Construction of functional grain boundary clusters for casting large-size and high-quality monocrystalline silicon ingots

Qi Lei*^a, Liang He*^bc, Jianmin Li^bc, Yunfei Xu^bc, Wei Mao^c, Yufei Zhong^a, Jinbing Zhang*^a and Dongli Hu*^a
aZhejiang Engineering Research Center for Fabrication and Application of Advanced Photovoltaic Materials, School of Materials Science and Engineering, NingboTech University, Ningbo, 315100, China. E-mail: 529594405@qq.com; zhangjb@nbt.edu.cn; msehudl@zju.edu.cn
^bSchool of New Energy Science and Engineering, Xinyu University, Xinyu, 338004, China
^cNational Photovoltaic Engineering Research Center, LDK Solar Co., Ltd., Xinyu 338032, China

First published on 13th August 2024

---

---

1 Introduction

Due to the rapid development of photovoltaic (PV) technology and its expanding applications, there is an increasing demand for high-performance and low-cost solar cells. Casting monocrystalline silicon (mono-Si) technology offers a promising low-cost method, but its crystal quality and solar cell efficiency still lag behind those of the Czochralski (CZ) method.^1–3 Therefore, improving the crystal quality of cast mono-Si is crucial for its mass production. Extensive research has been conducted on various aspects affecting the efficiency and quality of solar cells, including the differences in cell efficiency between cast mono-Si and CZ mono-Si, the impact of seed recycling and metal contamination on the formation of subgrain boundaries, the effects of different seed orientations on cell efficiency, and the defect phenomena induced by various seed junctions.^4–8 Lan et al. investigated the effects of different tilt angles of the seed junctions on defects in cast mono-Si ingots, finding that the optimal angle ranges from 10° to 30°.^9,10 In addition to these issues, another key challenge is to suppress the generation of multicrystalline silicon (mc-Si) near the crucible wall and its expansion into the mono-Si region. This challenge complicates the production of intact mono-Si ingots under industrial conditions and increases dislocation density in the mono-Si. One potential solution involves controlling the solid–liquid interface during crystallization to increase the mono-Si area and improve the distribution of crystal defects. Previous studies have demonstrated that precisely controlling the horizontal and vertical temperature distributions in a directional solidification furnace can form a flat or slightly convex solid–liquid interface.^11–13 This interface helps cast ingots with large mono-Si regions and low dislocation densities, inhibits disordered mc-Si growth near the crucible sidewalls, and reduces thermal stresses caused by horizontal temperature differences.^14,15 However, most prior research has focused on small-sized cast mono-Si ingots. In the photovoltaic industry, producing large-sized monocrystalline ingots is essential to further reduce production costs. Currently, G7 or larger cast mono-Si ingots suffer from uneven horizontal temperature fields, uneven growth interfaces near the crucible sidewalls, and difficulties in controlling the growth direction of mc-Si, which tends to overgrow into the mono-Si region.¹⁶ This overgrowth reduces the mono-Si area and significantly deteriorates the crystal quality of the ingot's side regions.

Therefore, some methods have been proposed to prevent the growth of mc-Si grains near the crucible wall or to change their growth direction, thereby increasing the mono-Si area of large-size cast mono-Si ingots. Kutsukake et al. proposed the use of Σ5 functional boundaries for casting mono-Si, effectively inhibiting the growth of mc-Si grains near the crucible sidewalls during the initial growth stage.^17–21 However, as the growth height increases, peripheral mc-Si grains still tend to grow towards the interior of the ingot, rapidly decreasing the mono-Si region and making it difficult to obtain high-quality ingots with nearly complete mono-Si regions. Takahashi et al. addressed this issue by forming smart functional defect regions using a combination of various grain boundaries, including small-angle, tilted, random, and Σ3 boundaries.²² They utilized high-density dislocation regions to absorb impurities, hinder dislocation multiplication, and mitigate plastic deformation to release stresses. To further improve internal crystal quality, Zhang et al. created low-energy Σ13 grain boundaries by adjusting the angle of neighboring <100> seeds, thereby reducing the generation of twins and dislocations through the use of functional grain boundaries.²³ Many previous studies on the use of grain boundaries to inhibit the nucleation of mc-Si grains from the crucible wall have focused primarily on small-size cast mono-Si ingots. Research on inhibiting mc-Si growth and improving defect distribution in large-size cast mono-Si ingots is limited. Furthermore, the methods proposed thus far cannot be directly applied to the industrial production of large-size mono-Si ingots. Therefore, there is a critical need for innovative approaches that can be effectively implemented in the industrial process to inhibit mc-Si growth and improve defect distribution in large-size cast mono-Si ingots.

In this study, we propose a novel approach using functional grain boundary clusters (FGBCs) for casting G7-type large-size mono-Si ingots. This method aims to effectively increase the proportion of the mono-Si area and improve the crystal defect distribution near the sides of the ingot. By constructing multiple parallel large-angle grain boundaries, the method prevents the overgrowth of mc-Si from the crucible sidewalls to the ingot center. This significantly expands the mono-Si region and substantially enhances the defect distribution, resulting in higher quality ingots.

2 Experiment

2.1 Seed layout

The seed layout schematics are shown in Fig. 1(a) and (b). For the reference ingot, square mono-Si seeds were arranged in a 6 × 6 array at the bottom of a G7 quartz crucible which had an inner silicon nitride coating. The crucible dimensions were 1200 mm × 1200 mm × 480 mm. The side orientation angle difference between adjacent square seeds was approximately 15° (Fig. 1(a)). For the experimental ingot, in order to block the overgrowth of mc-Si near the crucible sidewall, we set three layers of seed strips at the periphery of the square seeds, and the side orientation angle difference between adjacent seed strips was approximately 45° for the formation of the FGBC. The actual seed pavement is shown in Fig. 1(c), The dimensions of the square seeds are 168 mm × 168 mm × 25 mm, which derived from <100> oriented CZ mono-Si. The outer region of the square seed area was covered with three layers of seed strips (Fig. 1(d)), also <100> oriented, each measuring 168 mm × 5 mm × 25 mm, which is expected to induce three sets of parallel grain boundaries (GB I, GB II, GB III) during the seeding process to form the FGBC.

After the seeds were arranged, the crucible was filled with a mixture of 50% polysilicon and 50% recycled silicon, with a total charge weight of about 1100 kg. The casting process of the experimental silicon ingot was conducted in a JZ660 directional solidification furnace, manufactured by Beijing JYT Technology Co., Ltd, and the experimental and reference ingots use the same growth process. To precisely control the seeding process, a quartz rod was used during the melting stage to detect and maintain the residual height of the seeds between 12 and 15 mm. The average crystallization rate of the silicon ingots was approximately 0.8 cm h⁻¹.

2.2 Sample analysis

After the growth of the silicon ingot, it was cut into 36 square silicon bricks and 20 edge silicon bricks. The dimensions of the square silicon bricks were 166 mm × 166 mm × 350 mm, while the edge silicon bricks measured 166 mm × 80 mm × 350 mm. The minority carrier lifetime distribution on the longitudinal section of the silicon bricks was tested by using a microwave photoconductance decay tester (MW-PCD, Semilab, WT-2000). The defect distribution in the silicon bricks and wafers was detected using photoluminescence imaging equipment (PL mapping, BT Imaging, LIS-R1). The grain morphology and defect distribution in the samples were examined by etching with a mixed acid solution (CH₃COOH/HF/HNO₃/H₂O = 1.14:4:1:2). The distribution of dislocations was observed using an optical microscope (OM, Nikon, Eclipse-L150) and a scanning electron microscope (SEM, ZEISS, Gemini 300). Finally, the crystal orientation and grain boundary characteristics of the samples were tested using an electron backscatter diffraction detector (EBSD, Oxford Instruments, C-Nano) attached to the SEM. The solar cells were fabricated and tested for photovoltaic parameters using the passivated emitter rear cell (PERC) process to assess the electrical properties and crystal quality of the ingots.

3 Results and discussion

3.1 Mono-Si proportion analysis

The surface of the mono-Si ingot grown with FGBC is shown in Fig. 2(a). The mono-Si region is enclosed within a black box, while the surrounding area is mc-Si. It can be observed that the FGBC well separates the mono-Si region from the mc-Si region. In the IR image of the ingot's longitudinal section, no mc-Si morphology is observed, indicating that the mc-Si in the edge region has not overgrown into the mono-Si region (Fig. 2(c)). In contrast, the surfaces of normal cast mono-Si ingots without FGBC show a significant reduction in the mono-Si region (Fig. 2(b)). The mc-Si overgrow into the mono-Si region from the crucible sidewalls at various depths. The infrared image of the longitudinal section of the ingot reveals a large amount of mc-Si on both sides, which expands with increasing ingot height (Fig. 2(d)). A possible explanation for this overgrowth is that the mc-Si near the sidewalls contains grains with different orientations.²⁴ When these grains encounter <100> oriented mono-Si, the grains with faster lateral growth rates gradually overgrow the mono-Si region. Additionally, under certain high undercooling conditions, some grains form twins along the {111} plane when creating grain boundaries with <100> oriented grains. These twins can rapidly overgrow into the mono-Si region, significantly reducing its area.²⁵

To assess the effect of the FGBC on the mono-Si proportion of mono-Si ingots, the height of mono-Si in all silicon bricks needs to be measured after ingot squaring. Fig. 3(a) shows an infrared image of a silicon brick containing mc-Si. The exact height of the mono-Si can be determined by identifying the position of the cross section that does not contain mc-Si. Once the mc-Si regions are removed, the sliced wafer is entirely mono-Si. Fig. 3(b) and (c) illustrate the distribution of the height of mono-Si in the ingots with and without FGBC, respectively. In the ingot with FGBC, only the upper regions of the corner bricks are overgrown by mc-Si. This invasion is mainly due to the inhomogeneous temperature field in the casting furnace, which leads to overcooling at the corner region of the ingot, resulting in a concave growth interface towards the melt. The curvature of this growth interface causes the FGBC to lose its barrier effect. In contrast, the height of mono-Si in other regions of the ingot with FGBC is consistent with the overall ingot height, indicating that the FGBC effectively increases the mono-Si proportion where the growth interface is not concave. Conversely, in the ingot without FGBC, both the edge and corner bricks are significantly overgrown by mc-Si, leading to a substantial reduction in the mono-Si height in the edge bricks. Calculations show that the mono-Si proportion of the ingot with FGBC is about 95%, whereas the mono-Si proportion of the ingot without FGBC is approximately 60%. Therefore, the use of the FGBC significantly increases the mono-Si proportion in cast mono-Si ingots, thereby enhancing the yield of mono-Si ingots.

In addition to the mono-Si proportion, the minority carrier lifetime distribution is also an important characterization for assessing ingot quality. Fig. 4(a) and (b) show the minority carrier lifetime distributions in the longitudinal sections of ingots with and without FGBC, respectively. The red regions indicate much lower minority carrier lifetimes, mainly due to the diffusion and accumulation of metallic elements. Yellow filamentary regions with low minority carrier lifetimes, mostly due to grain boundaries, dislocations, impurities, and other crystal defects, are present in the blue areas. The minority carrier lifetime distribution in the ingot with FGBC indicates a small variance between the ingot edge area and the middle region. This suggests that the FGBC reduces the amount of crystal defects in the edge area. The ingot without FGBC, on the other hand, exhibits more low minority carrier lifetime areas, primarily in the edge region, suggesting a higher concentration of crystal defects than in the central region. Overall, the FGBC not only increases the mono-Si proportion but also significantly improves the minority carrier lifetime distribution and reduces crystal defects near the edge region. This enhancement contributes to improved crystal quality in the edge region of cast mono-Si ingots.

3.2 Defect distribution

In order to confirm the role of the FGBC in blocking the overgrowth of mc-Si and suppressing the generation of crystal defects, we selected edge bricks containing FGBC for analysis. Fig. 5(a) shows the longitudinal cross-section of the edge brick containing mc-Si, FGBC and mono-Si. Among them, multiple grain boundaries growing along the solidification direction can be observed in the FGBC region, as shown in the red box. Some of the grain boundaries merge in the upper-middle region, which may be caused by the merging of the grain boundaries with the neighboring ones after the distortion of the grain boundaries occurs, as confirmed from Fig. 5(c). This suggests that the width of the seed strips determines the distance of the grain boundaries in the FGBC, and too small a distance may allow the adjacent grain boundaries to intersect prematurely, thus reducing the number of grain boundaries used for blocking. Based on the grain morphology of the mc-Si region in the edge brick, it can be deduced that the growth interface near the crucible wall is concave towards the melt, while the growth interface in the FGBC region at a certain distance from the crucible sidewall tends to be flat. Thus, the outermost part of the FGBC region not being in the concave interface region ensures that the mc-Si in the edge brick is always blocked by the grain boundaries in the FGBC throughout the growth process. Fig. 5(b) shows the PL image of the longitudinal section of the edge brick, revealing a large number of crystal defects within the FGBC, with a defect density even higher than that of the neighboring mc-Si regions. Nevertheless, the mono-Si region was not affected by these defects in the FGBC, and the growth process did not result in a notable rise in the defect density in the nearby mono-Si region. These findings verify that although there is a larger density of crystal defects in the FGBC, these defects are efficiently kept from propagating into the mono-Si area. Furthermore, the crystal defects are not readily generated in the mono-Si region since these defects also aid in the release of stress.²⁶ Thus, by preventing mc-Si overgrowth and regulating the propagation of crystal defects, the FGBC is essential to preserving the integrity and quality of the mono-Si area.

To better observe the interaction between grain boundaries and mc-Si within the FGBC, cross sections at different growth heights in the edge brick were analyzed. Fig. 5(c) shows the morphology of the cross section at different heights in the edge brick. At H = 60 mm, the three grain boundaries formed by the seed strips and the square seed after seeding are nearly parallel to each other. The growth region of the outermost seed strip is overgrown by mc-Si, but the mc-Si is blocked by the outermost grain boundary and does not enter the growth region of the middle seed strip. At H = 120 mm and H = 180 mm, the outermost grain boundaries are still not penetrated by mc-Si. However, all three grain boundaries are distorted to varying degrees, likely due to structural changes under the combined squeezing effect of mc-Si and dislocations. Fig. 5(d) shows the distribution of PL defects in the cross sections at different heights of the edge brick. At H = 60 mm, black clusters consisting of numerous defects are visible within the FGBC. At H = 120 mm, defects confined inside the FGBC further propagate, increasing the area and density of the black clusters. However, these defects do not significantly enter the mono-Si region, as the innermost grain boundary effectively blocks their overgrowth. At H = 180 mm, the area of the black clusters in the FGBC no longer increases, and the defect density decreases, similar to the observation in the upper region in Fig. 5(b). This may be due to the annihilation of similar types of dislocations in the high-density dislocation region during the process of slipping and climbing.²⁷ The defective regions in the mono-Si region remain minimal, with defects well confined within the FGBC, and do not breach the grain boundaries formed by the seed strips during growth. Overall, these observations indicate that the FGBC effectively confines defects within its boundaries, preventing their spread into the mono-Si region, and maintains the quality of the mono-Si ingot throughout the growth process.

To analyze the effect of the FGBC on defect distribution in the mono-Si regions, the mono-Si bricks from the edge and center of the ingot were cut into wafers after removing the defective regions at the bottom and top. Wafers were collected at certain intervals and tested for PL images. The black filament regions in the PL images indicate the presence of crystal defects. The crystal quality of the wafers can be quantified by calculating the ratio of the black filament regions to the total wafer area using software. Fig. 6(a) and (b) show the PL images of mono-Si wafers from the center and edge zones of an ingot grown with FGBC, respectively. The difference in defect distribution between the edge and the center zones is small. The sources of dislocations are mainly generated from the corner regions of the wafers and expand during the subsequent growth process. Even at the top of the ingot, the black filament ratio in the wafer from the edge zone is only slightly higher than that from the center zone. In the regions of the wafers near the FGBC (indicated by arrows) from the edge zone brick, the black filament regions do not significantly increase compared to those from the center zone brick when H < 210 mm. This is mainly attributed to the FGBC in the edge bricks effectively blocking the expansion of crystal defects. Fig. 6(c) shows the PL image of a mono-Si wafer from the edge zone of an ingot without FGBC. At H = 110 mm, the mono-Si region has been overgrown by mc-Si from the crucible wall, generating a visible defective region (highlighted in the red circle). Due to the lack of blocking by grain boundaries at the periphery of the mono-Si region, the mc-Si rapidly expands into the mono-Si region, significantly increasing the area of the black filament region. Therefore, the use of the FGBC effectively improves the crystal quality of the edge region of cast mono-Si ingots by increasing the proportion of mono-Si and improving the defect distribution.

3.3 Barrier mechanism analysis

Fig. 7 shows the defect distribution within the FGBC at H = 60 mm. Three grain boundaries induced by three seed strips and square seeds are visible in the red box in Fig. 7(a). The growth region of the outermost seed strip is overgrown by mc-Si due to the absence of a grain boundary barrier, creating an intersection region between mono-Si and mc-Si. The area within the red box is etched using mixed acid, revealing numerous etching pits in the mono-Si and mc-Si intersection region. A small number of linear dislocation clusters are also present in the growth region of the middle seed strip, while no significant dislocation clusters are observed in the growth region of the inner seed strip. Fig. 7(b1)–(b3) illustrate the microscopic morphology of the mono-Si and mc-Si junction region, including twins and other mc-Si formed through extrusion growth. Due to the much lower grain boundary energy of the twin boundaries,²⁸ almost no dislocation regions are generated within the individually overgrown twins (Fig. 7(b1)). However, as shown in Fig. 7(b2) and (b3), a large number of dislocation clusters are generated when the twin intersects with the mc-Si. This is because other types of grain boundaries undergo reactions and form grain boundaries with higher grain boundary energy when they intersect with the twin boundaries.²⁹ Grain boundaries with higher energy are highly prone to emit dislocations during the growth process due to their structural instability. Fig. 7(b4), (b5), and (b6) show the microscopic morphology near grain boundaries GB I, GB II, and GB III, respectively. The morphology of GB I is slightly deformed due to the extrusion of mc-Si and dislocations (Fig. 7(b4)), and a bundle of linear dislocations can be seen blocked by GB I, demonstrating that the large-angle grain boundary can effectively block the movement of dislocations. Moreover, micro-twins are observed to be generated on the grain boundary and terminate on another grain boundary in Fig. 7(b5) and (b6). However, no significant dislocation clusters are observed in the growth region of the inner strip and the mono-Si region. GB II and GB III are almost in a straight-line morphology, suggesting that their structures are relatively stable. Overall, the observations indicate that FGBC effectively confines dislocations within its boundaries, preventing their spread into the mono-Si region. This behavior highlights the potential of FGBC to maintain the quality of mono-Si regions by blocking mc-Si overgrowth and controlling defect propagation.

We selected the FGBC area in Fig. 7 for EBSD testing and analyzed the grain orientation and grain boundary properties. Fig. 8(a), (b) and (c) respectively show the SEM images and local EBSD inverse pole images of GB I, GB II, and GB III regions in Fig. 7. The orientation differences between grains on both sides of GB I and GB II are about 44° (Fig. 8(a) and (b)), which is very close to the angular difference between the sides of neighboring seed strips during processing. The distortion of GB I results in an angular deviation of about 0.3° over the length of the illustration, suggesting that GB I already exhibits structural instability at this growth height. The difference in orientation between the two sides of the linear dislocation blocked by GB I is about 0.19°, so it can be regarded as a subgrain boundary or small-angle GB. The growth patterns of GB II and GB III are relatively stable, and the misorientation angles on both sides are kept within 0.1° (Fig. 8(b) and (c)), suggesting that GB II and GB III are more structurally stable than GB I at this growth height. Consequently, the probability of dislocations generated at these grain boundaries is lower. Overall, these EBSD analyses indicate that while GB I shows signs of structural instability and higher dislocation activity, GB II and GB III maintain their structural integrity, effectively minimizing the generation and propagation of dislocations within the FGBC. This reinforces the role of stable grain boundaries in enhancing the quality of mono-Si regions.

Fig. 9 shows the defect distribution within the FGBC at H = 120 mm. Compared to H = 60 mm, the three grain boundaries within the FGBC exhibit severe deformation (Fig. 9(a)). The penetration depth of mc-Si completely occupies the width of the growth region of the outermost seed strip, but GB I can still block the invasion of mc-Si. After etching the region within the red box with mixed acid, it was found that the area of dislocation clusters within the FGBC significantly increased compared to H = 60 mm, especially in the growth regions of the outermost and middle seed strips. Fig. 9(b1)–(b3) show the microstructure at the junction of mono-Si and mc-Si regions. No obvious dislocation clusters were observed in the isolated twins (Fig. 9(b1)), which is similar to the observation in Fig. 7. The reaction of Σ3 grain boundaries with GB I will alter the structure of GB I to generate new grain boundaries, which may no longer grow vertically, when the twins come into contact with GB I or other overgrowth mc-Si during growth. If the expansion inside the FGBC results in further reactions with the GB II and GB III grain boundaries, it will not be advantageous to stop the overgrowth of mc-Si. As a result, a weakened barrier effect may arise from the FGBC induced by a limited width or an excessively small number of seed strips. At the point where the twin boundaries inflection occurs, linear dislocations can be seen originating from the grain boundaries (Fig. 9(b2) and (b3)), indicating that dislocation formation is triggered by areas of stress concentration created by grain boundary reactions. GB I, GB II, and GB III's surrounding microstructure is depicted in Fig. 9(b4), (b5) and (b6), in that order. The morphologies of GB I and GB II at H = 120 mm are significantly deformed, in contrast to the observation at H = 60 mm (Fig. 9(b4) and (b5)). The distorted grain boundaries cause a large number of dislocations to form dislocation clusters, which suggests that the grain boundaries induced by seed strips also emit dislocations during growth. In spite of this, the deformed GB II and GB III may still efficiently prevent dislocations from moving, hence limiting the dislocation clusters to the seed strip growth regions. Meanwhile, the undistorted GB III effectively blocks the dislocation clusters from entering the mono-Si region (Fig. 9(b6)). The above results indicate that the FGBC undergoes gradual distortion and generates numerous dislocations due to the overgrowth of mc-Si during growth. However, multiple large-angle grain boundaries in the FGBC can repeatedly block dislocations or dislocation clusters, thus maximizing the restriction of dislocation propagation into the mono-Si region.

We selected the FGBC area in Fig. 9 for EBSD testing and analyzed the grain orientation and grain boundary properties. Fig. 10(a), (b) and (c) respectively show the SEM images and local EBSD inverse pole images of GB I, GB II, and GB III regions in Fig. 9. The orientation difference between the grains on either side of GB I is about 42° (Fig. 10(a)), which is slightly reduced compared to H = 60 mm. The deformation of the grain boundaries also leads to their angular deviation of more than 1° over a range of several hundred micrometers, likely due to the grain boundary reaction that occurs when GB I intersects the overgrown mc-Si. The distortion of GB II is relatively small, and since it does not intersect other grain boundaries, the angular deviation of GB II is only about 0.1° over the illustrated length of several hundred micrometers (Fig. 10(b)). The orientation difference on both sides of line dislocations on GB II was measured, revealing that one dislocation has an orientation difference of 1.29°, while another dislocation has a difference of only 0.17°. These line dislocations can be regarded as small-angle grain boundaries, and the varying angular differences reflect differences in dislocation density.³⁰ Small-angle grain boundaries can be considered as consisting of a set of parallel edge dislocations or intersecting screw dislocations, with dislocation density increasing with the angle of the small-angle grain boundaries. Compared to H = 60 mm, the orientation difference between the two sides of GB III does not change significantly, and the distortion of the grain boundary remains very small, with a deviation of only about 0.1° over several hundred micrometers (Fig. 10(c)), and no linear dislocations are observed nearby. These results indicate that the distortion and deformation of grain boundaries during the growth of the FGBC are closely related to the generation of dislocations. The farther a region within the FGBC is from the mc-Si region, the more stable the grain boundary structure is and the less likely dislocations will be generated. Therefore, if the growth height of the mono-Si ingot is substantial, a small number of grain boundaries within the FGBC may not be able to continuously block the overgrowth of mc-Si and dislocations.

3.4 Solar cell performance

To evaluate the improvement brought by the FGBC in the production of cast mono-Si solar cells, we collected silicon bricks from the center and edge of cast mono-Si ingots and cut them into mono-Si wafers, which were fabricated into solar cells on a PERC solar cell production line. The monocrystalline proportion of these wafers were all 100%, and the ratio of the PL defect area was less than 10%. The performance parameters of the solar cells are shown in Table 1. The absolute cell efficiency of wafers from the edge of the ingots using the FGBC increased by approximately 0.3% compared to wafers not using the FGBC. The absolute efficiency difference is narrowed from 0.5% to 0.2% compared to the solar cells prepared with center region wafers, which is mainly attributed to the significant improvement in the distribution of defects in the wafers from the edge of the ingots. Therefore, the application of the FGBC not only significantly improves the yield of cast mono-Si ingots but also significantly improves the efficiency distribution of solar cells, providing a reliable solution for the scale-up of cast mono-Si.

---

4 Conclusions

The study investigates the impact of the FGBC on the quality and performance of cast mono-Si ingots and solar cells. The FGBC technique creates a barrier between mc-Si and mono-Si regions, effectively preventing mc-Si overgrowth and significantly improving the crystal quality of the ingot. Infrared imaging and defect analysis reveal that ingots grown with FGBC show a substantial increase in the proportion of mono-Si (95% with FGBC compared to 60% without FGBC). This is due to the FGBC's ability to maintain a flat growth interface and block the spread of mc-Si, particularly in the edge regions of the ingots. Further, minority carrier lifetime distributions indicate that the FGBC minimizes crystal defects in the edge regions, leading to more uniform and higher-quality crystal structures throughout the ingot. The FGBC confines defects within its boundaries, preventing them from spreading into the mono-Si regions and maintaining the structural integrity of the grain boundaries. Performance tests of solar cells reveal that cells derived from wafers grown with FGBC had an abs. 0.3% higher efficiency than those grown without FGBC, which considerably narrows the efficiency difference between edge and center wafers. Overall, the application of the FGBC in cast mono-Si production enhances the yield and efficiency distribution of solar cells, providing a viable solution for the large-scale application of cast mono-Si in the PV industry.

Data availability

Due to the involvement of participant privacy, the dataset for this study is available only upon reasonable request. Research data can be obtained by contacting the corresponding author (email: E-mail: 529594405@qq.com) and will require signing a data use agreement.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work is supported by the Major Discipline Academic and Technical Leaders Training Program of Jiangxi Province (20232BCJ22029), the Yongjiang Talent Introduction Programme (2021A-147-G and 2022A-095-G) and the Ningbo Science and Technology Project (No. 2022-DST-004).

References

1. D. Hu, S. Yuan, J. Zhang, H. Chen, C. Wang, L. Wang, X. Yu and D. Yang, Higher quality composite silicon ingot with monocrystalline and multicrystalline bricks, Sol. Energy, 2022, 244, 484–489 CrossRef CAS .
2. Q. Lei, L. He, C. Tang, S. Liu, X. He, X. Li, Y. Xu, W. Mao, J. Li and L. Zhou, Growth of low-cost and high-quality monocrystalline silicon ingots by using split recycled seeds, Mater. Sci. Semicond. Process., 2022, 138, 106318 CrossRef CAS .
3. P. Wang, C. Cui, D. Yang and X. Yu, Seed-Assisted Growth of Cast-Mono Silicon for Photovoltaic Application: Challenges and Strategies, Sol. RRL, 2020, 4, 1900486 CrossRef CAS .
4. M. Becker, E. Pihan, F. Guittonneau, L. Barrallier, G. Regula, H. Ouaddah, G. Reinhart and N. Mangelinck-Noël, Investigation of subgrains in directionally solidified cast mono-seeded silicon and their interactions with twin boundaries, Sol. Energy Mater. Sol. Cells, 2020, 218, 110817 CrossRef CAS .
5. L. He, S. Yuan, Y. Xu, Q. Lei, W. Mao, H. Luo, X. He, X. Li, L. Wang, D. Yang and Y. Qian, Evaluation of large-scale recycled seed for cast monocrystalline silicon: Defect multiplication mechanisms and feasibility, Sol. Energy Mater. Sol. Cells, 2021, 230, 111266 CrossRef CAS .
6. D. Hu, D. S. Yuan, L. He, H. Chen, Y. Wan, X. Yu and D. Yang, Higher quality monolike cast silicon with induced grain boundaries, Sol. Energy Mater. Sol. Cells, 2015, 140, 121–125 CrossRef CAS .
7. W. Su, X. Han, C. Chang, Z. Zhang, Z. Zhang, W. Yang, J. Li and J. Wang, Impact of seed orientation on dislocation propagation in cast seed-assisted monocrystalline silicon, Sol. Energy Mater. Sol. Cells, 2021, 230, 111084 CrossRef CAS .
8. S. Liu, X. Huang, Y. Wang, M. Xia, Q. Lei and N. Zhou, Suppression of dislocations and twins by inducing asymmetrical grain boundaries for casting high-quality monocrystalline silicon ingot, Vacuum, 2022, 206, 111533 CrossRef CAS .
9. Y. C. Wu, A. Lan, C. F. Yang, C. W. Hsu, C. M. Lu, A. Yang and C. W. Lan, Effect of Growth Conditions on the Quality of Silicon Ingots, Cryst. Growth Des., 2016, 16, 6641–6647 CrossRef CAS .
10. C. Y. Lan, Y. C. Wu, A. Lan, C. F. Yang, C. Hsu, C. M. Lu, A. Yang and C. W. Lan, Investigation of Defect Formation in Silicon Crystals, J. Cryst. Growth, 2017, 475, 136–143 CrossRef CAS .
11. Q. H. Yu, L. J. Liu, W. C. Ma, G. X. Zhong and X. M. Huang, Influence of Growth Parameters on the Quality of Silicon Crystals, J. Cryst. Growth, 2012, 358, 5–11 CrossRef CAS .
12. B. Gao and K. Kakimoto, Numerical Simulation of Silicon Crystal Growth, J. Cryst. Growth, 2013, 384, 13–20 CrossRef CAS .
13. X. Liu, Y. F. Dang, H. Tanaka, Y. Fukuda, K. Kutsukake, T. Kojima, T. Ujihara and N. Usami, Optimization of Silicon Crystal Growth Processes, ACS Omega, 2022, 7, 6665–6673 CrossRef CAS PubMed .
14. W. J. Su, X. M. Han, C. B. Chang, Z. Q. Zhang, Z. Z. Zhang, W. Yang, J. L. Li and J. F. Wang, Investigation of Silicon Wafer Quality for Solar Cells, Sol. Energy Mater. Sol. Cells, 2021, 230, 111084 CrossRef CAS .
15. M. C. Schubert, F. Schindler, J. Benick, S. Riepe, P. Krenckel, A. Richter, R. Müller, B. Hammann and S. Nold, Efficiency Improvement in Silicon Solar Cells Through Advanced Materials, Sol. Energy Mater. Sol. Cells, 2021, 219, 110789 CrossRef CAS .
16. W. Mao, Q. Lei, L. He, J. Li, Y. Xu and Y. Liu, Upgrade and performance evaluation of G8+ large-size cast mono-Si ingot, Sol. Energy Mater. Sol. Cells, 2024, 273, 112951 CrossRef CAS .
17. K. Kutsukake, N. Usami, Y. Ohno, Y. Tokumoto and I. Yonenaga, Mono-like silicon growth using functional grain boundaries to limit area of multicrystalline grains, IEEE J. Photovolt., 2014, 4(1), 84–87 Search PubMed .
18. K. Kutsukake, T. Abe, N. Usami, K. Fujiwara, K. Morishita and K. Nakajima, Impact of Grain Boundaries on Silicon Crystal Growth, Scr. Mater., 2011, 65, 556–559 CrossRef CAS .
19. K. Kutsukake, N. Usami, Y. Ohno, Y. Tokumoto and I. Yonenaga, Stress Distribution in Silicon Wafers, Appl. Phys. Express, 2013, 6, 025505 CrossRef .
20. I. Taisho, T. Isao and N. Uritaka, Effects of grain boundary structure controlled by artificially designed seeds on dislocation generation, Jpn. J. Appl. Phys., 2017, 56, 075501 CrossRef .
21. S. Riepe, et al., Enhanced material quality in SMART mono-Si block cast ingots by introduction of functional defects, in Proceedings of the 36th EUPVSEC, Marseille, 2019 Search PubMed .
22. I. Takahashi, S. Joonwichien, T. Iwata and N. Usami, Seed manipulation for artificially controlled defect technique in new growth method for quasi-monocrystalline Si ingot based on casting, Appl. Phys. Express, 2015, 8, 105501 CrossRef .
23. F. Zhang, X. G. Yu, C. Liu, S. Yuan, X. D. Zhu, Z. X. Zhang, L. Huang, Q. Lei, D. L. Hu and D. R. Yang, Enhancing Silicon Solar Cell Efficiency with Novel Materials, Sol. Energy Mater. Sol. Cells, 2019, 200, 109985 CrossRef CAS .
24. M. Trempa, C. Reimann, J. Friedrich, G. Müller and D. Oriwol, Mono-crystalline growth in directional solidification of silicon with different orientation and splitting of seed crystals, J. Cryst. Growth, 2012, 351, 131–140 CrossRef CAS .
25. D. Hu, J. Zhang, S. Yuan, H. Chen, C. Wang, Q. Lei and D. Yang, Study on growth behavior of twins in cast monocrystalline silicon, Silicon, 2023, 15, 1993–2002 CAS .
26. C. W. Lan, C. F. Yang, A. Lan, M. Yang, A. Yu, H. P. Hsu, B. Hsu and C. Hsu, Engineering silicon crystals for photovoltaics, CrystEngComm, 2016, 18, 1474–1485 RSC .
27. L. C. Chuang, K. Maeda, H. Morito, K. Nagano and N. Usami, In situ observation of interaction between grain boundaries during directional solidification of Si, Scr. Mater., 2018, 148, 37–41 CrossRef CAS .
28. T. F. Ciszek, G. H. Schwuttke and K. H. Yang, Directionally solidified solar-grade silicon using carbon crucibles, J. Cryst. Growth, 1979, 46, 527–533 CrossRef CAS .
29. M. G. Tsoutsouva, B. T. Riberi, G. Regula, J. S. Lecomte, V. Favier and G. Rémond, In situ investigation of the structural defect generation and evolution during the directional solidification of seeded growth Si, Acta Mater., 2016, 115, 210–223 CrossRef CAS .
30. R. Hu, J. Zhang and S. Yuan, et al., Study on subgrain boundaries in cast monocrystalline silicon, Sol. Energy Mater. Sol. Cells, 2024, 271, 112845 CrossRef CAS .

This journal is © The Royal Society of Chemistry 2024