work looking at ways to reduce the recombination of crystallographic defects, the reality is that these defects are still dominant in many forms of cheaper silicon wafers.

A topic of recent interest in the area of hydrogen passivation for solar cell applications is the use of charge state control, which has not been considered in the aforementioned studies. Recent work by these authors has found that laser illuminated hydrogenation can significantly improve the performance of heavily dislocated cells above that achieved by typical belt furnace hydrogenation. It was hypothesized that this was due to the laser illumination increasing the concentrations of the minority charge states of hydrogen for improved passivation [ 23]. Atomic hydrogen can exist in three charge states, namely the positive (H⁺), negative (H⁻) and neutral (H⁰) charge states, which have large differences in diffusion and bonding mechanisms [ 24– 27]. In a p-type silicon wafer at room temperature and in the dark, H⁺ is the dominant charge state of hydrogen [ 28]. It’s positive charge will prevent it diffusing far/quickly due to its coulombic attraction to the ionized p-type dopant atoms with fixed negative charge [ 24]. Furthermore, H⁺ has no electron, leaving it with no mechanism for bonding to positively charged defects or dangling bonds [ 29]. Converting the hydrogen to the minority charge states should therefore be beneficial for two reasons: ① H⁰ can diffuse quickly being unaffected by coulombic interactions [ 26] which should enable hydrogen to penetrate further into the bulk quicker, and ② hydrogen in n-type, which exists primarily as H⁻ (and H⁰), has been shown capable of passivating defects that can’t be passivated by the same hydrogen process in p-type when the hydrogen would be almost entirely as H⁺ [ 30], likely due to the lack of electrons. In a silicon wafer, the minority charge states of hydrogen can be increased by generating a large amount of excess minority carriers, shifting the Fermi level closer to the middle of the band-gap and therefore making the silicon act intrinsic [ 8, 31].

Two methods of achieving a shift of the Fermi level toward the middle of the band-gap are an increase in temperature and the use of high intensity illumination [ 26, 27]. Increasing the temperature has the benefits of increasing the concentration of minority hydrogen charge species, the reactivity, and the diffusivity of hydrogen [ 25]. However, increasing the temperature will also lead to a more rapid dissociation of hydrogen-defect complexes [ 32]. Therefore it is crucial to control the temperature within an optimal window for the most effective hydrogen passivation. Doing so is complicated, given that hydrogenation is generally performed simultaneously with the formation of metal contacts. For example, the thermal process used is typically dictated by the requirements to form a back surface field (BSF) [ 33, 34]; the optimal metallization temperatures for the penetration of metal contacts through dielectric layers and contact formation [ 35, 36]; and/or metal sintering processes [ 4, 37]. For passivating the defects within a screen printed silicon solar cell, a typical commercial hydrogenation process occurs during the high temperature (~800°C) fast-firing (~2s above 700°C) of the metal contacts with a hydrogenated SiN_x antireflection coating (ARC) [ 8, 21– 12, 31]. Although this firing step significantly reduces recombination within the cell, multi-crystalline silicon solar cells with grain boundaries and dislocations still suffer from lower efficiency than Cz-mono cells. Generally, this process is insufficient for the complete passivation of dislocations.

Laser hydrogenation post fabrication allows for greater control over the sample temperature and illumination, with high intensities possible for maximum generation of the minority charge states. Previous results have demonstrated substantial improvements in cell efficiency by applying such a post-cell-fabrication laser hydrogenation step to wafers with a high density of crystal defects [ 38, 39]. This work clearly indicates that the hydrogenation achieved during standard screen print firing is not sufficient. However, a lack of control conditions did not allow for accurate conclusions to be drawn on whether the charge state of the hydrogen was in fact important during this process. In all cases the samples were laser illuminated thereby resulting in large quantities of minority charge states in all conditions. Here, we use similar laser illumination of varying intensities, this time including a control anneal process with no laser illumination so that the hydrogen should almost entirely exist as H⁺ in the p-type wafer to ascertain whether the laser illumination and minority charge states do in fact provide added benefit.

Experimental method

An experiment was conducted to test hydrogenation processes with charge state control on complete cells. The substrates used in the experiment were four 156 mm × 156 mm 1 W·cm random pyramid textured cast-mono crystalline silicon wafers. The samples were selected from adjacent locations from the ingot to ensure that crystal defects were closely matched. After a standard Radio Corporation of America (RCA) clean with a final dip in HF, the samples were diffused in a Tempress Systems tube furnace using a POCl₃ process (795°C and 28 min for the pre-deposition step, 885°C and 30 min for the drive-in step) to achieve a phosphorous emitter with a final sheet resistance of 65 W/□. After diffusion the front side of the wafers were deposited with a SiN_x:H layer of thickness 75 nm and refractive index 2.08 via PECVD. The wafers were then cleaved into 39 mm × 39 mm tokens with the location and orientation of each carefully matched to allow comparison between sets of sister tokens (i.e. samples with very closely matched crystallographic properties due to their original proximity within the ingot). After cleaving, a set of 4 sister tokens were screen-printed with an aluminum (Monocrystal PASE1203) full area back contact and a silver (DuPont PV17) front H-bar contact grid and then fired in an industrial Sierratherm infrared fast firing belt furnace with a peak temperature of 800°C. Edge isolation was achieved by laser cutting and cleaving the samples from the rear, resulting in a final cell area of 7.84 cm². The samples were then characterized using a custom built 1-Sun light I-V and dark I-V tester, photoluminescence (PL) imaging in a BT Imaging tool and LBIC (light beam induced current) imaging at 981 nm using a Semilab WT-2000 tool.

After the final cell processing and characterization the samples received an additional anneal process to expose them to a specific temperature for a range of times while varying the laser illumination intensity for varied charge-states of the interstitial hydrogen atoms within the silicon. This was done to evaluate the combined impact these conditions have on the ability to passivate recombination within the finished device, in particular the potential impact of hydrogen mobility and reactivity. This treatment involved heating the devices on a hotplate for 4 min with the samples removed after each minute, quenched by placing on a damp cloth. During this heat treatment, different illumination levels (including no illumination) were used to generate different carrier concentrations within the respective devices, which in turn varied the respective concentrations of the minority charge states for the hydrogen atoms in the silicon. The absence of illumination of the device results in virtually all the interstitial hydrogen within the p-type silicon being in the positive charge state (H⁺) while at the other extreme, the highest illumination gives the lowest concentration of H⁺ and correspondingly highest concentration of H⁻. The illumination source for the devices on the hotplate was a 938 nm laser with a spot size that entirely covered the sample. The laser illumination resulted in an increase of the substrate temperature that was accounted for by lowering the hotplate set point such that the final cell temperature was 573 K in all cases. Three different laser intensities were used with the intensity measured using a calibrated photodiode. A summary of the processing conditions applied to each sample is shown in Table 1. Subsequent to this experiment, improvements in the equipment used to measure the temperature revealed some slight differences; these are recorded in the table as the actual cell temperature. PL image ratio maps were created by using software to align the open circuit PL images before and after annealing, and divide the image after anneal by the prior image to visualize the improvements.

Results and discussion

The variations in 1-Sun I-V parameters of the cells after each minute of annealing are shown in Fig. 1 and demonstrate a similarly significant improvement in cell performance for all conditions.

Following the 4-min thermal anneal process applied to the heavily dislocated cast wafers, the efficiency of all cells increased by roughly 2% absolute. Most of the enhancement occurred within the first minute, however the efficiency continued to increase slightly over the 4 min of processing. The main drivers for the improvement in efficiency were a substantial increase in open circuit voltage, V_oc (20–25 mV increase), as well as fill factor, FF (3%–4% absolute increase). A variable increase in short circuit current, J_sc (between 0.1 and 0.7 mA/cm²) was also achieved although it is likely that some of this variation was due to instabilities in the measurement. Somewhat surprisingly from the lumped I-V parameters it was difficult to see any particular enhancement due specifically to the application of the laser.

The improvement in the V_oc of the cells by 20–25 mV indicates a greater than 50% reduction in the total recombination for the device. Calibrated open-circuit PL voltage maps [ 41] of cells with and without laser application at various process stages (Fig. 2) provide further insight into this improvement. These images highlight two distinct regions within the wafers: ① good quality with few crystal defects in the top left of each sample, and ② poor quality, highly defected regions in the bottom right. After thermal annealing the PL signal was found to increase in both of these regions; indicating a similar overall reduction in recombination within the sample for the samples with and without illumination.

The localized improvement can be better visualized by examining the PL ratio image (Fig. 3) created by comparing the image before and after a 4 min hydrogenation process on a cell with the same conditions as sister 4. This image reveals that the reduction in recombination occurred in regions that were heavily impacted by crystallographic defects, as the PL counts in these regions increased up to 4 times, while some of the non-dislocated regions do not appear to have improved at all. This sample exhibits a different dislocation pattern to those in Fig. 2 since it is a different token derived from the same large wafer and hence should respond identically.

A similar reduction in recombination in the defected region was realized under short circuit conditions. The LBIC images (Fig. 4) measured at 981 nm of the same sample as shown in Fig. 3 show a noticeable reduction in the recombination associated with the dislocations in the bulk of the wafer. Similar to the changes in the PL images it is likely that these improvements are due to hydrogen passivation of those defected regions.

These enhancements are significant considering the wafers had already received the standard commercial hydrogenation that occurs during the contact firing process. The enhancement is most likely due to further hydrogen passivation of the cells during the thermal anneal, as the thermal anneal at these temperatures is unlikely causing any structural change in BSF or emitter.

Based on previous hypotheses [ 23], it was an unexpected result that the illumination during annealing did not appear to have any significant effect on the quality of the hydrogenation process. It is very clear that every set of conditions used in this experiment, regardless of the distribution of the respective charge state concentrations, gave great improvement in the passivation of the recombination in the p-type wafer compared to the conditions used for the normal contact firing process at a much higher temperature. The only obvious structural difference of the complete cells processes here compared with devices during contact firing is the lack of the p⁺ layer at the rear (BSF) which only forms during the cool down following the contact firing at ~800°C. It is possible that during firing the molten Aluminum acts as a hydrogen sink without a formed BSF to protect it during firing. When post annealing is performed, a BSF exists which could possibly shield any charged hydrogen from the metal at the rear allowing more hydrogen to build up in the bulk for improved passivation. If this is the case, it is perhaps fortunate that H⁰ will never be the dominant charge state [ 26] so that almost all of the hydrogen will be charged either H⁺ or H⁻ and therefore be affected by electric fields making it less able to diffuse through the p⁺ layer to the metal-silicon interface.

It was also expected that the minority charge states would have an advantage with regard to reactivity and ability to passivate, however with no obvious difference in passivation between processing conditions this does not appear to be the case. It is suspected that the dislocation clusters are very sensitive to hydrogen passivation, and therefore hydrogen in all charge states has a strong reactivity with the dislocation clusters. Further, all samples were re-measured after 6 months and the electrical characteristics had remained stable in all cases, indicating that the charge state of the hydrogen does not provide any added benefit to the stability of the hydrogen passivation either. Future work will investigate the stability of these hydrogenated cells in operating conditions.

The dramatic reduction of recombination in the dislocation clusters is encouraging, however these regions still appear as a dominant form of recombination in the images of Figs. 2 and 4. This indicates that although this hydrogenation was an improvement upon previous techniques, further improvements are likely still possible.

The cells used in this work suffered from a particularly poor FF, which was also significantly improved by the annealing processes. While some improvement in FF could be expected from an increase in V_oc this alone cannot explain a ~4% abs increase. To further understand the improvement in FF the cells were tested using suns-V_oc. The pseudo FF (pFF) was extracted from the suns-V_oc curves, yielding a metric for changes in FF due to mechanisms other than series resistance. In addition to this, the FF limit (iFF) due to the 1-Sun V_oc (assuming J₀₁ dominates with an ideality factor of n = 1) extracted from the 1-Sun I-V curve was also calculated. A comparison between the change in FF, iFF and pFF due to annealing (Fig. 5) found that the dominant change in FF was due to series resistance which accounted for ~3% abs increase in FF of the cells. In comparison, the annealing resulted in a change in iFF of ~0.5% abs and in pFF of 1% to 1.5% abs, which are also significant to a lesser extent. The latter is likely due to the reduced ideality factor, which appears to have resulted from reduced junction recombination achieved by the passivation of the defects.

R_s images taken using photoluminescence techniques (not shown), revealed that the R_s of the finished cells was highly non-uniform and likely related to the contact resistance between the front silver electrode and front surface emitter. It is not well understood at this stage either why the regions of high contact resistance resulted following the standard screen-printed contact firing nor why the increased thermal processing at ~350°C appeared to significantly reduce such series resistance.One possibility could be that the duration of firing for screen-printed contacts was slightly less than optimal such that the increased thermal dose from the additional hydrogenation process improved the results.

Conclusions

In this paper, it was shown that hydrogenation processes can dramatically increase the performance of screen-printed cast-mono crystalline silicon solar cells, leading to an absolute efficiency enhancement of as much as 2% (more than 10% in relative terms) over and above that achieved by standard belt furnace hydrogenation in conjunction with the firing of screen-printed metal contacts. Investigation into the mechanisms suggested that the efficiency enhancement was due to reduced recombination in dislocation clusters, likely due to passivation of those regions with hydrogen, and a large improvement in FF, primarily due to a reduction in series resistance. The laser illumination, used to control the charge state of the interstitial hydrogen atoms within the silicon, did not appear to make a significant difference, suggesting that the hydrogen concentration was more important than their charge state and that hydrogen in all charge states can react well with the type of crystallographic defects found in these cast mono wafers as long as enough hydrogen exists in the bulk. However it was very clear that the hydrogenation achieved during the standard screen-printed contact firing was insufficient and that greatly improved passivation of the defects in the wafers could be achieved using a 350°C anneal.

References

[1] Takahashi I, Usami N, Kutsukake K, Stokkan G, Morishita K, Nakajima K. Generation mechanism of dislocations during directional solidification of multicrystalline silicon using artificially designed seed. Journal of Crystal Growth, 2010, 312(7): 897–901

[2] Yang K, Schwuttke G H, Ciszek T F. Structural and electrical characterization of crystallographic defects in silicon ribbons. Journal of Crystal Growth, 1980, 50(1): 301–310

[3] Gu X, Yu X, Guo K, Chen L, Wang D, Yang D. Seed-assisted cast quasi-single crystalline silicon for photovoltaic application: towards high efficiency and low cost silicon solar cells. Solar Energy Materials and Solar Cells, 2012, 101(2): 95–101

[4] Breitenstein O, Rakotoniaina J P, Al Rifai M H, Werner M. Shunt types in crystalline silicon solar cells. Progress in Photovoltaics: Research and Applications, 2004, 12(7): 529–538

[5] Hanoka J, Bell O. Electron-beam-induced currents in semiconductors. Annual Review of Materials Science, 1981, 11(1): 353–380

[6] Sopori B, Zhang Y, Ravindra N M. Silicon device processing in H-ambients: H-diffusion mechanisms and influence on electronic properties. Journal of Electronic Materials, 2001, 30(12): 1616–1627

[7] Haunschild J, Glatthaar M, Demant M, Nievendick J, Motzko M, Rein S, Weber E R. Quality control of as-cut multicrystalline silicon wafers using photoluminescence imaging for solar cell production. Solar Energy Materials & Solar Cells, 2010, 94(12): 2007–2012

[8] Sopori B. Silicon solar-cell processing for minimizing the influence of impurities and defects. Journal of Electronic Materials, 2002, 31(10): 972–980

[9] Sopori B L, Deng X, Benner J P, Rohatgi A, Sana P, Estreicher S K, Park Y K, Robertson M A. Hydrogen in silicon: a discussion of diffusion and passivation mechanisms. In: IEEE 1st World Conference on Photovoltaic Energy, 1994, 2(95):159–169

[10] Divigalpitiya W M R, Morrison S R, Vercruysse G, Praet A, Gomes W P. Hydrogen passivation of dislocations in silicon. Solar Energy Materials, 1987, 15(2): 141–151

[11] Dubé C, Hanoka J I. Hydrogen passivation of dislocations in silicon. Applied Physics Letters, 1984, 45(10): 1135–1137

[12] Weronek K, Weber J, Queisser H J. Hydrogen passivation of the dislocation-related D-band luminescence in silicon. physica status solidi (a), 1993, 137(2): 543–548

[13] Benton J L, Doherty C J, Ferris S D, Flamm D L, Kimerling L C, Leamy H J. Hydrogen passivation of point defect in silicon. Applied Physics Letters, 1980, 36(8): 670–671

[14] Hallam B, Sugianto A, Mai L, Xu G Q, Chan C, Abbott M, Wenham S, Uruena A, Aleman M, Poortmans J. Hydrogen passivation of laser-induced defects for silicon solar cells. In: Proceedings of IEEE 40th Photovoltaic Specialist Conference, 2014

[15] Abbott M, Cousins P, Chen F, Cotter J. Laser-induced defects in crystalline silicon solar cells. In: Proceedings of IEEE Photovoltaic Specialist Conference, 2005

[16] Tan J, Cuevas A, Macdonald D, Trupke T, Bardos R, Roth K. On the electronic improvement of multi-crystalline silicon via gettering and hydrogenation. Progress in Photovoltaics: Research and Applications, 2008, 16(2): 129–134

[17] Martinuzzi S, Périchaud I, Warchol F. Hydrogen passivation of defects in multicrystalline silicon solar cells. Solar Energy Materials and Solar Cells, 2003, 80(3): 343–353

[18] Sheoran M, Upadhyaya A, Rohatgi A. Bulk lifetime and efficiency enhancement due to gettering and hydrogenation of defects during cast multicrystalline silicon solar cell fabrication. Solid-State Electronics, 2008, 52(5): 612–617

[19] Song L, Wenham A, Wang S, Hamer P, Ahmmed M S, Hallam B, Mai L, Abbott M, Hawkes E R, Chong A M, Wenham S R. Laser enhanced hydrogen passivation of silicon wafers. International Journal of Photoenergy, 2015, 193892

[20] Van de Walle C G, Denteneer P J H, Bar-Yam Y, Pantelides S T. Theory of hydrogen diffusion and reactions in crystalline silicon. Physical Review B: Condensed Matter and Materials Physics, 1989, 39(15): 10791–10808

[21] Chang K, Chadi D. Hydrogen bonding and diffusion in crystalline silicon. Physical Review B: Condensed Matter and Materials Physics, 1989, 40(17): 11644–11653

[22] Herring C, Johnson N M, Van de Walle C G. Energy levels of isolated interstitial hydrogen in silicon. Physical Review B: Condensed Matter and Materials Physics, 2001, 64(12): 125209

[23] Johnson N M, Herring C. Hydrogen immobilization in silicon p-n junctions. Physical Review B: Condensed Matter and Materials Physics, 1988, 38(2): 1581–1584

[24] Fedders P A. Diffusion of hydrogen in different charge states in realistic models of a-Si:H. Physical Review B: Condensed Matter and Materials Physics, 2002, 66(19): 195308

[25] Hamer P, Hallam B, Wenham R, Abbott M. Manipulation of hydrogen charge states for passivation of P-type wafers in photovoltaics. IEEE Journal of Photovoltaics, 2014, 4(5): 1252–1260

[26] Sadoh T, Tsukamoto K, Baba A, Bai D, Kenjo A, Tsurushima T, Mori H, Nakashima H. Deep level of iron-hydrogen complex in silicon. Journal of Applied Physics, 1997, 82(8): 3828–3831

[27] Hallam B J, Hamer P G, Wenham S R, Abbott M A, Sugianto A, Wenham A M, Chan C E, Xu G Q, Kraiem J, Degoulange J, Einhaus R. Advanced bulk defect passivation for silicon solar cells. IEEE Journal of Photovoltaics, 2014, 4(1): 88–95

[28] Nakayashiki K, Rohatgi A, Ostapenko S, Tarasov I. Minority-carrier lifetime enhancement in edge-defined film-fed grown Si through rapid thermal processing-assisted reduction of hydrogen-defect dissociation. Journal of Applied Physics, 2005, 97(2): 024504

[29] Narasimha S, Rohatgi A, Weeber A W. An optimized rapid aluminum back surface field technique for silicon solar cells. IEEE Transactions on Electron Devices, 1999, 46(7): 1363–1370

[30] del Alamo J, Eguren J, Luque A. Operating limits of Al-alloyed high–low junctions for BSF solar cells. Solid-State Electron Devices, 1981, 24(5): 415–420

[31] Cheek G C, Mertens R P, Van Overstraeten R, Frisson L. Thick-film metallization for solar cell applications. IEEE Transactions on Electron Devices, 1984, 31(5): 602–609

[32] Hilali M M, Sridharan S, Khadilkar C, Shaikh A, Rohatgi A, Kim S. Effect of glass frit chemistry on the physical and electrical properties of thick-film Ag contacts for silicon solar cells. Journal of Electronic Materials, 2006, 35(11): 2041–2047

[33] Lennon A, Yao Y, Wenham S. Evolution of metal plating for silicon solar cell metallisation. Progress in Photovoltaics: Research and Applications, 2013, 21(7): 1454–1468

[34] Wilking S, Beckh C, Ebert S, Herguth A, Hahn G. Influence of bound hydrogen states on BO-regeneration kinetics and consequences for high-speed regeneration processes. Solar Energy Materials and Solar Cells, 2014, 131(58): 2–8

[35] Wenham A, Hallam B, Song L, Wang S, Abbott M, Chan C, Hamer P, Azmi A, Barnett A, Wenham S R. Efficiency enhancement for screen printed solar cells on Quasi-Mono wafers through advanced hydrogenation. In: Proceedings of the European PVSEC, 2015

[36] Hallam B, Tjahjono B, Trupke T, Wenham S. Photoluminescence imaging for determining the spatially resolved implied open circuit voltage of silicon solar cells. Journal of Applied Physics, 2014, 115(4): 044901

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag Berlin Heidelberg