Laser enhanced gettering of silicon substrates

Abstract

One challenge to the use of lightly-doped, high efficiency emitters on multicrystalline silicon wafers is the poor gettering efficiency of the diffusion processes used to fabricate them. With the photovoltaic industry highly reliant on heavily doped phosphorus diffusions as a source of gettering, the transition to selective emitter structures would require new alternative methods of impurity extraction. In this paper, a novel laser based method for gettering is investigated for its impact on commercially available silicon wafers used in the manufacturing of solar cells. Direct comparisons between laser enhanced gettering (LasEG) and lightly-doped emitter diffusion gettering demonstrate a 45% absolute improvement in bulk minority carrier lifetime when using the laser process. Although grain boundaries can be effective gettering sites in multicrystalline wafers, laser processing can substantially improve the performance of both grain boundary sites and intra-grain regions. This improvement is correlated with a factor of 6 further decrease in interstitial iron concentrations. The removal of such impurities from multicrystalline wafers using the laser process can result in intra-grain enhancements in implied open-circuit voltage of up to 40 mV. In instances where specific dopant profiles are required for a diffusion on one surface of a solar cell, and the diffusion process does not enable effective gettering, LasEG may enable improved gettering during the diffusion process.

Keywords

gettering / multicystaline / silicon / impurities / laser doping

Introduction

Progress in manufacturing and processing techniques has reduced the cost of solar technology over recent years. However, polysilicon costs and the cost of silicon wafers still amount to over 33% of the cost of a solar module. The majority of solar cells in the market are manufactured from crystalline silicon (c-Si) with over 65% dominated by p-type multicrystalline silicon (mc-Si) and high performance multicrystalline silicon (HP mc-Si) technology [ 1]. Although these materials can be manufactured with a low cost and high throughput when compared to their monocrystalline counterparts, the quality of mc-Si wafers are significantly limited by impurities. Increased amounts of impurities such as oxygen, iron, copper and various other transition metals, in addition to dislocations and grain boundaries, have a profound effect on the electrical characteristics of the material [ 2].

One technique to reduce impurities in the final solar cell is to use high temperature steps to segregate impurities into regions of the device away from the bulk, a process known as gettering. In solar cell manufacturing, gettering is typically achieved during phosphorus (P) diffusion and the formation of a heavily doped emitter [ 3– 7]. In this process, P is typically introduced to the Si wafer via either a gas phase source, such as phosphorus oxychloride (POCl₃), or a liquid spin-on or spray-on source consisting of phosphoric acid (H₃PO₄) [ 6]. Thermal diffusion of interstitial P atoms within the Si crystal lattice can occupy substitutional positions, ejecting impurities into the bulk via a kick-out reaction [ 8]. The increased mobility of impurities at elevated temperatures allows them to diffuse into and become captured in the phosphorous layers formed at the surface of the wafer [ 5]. These layers, along with the impurities can be subsequently removed using a chemical etching.

Although it may be apparent that heavily doped emitters are ideal for the gettering of transition metals, the performance of the emitter plays a major role in determining the final efficiency of a solar cell. Heavy doping of the emitter may generate a dead layer, a layer of electrically inactive interstitial P atoms. Although this layer has been suggested to be responsible for gettering [ 3], it can lead to an increased surface recombination velocity (SRV) and a poor blue response. Industry predictions by the ITRPV have suggested a movement toward lightly-doped emitters with increased sheet resistance (ρ_sh) of approximately 120 Ω/sq by 2020 [ 1]. These lightly-doped emitters, also known as high efficiency emitters, are beneficial in improving collection of carriers generated close to the front surface and keeping the emitter saturation current (J_0e) low. However, by reducing the emitter surface phosphorus concentration, gettering efficiency is reduced, thus, alternate forms of gettering may be required for high efficiency emitters.

Laser damage induced gettering has been studied extensively in the fabrication of integrated circuits in the semiconductor industry. Typically, laser damage is introduced to the rear-side of a wafer followed by an extended annealing process at a high temperature [ 9, 10]. The relaxation of stress during the annealing process generates a dislocation array, which propagates throughout the Si lattice. These laser induced dislocations have been proven to be effective nucleation sites for impurity precipitation. Subsequent active device fabrication can then be made at the defect free surface. Hayafuji et al. demonstrated increases in generation lifetime over two orders of magnitude on samples irradiated with laser pulses. Although this technique has proven to be effective for producing certain devices such as metal-oxide-semiconductor (MOS) capacitors [ 10], solar cells require the entire structure and bulk to be defect free. Other researchers [ 11, 12] have experimentally shown lifetime degradation of p-type silicon due to laser induced damage but also subsequent recovery if the laser damage is etched off [ 12– 14]. There is, however, limited information on the application of laser induced gettering on substrates used on solar cell manufacturing.

This paper investigates the application and effectiveness of laser induced gettering techniques on commercial silicon wafers used in the photovoltaics industry. The use of a laser in combination with methods of diffusion gettering is explored and the subsequent effectiveness of laser enhanced gettering (LasEG) is presented. Interstitial iron concentrations [Fe_i] are used as a metric for measuring the relative efficiency of the gettering processes.

Experiment and results

The impact and effectiveness of LasEG on the minority carrier lifetime of silicon substrates was investigated on lifetime test structures with double-sided passivation [ 15]. Analysis was made using quasi-steady-state photo conductance (QSS-PC) measurements and spatially resolved photoluminescence (PL) imaging.

Sample fabrication

For this experiment, a series of adjacent, 6-inch, 180 mm multicrystalline and Czochralski (Cz) grown p-type silicon wafers were used. These wafers had a nominal bulk resistivity1.3±0.2 W·cm and 1.6±0.1 W·cm, respectively. The Cz wafers were etched in an isotropic alkaline solution to remove 15 mm of silicon from each side, while the mc-Si wafers were acidic etched to remove 8–10 mm per side in order to remove any saw damage. A subsequent alkaline etch comprising of sodium hydroxide (NaOH) and isopropanol was used to texture the Cz wafers with a random pyramid surface. An acidic texturing bath comprising of nitric acid (HNO₃) and hydrofluoric acid (HF) was used to texture the mc-Si wafers. Wafers were split into two major groups, half of which received an 85% (w/w) Phosphoric Acid (H₃PO₄) spin-on dopant (SOD) source spun at 6000 rpm for 40 s on the rear. Wafers were then irradiated over the full rear area using a Nd:YAG 532 nm continuous wave (CW) Newport laser with a beam diameter of approximately 20 mm. Laser processing was performed using a galvanometer scanner with a scan speed of 500 mm/s, laser spacing of 125 mm and a power density of 8.6 MW/cm². During preliminary testing and optimization, other parameters were used. However, some parameters may cause excessive laser damage which required prolonged processing to remove while others were below the gettering thresholds [ 10]. The best case tested parameters were utilized. A subsequent dip in 2% HF (w/w) and 5 min rinse in de-ionised (DI) water was used to remove any residual dopant source. All wafers were then RCA cleaned. Both laser irradiated and non-irradiated wafers were split into 4 groups, A, B, C and D, as shown in Fig. 1.

Wafers in Group A were used as a control and did not receive any thermal treatment. Wafers in Group B underwent a thermal anneal with an N₂ ambient in an oxidation tube. The thermal profile of the annealing was modified to match the temperatures and durations of the recipe used at UNSW to create lightly-doped Emitters(LDE). Wafers in Group C underwent the LDE diffusion process, that is, a phosphosilicate glass (PSG) formation and pre-deposition step with a peak temperature of 770°C, a phosphorus diffusion step at 850°C and finally, a drive-in step at 880°C. The resulting emitter had a sheet resistance of approximately 110 Ω/sq. Finally, wafers in Group D underwent a screen print (SP) emitter diffusion, resulting in an emitter of approximately 70 Ω/sq. Wafers in Groups C and D were etched in 2% HF for 2 min to remove the PSG layer.

All samples were then subject to a single sided rear-etch in a solution comprising of hydrofluoric acid (HF), nitric acid (HNO₃) and acetic acid (CH₃COOH) in order to remove 3–4 mm of the rear surface. A short duration submersion in dilute potassium hydroxide (KOH) solution was used to remove any porous silicon that may have formed during the rear-etch process.

Samples were than RCA cleaned in preparation for silicon nitride deposition. A 75 nm layer of silicon nitride (SiN_x) with a refractive index of 2.08 [ 16] was deposited on both sides using a Roth & Rau MAiA remote plasma enhanced chemical vapor deposition (r-PECVD) tool at a temperature of 400°C.

Characterization

Recombination in the wafers was characterized using QSS-PC measured on a Sinton WCT-120 photo conductance bridge and analyzed using generalized methods [ 17] with the intrinsic recombination removed using the Richter model [ 18]. As a metric for wafer performance, the bulk minority carrier lifetime was obtained by a method of curve fitting to separate the bulk and surface lifetime components from the measured effective carrier lifetime [ 19]. This was extracted at an injection level of \Delta n=\mathrm{1}\times {\mathrm{10}}^{\mathrm{16}} cm⁻³.

Characterization and extraction of interstitial iron concentration was done using a method described by Macdonald et al. [ 20]. Photo conductance measurements taken prior to and directly after light soaking established {\tau }_{\mathrm{FeB}} and {\tau }_{\mathrm{F}{\mathrm{e}}_{\mathrm{i}}} , and the carrier lifetimes before and after the dissociation of iron-boron pairs (FeB) into interstitial iron (Fe_i) and substitutional boron (B_s) respectively. A 20 s light soak under 1-sun illumination using a large area 808 nm laser was used as a method for iron dissociation. An injection level of \Delta n=\mathrm{9.1}\times {\mathrm{10}}^{\mathrm{14}}{\mathrm{cm}}^{-\mathrm{3}} was used to extract iron concentrations to ensure operation away from the crossover point [ 21, 22] due to insensitivity of the technique to small changes in carrier lifetime. All lifetime measurements were taken at 9 points across the wafer to account for any spatial variations in sheet resistivity as well as diffusion and silicon nitride non-uniformity.

Spatially resolved photoluminescence (PL) imaging was used to give a spatial analysis of substrate performance and interstitial iron concentrations. Measurements were taken at a 1-sun illumination and an exposure of 0.5 s using a BT imaging LIS-R1 photoluminescence imaging tool [ 23]. Image deconvolution was achieved by application of a point spread function (PSF) using the Matlab based software PLPro [ 24] in order to correct for photon smearing [ 25]. Combined with carrier lifetime measurements, calibrated effective lifetime maps were generated to show performance across each substrate. The excess carrier concentration before and after iron-boron dissociation ( \Delta n_{\mathrm{0}},\Delta n_{\mathrm{1}}^{ } ) and the respective carrier lifetime {\tau }_{\mathrm{FeB}},{\tau }_{{\mathrm{Fe}}_{\mathrm{i}}} were calibrated with the change in pixel counts to generate maps of iron concentration [ 20, 26]. Furthermore, another metric for performance that was derived from PL imaging was the implied open-circuit voltage (V_oc) [ 27].

Since the dissociation rate of FeB pairs dominate over the association rate above temperatures of 350°C, wafers were left in the dark for at least two hours after the 400°C silicon nitride deposition. This ensured the relaxation of interstitial Fe into FeB pairs prior to characterization [ 28].

Electrochemical capacitance-voltage (ECV) was used to characterize the active phosphorus dopant concentration within the SP and LDE emitter as well as the laser doped regions, as demonstrated in Fig. 2.

Laser enhanced gettering of interstitial iron

The impact of the different processes on the area averaged concentration of interstitial iron (Fig. 3) demonstrates the gettering effect of the phosphorous doping and laser treatment. Substantial differences in [Fe_i] were measured due to LasEG and thermal processing. Without any processing, multicrystalline samples consisted of an average [Fe_i] = 3.6×10¹¹ atoms/cm³ while monocrystalline samples consisted of an average [Fe_i] = 2.2×10¹¹ atoms/cm³. Once annealed in nitrogen, the concentration of interstitial iron increased, with both Cz-mono samples and mc-Si samples reaching an average concentration of [Fe_i] = 5.2±0.5×10¹¹ atoms/cm³. This increase in concentration was likely due to the Fe in the precipitated form (Fe_P) [ 29] being released during the annealing at high temperatures above 780°C. Samples that underwent LDE emitter diffusion alone exhibited a decrease in [Fe_i] from the control sample by a factor of 4 and a factor of 1 in mc-Si and Cz materials respectively. Since the thermal profile of the diffusion was identical to the N₂ annealing, the reduction in [Fe_i] when compared to the annealed samples would be a result of the phosphorus diffusion. Further reductions in [Fe_i] down to 3.6×10¹⁰ atoms/cm³ for mc-Si samples and 3.06×10¹⁰ atoms/cm³ for Cz were observed on samples that underwent a screen-print diffusion. This result not only demonstrates the advantage of gettering through a heavy diffusion, but also the reduction in Fe_i gettering efficiency by 16.49% from 90.32% to 73.83% in mc-Si materials and 44.35% from 85.62% to 41.27% in Cz materials when applying the lighter diffusion.

In general, the incorporation of the laser process reduced the concentration of interstitial iron. The exception to this was on as-cut wafers without extended thermal treatment, in which no changes in interstitial iron concentrations were seen with the LasEG process. Annealing in nitrogen with the laser processed surface lowered the concentrations of [Fe_i] by a factor of 5 on both Cz and mc-Si wafers. Similarly, a factor of 5 decrease in [Fe_i] was observed on LDE diffused samples, beyond the [Fe_i] obtained just by using the identical diffusion process with LasEG. The final [Fe_i] of 4.1×10¹⁰ atoms/cm³ equates to an order of magnitude reduction through using the LasEG process. The screen print diffusion process resulted in the lowest iron concentration compared to all others. When used on mc-Si wafers, the laser process only marginally enhanced the gettering process if at all but on monocrystalline wafers the iron concentrations were reduced essentially to the detection limit.

Further information about the impact of the different gettering processes on the movement of iron within the samples is revealed in the PL images and related [Fe_i] maps (Fig. 4, LDE with and without LasEG). Without the LasEG process, high concentrations of interstitial iron can be seen decorating the GBs (Fig. 4(A-2)). With the LasEG process, however (Fig. 4(B-2)), in both intra-grain areas and grain boundaries the iron concentrations are observed to be lower. From the calibrated implied open-circuit voltage mapping, the LasEG processed sample exhibits higher overall voltages, particularly within the intra-grain regions (Fig. 4(A-1 and B-1)). An increase in local intra-grain regions of up to 40 mV in implied open-circuit voltage was observed.

Possible explanations for the enhanced gettering in samples processed with a laser are the availability of additional phosphorus and introduction of laser damage. The peak [P] of the LDE diffusion and laser doped region are 2.5±0.2×10¹⁹ cm⁻³ and 7.4±0.2×10¹⁹ cm⁻³ respectively while the screen print diffusion produces a peak of 1.5×10²⁰ cm⁻³. These results correlate well with the work by Phang and Macdonald who have demonstrated the effectiveness of P gettering at peak concentrations greater than 1×10²⁰ cm⁻³. However, at below 5×10¹⁹ cm⁻³, the gettering efficiency decreases to below 50% [ 7]. At the measured concentration, phosphorus within the heavily doped regions generated by the LasEG process should evidently enhance the amount of Fe_i that would be gettered when combined with an LDE diffusion, however less effectively when applied with a SP diffusion.

An additional possibility is that laser induced dislocations generate a defect strain field, creating an effective gettering layer close to the surface of the silicon [ 10]. Metal impurity transported to the getter region via relaxation remove the mobile Fe_i from the bulk of the wafer. In addition to gettering via a reduced nucleation barrier, the elastic interactions which come about due to the size of both the Si atom and the covalent radius of the impurity cause defect segregation to such strain fields [ 30]. By creating a full area damage layer, the effective diffusion length required by impurities to migrate to the nearest gettering site is reduced. Metallic impurities, such as iron preferentially decorate at the artificial, laser doped regions instead of the nearest grain boundaries. In wafers with emitters that do not effectively getter metallic impurities, the LasEG process can be used to effectively collect impurities at the rear-surface and be subsequently removed. The decoration of the grain boundaries by the Fe_i in the non-lasered sample may also be explained by the relaxation gettering mechanism [ 26, 30]. In this case, the mechanism is driven primarily by structural defects within the silicon which act as effective getter sites [ 31]. Due to the lower nucleation barrier present at these sites, metallic impurities such as Fe_i are driven toward and readily precipitate once the solubility limit is reached. The diffusion length of these impurities toward gettering sites is dependent on the time and temperature of the thermal process. As the wafer cool, impurity diffusivity decreases, causing the concentration of iron within larger grains to remain relatively uniform.

It is worth noting that if the samples studied above were integrated into standard screen print solar cells, there would be an additional impact of the BSF formation process and Al layer on the rear side. It has been shown that Al layers can provide very effective sinks for gettering of interstitial iron. However, in the case of rapid contact firing this could only be expected to extend to the region close to that surface [ 32]. This may help to recover some of the gettering efficiency loss seen by the LDE emitter but the extent of this is expected to be minimal and would need to be determined experimentally. Furthermore, the transition to the PERC cell technology with localized Al-alloyed regions would reduce the effectiveness of Al gettering even further.

Impact of gettering process on bulk lifetime

The previous results have demonstrated that iron can be gettered through thermal diffusion and the gettering can be enhanced through using laser processes. However, the overall minority carrier lifetime can be affected by other impurities, laser induced dislocations, thermal processes, and various other aspects.

The impact of the thermal and laser processes on the bulk minority carrier lifetimes of both Cz and mc-Siwafers is illustrated in Fig. 5. Without thermal treatment, there is no significant difference between samples with and without laser irradiation. After thermal annealing in nitrogen ambient, degradation in both mc-Si and Cz recombination lifetime from the control of 50% and 21.5% respectively is observed on samples without laser processing. After LDE diffusion, samples that underwent laser treatment retained their performances. Cz and mc-Si samples that underwent the LasEG process both exhibited a 45% higher life time from 84 to 122 ms and from 86 to 126 ms, respectively. Mc-Si wafers that underwent SP diffusion exhibited an 18% increase (from 138 to 169 ms) while the Czwafers showed an unexpected 47% decrease (from 232 to 158 ms).

The behavior exhibited by sample minority carrier lifetime mostly correlates with the reduction in [Fe_i] highlighted in Fig. 3. The dissolution of precipitated iron during inert gas annealing resulted in a reduction in {\tau }_{\mathrm{bulk}} on both mc-Si and Cz wafers whereas other thermally processed samples achieved significantly higher lifetime compared to the as-cut material. In some cases, certain conditions resulted in higher bulk lifetime than other conditions with similar or greater amounts of iron. In particular, the monocrystalline samples processed with a screen print diffusion had a lower than expected lifetime. PL images of those samples (not shown here) revealed a pattern of recombination consistent with sample contamination from an extrinsic source. Such a contamination would be expected to introduce SRH defects into the bulk and thus reduce the bulk lifetime [ 33]. Laser damage would have a similar impact [ 13] and while it was not visible on the PL images it is not possible to fully rule it out. The difference in lifetime between LDE diffused wafers with and without laser processing however, follows closely with the trends in [Fe_i]. The factor of 5 reductions in iron on both mc-Si and Cz wafers contributed to the 45% increase in carrier lifetime. Importantly, for this high efficiency diffusion a higher bulk lifetime was achieved for the laser processed samples, indicating that the benefits of the enhanced getter outweighed any reduction due to the introduction of laser induced defects.

Conclusions

The impact of rear-side laser doping on the performance of silicon wafers and impurity gettering capabilities of thermal diffusions were analyzed. Comparisons between various thermal processes such as annealing in nitrogen and lightly-doped emitter diffusion further showed significant enhancements in bulk carrier lifetime when applying the laser process. Reduction in interstitial iron concentrations by a factor of 5 were observed on laser processed multicrystalline wafers, suggesting improvements in gettering efficiency. Furthermore, photoluminescence data both qualitatively and quantitatively indicated removal of iron from both grain boundaries and intra-grain regions. It was observed that by introducing artificial, heavily doped, dislocated regions, various mechanisms such as relaxation gettering and segregation gettering allow impurities to preferentially migrate and decorate such areas. Results showed improvements of up to 40 mV within localized intra-grain regions of mc-Si wafers.

Larger iron reductions of over half an order of magnitude were recorded on LDE diffused mc-Si and Cz wafers, corresponding to 45% higher bulk minority carrier lifetime when compared to the diffusion alone. Overall, this laser enhanced gettering (LasEG) process demonstrates significant gettering potential in low-cost multicrystalline wafers with lightly-doped high-efficiency emitters.

References

[1] International Technology Roadmap for Photovoltaic (ITRPV). International Technology Roadmap for Photovoltaic Results 2015, 7th edition. Frankfurt. 2016,

[2] Weber E. Transition metals in silicon. Applied Physics A, Materials Science & Processing, 1983, 30(1): 1–22

[3] Cuevas A, Macdonald D, Kerr M, Samundsett C, Sloan A, Shea S, Leo A, mrcarica M, Winderbaum S. Evidence of impurity gettering by industrial phosphorus diffusion.In: IEEE 28th Photovoltaic Specialists Conference (PVSC). 2000: 244–247

[4] Khedher N, Hajji M, Hassen M, Ben Jaballah A, Ouertani B, Ezzaouia H, Bessais B, Selmi A, Bennaceur R. Gettering impurities from crystalline silicon by phosphorus diffusion using a porous silicon layer. Solar Energy Materials and Solar Cells, 2005, 87(1–4): 605–611

[5] Bentzen A, Holt A, Kopecek R, Stokkan G, Christensen J, Svensson G. Gettering of transition metal impurities during phosphorus emitter diffusion in multicrystalline silicon solar cell processing. Journal of Applied Physics, 2006, 99(9): 093509

[6] Mayandi J, Marstein E S, Syre M, Olaisen B R, Thomassen B, Holt A, Vines L. The effect of P concentration on gettering of multicrystalline Si wafers. European Photovoltaic Solar Energy Conference (EU PVSEC), Hamburg, Germany, 2009, 2160–2163

[7] Phang S, Macdonald D. Effect of boron codoping and phosphorus concentration on phosphorus diffusion gettering. In: 2014 IEEE 40th Photovoltaics Specialist Conference (PVSC). Denver, USA, 2014, 64–69

[8] Hahn G, Joos S. State-of-the-art industrial crystalline silicon solar cells. In: Willeke G P, Weber F R, ed. Advances in Photovoltaics: Part 3. Academic Press, 2014, 1–72

[9] Yang K H, Schwuttke G H. Minority carrier lifetime improvement in silicon through laser damage gettering. physica status solidi (a), 1980, 58(1): 127–134

[10] Hayafuji Y, Yanada T, Aoki Y. Laser damage gettering and its application to lifetime improvement in silicon. Journal of the Electrochemical Society, 1981, 128(9): 1975–1980

[11] Matsuoka Y, Usami A, Shimazaki K. Effect of carrier lifetime on laser-induced damage in silicon. Journal of Physics D: Applied Physics, 1975, 8(1): 1–2

[12] Ametowobla M, Bilger G, Köhler J R, Werner J H. Laser induced lifetime degradation in p-type crystalline silicon. Journal of Applied Physics, 2012, 111(11): 114515

[13] Abbott M, Cousins P, Chen E, Cotter J. Laser-induced defects in crystalline silicon solar cells. In: 2005 IEEE 31st Photovoltaics Specialist Conference (PVSC). Lake Buena Vista, USA, 2005, 1241–1244

[14] Thorstensen J, Foss E. Investigation of depth of laser damage to silicon as function of wavelength and pulse duration. Energy Procedia, 2013, 38: 794–800

[15] Sinton R A, Cuevas A. A quasi-steady-state open-circuit voltage method for solar cell characterization. In: 16th European Photovoltaic Solar Energy Conference (EU PVSEC). Glasgow, UK, 2000: 1–4

[16] Hameiri Z, Borojevic N, Mai L, Nandakumar N, Kim K, Winderbaum S. Should the refractive index at 633 nm be used to characterize silicon nitride films? In: Proceedings of the 43rd IEEE Photovoltaic Specialists Conference (PVSC). Portland, USA, 2016

[17] Sinton R A, Cuevas A. Contactless determination of current–voltage characteristics and minority-carrier lifetimes in semiconductors from quasi-steady-state photoconductance data. Applied Physics Letters, 1996, 69(17): 2510–2512

[18] Richter A, Werner F, Cuevas A, Schmidt J, Glunz S W. Improved parameterization of Auger recombination in silicon. Energy Procedia, 2012, 27: 88–94

[19] Cuevas A, Macdonald D. Measuring and interpreting the lifetime of silicon wafers. Solar Energy, 2004, 76(1–3): 255–262

[20] Macdonald D, Tan J, Trupke T. Imaging interstitial iron concentrations in boron-doped crystalline silicon using photoluminescence. Journal of Applied Physics, 2008, 103(7): 073710

[21] Birkholz J E, Bothe K, Macdonald D, Schmidt J. Electronic properties of iron-boron pairs in crystalline silicon by temperature- and injection-level-dependent lifetime measurements. Journal of Applied Physics, 2005, 97(10): 103708

[22] Macdonald D, Geerligs L J, Azzizi A. Iron detection in crystalline silicon by carrier lifetime measurements for arbitrary injection and doping. Journal of Applied Physics, 2004, 95(3): 1021–1028

[23] Trupke T, Bardos R, Schubert M, Warta W. Photoluminescence imaging of silicon wafers. Applied Physics Letters, 2006, 89(4): 044107

[24] Payne D, Wenham S. Evaluating the accuracy of point spread function deconvolutions applied to luminescence images. In: Proceedings of the 43rd IEEE Photovoltaic Specialists Conference (PVSC). Portland, USA, 2016

[25] Teal A, Juhl M. Correcting the inherent distortion in luminescence images of silicon solar cells. In: Proceedings of the 42nd IEEE Photovoltaic Specialists Conference (PVSC). New Orleans, USA, 2015

[26] Liu A, Walter D, Phang S P, Macdonald d. Investigating internal gettering of iron at grain boundaries in multicrystalline silicon via photoluminescence imaging. IEEE Journal of Photovoltaics, 2012, 2(4): 479–484

[27] 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

[28] Reiss J H, King R R, Mitchell K W. Characterization of diffusion length degradation in Czochralski silicon solar cells. Applied Physics Letters, 1996, 68(23): 3302–3304

[29] Macdonald D, Cuevas A, Kinomura A, Nakano Y, Geerligs L J. Transition-metal profiles in a multicrystalline silicon ingot. Journal of Applied Physics, 2005, 97(3): 033523

[30] Seibt M, Kveder V. Gettering processes and the role of extended defects. In Advanced Silicon Materials for Photovoltaic Applications, 2012: 127–188

[31] Nouri H, Bouaïcha M, Ben Rabha M, Bessaïs B. Gettering effect in grain boundaries of multi-crystalline silicon. Physica Status Solidi. C, Current Topics in Solid State Physics, 2012, 9(10–11): 1937–1941

[32] Phang S P, Macdonald d. Direct comparison of boron, phosphorus, and aluminum gettering of iron in crystalline silicon. Journal of Applied Physics, 2011, 109(7): 073521

[33] Abbott M, Poplavskyy D, Scardera G, Inns D, Lemmi F, McIntosh K, Baker-Finch C S. Iron contamination in silicon solar cell production environments. In: Proceedings of the 40th IEEE Photovoltaic Specialists Conference (PVSC). Denver, USA, 2014

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag Berlin Heidelberg