POCl₃ diffusion for industrial Si solar cell emitter formation

Abstract

POCl₃ diffusion is currently the de facto standard method for industrial n-type emitter fabrication. In this study, we present the impact of the following processing parameters on emitter formation and electrical performance: deposition gas flow ratio, drive-in temperature and duration, drive-in O₂ flow rate, and thermal oxidation temperature. By showing their influence on the emitter doping profile and recombination activity, we provide an overall strategy for improving industrial POCl₃ tube diffused emitters.

Graphical abstract

Keywords

POCl₃ diffusion / emitter recombination / oxidation / silicon

1 Introduction

Phosphorus diffusion has been the de facto standard method for forming electron collectors for p-type crystalline Si solar cells since the 1970s. Currently, 95% of all industrial solar cells are fabricated from p-type substrates and the industry technology roadmap for photovoltaics (ITRPV) predicts that this dominance will continue with over 65% of industrial silicon solar cells still being p-type by 2026 [1]. The popularity of POCl₃ diffusion can be attributed to the low costs, good stability, relative simplicity, and high throughput of the available production equipment. In addition to forming the n-type region, the POCl₃ process has the additional benefit of improving substrate bulk minority carrier lifetime through gettering [2–7] which has been particularly important on lower cost substrates such as multi-crystalline wafers. The POCl₃ process continues to be optimised to support higher efficiency solar cells. The ITRPV predicts that the emitter sheet resistance (r _sh) will increase from 90 to 140 W/sq in the next 10 years and maintaining a high uniformity is quite challenging for these lightly diffused emitters. In addition the emitter saturation current density (J _0e) will reduce from 150 to 50 fA/cm² [1]. These developments are partly enabled by improvements in Ag screen printing pastes which allow for contacting of these increasingly shallower emitters. In addition, the tube furnace manufacturers continue to increase the throughput of their tools e.g. by decreasing the wafer pitch. Consequently, it is important to continually develop the POCl₃ diffusion process.

The POCl₃ process has been studied extensively by various research groups [8–17]. When phosphorus diffusion was initially used in Si solar cell fabrication, solid P₂O₅ sources were used as the dopant source [18]. Afterwards, liquid POCl₃ source diffusion was adapted widely due to its better control, uniformity and higher throughput. One-step diffusion recipes were common in which a single temperature and continuous flow of dopant gas were used to both deposit phosphosilicate glass (PSG) and drive dopants to the desired depth [11, 19]. This is a fast process but it tends to create an excessively heavily doped emitter to deteriorate the electrical performance [19, 20]. After the one-step process, an additional thermal oxidation was sometimes performed to drive the dopants further into the silicon and form a deeper junction [18]. POCl₃ diffusion can also be performed in a two-step manner: a PSG deposition step followed by a drive-in step (potentially at different temperatures). During the process, POCl₃ gas is only allowed in the PSG deposition step, and subsequently dopants are moved deeper from the PSG to the silicon substrate in the drive-in step. This paper focuses on the two-step process, due to its good trade-off between time and performance.

Conventional two-step POCl₃ diffusion can be broken down in two phases: ①deposition of the PSG layer and ② a subsequent drive-in to move the phosphorous deeper into the silicon. To further reduce the J _0e, an additional thermal oxidation step can be applied (at the expense of a longer process time). During each of these steps, the processing temperature and the process gas flows are varied to control the formation of the emitter. To give a clear view of the diffusion process a schematic temperature-time profile and gas flow-time profile are shown in Fig. 1. The details of each step are described in greater detail in Section 3 and 4.

The emitter plays a major role in determining the overall efficiency of a silicon solar cell [11, 21–24]. A high-efficiency emitter must minimize both resistive losses (contact and lateral) and recombination losses (at the surface as well as in the diffused region). Emitter recombination is typically characterized by J_0e and can be separated into 3 distinct components: Auger recombination, Shockley-Read-Hall (SRH) recombination in the diffused region, and surface SRH recombination [25]. Auger recombination is a three particle intrinsic recombination process and the recombination rate approximately scales with the square of the active doping concentration. Emitter SRH recombination scales linearly with the active defect density in the emitter region, which is the result of incomplete ionization of dopants in the heavily doped emitters [26]. Surface SRH recombination scales linearly with the surface defect density and is also strongly affected by the surface doping concentration [25, 27–29]. In this paper, we demonstrate how to minimize the SRH recombination from both the emitter bulk and surface by properly optimising the POCl₃ diffusion process.

In the past few decades, many research groups have published their investigations regarding particular key processing parameters, but there are very few reports presenting a systematic study about the overall optimisation strategy. This is particularly important with the continual processing tool development which allows for greater control of the process metrics. This paper demonstrates an overall strategy to control POCl₃ diffusion in a commercial tube furnace for solar cell fabrication, by showing how each processing parameters can influence the active doping profiles and the emitter electrical properties. We also demonstrate how to identify major emitter recombination using numerical modeling to give a clear indication about future development, using experimental data from our laboratories as an example.

2 Method

2.1 Sample preparation

The samples used in this work were 156 mm × 156 mm pseudo-square Czochralski (Cz) grown p-type silicon wafers with a nominal bulk resistivity of 1.6 W·cm. Both surfaces were textured with upright random pyramids using alkaline etching. After RCA clean with a final HF dip, the wafers were loaded into a quartz tube diffusion furnace (TS8603, Tempress) at 750°C to start the POCl₃ diffusion process. Experimental splits (see Table 1) were performed to investigate the influence of the following processing parameters: gas flow ratio during PSG deposition, temperature and duration and O₂ flow rate during drive-in, and oxidation temperature. The total gas flow rate in the tube was set to 7.25 slm (standard liter per minute) for PSG deposition and drive-in and 15 slm for oxidation. The temperature ramp rate was set to 10°C/min for all heating and cooling steps. Finally, a 75 nm SiN _x film with a refractive index of 2.05 was deposited by plasma-enhanced chemical vapor deposition (PECVD, MAiA XS, Roth & Rau). Lifetime samples with a symmetric n⁺pn ⁺ structure were coated by SiN _x on both sides, while solar cell samples were only single side coated. Lifetime samples were subsequently fired using a standard industrial metallization co-firing process (7KX-70C96-5LIR, Sierra Therm) with a peak set-temperature of 855°C (time over 700°C was ~ 2– 3 s). Each full sized cell was laser cut into 3.9 cm by 3.9 cm small tokens, and then screen printed. A full area Al print (Monocrystal PASE1203) was used at the rear while at the front side a single busbar grid pattern (finger width 100 µm and spacing of 2.3 mm) was printed using commercial Ag paste (DuPont PV17). Subsequently the samples were fired with the same process as the lifetime samples. Finally each sample was cut (for edge isolation) into 2.9 cm by 2.9 cm tokens using a Nd:YAG 1064 nm laser.

2.2 Characterization

Sheet resistance (r _sh) was measured at 49 points equally spaced over the wafer by a four-point-probe (SheRRescan, Mechatronics).The depth dependent electrically active doping profiles of each emitter were measured using an electrochemical capacitance-voltage (ECV, CVP21, WEP) analyzer, calibrated using the r _sh to account for the surface roughness [30]. Injection level dependent effective minority carrier lifetime measurements were performed at 9 points across the sample using a Sinton WCT-120 lifetime tester [31]. Lifetime data was analyzed to extract the J _0e value at an injection level of 1 × 10¹⁶ cm^{− 3} using the Kane and Swanson method [32]. Thermal oxide thickness was measured by spectral ellipsometry on polished wafers processed in parallel with the textured wafers. The wavelength dependent external quantum efficiency (EQE) over the wavelength range of 350–1300 nm was measured by a PV Measurements QEX7 Spectral Response System, while reflectance (R) was measured by a Perkin-Elmer Lambda 950 UV-vis-NIR spectrophotometer. Spectral absorption data (A) for the silicon nitride films reported by Duttagupta et al. was used [33] to calculate internal quantum efficiency using the following equation:

\mathrm{I}\mathrm{Q}\mathrm{E}={\displaystyle \frac{\mathrm{E}\mathrm{Q}\mathrm{E}}{\mathrm{1}-R-A}}.

Emitter modeling

The attribution of Auger recombination, emitter and surface SRH recombination to the overall J _0e was simulated using EDNA2 [34–36]. EDNA2 assumes quasi-neutrality in the diffused region, and allows the use of semiconductor models. In this work, we applied the following models: Richter’s Auger recombination model, Klaassen’s mobility model, Pässler’s E _gi model with an E _gi multiplier of 1.00547, Altermatt’s dopant ionisation model, Sentuarus’s DOS model, Schenk’s BGN model and Fermi-Dirac statistics. The full list of references for these models can be found in Ref. [36]. Experimental active doping profiles were imported to the program and the surface dark saturation current density (J _0s) and emitter minority carrier lifetime (t _p0) were varied to fit the experimental J _0e (original data divided by 1.7 to account for the difference in surface area between a textured and a planar surface [37]) and the experimental IQE iteratively. In this way the contribution from the above three recombination to the J _0e can be simulated to identify the major loss process.

3 Results

3.1 PSG deposition

During the first deposition step, Si wafers are exposed to an atmosphere of POCl₃, O₂ and N₂, where POCl₃ and O₂ react with the Si substrate forming a mixture of P₂O₅ and SiO₂ [38–40]. This so-called phosphorus-silicate glass (PSG) acts as a dopant source during the subsequent high temperature drive-in process [41]. Various studies have demonstrated that the POCl₃:O₂ ratio and deposition temperature have a significant impact on the resulting doping profiles and the J_0e [10, 38, 42–45]. In particular a higher POCl₃:O₂ ratio and a higher deposition temperature increase the phosphorus concentration in the PSG layer, consequently reducing the r_sh due to higher surface doping concentration [10, 19, 46, 47]. Figure 2 illustrates the impact of POCl₃:O₂ ratio during the deposition phase (followed by 880°C drive-in with no O₂) on the final phosphorus doping profiles. A high POCl₃:O₂ ratio resulted in a higher surface active doping concentration and a deeper junction, i.e. more dopants diffused into the substrate. The amount of dopants has a substantial influence of the J_0e value. J_0e values of SiN _x passivated samples (Fig. 2) increased from 90 to 150 fA/cm² when POCl₃:O₂ ratio was increased from 350:600 to 450:600, and r_sh reduced from 83 to 68 W/sq. Although a lower POCl₃:O₂ ratio results in lower J_0e values, it typically also causes a higher cross-wafer variation of r_sh which can have negative effects on a cell level e.g. contacting issues [15]. In our cases, the standard deviation of r_sh increased from 2.0 to 5.2 W/sq while reducing the POCl₃:O₂ ratio from 450:600 to 300:600.

3.2 Drive-in

After PSG deposition POCl₃ is no longer fed into the main gas stream leaving the PSG layer as the only dopant source for each wafer. During the drive-in step, phosphorus dopants continue diffusing from the PSG layer further into the Si against the dopant concentration gradient. The ambient temperature determines the dopant diffusivity which allows the formation of a deep junction at high temperatures, and also determines the solid solubility limit of phosphorus in Si which defines the surface doping concentration at a certain temperature as shown by Tsai et al.[12]. While in the case of an extended drive-in, dopants are only diffused further into the Si to form a deeper junction without changing the surface concentration. By doing so, the amount of dopants can be controlled to ensure sufficient electrical conductivity for high fill factor (FF) on finished devices. The impact of the drive-in time and temperature (with deposition of POCl₃:O₂ = 400:600, 795°C, 25 min), is shown in Fig. 3. Higher temperature resulted in a higher active surface doping concentration, a deeper junction and an increase in J _0e from 58 to 220 fA/cm². A longer drive-in duration resulted in a deeper junction depth (from 0.35 to 0.45 µm) while increasing the duration from 40 to 80 min. The surface doping concentration remained constant since it is only determined by the temperature when the silicon is saturated with phosphorus, assuming the composition in the PSG layer is consistent [12, 48]. As a result, r _sh was reduced from 145 to 110 W/sq, and J _0e increased from 55 to 85 fA/cm² due to a higher amount of total dopants.

The O₂ flow rate used during the drive-in step is another important metric to determine the emitter properties [15, 19, 49]. Varying the O₂ flow during the drive-in step has a substantial impact on the active doping profiles (Fig. 4). The surface active doping concentration reduced from 5 × 10²⁰ cm^{− 3} to 5 × 10¹⁹ cm^{− 3} and the junction depth reduced from 0.55 to 0.4 mm, when increasing O₂ flow rate from 0 to 200 sccm. As a result, J _0e was substantially reduced from 120 to 53 fA/cm².

Another metric used to characterize the emitter recombination is the internal quantum efficiency (IQE) measured on a finished solar cell [45, 50, 51]. This describes the probability of light generated carriers being collected and contributing to the current as a function of wavelength. Due to the high absorption coefficient for short wavelength light, the IQE in this wavelength range can be used to qualify the electronic quality of the emitter. Figure 5 depicts a clear increase of the IQE for shorter wavelengths when higher O₂ flow rates were used during the drive-in step.

3.3 Thermal oxidation

Thermal oxidation is a well-known method to passivate c-Si surfaces as the thermal silicon oxide provides an excellent surface passivation due to its low interface defect density [25, 52–56], hence, it is used for some (mainly laboratory) high efficiency Si solar cells [57]. A thermal oxide can be grown in a tube furnace and this can e.g. be done after stripping the PSG layer. Such a process is not very popular in industry as lengthy high temperature processing steps are not compatible with the solar-grade silicon wafers (e.g. multicrystalline silicon) which are commonly used in industry. In addition the formation of a thermal oxide also changes the phosphorus dopant profile [16, 58, 59], which can cause contacting problems for screen printed solar cells although the use of selective emitter structures may overcome this issue [60, 61].

The impact of a post-diffusion thermal oxidation step on the doping profile is shown in Fig. 6. When a high temperature process (930°C for 15 min) was used on a typical industrial emitter profile (Fig. 6 (a)) the growth of a 50 nm thermal silicon oxide film resulted in a reduction in the surface phosphorus concentration from 4 × 10²⁰ to 3 × 10¹⁹ cm⁻³ and an increase in junction depth from 0.6 to 0.8 µm. As a result, J_0e was reduced from 83 to 45 fA/cm². This effect was much more dramatic than a lower temperature oxidation (Fig. 6 (b)) which resulted in a 5 nm thermal silicon oxide and only a minor reduction in surface doping concentration from 4 × 10²⁰ to 2 × 10²⁰ cm⁻³. With this minor change in emitter doping profiles, r_sh increased slightly from 69 to 75 W/sq after the 830°C oxidation. Keeping all these controlled properties, the thin thermal oxide growth made a significant difference as J_0e was reduced substantially from 88 to 55 fA/cm². In addition, the low surface doping concentration after the 930°C oxidation does not allow for low contact resistance by the screen printing technology, which is widely used in industrial solar cell fabrication. This highlights the potential advantage provided by the 830°C oxidation, where low J_0e can be achieved and an active surface concentration of 2 × 10²⁰ cm⁻³ still allows a low specific contact resistance value down to 4 mW·cm² using Dupont PV17A paste [62]. In addition to J_0e, the IQE shown in Fig. 7 shows that low temperature oxidation allows a higher carrier collection probability, indicating an overall reduction of emitter recombination.

3.4 Emitter recombination modeling

To demonstrate how the dominant recombination type can be determined by using EDNA2 modeling, the 830°C sample set was used as an example. J_0e breakdown and the change of t_p0, S_p0 and J_0s of the as-diffused and oxidized samples are presented in Fig. 8 and Table 2, respectively. The major J_0e reduction is attributed to the reduction in emitter SRH recombination by 63%, and the reduction in surface SRH recombination by 57%. To support this, the fitting values show that t_p0 increases from 50 to 120 ns, and J_0s decreases from 30 to 13 fA/cm² after the 830°C thermal oxidation. The change of these two parameters indicates the reduction in emitter bulk defects and surface defects after oxidation, resulting in an overall reduction in J_0e.

4 Discussion

In this section, we will explain how the POCl₃ diffusion process changes the doping profiles and how the emitter recombination can be influenced as a result of changes in doping profiles.

4.1 Auger recombination

Auger recombination is determined by the active doping concentration which can be determined by the ECV method, and it roughly scales with the square of the active doping concentration. In each stage of the process, active doping profiles can be changed by the key processing parameters as presented in Section 3.

Figure 2 shows that a higher POCl₃:O₂ ratio during deposition results in a higher surface doping concentration and a deeper junction. A high POCl₃ flow rate enhances the concentration gradient across the PSG/Si interface, resulting in a higher phosphorus flux diffusing into the Si substrate, hence, a higher active doping concentration. As a result, the active dopant dose increased from 5.51 × 10¹⁸ to 2.33 × 10¹⁹ cm^{− 2} with increasing the temperature, calculated by integrating the area under the active doping profiles. Therefore the Auger recombination is enhanced by increasing POCl₃:O₂ ratio during PSG deposition.

During the subsequent drive-in step, a higher temperature and a longer duration can enhance the active doping concentration as shown in Fig. 3. The ambient temperature determines the dopant diffusivity which allows the formation of a deep junction at high temperatures, and also determines the solid solubility limit of phosphorus in Si which defines the surface doping concentration at a certain temperature as shown by Tsai et al.[12].Therefore, the surface active doping concentration increased from 2 × 10²⁰‚to 5 × 10²⁰ cm^{− 3} and the junction depth increased from 0.35 to 0.55 µm while the drive-in temperature increased from 870°C to 900°C. As a result, the active dopant dose increased from 4.19 × 10¹⁸ to 1.80 × 10¹⁹ cm^{− 2} with increasing temperature. While in the case of an extended drive-in, dopants are only diffused further into the Si to form a deep junction (increased from 0.35 to 0.45 µm in this case) but keeping the surface concentration constant at 2 × 10²⁰ cm^{− 3} [12]. The active dopant dose increased from 4.19 × 10¹⁸ to 5.51 × 10¹⁸ cm^{− 2}. Besides temperature and time, the O₂ flow rate used during the drive-in step is another important metric to determine the emitter properties [15, 19, 49]. Varying the O₂ flow during the drive-in step has a substantial impact on the active doping profiles. At higher O₂ flows an interfacial oxide is grown between the PSG layer and the Si substrate which effectively oxidizes the heavily doped Si near the surface and acts as a barrier layer to slow down the dopant flux diffusing from the PSG layer into the substrate. Figure 4 demonstrates the reduction in both surface doping concentration and junction depth, hence, the active dopant dose reduces from 1.43 × 10¹⁹ to 3.98 × 10¹⁸ cm^{− 2}, reducing Auger recombination for samples fabricated using high O₂ flow rate during the drive-in step. In summary, a higher temperature, a longer process and a lower O₂ flow rate during the drive-in step can result in a high active doping concentration and hence increases the Auger recombination.

The thermal oxidation process after POCl₃ diffusion oxidizes the emitter surface and changes the active doping profiles depending on the oxidation temperature (Fig. 6) [63, 64]. It can be seen that a 930°C oxidation reduced the active dopant dose from 1.34 × 10¹⁹ to 7.19 × 10¹⁸ cm^{− 2} whereas the 830°C oxidation reduced the active dopant dose from 1.25 × 10¹⁹ to 9.60 × 10¹⁸ cm^{− 2}. This is in good agreement with Bazer-Bachi et al. who demonstrated that thermal oxidation reduced the total amount of dopants, hence, reduces Auger recombination [63, 64].

In addition to the Auger recombination, band gap narrowing (BGN) can be caused by high active doping concentrations. BGN is detrimental to the effective minority carrier lifetime and J _0e [65–68], hence, reducing the active dopants as discussed above can reduce the adverse impact by BGN on the emitter quality.

4.2 Emitter and surface SRH recombination

The emitter and surface SRH recombination are the other two major contributing factors to the overall emitter recombination. They are determined by the active defect density in the diffused region and the surface defect density, respectively.

During the PSG deposition, the active phosphorus doping concentration can reach its solid solubility limit at high POCl₃:O₂ ratios, in some cases a flat plateau can be formed [10, 40, 62]. In this case, there will be a discrepancy between the active and total phosphorus concentration [69, 70]. The total phosphorus concentration has a substantial influence on the emitter saturation current density as inactive phosphorus atoms form precipitates with a size in the range of 0.1 to 0.2 µm which act as point defects and therefore significantly increase the SRH recombination in the emitter, which has been recently demonstrated as a major limitation in some industrial phosphorus diffused emitters [8, 11, 71–73]. As a result, minority carrier lifetime in the heavily doped region can be substantially reduced [9, 14, 20, 21, 73, 74]. At the worst case, misfit dislocations induced by heavy diffusion could extend to the bulk region, resulting in electrical defects to degrade the device performance [13].

The drive-in temperature not only determines the dopant diffusivity during this step, but also defines its solid solubility limit and how many dopants are electrically active [12]. It is reported that when silicon is saturated with phosphorus, a precipitated phosphorus phase can exist at the surface and the surface concentration is determined only by the temperature [48]. At a higher temperature, the higher diffusivity allows a higher total doping concentration and this temperature determines how many of these dopants are beyond the solid solubility limit to become inactive. As discussed previously, these inactive dopants contribute to emitter SRH recombination and they are detrimental to the minority carrier lifetime in the doped region. In the case of varying O₂ flow rates, the surface doping concentration can be reduced by a higher O₂ flow rate (Fig. 3), since the heavily doped region near the surface can be oxidized and the dopant flux can be significantly reduced by the barrier effect of the interfacial oxide. It has been shown that most of the inactive dopants are located at the first few tens of nm from the diffused surface [38, 62, 75]. By reducing these inactive-dopant-rich region, emitter SRH recombination can be effectively suppressed, hence, reducing the overall J _0e depicted in Fig. 4.

In addition to oxidizing the heavily doped near surface region, the thermal oxidation step can also reduce emitter recombination by altering the dopant profile thereby reducing the non-active phosphorus concentration [16]. This impact is more pronounced for the 930°C oxidation as shown in Fig. 6, where the surface active doping concentration was reduced from 4 × 10²⁰ to 3 × 10¹⁹ cm^{− 3}. Additionally, thermal oxide can provide effective surface passivation to reduce the surface defect density [25, 52–56]. As a result, both surface and emitter SRH recombination can be substantially reduced by thermal oxidation, showing an overall J _0e reduction as shown in Fig. 6. This is further supported by a significant increase in short-wavelength IQE after the 830°C oxidation as shown in Fig. 7. Emitter modeling by EDNA2 was used to further analyze the dominant loss mechanism of the as-diffused and oxidized samples. As shown in Fig. 8, the emitter and surface SRH recombination contribute to more than 67% of the overall emitter recombination for the as-diffused samples, hence, reducing the SRH recombination should be the focus for improving industrial emitters. After thermal oxidation, although the active surface doping profile was slightly changed, the emitter and the surface SRH recombination was reduced drastically as discussed in Section 3.4. These improvements can most likely be attributed to a decrease in emitter bulk defects such as reduced inactive dopants and/or improved surface passivation. This is supported by the reduced J _0s and the increased t _p0 values at active surface doping concentration of 2 × 10²⁰ cm^{− 3}, i.e. the oxidized samples in this case. In summary, the 830°C oxidation effectively reduced the emitter bulk defects and surface defects, contributing to an overall reduction in J _0e.

5 Conclusions

POCl₃ diffusion is widely used in industrial solar cell fabrication due to its high throughput, impurity gettering capability, and reliability. The continual development of high performance emitters is of interest for industrial high efficiency solar cells. Although POCl₃ diffusion has been used for decades, there are very few reports in the literature that include the influence of each key processing parameter on the emitter performance and provide an overall optimization guide. In this paper, the following key processing parameters were identified: the PSG deposition gas ratio, the drive-in conditions including temperature, duration and O₂ flow rate, and the thermal oxidation temperature after POCl₃ diffusion. From these results, we provide an overall guideline to demonstrate how emitter formation could be controlled to fulfill different device application, ensuring glow recombination and high electrical conductivity emitters. As a clear guideline, the impacts of the above processing parameters on emitter doping profile and emitter saturation current density are listed in Table 3. By controlling each of these processing parameters, we demonstrated a pathway to improve the emitter performance by identifying the dominate emitter recombination mechanism and providing a viable solution to tackle both emitter SRH recombination and surface SRH recombination.

References

[1] International Technology Roadmap for Photovoltaic. (ITRPV.net) Results 2015 Version 2. 2015

[2] Lossen J, Lauer K, Mittelstädt L, Dauwe S, Beneking C. Making use of silicon wafers with low lifetimes by adequate POCl₃ diffusion. In: The 20th European Photovolatic Solar Energy Conference. Barcelona, Spain, 2005

[3] Manshanden P, Geerligs L J. Improved phosphorous gettering of multicrystalline silicon. Solar Energy Materials and Solar Cells, 2006, 90(7–8): 998–1012

[4] Härkönen J, Lempinen V P, Juvonen T, Kylmäluoma J. Recovery of minority carrier lifetime in low-cost multicrystalline silicon. Solar Energy Materials and Solar Cells, 2002, 73(2): 125–130

[5] Joonwichien S, Takahashi I, Matsushima S, Usami N. Enhanced phosphorus gettering of impurities in multicrystalline silicon at low temperature. Energy Procedia, 2014, 55: 203–210

[6] Vacek V, Žáček S, Horsák I. Phosphorus and boron diffusion gettering of iron in monocrystalline silicon. Journal of Applied Physics, 2011,109(109): 093505

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

[8] Cerofolini G F. Polignano M L, Nava F, Ottaviani G. On the mechanism responsible for phosphorus inactivation in heavily doped silicon. Thin Solid Films, 1982, 97(4): 363–367

[9] Solmi S. Parisini A, Angelucci R, Armigliato A, Nobili D. Dopant and carrier concentration in Si in equilibrium with monoclinic SiP precipitates. Physical Review B, 1996, 53(12): 7836–7841

[10] Dastgheib-Shirazi A, Steyer M, Micard G, Wagner H, Altermatt P P. Relationships between diffusion parameters and phosphorus precipitation during the POCl₃ diffusion process. Energy Procedia, 2013, 38(0): 254–262

[11] Ostoja P, Guerri S, Negrini P, Solmi S. The effects of phosphorus precipitation on the open-circuit voltage in N⁺/P silicon solar cells. Solar Cells, 1984, 11(1): 1–12

[12] Tsai J C C. Shallow phosphorus diffusion profiles in silicon. Proceedings of the IEEE, 1969, 57(9): 1499–1506

[13] Cousins P J, Cotter J E. The influence of diffusion-induced dislocations on high efficiency silicon solar cells. IEEE Transactions on Electron Devices, 2006, 53(3): 457–464

[14] Kimmerle A, Wolf A, Belledin U, Biro D. Modelling carrier recombination in highly phosphorus-doped industrial emitters. Energy Procedia, 2011, 8: 275–281

[15] Wolf A, Kimmerle A, Werner S, Maier S, Belledin U, Meier S, Biro D. Status and perspective of emitter formation by POCl₃-diffusion. 2015

[16] Bazer-Bachi B, Fourmond E, Papet P, Bounaas L, Nichiporu k O. Higher emitter quality by reducing inactive phosphorus. Solar Energy Materials and Solar Cells, 2012, 105(10): 137–141

[17] Altermatt P P, Schumacher J O, Cuevas A, Kerr M J, Glunz S W. Numerical modeling of highly doped Si: P emitters based on Fermi–dirac statistics and self-consistent material parameters. Journal of Applied Physics, 2002, 92(6): 3187–3197

[18] Tannenbaum E. Detailed analysis of thin phosphorus-diffused layers in p-type silicon. Solid-State Electronics, 1961, 2(2–3): 123–132

[19] Morris B L, Katz L E. Reduction of excess phosphorus and elimination of defects in phosphorus emitter diffusions. Journal of the Electrochemical Society, 1978, 125(5): 762–765

[20] Nobili D. Precipitation as the phenomenon responsible for the electrically inactive phosphorus in silicon. Journal of Applied Physics, 1982, 53(3): 1484–1491

[21] Komatsu Y, Vlooswijk A H G, Stassen A F, Venema P, Meyer C. Sophistication of doping profile manipulation-emitter performance improvement without additional process step. The 25th European Photovoltaic Solar Energy Conference and Exhibition-5th World Conference on Photovoltaic Energy Conversion. Valenia, Spain, 2010, 6: 10

[22] Burrows M Z. Meisel A, Scardera G, Lemm i F. Front metal and diffusion optimization for selective emitter. In: 38th IEEE Photovoltaic Specialists Conference (PVSC). 2012: 002138–002141

[23] Zhao J, Wang A, Green M A. Emitter design for high-efficiency silicon solar cells. Part I: Terrestrial cells. Progress in Photovoltaics: Research and Applications, 1993, 1(3): 193–202

[24] Cuevas A, Balbuena M. Thick-emitter silicon solar cells. In: Photovoltaic Specialists Conference. 1988, 1: 429–434

[25] Kerr M J. Surface, emitter and bulk recombination in silicon and development of silicon nitride passivated solar cells. Dissertation for the Doctoral Degree. Canberra: Australian National University, 2002

[26] Altermatt P P, Schenk A, Heiser G. A simulation model for the density of states and for incomplete ionization in crystalline silicon. I. Establishing the model in Si: P. Journal of Applied Physics, 2006, 100(11): 113714

[27] Cuevas A, Kerr M J, Schmidt J. Passivation of crystalline silicon using silicon nitride. In: Proceedings of 3rd World Conference on Photovoltaic Energy Conversion, 2003, 1: 913–918

[28] Cuevas A, Basore P A, Giroult-Matlakowski G, Dubois C. Surface recombination velocity of highly doped n-type silicon. Journal of Applied Physics, 1996, 80(6): 3370–3375

[29] Kerr M J, Schmidt J, Cuevas A, Bultman J H. Surface recombination velocity of phosphorus-diffused silicon solar cell emitters passivated with plasma enhanced chemical vapor deposited silicon nitride and thermal silicon oxide. Journal of Applied Physics, 2001, 89(7): 3821–3826

[30] Bock R, Altermatt P P, Schmidt J. Accurate extraction of doping profiles from electrochemical capacitance voltage measurements. In: European Photovoltaic Solar Energy Conference. Valencia, Spain. 2008:1(5)

[31] Sinton R A, Cuevas A. A quasi-steady-state open-circuit voltage method for solar cell characterization. In: Proceedings of the 16th European Photovoltaic Solar Energy Conference. 2000, 1152

[32] Kane D E, Swanson R M. Measurement of the emitter saturation current by a contactless photoconductivity decay method. In: IEEE Photovoltaic Specialists Conference. 1985, 18: 578–583

[33] Duttagupta S, Ma F, Hoex B, Mueller T, Aberle A G. Optimised antireflection coatings using silicon nitride on textured silicon surfaces based on measurements and multidimensional modelling. Energy Procedia, 2012, 15(17): 78–83

[34] PVLIGHTHOUSE.

[35] Abbott M, Scardera G, Mcintosh K R, Meisel A. Simulation of emitter doping profiles formed by industrial POCl₃ processes. In: The 39th Photovoltaic Specialists Conference (PVSC). 2013: 1383–1388

[36] Abbott M D. An examination of three common assumptions used to simulate recombination in heavily doped silicon. In: Proceedings of the 28th EU PVSEC. Paris, France. 2013: 1672–1679

[37] McIntosh K R, Johnson L P. Recombination at textured silicon surfaces passivated with silicon dioxide. Journal of Applied Physics, 2009, 105(12): 124520

[38] Wagner H, Dastgheib-Shirazi A, Chen R, Dunham S T, Kessler M. Improving the predictive power of modeling the emitter diffusion by fully including the phosphsilicate glass (PSG) layer. In: Photovoltaic Specialists Conference (PVSC). 2011: 002957–002962

[39] Chen R, Wagner H, Dastgheib-Shirazi A, Kessle r M. Understanding coupled oxide growth and phosphorus diffusion in POCl₃ deposition for control of phosphorus emitter diffusion. In: Photovoltaic Specialists Conference (PVSC). 2012: 000213–000216

[40] Micard G, Dastgheib-Shirazi A, Altermatt P P, Solar T. Advances in the understanding of phosphorus silicate glass (PSG) formation for accurate process simulation of phosphorus diffusion. In: The 27th European Photovoltaic Solar Energy Conference. Frankfurt, Germany. 2012

[41] Hu S M, Fahey P, Dutton R W. On models of phosphorus diffusion in silicon. Journal of Applied Physics, 1983, 54(12): 6912–6922

[42] Rothhardt P, Keding R, Belledin U, Wolf A, Biro D. Control of phosphorus doping profiles for co-diffusion processes. In: Proceedings of the 27th European Photovoltaic Solar Energy Conference and Exhibition. 2012

[43] Safiei A, Windgassen H, Wolter K, Kurz H. Emitter profile tailoring to contact homogeneous high sheet resistance emitter. Energy Procedia, 2012, 27: 432–437

[44] Rothhardt P, Demberger C, Wolf A, Biro D. Co-diffusion from APCVD BSG and POCl₃ for industrial n-type solar cells. Energy Procedia, 2013, 38: 305–311

[45] Lee H J, Kang M G, Choi S J, Kang G H, Myoung J M. Characteristics of silicon solar cell emitter with a reduced diffused phosphorus inactive layer. Current Applied Physics, 2013, 13(8): 1718–1722

[46] Duffy M C, Barson F, Fairfield J M, Schwuttke G H. Effects of high phosphorus concentration on diffusion into silicon. Journal of the Electrochemical Society, 1968, 115(1): 84–88

[47] Steyer M, Dastgheib-Shirazi A, Hahn G, Terheiden B. New method for determination of electrically inactive phosphorus in n-type emitters. Energy Procedia, 2015, 77: 316–320

[48] Kooi E. Formation and composition of surface layers and solubility limits of phosphorus during diffusion in silicon. Journal of the Electrochemical Society, 1964, 111(12): 1383–1387

[49] Kumar D, Saravanan S, Suratkar P. Effect of oxygen ambient during phosphorous diffusion on silicon solar cell. Journal of Renewable and Sustainable Energy, 2012, 4(3): 033105

[50] de Rose R, Zanuccoli M, Magnone P, Frei M, Sangiorgi E. Understanding the impact of the doping profiles on selective emitter solar cell by two-dimensional numerical simulation. IEEE Journal of Photovoltaics, 2013, 3(1): 159–167

[51] Carroll A F. Screen printed metal contacts to Si solar cells-formation and synergistic improvements. In: The 39th Photovoltaic Specialists Conference (PVSC). 2013: 3435–3440

[52] Biro D, Mack S, Wolf A, Lemke A, Belledin U. Thermal oxidation as a key technology for high efficiency screen printed industrial silicon solar cells. In: Proceedings of the 34th IEEE Photovoltaic Specialists Conference. PA, USA, 2009: 1594–1599

[53] Schultz O, Mette A, Hermle M, Glunz S W. Thermal oxidation for crystalline silicon solar cells exceeding 19% efficiency applying industrially feasible process technology. Progress in Photovoltaics: Research and Applications, 2008, 16(4): 317–324

[54] Schultz O, Hofmann M, Glunz S W, Willeke G P. Silicon oxide/silicon nitride stack system for 20% efficient silicon solar cells. In: The 31st IEEE Photovoltaic Specialists Conference. 2005: 872–876

[55] Ortega P, Vetter M, Bermejo S, Alcubilla R. Very low recombination phosphorus emitters for high efficiency crystalline silicon solar cells. Semiconductor Science and Technology, 2008, 23(12): 125032

[56] Janssens T, Posthuma N E, van Kerschaver E, Baer t K. Advanced phosphorous emitters for high efficiency Si solar cells. In: The 34th IEEE Photovoltaic Specialists Conference (PVSC). 2009: 000878–000882

[57] Blakers A W, Wang A H, Milne A M, Zhao J H, Green Martin A. 22.8% efficient silicon solar cell. Applied Physics Letters, 1989, 55(13): 1363–1365

[58] Masetti G, Solmi S, Soncini G. On phosphorus diffusion in silicon under oxidizing atmospheres. Solid-State Electronics, 1973, 16(12): 1419–1421

[59] Florakis A, Janssens T, Posthuma N, Delmotte J, Douhard B. Simulation of the phosphorus profiles in a c-Si solar cell fabricated using POCl₃ diffusion or ion implantation and annealing. Energy Procedia, 2013, 38: 263–269

[60] Hallam B, Wenham S, Chong C M, Sugianto A, Mai L. Record large-area p-type CZ production cell efficiency of 19.3% based on LDSE technology. IEEE Journal of Photovoltaics, 2011, 1(1): 43–48

[61] Lee E, Lee H, Choi J, Oh D, Shim J. Improved LDSE processing for the avoidance of overplating yielding 19.2% efficiency on commercial grade crystalline Si solar cell. Solar Energy Materials and Solar Cells, 2011, 95(12): 3592–3595

[62] Cooper I B, Tate K, Carroll A F, Mikeska K R, Reedy R C. Low resistance screen-printed Ag contacts to POCl₃ emitters with low saturation current density for high efficiency Si solar cells. In: Photovoltaic Specialists Conference (PVSC). 2012: 003359–003364

[63] Kerr M J, Cuevas A. General parameterization of Auger recombination in crystalline silicon. Journal of Applied Physics, 2002, 91(4): 2473–2480

[64] Richter A, Glunz S W, Werner F, Schmidt J, Cuevas A. Improved quantitative description of auger recombination in crystalline silicon. Physical Review B, 2012, 86(16): 165–202

[65] Schenk A. Finite-temperature full random-phase approximation model of band gap narrowing for silicon device simulation. Journal of Applied Physics, 1998, 84(7): 3684–3695

[66] Banghart E K, Gray J L. Extension of the open-circuit voltage decay technique to include plasma-induced bandgap narrowing. IEEE Transactions on Electron Devices, 1992, 39(5): 1108–1114

[67] Altermatt P P, Schenk A, Geelhaar F, Heiser G. Reassessment of the intrinsic carrier density in crystalline silicon in view of band-gap narrowing. Journal of Applied Physics, 2003, 93(3): 1598–1604

[68] Mäckel H, Varner K. On the determination of the emitter saturation current density from lifetime measurements of silicon devices. Progress in Photovoltaics: Research and Applications, 2013, 21(5): 850–866

[69] Fahey P M, Griffin P, Plummer J. Point defects and dopant diffusion in silicon. Reviews of Modern Physics, 1989, 61(2): 289–384

[70] Dastgheib-Shirazi A, Steyer M, Micard G, Wagner H. Effects of process conditions for the n⁺-emitter formation in crystalline silicon. In: Photovoltaic Specialists Conference (PVSC). 2012: 001584–001589

[71] Tsai J C C. Point defect generation during phosphorus diffusion in silicon I. Journal of the Electrochemical Society, 1987, 134(6): 1508–1518

[72] Min B, Wagner H, Dastgheib-Shirazi A, Kimmerle A, Ku rz H. Heavily doped Si:P emitters of crystalline Si solar cells: recombination due to phosphorus precipitation. physica status solidi (RRL) – Rapid Research Letters, 2014, 8(8): 680–684

[73] Min B, Wagner H, Altermatt P P, Dastgheib-Shiraz i A. Limitation of industrial phosphorus-diffused emitters by SRH recombination. Energy Procedia, 2014, 55: 115–120

[74] Schmidt P F, Stickler R. Silicon phosphide precipitates in diffused silicon. Journal of the Electrochemical Society, 1964, 111(10): 1188–1189

[75] Komatsu Y, Vlooswijk A H G, Stassen A F, Venema P, Meyer C. Sophistication of doping profile manipulation-emitter performance improvement without additional process step. In: The 25th European Photovoltaic Solar Energy Conference and Exhibition-5th World Conference on Photovoltaic Energy Conversion. Valencia, Spain, 2010, 6: 10

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag Berlin Heidelberg