Recycling a variety of polysilicon qualified materials for semiconductor.

1, Photovoltaic end demand: The demand for photovoltaic installed capacity is strong, and the demand for polysilicon is reversed based on the installed capacity forecast

1.1. Polysilicon consumption: The global consumption volume is increasing steadily, mainly for photovoltaic power generation

The past ten years, the global polysilicon consumption has continued to rise, and China’s proportion has continued to expand, led by the photovoltaic industry. From 2012 to 2021, the global polysilicon consumption generally showed an upward trend, rising from 237,000 tons to about 653,000 tons. In 2018, China’s 531 photovoltaic new policy was introduced, which clearly reduced the subsidy rate for photovoltaic power generation. The newly installed photovoltaic capacity fell by 18% year-on-year, and the demand for polysilicon was affected. Since 2019, the state has introduced a number of policies to promote the grid parity of photovoltaics. With the rapid development of the photovoltaic industry, the demand for polysilicon has also entered a period of rapid growth. During this period, the proportion of China’s polysilicon consumption in the total global consumption continued to rise, from 61.5% in 2012 to 93.9% in 2021, mainly due to China’s rapidly developing photovoltaic industry. From the perspective of the global consumption pattern of different types of polysilicon in 2021, silicon materials used for photovoltaic cells will account for at least 94%, of which solar-grade polysilicon and granular silicon account for 91% and 3%, respectively, while electronic-grade polysilicon that can be used for chips accounts for 94%. The ratio is 6%, which shows that the current demand for polysilicon is dominated by photovoltaics. It is expected that with the warming of the dual-carbon policy, the demand for photovoltaic installed capacity will become stronger, and the consumption and proportion of solar-grade polysilicon will continue to increase.

1.2. Silicon wafer: monocrystalline silicon wafer occupies the mainstream, and continuous Czochralski technology develops rapidly

The direct downstream link of polysilicon is silicon wafers, and China currently dominates the global silicon wafer market. From 2012 to 2021, the global and Chinese silicon wafer production capacity and output continued to increase, and the photovoltaic industry continued to boom. Silicon wafers serve as a bridge connecting silicon materials and batteries, and there is no burden on production capacity, so it continues to attract a large number of companies to enter the industry. In 2021, Chinese silicon wafer manufacturers had significantly expanded production capacity to 213.5GW output, which drived the global silicon wafer production to increase to 215.4GW. According to the existing and newly increased production capacity in China, it is expected that the annual growth rate will maintain 15-25% in the next few years, and China’s wafer production will still maintain an absolute dominant position in the world.

Polycrystalline silicon can be made into polycrystalline silicon ingots or monocrystalline silicon rods. The production process of polycrystalline silicon ingots mainly includes casting method and direct melting method. At present, the second type is the main method, and the loss rate is basically maintained at about 5%. The casting method is mainly to melt the silicon material in the crucible first, and then cast it in another preheated crucible for cooling. By controlling the cooling rate, the polycrystalline silicon ingot is cast by the directional solidification technology. The hot-melting process of the direct-melting method is the same as that of the casting method, in which the polysilicon is directly melted in the crucible first, but the cooling step is different from the casting method. Although the two methods are very similar in nature, the direct melting method only needs one crucible, and the polysilicon product produced is of good quality, which is conducive to the growth of polycrystalline silicon ingots with better orientation, and the growth process is easy to automate, which can make the internal position of the crystal Error reduction. At present, the leading enterprises in the solar energy material industry generally use the direct melting method to make polycrystalline silicon ingots, and the carbon and oxygen contents are relatively low, which are controlled below 10ppma and 16ppma. In the future, the production of polycrystalline silicon ingots will still be dominated by the direct melting method, and the loss rate will remain around 5% within five years.

The production of monocrystalline silicon rods is mainly based on the Czochralski method, supplemented by the vertical suspension zone melting method, and the products produced by the two have different uses. The Czochralski method uses graphite resistance to heat polycrystalline silicon in a high-purity quartz crucible in a straight-tube thermal system to melt it, then insert the seed crystal into the surface of the melt for fusion, and rotate the seed crystal while inverting the crucible. , the seed crystal is slowly raised upward, and monocrystalline silicon is obtained through the processes of seeding, amplification, shoulder turning, equal diameter growth, and finishing. The vertical floating zone melting method refers to fixing the columnar high-purity polycrystalline material in the furnace chamber, moving the metal coil slowly along the polycrystalline length direction and passing through the columnar polycrystalline, and passing a high-power radio frequency current in the metal coil to make Part of the inside of the polycrystalline pillar coil melts, and after the coil is moved, the melt recrystallizes to form a single crystal. Due to the different production processes, there are differences in production equipment, production costs and product quality. At present, the products obtained by the zone melting method have high purity and can be used for the manufacture of semiconductor devices, while the Czochralski method can meet the conditions for producing single crystal silicon for photovoltaic cells and has a lower cost, so it is the mainstream method. In 2021, the market share of the straight pull method is about 85%, and it is expected to increase slightly in the next few years. The market shares in 2025 and 2030 are predicted to be 87% and 90% respectively. In terms of district melting single crystal silicon, the industry concentration of district melting single crystal silicon is relatively high in the world. acquisition), TOPSIL (Denmark) . In the future, the output scale of molten single crystal silicon will not increase significantly. The reason is that China’s related technologies are relatively backward compared with Japan and Germany, especially the capacity of high-frequency heating equipment and crystallization process conditions. The technology of fused silicon single crystal in large diameter area requires Chinese enterprises to continue to explore by themselves.

Czochralski method can be divided into continuous crystal pulling technology (CCZ) and repeated crystal pulling technology (RCZ). At present, the mainstream method in the industry is RCZ, which is in the transition stage from RCZ to CCZ. The single crystal pulling and feeding steps of RZC are independent of each other. Before each pulling, the single crystal ingot must be cooled and removed in the gate chamber, while CCZ can realize feeding and melting while pulling. RCZ is relatively mature, and there is little room for technological improvement in the future; while CCZ has the advantages of cost reduction and efficiency improvement, and is in a stage of rapid development. In terms of cost, compared with RCZ, which takes about 8 hours before a single rod is drawn, CCZ can greatly improve production efficiency, reduce crucible cost and energy consumption by eliminating this step. The total single furnace output is more than 20% higher than that of RCZ. Production cost is more than 10% lower than RCZ. In terms of efficiency, CCZ can complete the drawing of 8-10 single crystal silicon rods within the life cycle of the crucible (250 hours), while RCZ can only complete about 4, and the production efficiency can be increased by 100-150%. In terms of quality, CCZ has more uniform resistivity, lower oxygen content, and slower accumulation of metal impurities, so it is more suitable for the preparation of n-type single crystal silicon wafers, which are also in a period of rapid development. At present , some Chinese companies have announced that they have CCZ technology , and the route of granular silicon-CCZ-n-type monocrystalline silicon wafers has been basically clear, and has even begun to use 100% granular silicon materials. . In the future, CCZ will basically replace RCZ, but it will take a certain process.

The production process of monocrystalline silicon wafers is divided into four steps: pulling, slicing, slicing, cleaning and sorting. The emergence of the diamond wire slicing method has greatly reduced the slicing loss rate. The crystal pulling process has been described above. The slicing process includes truncation, squaring, and chamfering operations. Slicing is to use a slicing machine to cut the columnar silicon into silicon wafers. Cleaning and sorting are the final steps in the production of silicon wafers. The diamond wire slicing method has obvious advantages over the traditional mortar wire slicing method, which is mainly reflected in the short time consumption and low loss. The speed of diamond wire is five times that of traditional cutting. For example, for single-wafer cutting, traditional mortar wire cutting takes about 10 hours, and diamond wire cutting only takes about 2 hours. The loss of diamond wire cutting is also relatively small, and the damage layer caused by diamond wire cutting is smaller than that of mortar wire cutting, which is conducive to cutting thinner silicon wafers. In recent years, in order to reduce cutting losses and production costs, companies have turned to diamond wire slicing methods, and the diameter of diamond wire bus bars is getting lower and lower. In 2021, the diameter of the diamond wire busbar will be 43-56 μm, and the diameter of the diamond wire busbar used for monocrystalline silicon wafers will decrease greatly and continue to decline. It is estimated that in 2025 and 2030, the diameters of the diamond wire busbars used to cut monocrystalline silicon wafers will be 36 μm and 33 μm, respectively, and the diameters of the diamond wire busbars used to cut polycrystalline silicon wafers will be 51 μm and 51 μm, respectively. This is because there are many defects and impurities in polycrystalline silicon wafers, and thin wires are prone to breakage. Therefore, the diameter of the diamond wire busbar used for polycrystalline silicon wafer cutting is larger than that of monocrystalline silicon wafers, and as the market share of polycrystalline silicon wafers gradually decreases, it is used for polycrystalline silicon The reduction in the diameter of the diamond wire busbars cut by slices has slowed down.

At present, silicon wafers are mainly divided into two types: polycrystalline silicon wafers and monocrystalline silicon wafers. Monocrystalline silicon wafers have the advantages of long service life and high photoelectric conversion efficiency. Polycrystalline silicon wafers are composed of crystal grains with different crystal plane orientations, while single crystal silicon wafers are made of polycrystalline silicon as raw materials and have the same crystal plane orientation. In appearance, polycrystalline silicon wafers and single crystal silicon wafers are blue-black and black-brown. Since the two are cut from polycrystalline silicon ingots and monocrystalline silicon rods, respectively, the shapes are square and quasi-square. The service life of polycrystalline silicon wafers and monocrystalline silicon wafers is about 20 years. If the packaging method and use environment are suitable, the service life can reach more than 25 years. Generally speaking, the lifespan of monocrystalline silicon wafers is slightly longer than that of polycrystalline silicon wafers. In addition, monocrystalline silicon wafers are also slightly better in photoelectric conversion efficiency, and their dislocation density and metal impurities are much smaller than those of polycrystalline silicon wafers. The combined effect of various factors makes the minority carrier lifetime of single crystals dozens of times higher than that of polycrystalline silicon wafers. Thereby showing the advantage of conversion efficiency. In 2021, the highest conversion efficiency of polycrystalline silicon wafers will be around 21%, and that of monocrystalline silicon wafers will reach up to 24.2%.

In addition to long life and high conversion efficiency, monocrystalline silicon wafers also have the advantage of thinning, which is conducive to reducing silicon consumption and silicon wafer costs, but pay attention to the increase in fragmentation rate. The thinning of silicon wafers helps reduce manufacturing costs, and the current slicing process can fully meet the needs of thinning, but the thickness of silicon wafers must also meet the needs of downstream cell and component manufacturing. In general, the thickness of silicon wafers has been decreasing in recent years, and the thickness of polycrystalline silicon wafers is significantly larger than that of monocrystalline silicon wafers. Monocrystalline silicon wafers are further divided into n-type silicon wafers and p-type silicon wafers, while n-type silicon wafers mainly include TOPCon Battery usage and HJT battery usage. In 2021, the average thickness of polycrystalline silicon wafers is 178μm, and the lack of demand in the future will drive them to continue to thin. Therefore, it is predicted that the thickness will decrease slightly from 2022 to 2024, and the thickness will remain at about 170μm after 2025; the average thickness of p-type monocrystalline silicon wafers is about 170μm, and it is expected to drop to 155μm and 140μm in 2025 and 2030. Among the n-type monocrystalline silicon wafers, the thickness of the silicon wafers used for HJT cells is about 150μm, and the average thickness of n-type silicon wafers used for TOPCon cells is 165μm. 135μm.

In addition, the production of polycrystalline silicon wafers consumes more silicon than monocrystalline silicon wafers, but the production steps are relatively simple, which brings cost advantages to polycrystalline silicon wafers. Polycrystalline silicon, as a common raw material for polycrystalline silicon wafers and monocrystalline silicon wafers, has different consumption in the production of the two, which is due to the differences in the purity and production steps of the two. In 2021, the silicon consumption of polycrystalline ingot is 1.10 kg/kg. It is expected that the limited investment in research and development will lead to small changes in the future. The silicon consumption of the pull rod is 1.066 kg/kg, and there is a certain room for optimization. It is expected to be 1.05 kg/kg and 1.043 kg/kg in 2025 and 2030, respectively. In the single crystal pulling process, the reduction of the silicon consumption of the pulling rod can be achieved by reducing the loss of cleaning and crushing, strictly controlling the production environment, reducing the proportion of primers, improving the precision control, and optimizing the classification and processing technology of degraded silicon materials. Although the silicon consumption of polycrystalline silicon wafers is high, the production cost of polycrystalline silicon wafers is relatively high because polycrystalline silicon ingots are produced by hot-melting ingot casting, while monocrystalline silicon ingots are usually produced by slow growth in Czochralski single crystal furnaces, which consumes relatively high power. Low. In 2021, the average production cost of monocrystalline silicon wafers will be about 0.673 yuan/W, and that of polycrystalline silicon wafers will be 0.66 yuan/W.

As the thickness of the silicon wafer decreases and the diameter of the diamond wire busbar decreases, the output of silicon rods/ingots of equal diameter per kilogram will increase, and the number of single crystal silicon rods of the same weight will be higher than that of polycrystalline silicon ingots. In terms of power, the power used by each silicon wafer varies according to the type and size. In 2021, the output of p-type 166mm size monocrystalline square bars is about 64 pieces per kilogram, and the output of polycrystalline square ingots is about 59 pieces. Among the p-type single crystal silicon wafers, the output of 158.75mm size monocrystalline square rods is about 70 pieces per kilogram, the output of p-type 182mm size single crystal square rods is about 53 pieces per kilogram, and the output of p-type 210mm size single crystal rods per kilogram is about 53 pieces. The output of the square bar is about 40 pieces. From 2022 to 2030, the continuous thinning of silicon wafers will undoubtedly lead to an increase in the number of silicon rods/ingots of the same volume. The smaller diameter of the diamond wire busbar and medium particle size will also help reduce cutting losses, thereby increasing the number of wafers produced. quantity. It is estimated that in 2025 and 2030, the output of p-type 166mm size monocrystalline square rods is about 71 and 78 pieces per kilogram, and the output of polycrystalline square ingots is about 62 and 62 pieces, which is due to the low market share of polycrystalline silicon wafers It is difficult to cause significant technological progress. There are differences in the power of different types and sizes of silicon wafers. According to the announcement data for the average power of 158.75mm silicon wafers is about 5.8W/piece, the average power of 166mm size silicon wafers is about 6.25W/piece, and the average power of 182mm silicon wafers is about 6.25W/piece. The average power of the size silicon wafer is about 7.49W/piece, and the average power of the 210mm size silicon wafer is about 10W/piece.

In recent years, silicon wafers have gradually developed in the direction of large size, and large size is conducive to increasing the power of a single chip, thereby diluting the non-silicon cost of cells. However, the size adjustment of silicon wafers also needs to consider upstream and downstream matching and standardization issues, especially the load and high current issues. At present, there are two camps in the market regarding the future development direction of silicon wafer size, namely 182mm size and 210mm size. The proposal of 182mm is mainly from the perspective of vertical industry integration, based on the consideration of the installation and transportation of photovoltaic cells, the power and efficiency of modules, and the synergy between upstream and downstream; while 210mm is mainly from the perspective of production cost and system cost. The output of 210mm silicon wafers increased by more than 15% in the single-furnace rod drawing process, the downstream battery production cost was reduced by about 0.02 yuan/W, and the total cost of power station construction was reduced by about 0.1 yuan/W. In the next few years, it is expected that silicon wafers with a size below 166mm will be gradually eliminated; the upstream and downstream matching problems of 210mm silicon wafers will be gradually solved effectively, and cost will become a more important factor affecting the investment and production of enterprises. Therefore, the market share of 210mm silicon wafers will increase. Steady rise; 182mm silicon wafer will become the mainstream size in the market by virtue of its advantages in vertically integrated production, but with the breakthrough development of 210mm silicon wafer application technology, 182mm will give way to it. In addition, it is difficult for larger-sized silicon wafers to be widely used in the market in the next few years, because the labor cost and installation risk of large-sized silicon wafers will greatly increase, which is difficult to be offset by the savings in production costs and system costs. . In 2021, silicon wafer sizes on the market include 156.75mm, 157mm, 158.75mm, 166mm, 182mm, 210mm, etc. Among them, the size of 158.75mm and 166mm accounted for 50% of the total, and the size of 156.75mm decreased to 5%, which will be gradually replaced in the future; 166mm is the largest size solution that can be upgraded for the existing battery production line, which will be the largest size in the past two years. In terms of transition size, it is expected that the market share will be less than 2% in 2030; the combined size of 182mm and 210mm will account for 45% in 2021, and the market share will increase rapidly in the future. It is expected that the total market share in 2030 will exceed 98%.

In recent years, the market share of monocrystalline silicon has continued to increase, and it has occupied the mainstream position in the market. From 2012 to 2021, the proportion of monocrystalline silicon rose from less than 20% to 93.3%, a significant increase. In 2018, the silicon wafers on the market are mainly polycrystalline silicon wafers, accounting for more than 50%. The main reason is that the technical advantages of monocrystalline silicon wafers cannot cover the cost disadvantages. Since 2019, as the photoelectric conversion efficiency of monocrystalline silicon wafers has significantly exceeded that of polycrystalline silicon wafers, and the production cost of monocrystalline silicon wafers has continued to decline with technological progress, the market share of monocrystalline silicon wafers has continued to increase, becoming the mainstream in the market. product. It is expected that the proportion of monocrystalline silicon wafers will reach about 96% in 2025, and the market share of monocrystalline silicon wafers will reach 97.7% in 2030. (Report source: Future Think Tank)

1.3. Batteries: PERC batteries dominate the market, and the development of n-type batteries pushes up product quality

The midstream link of the photovoltaic industry chain includes photovoltaic cells and photovoltaic cell modules. The processing of silicon wafers into cells is the most important step in realizing photoelectric conversion. It takes about seven steps to process a conventional cell from a silicon wafer. First, put the silicon wafer into hydrofluoric acid to produce a pyramid-like suede structure on its surface, thereby reducing the reflectivity of sunlight and increasing light absorption; the second is Phosphorus is diffused on the surface of one side of the silicon wafer to form a PN junction, and its quality directly affects the efficiency of the cell; the third is to remove the PN junction formed on the side of the silicon wafer during the diffusion stage to prevent short circuit of the cell; A layer of silicon nitride film is coated on the side where the PN junction is formed to reduce light reflection and at the same time increase efficiency; the fifth is to print metal electrodes on the front and back of the silicon wafer to collect minority carriers generated by photovoltaics; The circuit printed in the printing stage is sintered and formed, and it is integrated with the silicon wafer, that is, the cell; finally, the cells with different efficiencies are classified.

Crystalline silicon cells are usually made with silicon wafers as substrates, and can be divided into p-type cells and n-type cells according to the type of silicon wafers. Among them, n-type cells have higher conversion efficiency and are gradually replacing p-type cells in recent years. P-type silicon wafers are made by doping silicon with boron, and n-type silicon wafers are made of phosphorus. Therefore, the concentration of boron element in the n-type silicon wafer is lower, thereby inhibiting the bonding of boron-oxygen complexes, improving the minority carrier lifetime of the silicon material, and at the same time, there is no photo-induced attenuation in the battery. In addition, the n-type minority carriers are holes, the p-type minority carriers are electrons, and the trapping cross-section of most impurity atoms for holes is smaller than that of electrons. Therefore, the minority carrier lifetime of the n-type cell is higher and the photoelectric conversion rate is higher. According to laboratory data, the upper limit of the conversion efficiency of p-type cells is 24.5%, and the conversion efficiency of n-type cells is up to 28.7%, so n-type cells represent the development direction of future technology. In 2021, n-type cells (mainly including heterojunction cells and TOPCon cells) have relatively high costs, and the scale of mass production is still small. The current market share is about 3%, which is basically the same as that in 2020.

In 2021, the conversion efficiency of n-type cells will be significantly improved, and it is expected that there will be more room for technological progress in the next five years. In 2021, the large-scale production of p-type monocrystalline cells will use PERC technology, and the average conversion efficiency will reach 23.1%, an increase of 0.3 percentage points compared with 2020; the conversion efficiency of polycrystalline black silicon cells using PERC technology will reach 21.0%, compared with 2020. Annual increase of 0.2 percentage points; conventional polycrystalline black silicon cell efficiency improvement is not strong, the conversion efficiency in 2021 will be about 19.5%, only 0.1 percentage point higher, and the future efficiency improvement space is limited; the average conversion efficiency of ingot monocrystalline PERC cells is 22.4% , which is 0.7 percentage points lower than that of monocrystalline PERC cells; the average conversion efficiency of n-type TOPCon cells reaches 24%, and the average conversion efficiency of heterojunction cells reaches 24.2%, both of which have been greatly improved compared with 2020, and the average conversion efficiency of IBC cells reaches 24.2%. With the development of technology in the future, battery technologies such as TBC and HBC may also continue to make progress. In the future, with the reduction of production costs and the improvement of yield, n-type batteries will be one of the main development directions of battery technology.

From the perspective of battery technology route, the iterative update of battery technology has mainly gone through BSF, PERC, TOPCon based on PERC improvement, and HJT, a new technology that subverts PERC; TOPCon can be further combined with IBC to form TBC, and HJT can also be combined with IBC to become HBC. P-type monocrystalline cells mainly use PERC technology, p-type polycrystalline cells include polycrystalline black silicon cells and ingot monocrystalline cells, the latter refers to the addition of monocrystalline seed crystals on the basis of conventional polycrystalline ingot process, directional solidification After that, a square silicon ingot is formed, and a silicon wafer mixed with single crystal and polycrystalline is made through a series of processing processes. Because it essentially uses a polycrystalline preparation route, it is included in the category of p-type polycrystalline cells. The n-type cells mainly include TOPCon monocrystalline cells, HJT monocrystalline cells and IBC monocrystalline cells. In 2021, the new mass production lines will still be dominated by PERC cell production lines, and the market share of PERC cells will further increase to 91.2%. As the product demand for outdoor and household projects has concentrated on high-efficiency products, the market share of BSF batteries will drop from 8.8% to 5% in 2021.

1.4. Modules: The cost of the cells accounts for the main part, and the power of the modules depends on the cells

The production steps of photovoltaic modules mainly include cell interconnection and lamination, and cells account for a major part of the total cost of the module. Since the current and voltage of a single cell are very small, the cells need to be interconnected through bus bars. Here, they are connected in series to increase the voltage, and then connected in parallel to obtain high current, and then the photovoltaic glass, EVA or POE, battery Sheet, EVA or POE, back sheet are sealed and heat pressed in a certain order, and finally protected by aluminum frame and silicone sealing edge. From the perspective of component production cost composition, material cost accounts for 75%, occupying the main position, followed by manufacturing cost, performance cost and labor cost. The cost of materials is led by the cost of cells. According to announcements from many companies, cells account for about 2/3 of the total cost of photovoltaic modules.

Photovoltaic modules are usually divided according to cell type, size, and quantity. There are differences in the power of different modules, but they are all in the rising stage. Power is a key indicator of photovoltaic modules, representing the module’s ability to convert solar energy into electricity. It can be seen from the power statistics of different types of photovoltaic modules that when the size and number of cells in the module are the same, the power of the module is n-type single crystal > p-type single crystal > polycrystalline; The larger the size and quantity, the greater the power of the module; for TOPCon single crystal modules and heterojunction modules of the same specification, the power of the latter is greater than that of the former. According to CPIA forecast, module power will increase by 5-10W per year in the next few years. In addition, module packaging will bring a certain power loss, mainly including optical loss and electrical loss. The former is caused by the transmittance and optical mismatch of packaging materials such as photovoltaic glass and EVA, and the latter mainly refers to the use of solar cells in series. The circuit loss caused by the resistance of the welding ribbon and the bus bar itself, and the current mismatch loss caused by the parallel connection of the cells, the total power loss of the two accounts for about 8%.

1.5. Photovoltaic installed capacity: The policies of various countries are obviously driven, and there is huge space for new installed capacity in the future

The world has basically reached a consensus on net zero emissions under the environmental protection goal, and the economics of superimposed photovoltaic projects have gradually emerged. Countries are actively exploring the development of renewable energy power generation. In recent years, countries around the world have made commitments to reduce carbon emissions. Most of the major greenhouse gas emitters have formulated corresponding renewable energy targets, and the installed capacity of renewable energy is huge. Based on the 1.5℃ temperature control target, IRENA predicts that the global installed renewable energy capacity will reach 10.8TW in 2030. In addition, according to WOODMac data, the level cost of electricity (LCOE) of solar power generation in China, India, the United States and other countries is already lower than the cheapest fossil energy, and will further decline in the future. The active promotion of policies in various countries and the economics of photovoltaic power generation have led to a steady increase in the cumulative installed capacity of photovoltaics in the world and China in recent years. From 2012 to 2021, the cumulative installed capacity of photovoltaics in the world will increase from 104.3GW to 849.5GW, and the cumulative installed capacity of photovoltaics in China will increase from 6.7GW to 307GW, an increase of over 44 times. In addition, China’s newly installed photovoltaic capacity accounts for more than 20% of the world’s total installed capacity. In 2021, China’s newly installed photovoltaic capacity is 53GW, accounting for about 40% of the world’s newly installed capacity. This is mainly due to the abundant and uniform distribution of light energy resources in China, the well-developed upstream and downstream, and the strong support of national policies. During this period, China has played a huge role in photovoltaic power generation, and the cumulative installed capacity has accounted for less than 6.5%. jumped to 36.14%.

Based on the above analysis, CPIA has given the forecast for newly increased photovoltaic installations from 2022 to 2030 all over the world. It is estimated that under both optimistic and conservative conditions, the global newly installed capacity in 2030 will be 366 and 315GW respectively, and newly installed capacity of China will be 128. , 105GW. Below we will forecast the demand for polysilicon based on the scale of newly installed capacity each year.

1.6. Demand forecast of polysilicon for photovoltaic applications

From 2022 to 2030, based on CPIA’s forecast for the global newly increased PV installations under both optimistic and conservative scenarios, the demand for polysilicon for PV applications can be predicted. Cells are a key step to realize photoelectric conversion, and silicon wafers are the basic raw materials of cells and the direct downstream of polysilicon, so it is an important part of polysilicon demand forecasting. The weighted number of pieces per kilogram of silicon rods and ingots can be calculated from the number of pieces per kilogram and the market share of silicon rods and ingots. Then, according to the power and market share of silicon wafers of different sizes, the weighted power of the silicon wafers can be obtained, and then the required number of silicon wafers can be estimated according to the newly installed photovoltaic capacity. Next, the weight of the required silicon rods and ingots can be obtained according to the quantitative relationship between the number of silicon wafers and the weighted number of silicon rods and silicon ingots per kilogram. Further combined with the weighted silicon consumption of silicon rods/silicon ingots, the demand for polysilicon for newly installed photovoltaic capacity can be finally obtained.

According to the forecast results, the global demand for polysilicon for new photovoltaic installations in the past five years will continue to rise, peaking in 2027, and then declining slightly in the next three years. It is estimated that under optimistic and conservative conditions in 2025, the global annual demand for polysilicon for photovoltaic installations will be 1,108,900 tons and 907,800 tons respectively, and the global demand for polysilicon for photovoltaic applications in 2030 will be 1,042,100 tons under optimistic and conservative conditions. , 896,900 tons. According to China’s  proportion of global photovoltaic installed capacity,  China’s demand for polysilicon for photovoltaic use in 2025 is expected to be 369,600 tons and 302,600 tons respectively under optimistic and conservative conditions, and 739,300 tons and 605,200 tons overseas respectively.

2, Semiconductor end demand: The scale is much smaller than the demand in the photovoltaic field, and future growth can be expected

In addition to making photovoltaic cells, polysilicon can also be used as a raw material for making chips and is used in the semiconductor field, which can be subdivided into automobile manufacturing, industrial electronics, electronic communications, home appliances and other fields. The process from polysilicon to chip is mainly divided into three steps. First, the polysilicon is drawn into monocrystalline silicon ingots, and then cut into thin silicon wafers. Silicon wafers are produced through a series of grinding, chamfering and polishing operations. , which is the basic raw material of the semiconductor factory. Finally, the silicon wafer is cut and laser engraved into various circuit structures to make chip products with certain characteristics. Common silicon wafers mainly include polished wafers, epitaxial wafers and SOI wafers. Polished wafer is a chip production material with high flatness obtained by polishing the silicon wafer to remove the damaged layer on the surface, which can be directly used to make chips, epitaxial wafers and SOI silicon wafers. Epitaxial wafers are obtained by epitaxial growth of polished wafers, while SOI silicon wafers are fabricated by bonding or ion implantation on polished wafer substrates, and the preparation process is relatively difficult.

Through the demand for polysilicon on the semiconductor side in 2021, combined with the agency’s forecast of the growth rate of the semiconductor industry in the next few years, the demand for polysilicon in the semiconductor field from 2022 to 2025 can be roughly estimated. In 2021, the global electronic-grade polysilicon production will account for about 6% of the total polysilicon production, and solar-grade polysilicon and granular silicon will account for about 94%. Most electronic-grade polysilicon is used in the semiconductor field, and other polysilicon is basically used in the photovoltaic industry. . Therefore, it can be assumed that the amount of polysilicon used in the semiconductor industry in 2021 is about 37,000 tons. In addition, according to the future compound growth rate of the semiconductor industry predicted by FortuneBusiness Insights, the demand for polysilicon for semiconductor use will increase at an annual rate of 8.6% from 2022 to 2025. It is estimated that in 2025, the demand for polysilicon in the semiconductor field will be around 51,500 tons. (Report source: Future Think Tank)

3, Polysilicon import and export: imports far exceed exports, with Germany and Malaysia accounting for a higher proportion

In 2021, about 18.63% of China ‘s polysilicon demand will come from imports, and the scale of imports far exceeds the scale of exports. From 2017 to 2021, the import and export pattern of polysilicon is dominated by imports, which may be due to the strong downstream demand for photovoltaic industry that has developed rapidly in recent years, and its demand for polysilicon accounts for more than 94% of the total demand; In addition, the company has not yet mastered the production technology of high-purity electronic-grade polysilicon, so some polysilicon required by the integrated circuit industry still needs to rely on imports. According to the data of the Silicon Industry Branch, the import volume continued to decline in 2019 and 2020. The fundamental reason for the decline in polysilicon imports in 2019 was the substantial increase in production capacity, which rose from 388,000 tons in 2018 to 452,000 tons in 2019. At the same time, OCI, REC, HANWHA Some overseas companies, such as some overseas companies, have withdrawn from the polysilicon industry due to losses, so the import dependence of polysilicon is much lower; although production capacity has not increased in 2020 , the impact of the epidemic has led to delays in the construction of photovoltaic projects, and the number of polysilicon orders has decreased in the same period. In 2021, China’s photovoltaic market will develop rapidly, and the apparent consumption of polysilicon will reach 613,000 tons, driving the import volume to rebound. In the past five years , China’s net polysilicon import volume has been between 90,000 and 140,000 tons, of which about 103,800 tons in 2021. It is expected that China’s net polysilicon import volume will remain around 100,000 tons per year from 2022 to 2025.

China’s polysilicon imports mainly come from Germany, Malaysia, Japan and Taiwan, China, and the total imports from these four countries will account for 90.51% in 2021. About 45% of China ‘s polysilicon imports come from Germany, 26% from Malaysia, 13.5% from Japan, and 6% from Taiwan. Germany owns the world’s polysilicon giant WACKER, which is the largest source of overseas polysilicon, accounting for 12.7% of the total global production capacity in 2021; Malaysia has a large number of polysilicon production lines from South Korea’s OCI Company, which originates from the original production line in Malaysia of TOKUYAMA, a Japanese company acquired by OCI. There are factories and some factories that OCI moved from South Korea to Malaysia. The reason for the relocation is that Malaysia provides free factory space and the cost of electricity is one-third lower than that of South Korea; Japan and Taiwan, China have TOKUYAMA , GET and other companies, which occupy a large share of polysilicon production. a place. In 2021, the polysilicon output will be 492,000 tons, which the newly installed photovoltaic capacity and chip production demand will be 206,400 tons and 1,500 tons respectively, and the remaining 284,100 tons will be mainly used for downstream processing and exported overseas. In the downstream links of polysilicon, silicon wafers, cells and modules are mainly exported, among which the export of modules is particularly prominent. In 2021, 4.64 billion silicon wafers and 3.2 billion photovoltaic cells had been exported from China, with a total export of 22.6GW and 10.3GW respectively, and the export of photovoltaic modules is 98.5GW, with very few imports. In terms of export value composition, module exports in 2021 will reach US$24.61 billion, accounting for 86%, followed by silicon wafers and batteries. In 2021, the global output of silicon wafers, photovoltaic cells, and photovoltaic modules will reach 97.3%, 85.1%, and 82.3%, respectively. It is expected that the global photovoltaic industry will continue to concentrate in China within the next three years, and the output and export volume of each link will be considerable. Therefore, it is estimated that from 2022 to 2025, the amount of polysilicon used for processing and producing downstream products and exported abroad will gradually increase. It is estimated by subtracting overseas production from overseas polysilicon demand. In 2025, polysilicon produced by processing into downstream products will be estimated to export 583,000 Tons to foreign countries from China

4, Summary and Outlook

The global polysilicon demand is mainly concentrated in the photovoltaic field, and the demand in the semiconductor field is not an order of magnitude. The demand for polysilicon is driven by photovoltaic installations, and is gradually transmitted to polysilicon through the link of photovoltaic modules-cell-wafer, generating demand for it. In the future, with the expansion of global photovoltaic installed capacity, the demand for polysilicon is generally optimistic. Optimistically, China and overseas newly increased PV installations causing the demand for polysilicon in 2025 will be 36.96GW and 73.93GW respectively, and the demand under conservative conditions will also reach 30.24GW and 60.49GW respectively. In 2021, the global polysilicon supply and demand will be tight, resulting in high global polysilicon prices. This situation may continue until 2022, and gradually turn to the stage of loose supply after 2023. In the second half of 2020, the impact of the epidemic began to weaken, and downstream production expansion drove the demand for polysilicon, and some leading companies planned to expand production. However, the expansion cycle of more than one and a half years resulted in the release of production capacity at the end of 2021 and 2022, resulting in a 4.24% increase in 2021. There is a supply gap of 10,000 tons, so prices have risen sharply. It is predicted that in 2022, under the optimistic and conservative conditions of photovoltaic installed capacity, the supply and demand gap will be -156,500 tons and 2,400 tons respectively, and the overall supply will still be in a state of relatively short supply. In 2023 and beyond, the new projects that started construction at the end of 2021 and early 2022 will start production and achieve a ramp-up in production capacity. Supply and demand will gradually loosen, and prices may be under downward pressure. In the follow-up, attention should be paid to the impact of the Russian-Ukrainian war on the global energy pattern, which may change the global plan for newly installed photovoltaic capacity, which will affect the demand for polysilicon.

(This article is only for the reference of UrbanMines’ customers and does not represent any investment advice）