Introduction

About 35% of current global photovoltaics production uses B-doped monocrystalline silicon (c-Si) [1]. Advances in the Czochralski (Cz) process, i.e. the crystal growth technique, have resulted in ever better material quality and ever lower prices. This in turn supports the roll-out of high-efficiency passivated emitter and rear cells (PERC), which already yield efficiencies up to 21.5% to 22.0% in mass-production when using Cz-Si [ 2,3,4,5 ].

Yet, despite all advances, industrial B-doped Cz-Si still contains significant amounts of interstitial oxygen. This, in combination with the boron doping, results in light-induced degradation (LID) of the carrier lifetime and in turn of solar cell efficiency.

BO-related LID was first reported in 1973 [6] and was studied in detail in the late 1990’s and early 2000’s. As it is firmly linked to the simultaneous presence of both boron and oxygen [7,8], two straightforward ways to avoid it are to either reduce the interstitial oxygen content or to replace boron as a dopant (e.g. with Ga for p-type or with P for n-type wafers) [7,9].

One way to produce low-oxygen Cz-Si is to employ a strong magnetic field during the silicon crystal growth process [10]. However, this technology has not entered mass-production probably due to too high production costs. Replacing boron by gallium is not as straightforward as it sounds as well: using standard Czochralski processes for Ga-doped ingots results in a very strong increase of doping concentration along the ingot.

A third option to mitigate LID due to boron-oxygen-related defects is the permanent deactivation of these defects. This can be done by generating excess carriers at elevated temperature, e.g. by illumination. The resulting regeneration of solar cell characteristics as well as carrier lifetime was first reported in 2006 [11]. The process can be applied to all B-doped Cz-Si materials and solar cells, however, the effectiveness of regeneration depends on a variety of material and process parameters.

In this work, we evaluate the carrier lifetime as well as the solar cell efficiency potential of current industry standard B-doped Cz-Si. For this, we use dedicated test samples for lifetime measurements as well as our in-house PERC+ solar cell baseline process, which typically yields energy conversion efficiencies around 21.5% [12].

We measure lifetime and efficiency after full LID due to BO-related defects and after applying an optimized, lab-type regeneration treatment.

We compare the performance of the industry standard B-doped Cz-Si to that of industrial Ga-doped Cz-Si as well as to B-doped Cz-Si wafers with ultra-low oxygen content. The latter results from an advanced pulling technology that is developed by LONGi as a potential candidate for future mass production for high-quality and low-cost Cz-Si wafers.

We use the lifetimes measured on dedicated test samples as an input parameter for device simulations and compare the results with the measured solar cell characteristics. In particular, we measure lifetime and solar cell efficiency in three different states: (i) directly after processing, (ii) after 24 h of illumination at room temperature (which results in complete LID for the industry standard B-doped Cz-Si), and (iii) after applying the regeneration treatment.

Through this, we essentially assess the effectiveness of the permanent deactivation of BO-related defects as well as the extent – if there is any – of LID in the advanced Cz-Si wafers with Ga doping or ultra-low oxygen content. In addition, we investigate whether there are any other lifetime limiting defects present in industry standard B-doped Cz-Si that are not observed in Ga-doped or oxygen-lean B-doped Cz-Si.

Excellent stable bulk lifetimes of the different Cz-Si wafer materials

We study four different Czochralski-grown (Cz) silicon materials from LONGi Clean Energy Technology Co. Two of the materials are industrial B-doped Cz-Si with “standard” interstitial oxygen concentrations of 12 and 16 ppma, respectively. The other two materials are assessed as industrial-scale LID-free options. The first is B-doped Cz-Si with an extremely low interstitial oxygen concentration of 2.6 ppma and the second is Ga-doped Cz-Si from an industrial-type puller. The ultra-low oxygen concentration of 2.6 ppma was achieved by using an advanced Cz pulling technology developed by LONGi which reduces the oxygen dissolution into the silicon melt during crystallization.

Table I summarizes the resistivity and interstitial oxygen concentration (measured by Fourier-transform infrared spectroscopy, FTIR, using the ASTM F121-83 standard) of the four materials. The wafers are 156.75 × 156.75 mm2 (M2) and have an initial thickness of about 180 µm.

Table I: Dopant species, resistivity ρ, and interstitial oxygen concentration [Oi] of the four different Cz-Si materials investigated in this work.

Table I: Dopant species, resistivity ρ, and interstitial oxygen concentration [Oi] of the four different Cz-Si materials investigated in this work.

To measure the bulk lifetime, we process symmetrical test structures. First, we remove the saw damage and clean the wafers. The wafers then go through the same POCl3 diffusion we use for our baseline PERC+ solar cell process. On the one hand, this acts as a gettering step. On the other, it adds to the thermal history of the wafers and keeps it close to that of the PERC+ solar cells. Subsequently, the resulting n+ regions on both surfaces are etched off and we clean the wafers and deposit an AlOx/SiNy stack on the front and the rear for optimal surface passivation.

Finally, the lifetime samples are fired in a belt-firing furnace. Note that we have two groups, which are fired at two different belt speeds: 5.6 m/min, which is the speed at which the solar cells are fired, and 6.8 m/min, in order to obtain optimum lifetimes after regeneration [13].

After firing, the lifetime samples are in the “as-processed” state. We then measure the carrier lifetimes with a WCT-120 Lifetime Tester from Sinton Instruments, both in quasi-steady-state and photoconductance decay mode.

Next, we illuminate the samples for 24 h at about 0.1 suns and room temperature to activate the boron-oxygen-related defects. With regard to the bulk resistivity of up to 2.1 Ω cm, we chose 24 h to reach saturation of LID [14]. At this point, the lifetime is measured again.

We then apply the regeneration treatment and measure the lifetime for a third time. Our lab-type regeneration process consists of annealing on a hotplate set to 185 °C and simultaneously illuminating with a halogen lamp at a light intensity of about 1 sun for 15 min. Please note that we verified that no further increase of the lifetime is observed when annealing under illumination for longer times.

The lifetimes of the four different materials in the three different states are plotted in Figs. 1 and 2. Figures 1(a) and 2(a) depict the effective lifetimes of inherently LID-free Cz-Si materials, while Figs. 1(b) and 2(b) show the significant changes that happen in B-doped Cz-Si with an industrially typical interstitial oxygen concentration.

Fig 1: Effective lifetimes of (a) the 2.1 Ω cm, ultra-low [Oi] B-doped Cz-Si and (b) the 1.7 Ω cm industry standard B-doped Cz-Si as a function of excess carrier concentration. The blue triangles refer to the as-processed state, the red circles to the degraded state, and the green diamonds to the regenerated state. Open symbols denote a belt speed during fast-firing of 5.6 m/min, while filled symbols represent 6.8 m/min belt speed. The solid black line marks the intrinsic lifetime limit according to Richter et al. [15] for the given resistivity.

Fig 1: Effective lifetimes of (a) the 2.1 Ω cm, ultra-low [Oi] B-doped Cz-Si and (b) the 1.7 Ω cm industry standard B-doped Cz-Si as a function of excess carrier concentration. The blue triangles refer to the as-processed state, the red circles to the degraded state, and the green diamonds to the regenerated state. Open symbols denote a belt speed during fast-firing of 5.6 m/min, while filled symbols represent 6.8 m/min belt speed. The solid black line marks the intrinsic lifetime limit according to Richter et al. [15] for the given resistivity.

Note that Fig. 1 contains the materials with slightly higher resistivity (1.7 and 2.1 Ω cm), while Fig. 2 depicts the materials with lower resistivity (1.1 and 1.0 Ω cm). We group the materials like this because material quality, or rather solar cell potential with regard to material quality, is determined by lifetime as well as resistivity. One can therefore only easily compare lifetimes of materials with similar doping concentration.

Fig 2: Effective lifetimes of (a) the 1.0 Ω cn Ga-doped Cz-Si and (b) the 1.1 Ω cm industry standard B-doped Cz-Si as a function of excess carrier concentration. The wafers are fired at a belt speed of 6.8 m/min. The blue triangles refer to the as-processed state, the red circles to the degraded state, and the green diamonds to the regenerated state. The solid black line marks the intrinsic lifetime limit according to Richter et al. [15] for the given resistivity.Effective lifetimes of (a) the 1.0 Ω cn Ga-doped Cz-Si and (b) the 1.1 Ω cm industry standard B-doped Cz-Si as a function of excess carrier concentration. The wafers are fired at a belt speed of 6.8 m/min. The blue triangles refer to the as-processed state, the red circles to the degraded state, and the green diamonds to the regenerated state. The solid black line marks the intrinsic lifetime limit according to Richter et al. [15] for the given resistivity.

Fig 2: Effective lifetimes of (a) the 1.0 Ω cn Ga-doped Cz-Si and (b) the 1.1 Ω cm industry standard B-doped Cz-Si as a function of excess carrier concentration. The wafers are fired at a belt speed of 6.8 m/min. The blue triangles refer to the as-processed state, the red circles to the degraded state, and the green diamonds to the regenerated state. The solid black line marks the intrinsic lifetime limit according to Richter et al. [15] for the given resistivity.Effective lifetimes of (a) the 1.0 Ω cn Ga-doped Cz-Si and (b) the 1.1 Ω cm industry standard B-doped Cz-Si as a function of excess carrier concentration. The wafers are fired at a belt speed of 6.8 m/min. The blue triangles refer to the as-processed state, the red circles to the degraded state, and the green diamonds to the regenerated state. The solid black line marks the intrinsic lifetime limit according to Richter et al. [15] for the given resistivity.

Note that Fig. 1 contains the materials with slightly higher resistivity (1.7 and 2.1 Ω cm), while Fig. 2 depicts the materials with lower resistivity (1.1 and 1.0 Ω cm). We group the materials like this because material quality, or rather solar cell potential with regard to material quality, is determined by lifetime as well as resistivity. One can therefore only easily compare lifetimes of materials with similar doping concentration.

The blue triangles in the graphs correspond to the measured lifetime after the fast firing step (as-processed). The red circles give the lifetime after illumination at room temperature for 24 hours, i.e. in the case of the B-doped samples after light-induced degradation. The green diamonds refer to the lifetime after applying the regeneration treatment (annealing under illumination at 185 °C for 15 min). In addition, all graphs contain the respective intrinsic lifetime according to the model of Richter et al. [15] (solid black line).

Figure 1(a) depicts the effective lifetimes of the 2.1 Ω cm B-doped material with ultra-low interstitial oxygen concentration of 2.6 ppma. As expected, we do not observe any light-induced degradation of the bulk carrier lifetime between the as-processed and the degraded state [ ]. However, applying the regeneration treatment improves the lifetime from around 2 ms to about 3.8 ms at an excess carrier concentration Δn = 1015 cm−3.

For the B-doped 1.7 Ω cm sample [Fig. 1(b)], we plot the lifetimes measured after firing at different belt speeds. The open symbols correspond to the effective lifetimes measured after firing at a belt speed of 5.6 m/min and the filled symbols correspond to a belt speed of 6.8 m/min. For the slower belt speed the as-processed lifetime (open blue triangles) at an excess carrier concentration of Δn = 1015 cm−3 is around 650 µs. After 24 h of illumination at room temperature, the lifetime drops to about 200 µs (degraded state, open red circles) and after application of the regeneration treatment, the lifetime increases to about 1.4 ms (regenerated state, open green diamonds).

Using a faster belt speed during the fast firing process notably increases the effective lifetime in all three states. As-processed (filled blue triangles), we measure 1.1 ms at Δn = 1015 cm−3. In the degraded state (filled red circles) the lifetime is around 390 µs and regeneration results in an effective lifetime of 3.2 ms at Δn = 1015 cm−3 (filled green diamonds), which is comparable to the lifetime measured on the ultra-low [Oi] Cz-Si.

While such a dependence of the lifetime on the belt speed during fast firing was already shown in Ref.  , it should be noted that we do not observe it for all B-doped Cz-Si materials. The ultra-low [Oi] Cz-Si, for example, yields similar lifetimes after firing at 5.6 m/min and 6.8 m/min, respectively (not shown here).

The fact that even the regenerated lifetime curves are below the intrinsic lifetime published by Richter et al. (solid black line) can be attributed to surface recombination. Assuming a surface recombination velocity of just 2 cm/s per side already closes the gap between measured lifetime and intrinsic model. This is a reasonable value for the used AlOx/SiNy stack.

Looking at the measured lifetimes of the Ga-doped Cz-Si [Fig. 2(a)], we see only small changes, as would be expected. 24 hours of illumination at room temperature (red circles) actually result in slightly higher lifetimes than measured in the as-processed state (blue triangles). After the regeneration treatment, however, we see a slight decrease of the lifetime, especially at excess carrier densities below 5 × 1015 cm−3 (green diamonds). This might be a bulk effect, but could also be related to a degradation of the surface passivation quality. However, it has to be proven with larger wafer numbers if the slight lifetime degradation after regeneration is statistically significant.

In contrast, for the industrial 1.1 Ω cm Cz-Si with an interstitial oxygen concentration of 12 ppma [Fig. 2(b)], the lifetime changes significantly. We measure about 600 µs in the as-processed state at Δn = 1015 cm−3, 210 µs after light-induced degradation, and around 1.4 ms after regeneration.

Comparing the injection-dependent bulk lifetimes measured after applying the regeneration treatment, we see that the industry standard B-doped Cz-Si wafers perform as well as the inherently LID-free materials with similar doping concentration. We would therefore also expect a similar solar cell performance after permanent deactivation of the BO-defect.

Higher PERC+ conversion efficiencies on completely LID-free Cz-Si wafer materials

We use the four above-mentioned Cz-Si materials to process bifacial PERC+ solar cells. The PERC+ process sequence is described in detail in [ ]. Here, we just highlight the basic process flow. The PERC+ process starts with saw-damage removal and wafer cleaning. We then apply a rear protection layer to act as etching and diffusion barrier during the following front surface texturing and phosphorus diffusion steps. Afterwards, the PSG and protection layer are removed and we apply a stack of aluminum oxide (AlOx) and plasma-enhanced chemical vapor deposited (PECVD) silicon nitride (SiNy) to passivate the rear side. PECVD-SiNy is also deposited on the front to passivate the P-doped emitter and act as antireflection coating.

To enable contact formation on the rear side, we locally remove the AlOx/SiNy stack using a picosecond laser with 532 nm wavelength before screen-printing Al fingers on the rear and Ag fingers on the front side. At the end, the metal pastes are co-fired in a belt-firing furnace with a belt speed of 5.6 m/min. A schematic of the final solar cell structure is shown in Fig. 3.

Figure 3: Schematic of a bifacial PERC+ solar cell.

Figure 3: Schematic of a bifacial PERC+ solar cell.

The IV curves of the solar cells are measured with a LOANA from pvtools. Similar to the dedicated lifetime samples, we measure the solar cells directly after processing, after illumination at room temperature for 24 h, as well as after applying the regeneration treatment.

The results are shown in Fig. 4. We plot the solar cell efficiencies for the different materials in the as-processed state relative to the efficiency of the industrial 1.7 Ω cm B-doped Cz-Si with 16 ppma [Oi] (green diamonds). The change due to illumination at room temperature and the change after applying the regeneration treatment are plotted relative to the respective as-processed efficiencies.

Fig 4: Efficiency difference for the PERC+ solar cells made from the four different Cz-Si materials in the as-processed state as well as after illumination at room temperature and after applying the regeneration treatment.

Fig 4: Efficiency difference for the PERC+ solar cells made from the four different Cz-Si materials in the as-processed state as well as after illumination at room temperature and after applying the regeneration treatment.

After processing, the efficiency of the PERC+ solar cells on 2.6 ppma [Oi] (blue squares) is 0.4%abs higher than of those using the industrial 16 ppma [Oi] Cz-Si. Similarly, the solar cells from the Ga-doped Cz-Si (open red circles) performs 0.3%abs better than the industry standard B-doped Cz-Si with similar doping concentration (purple triangles). On an absolute scale, the PERC+ conversion efficiencies range between 21.0% and 21.5% applying a 5 busbar design. Applying a busbar-less design and R&D type Ga wafers similar to material 4 in Table I, the best PERC+ efficiency is 22.1% [16].

As expected from the lifetime measurements, the solar cell efficiency of the ultra-low [Oi] B-doped as well as the Ga-doped cells is not affected by either illumination at room temperature or illumination at elevated temperature (within the measurement uncertainty) and hence completely LID free [16].

In contrast, the PERC+ solar cells from industrial B-doped Cz-Si degrade by 0.5%abs to 0.7%abs after 24 h illumination at room temperature, widening the gap to the LID-free materials to 0.8%abs to 1.1%abs. After applying the regeneration treatment, the measured efficiencies improve, however, only to a similar level as before LID, not higher.

The efficiency gap between the inherently LID-free materials and the industrial B-doped Cz-Si after regeneration seems to disagree with the lifetime measurements. There, the lifetime of the industrial B-doped materials after regeneration was similar to that of the ultra-low [Oi] and the Ga-doped Cz-Si, respectively.

Device simulations: Translating bulk lifetimes to solar cell efficiencies

In order to understand this discrepancy, we set up a device simulation using the conductive boundary model [19] as implemented by Quokka2 [20]. The input parameters, which are based on extensive characterization of our baseline PERC+ solar cell process, are summarized in Table II. The injection-dependent lifetimes of the B-doped Cz-Si wafers, as measured on the lifetime test structures, can be described by a Shockley-Read-Hall (SRH) defect at midgap with a factor of 10 between τn0 and τp0. The table lists the defect lifetimes for the as-processed state of the 1.7 Ω cm B-doped sample with 16 ppma [Oi] (material 1). Note that we used the lifetime measured on the designated test sample that was fired at the same belt speed as the solar cells, i.e. 5.6 m/min.

The front contact shadowing is calculated from the optical finger width and the layout of the front grid. The saturation current density of the emitter J0e is determined on symmetrical test wafers using the Kane and Swanson method. We do transmission line model (TLM) measurements to determine the specific contact resistances at the front and rear. The saturation current densities J0 at the contacts are based on lifetime measurements done on test samples with varying metallization fraction, both for the emitter and the base contact. The depth-dependent generation profile is calculated by parameterizing the measured reflectance of a PERC+ solar cell according to Brendel et al. [21].

Table II: Input parameters for the device simulations of PERC+ solar cells using the lifetime data of the B-doped Cz-Si with 16 ppma [Oi] in the as-processed state.

Table II: Input parameters for the device simulations of PERC+ solar cells using the lifetime data of the B-doped Cz-Si with 16 ppma [Oi] in the as-processed state.

We verify the simulation setup by inserting the effective lifetime measured in the as-processed state for material 1 (1.7 Ω cm B-doped with 16 ppma [Oi]) and comparing the result to measured IV parameters in the as-processed state. The results are summarized in Tab. III.

Efficiency n, open-circuit voltage Voc, and short-circuit current density Jsc match very well, while the measured fill factor FF is 0.45%abs higher than the simulated one. This is in part due to a higher pseudo fill factor pFF (0.3%abs) as well as a lower series resistance (~0.05 Ω cm2) but still it is a good match.

Table III: Simulated and measured IV parameters for the baseline PERC+ solar cell process and the 1.7 Ω cm B-doped Cz-Si with 16 ppma [Oi].

Table III: Simulated and measured IV parameters for the baseline PERC+ solar cell process and the 1.7 Ω cm B-doped Cz-Si with 16 ppma [Oi].

The results are shown in Fig. 5. For the as-processed state, we use the efficiency of the industrial 1.7 Ω cm B-doped Cz-Si with 16 pmma [Oi] (green diamonds) as the baseline value. The change due to illumination at room temperature and the change after applying the regeneration treatment are plotted relative to the respective as-processed efficiencies. Values corresponding to measured solar cell efficiencies are plotted in filled symbols, the results from the simulation are depicted by open symbols.

Fig 4: Measured (filled symbols, solid lines) and simulated (open symbols, dashed lines) difference in PERC+ cell efficiency Δη after illumination at room temperature and after applying the regeneration treatment.

Fig 4: Measured (filled symbols, solid lines) and simulated (open symbols, dashed lines) difference in PERC+ cell efficiency Δη after illumination at room temperature and after applying the regeneration treatment.

As mentioned above, the actual ultra-low [Oi] solar cells (orange circles) yield 0.3%abs higher efficiencies than the solar cells made on industrial B-doped Cz-Si (green diamonds) directly after processing. This difference is also predicted by the simulation when entering the respective effective lifetimes measured in the as-processed state.

Illumination at room temperature notably decreases the lifetime in the 16 ppma [Oi] B-doped Cz-Si. Inserting this degraded lifetime into the simulation yields a 0.7%abs loss compared to the as-processed state (open green diamonds). This is in good agreement with the degradation observed on actual PERC+ solar cells (filled green diamond). After regeneration, the measured PERC+ efficiency of the industrial B-doped Cz-Si improves to a similar level as before LID, whereas the device simulation with the regenerated lifetime predicts an increase by 0.4%abs compared to before LID.

The measured PERC+ efficiency of the 2.6 ppma [Oi] B-doped material, on the other hand, is stable both during the degradation and the regeneration treatment (within the measurement uncertainty). This is in accordance with the measured bulk lifetimes, which do not decrease after illumination at room temperature [see Fig. 1(b)]. While we do observe an increase of bulk lifetime for the 2.6 ppma B-doped Cz-Si after applying the regeneration treatment, this increase only translates to an efficiency gain of 0.02% in our device simulation, as the other recombination channels are much more dominant.

The comparison between measured solar cell efficiencies and efficiencies simulated on the basis of measured bulk lifetimes reveals a notable discrepancy for the regenerated state of the industrial B-doped Cz-Si. The difference of the measured bulk lifetimes in the degraded and the regenerated state should translate to an efficiency gain of 1.1%abs after regeneration. However, on the actual solar cells we only observe an increase by 0.7%abs.

Hence, in contrast to the experimental results, according to the simulation the regenerated industrial B-doped wafer material should enable identical PERC+ efficiencies as the ultra-low [Oi] wafer material. 

This suggests that for the industrial B-doped Cz-Si there is a difference in the regenerated bulk lifetime of the test wafers and the PERC+ solar cells. As the lifetime test wafers were processed in parallel to the solar cells – as far as possible – this raises the question as to where this difference might stem from.

Possible reasons for lower-than-expected efficiency after regeneration of industry-typical Cz-Si

Looking at the detailed processing sequence used in this work, we note three major differences between lifetime test structures and PERC+ solar cells: (1) the solar cells have a rear protection layer during the P-diffusion step, (2) during the fast-firing process, the solar cells have an emitter on the front side, and (3) the solar cells are fired with metal pastes on both surfaces. 
In contrast, for the lifetime test wafers (1) the n+ emitter first forms on both wafer sides. Subsequently, (2) the emitter is removed from both wafer surfaces, which are then passivated with an AlOx/SiNy stack. And (3) obviously we do not apply metal pastes to the lifetime samples.

Regarding the impact of a rear protection layer during P-diffusion, we have experimental results from a different study that makes it seem unlikely. In said experiment, the solar cells as well as the lifetime samples were diffused on both surfaces and the n+ emitter on the rear was then removed by a rear polishing step. In that experiment, too, the PERC+ cell efficiencies after processing and after regeneration were identical in contrast to the higher carrier lifetimes after regeneration measured on lifetime samples.

Regarding the presence of the n+ emitter during the fast firing step, we measure and simulate the effective lifetimes of non-metallized implied-Voc (iVoc) solar cell precursors. These are identical to PERC+ solar cells except for the laser contact opening and the screen-printing steps. For the simulation, we set the front metal recombination to be equal to that of the passivated emitter and the rear metal recombination equal to that of the passivated rear side. All other parameters are the same as for the solar cell simulation.

Figure 6 shows the effective lifetime of such an iVoc precursor. The blue triangles correspond to the as-processed state and the green diamonds to the regenerated state. The lines give the simulated effective lifetimes, i.e. the result of simulating an implied-Voc structure, when assuming the bulk lifetime measured in the as-processed (solid blue line) and in the regenerated state (dashed green line), respectively.

Fig 6: Measured (symbols) and simulated (lines) injection-dependent effective lifetimes of industry standard 1.7 Ω cm B-doped Cz-Si with a SiNy-passivated 115 Ω/sq. n+ emitter on the front side. Blue triangles and solid blue line correspond to the as-processed state, green diamonds and dashed green line to the regenerated state.

Fig 6: Measured (symbols) and simulated (lines) injection-dependent effective lifetimes of industry standard 1.7 Ω cm B-doped Cz-Si with a SiNy-passivated 115 Ω/sq. n+ emitter on the front side. Blue triangles and solid blue line correspond to the as-processed state, green diamonds and dashed green line to the regenerated state.

In the as-processed state, the measured effective lifetime (blue triangles) matches the simulated effective lifetime (dashed blue line) very well. The measured effective lifetime on the non-metallized precursor after regeneration (green diamonds), however, is notably lower than the simulated effective lifetime (solid green line). It thus seems as if the presence of the n+ emitter during the fast firing step negatively affects the bulk lifetime. 

 

A possible explanation for this effect could be related to hydrogen, which interacts with various defects in the silicon bulk. A major source of H is the SiNx-layer combined with the fast-firing step [22]. However, the presence of an n+ emitter affects the transport of hydrogen from the SiNx into the bulk [23]. This could result in different bulk lifetimes in the lifetime test wafers and the solar cell precursors.

We are currently investigating this further. 

References

[1] International Technology Roadmap for Photovoltaic (ITRPV) 2018

[2] https://www.pv-magazine.com/2018/01/19/longi-achieves-record-conversion-rate-of-20-41-for-perc-modules/

[3] JinkoSolar Holding Co., Ltd., Annual Report 2017

[4] https://www.pv-tech.org/news/aiko-solar-enters-volume-production-of-mono-perc-cell-with-22-conversion-ef

[5] Lee, B. G. et al. 2018, “Development of p-Cz PERC Solar Cells Approaching 23% Efficiency for Gigawatt-Level Production”, Proc. 35th EU PVSEC, Brussels, Belgium, pp. 308-400

[6] Fischer, H. and Pschunder, W. 1973, “Investigation of photon and thermal induced changes in silicon solar cells”, Proc. 10th IEEE PVSC, Palo Alto, CA, (IEEE New York 1973), p. 404.

[7] Schmidt, J., Aberle, A. and Hezel, R. 1997, “Investigation of carrier lifetime instabilities in Cz-grown silicon,”, Proc. 26th IEEE PVSC, Anaheim, CA, USA, pp. 13–18.

[8] Glunz, S. W. et al. 1998, “On the degradation of Cz-silicon solar cells”, Proc. 2nd WCPEC, Vienna, Austria (European Commission, Ispra, Italy, 1998), p. 1343.

[9] Glunz, S. W. et al. 1999 “Comparison of boron‐ and gallium‐doped p‐type Czochralski silicon for photovoltaic application”, Prog. Photovolt: Res. Appl., Vol. 7, No. 6, pp. 463-469.

[10] Mosel, F. et al. 2012, “Growth of high quality silicon mono ingots by the application of a magnetic CUSP field in Cz-puller”, Proc. 27th EU PVSEC, Frankfurt, Germany, p. 933.

[11] Herguth, A. et al. 2006, “Avoiding boron-oxygen related degradation in highly boron doped Cz silicon”, Proc. 21st EU PVSEC, Dresden, Germany, (Munich: WIP 2006), p. 530.

[12] Dullweber, T. et al. 2017, “Bifacial PERC+ Solar Cells and Modules: An Overview”, Proc. 33rd EU PVSEC, Amsterdam, The Netherlands, pp. 649-656.

[13] Walter, D., and Lim, B. and Schmidt, J. 2016 “Realistic efficiency potential of next-generation industrial Czochralski-grown silicon solar cells after deactivation of the boron–oxygen-related defect center”, Prog. Photovolt: Res. Appl., Vol. 24, No. 7, pp. 920-928.

[14] Palmer, D. W., Bothe, K. and Schmidt, J. 2007, “Kinetics of the electronically stimulated formation of a boron-oxygen complex in crystalline silicon”, Phys. Rev. B, Vol. 76, p. 035210.

[15] Richter, A. et al. 2012, “Improved quantitative description of Auger recombination in crystalline silicon”, Phys. Rev. B, Vol. 86, p. 165202.

[16] Lim, B. et al. 2018, “LID-Free PERC + Solar Cells with Stable Efficiencies Up to 22.1%”, Proc. 35th EU PVSEC, Brussels, Belgium, pp. 359-365.

[17] Walter, D. C. et al. 2014, “Lifetimes exceeding1msin1-Ω cm boron-dopedCz-silicon”, Solar Energy Materials and Solar Cells, Vol. 131, pp. 51-57.

[18] Dullweber, T. et al. 2016, “PERC+: Industrial PERC solar cells with rear Al grid enabling bifaciality and reduced Al paste consumption”, Prog. Photovolt: Res. Appl., Vol.  24, No. 12, pp. 1487-1498.

[19] Brendel, R. 2012, “Modeling solar cells with the dopant‐diffused layers treated as conductive boundaries”, Prog. Photovolt: Res. Appl., Vol. 20, No. 1, pp. 31-43.

[20] Fell, A. 2012, “A Free and Fast Three-Dimensional/Two-Dimensional Solar Cell Simulator Featuring Conductive Boundary and Quasi-Neutrality Approximations”, IEEE Transactions on Electron Devices, Vol. 60, No. 2, pp. 733–738.

[21] Brendel, R. et al. 2016, “Breakdown of the efficiency gap to 29% based on experimental input data and modeling”, Prog. Photovolt: Res. Appl., Vol. 24, No. 12, pp. 1475–1486.

[22] Kleekajai, S. et al. 2006, “Concentration and penetration depth of H introduced into crystalline Si by hydrogenation methods used to fabricate solar cells”, Journal of Applied Physics, Vol. 100, No. 9, p. 093517.

[23] Hamer, P. et al. 2018, “Modelling of hydrogen transport in silicon solar cell structures under equilibrium conditions”, Journal of Applied Physics, Vol. 123, No. 4, p. 043108.

Tags: pv celltech, c-si manufacturing, monocrystalline wafer, polysilicon, czochralski, perc, lid, longii green energy technology, isfh, leibniz universität

Comments