Patent Application
20110129402
Most of polysilicon for solar cell is produced by CVD (Chemical Vapor Deposition) reactor to deposit TCS on pure silicon rod at a temperature around 1,100° C., “Siemens Process”. However, lots of TCS turned into STC in the CVD reactor due to the HCl, which is produced from hydrogen-dechlorination of TCS, reacts again with TCS and produce STC. The amount of STC produced from the CVD reactor is about 15 MT of STC/1 MT polysilicon. Due to such huge amount, some small plants were shut down. Thermal converter, for hydrogenation of those STC at high temperature, has been commercially used for the “Siemens Process.” However, the conversion rate of STC to TCS from the thermal converter is only 20% around and it spends lot of electricity to maintain the reactor temperature around 1,000° C. Other means of converting STC to TCS is hydrogenation of STC in the presence of MGSI. This method operates around 500° C. But, the operation pressure is about 30 bar and the production rate and conversion is unstable below 35% and the STY is very low because of the low gas inlet velocity to maintain conversion rate. It is purpose of the current application to provide a reliable means to convert STC to TCS at a lower pressure, high conversion rate above 50% and high STY at the same time.
U.S. Pat. No. 2,595,620 to Wagner, et al. illustrates hydrogenation of halogenosilanes, especially hydrogenation of silicontetrachloride (STC) in the presence of MGSI (metallurgical grade silicon) based on the assumption as the following equations.
3SiCl4+Si+2H2→4SiHCl3
Reaction at a various temperatures between 400° C. and 500° C. were investigated. For atmospheric pressure, the concentration of TCS in the −78° C. condensed product from the reactor shows less than 25%. The TCS concentration increased with the retention time of STC and hydrogen in the reactor. This means the conversion depends on the ratio of bed height/SGV (specific gas velocity). In other words, the conversion of STC to TCS is higher when the bed height is higher and the feed rate of the gaseous reactants, hydrogen and STC, is lower. So, the over all production rates of TCS per volume is very low. Maximum concentration of TCS, 36%, was observed when the reaction pressure is 1,000 Pisa and retention time was 2.9 minutes. However, to maintain such conversion constantly, the height of the bed and the SGV of the gaseous reactants must be controlled. To control the bed height, the fluidized bed must be operated smoothly without channeling or slugging. In addition to this, the STY of this reaction system is very low to maintain high conversion rate by keeping long retention time. U.S. Pat. No. 4,676,967 to Breneman, et al. illustrates a process for the production of ultra high purity silane and silicon. Metallurgical silicon is initially reacted with hydrogen and silicon tetrachloride in a reaction zone maintained at a temperature of from about 400° C. to about 600° C. and at a pressure in the range of from about 300 to about 600 psi, to form trichloro-silane as follows;
3SiCl4+2H2+Si→4HSiCl3 (1)
with the reaction (1), mixture containing a yield of about 20˜30% by weight trichlorosilane on a hydrogen-free basis, and of about 0.5% dichlorosilane with the remainder being silicon tetrachloride together with impurities comprising mainly carryover metallurgical silicon powder, hydrogen chloride, metal halides essentially without undesired polysilanes were obtained. None of the method is disclosed therein to control the bed height and retention time.
PCT international publication No. WO2007/035106 to Andersen, et al. illustrates a method for the production of TCS (Trichlorosilane) by reaction of silicon, STC (Silicontetrachloride) and hydrogen gas at a temperature between 400° C. to 800° C. and at a pressure of 0.1 to 30 bar in a fluidized bed reactor, in a stirred bed reactor or in a solid bed reactor with MGSI containing manganese less than 50 ppmv. The MGSI with low manganese produced TCS over the conversion rate of 50%, which is higher than the known equilibrium conversion of 45% maximum. Generally to reach the equilibrium, the contact time must be longer. However, in this case they got conversion over 50% at a contact time is less than few seconds. It is totally contrary of the entire previous STC converter. Among the three types of reactors referred in the WO2007/035106, FBR is the best reactor type to realize uniform reaction condition if properly designed.
Even though the inventors claim that they used a FBR, it is a small laboratory scale reactor and their SGV is about 1 cm/sec. At this SGV, the MGSI bed does not move. The MGSI bed just expands. It is named as an ‘expanded bed’ (Fluidization Engineering, John Wiley & Sons, Inc., pp 1˜3, Daizo Kuni and Octave Levenspiel).
In other words, they tested the gas-solid reaction in some reactor, not in a real FBR (fluidized bed reactor). However, it is normal for testing a new reaction at early stage of technology development because running an unknown new reaction in a FBR directly is dangerous due to the scale of reactants, toxic gases, used.
Anyway, they found a reaction that behaves quite differently from all of the previous result that the reaction of hydrogenation of STC in the presence of MGSI is a slow equilibrium reaction and has a limitation of maximum conversion. But, their result is that even at the slow SGV, the contact time of the STC and hydrogen with the special MGSI is less than a second and the conversion is over 50%. It means the reaction is very fast and the mechanism is different. However, they did not disclose how to scale up and apply the invention to commercial scale of tens of thousand of TCS production by hydrogenation of STC.
The applicants of the current invention already developed a FBR reactor for very fast exothermic reaction of direct hydro chlorination of MGSI in a commercial scale, above 20,000 MT/YR TCS. The commercial scale FBR shows much stable operation than any other STC hydrogenation reactor and produce crude TCS with 95% purity among the liquefied products directly from the FBR.
It is the purpose of the current application to provide an industrially practical FBR to convert STC to TCS by reacting STC, hydrogen and MGSI at a lower pressure and high out put rate, STY.
It is another purpose of the current application to provide a method of producing TCS stably by hydrogenation of STC with the FBR disclosed.
Many preliminary works have been done to find the optimized structure of the FBR and the method of operation of the FBR.
Fluidized Bed Reactor (FBR) is selected for maximum mixing of ‘fluidizing bed materials.’ That enables uniform reaction condition inside of the ‘fluidizing bed.’
Among the various stage of the FBR, ‘slugging bed’ is known as ‘must be avoided’ because of their unstable bed behavior and many ‘entrainment of bed material’ to the exit gas stream (Fluidization Engineering, John Wiley & Sons, Inc., pp 1˜3, Daizo Kuni and Octave Levenspiel). When the phenomenon ‘slugging’ happens, the upper part of the gas-solid bed is pushed up-ward and is separated from the main bed. Therefore, when the ‘bed’ is operated as ‘slugging mode’, the heat transfer within the bed and between the bed and reactor wall surface decreases because the heat transfer coefficient of gas is normally lower than that of the solid material. This phenomenon is typical in a gas-solid FBR.
It is naturally concluded that maintaining the ‘bed’ of the reactant in a ‘bubbling bed’ mode is the first thing to be resolved because none of the prior arts disclosed what is the parameter that categories the boundary of ‘bubbling bed.’
The applicant started from this point with a transparent cold bed of a FBR as shown in the
The applicant found from his long experience of FBR operation that relative value of the ‘height of the fluidized bed of the solid particles’ and ‘internal diameter of the fluidizing vessel’ is the key parameter that categorize the boundary of ‘bubbling bed’ and ‘slugging bed.’ However, the ‘height of the fluidized bed of the solid particle’ varies depends on the SGV. So, ‘initial bed height of the charged solid particles’ is selected as one parameter.
As shown in the table 1, the ‘slugging’ does not occur within the SGV range lower than 30 cm/sec until the ‘initial bed height of the charged solid particles’ (H)/‘inner diameter of the fluidizing vessel’ (d1) reaches over 2. At the level of H/d1=2, the ‘height of the fluidized bed of the solid particles’ reaches five times of the vessel's inner diameter, d1, when SGV is 30 cm/sec. When H/d1 is higher than 3, slugging starts even at SGV of 10 cm/sec. At this moment, the ‘height of the fluidized bed of the solid particles reached around six times of the inner diameter of the fluidizing vessel. Upper section of the ‘fluidized bed of the solid particles’ is separated from the rest of the bed and is raised higher followed by collapse of the separated portion. As the H/d1 is higher than 4, ‘slugging’ accompanied with ‘entrainment’. So, the solid particles come out of the FBR.
The meaning of the above finding is that if some FBR operates with initial bed height charged higher than 4 times of the inner diameter, the possibility of ‘slugging’ the bed is very high even at the lower SVG of 10 cm/sec. Then mixing in the FBR reactor is not good and the reaction inside of the reactor is non-uniform. In other words, the reactor is not under control.
The other founding is that, when the slope of the expending section (9) is low, particles that leave the top surface (10) of the fluidized bed (11) accumulate on the inner surface of the expanding section (9). By trial and error, it was found that the angle (11) of the slope of the expanding section (9) from a vertical line should be smaller than 7 degrees.
Based on the above findings, the FBR (fluidized bed reactor) (20) for STC conversion to TCS, according to current application, is designed as shown in the
The key features of the FBR (20) according to current application are as follows;
For producing TCS rich silane mixture by the hydrogenation of STC in the presence of a special MGSI, which has manganese concentration less than 35 ppmw, the FBR (20) is operated as follows;
The FBR (20) is purged with vaporized liquid nitrogen properly before start up. The reactor is filled with proper initial charging materials (42), including but not limited with, non-porous silica or porous silica, such as Grace Davison 952, quartz powder, amorphous quartz powder, sand, zirconia or equivalent. Those materials should have elemental Si contents at least 99.8 wt %. Particle size, true density, and bulk density of the seed bed material is almost equivalent to that of the metallurgical silicon as shown in the Table 2.
Amount of initial charging material (42) introduced at the start up should be enough to fill the height (H) of the lower reactor section (21) with the dimension that is equivalent to one to three times of the internal diameter (D1) of the lower reactor section.
The FBR (20) system is purged and fluidized with vaporized liquid nitrogen introduced to the initial charging material (42) bed from the bottom through the gas distribution plate (24) at 100° C. in a ‘bubbling bed’ mode until the effluent gas contains moisture less than 0.1 ppm.
Then, the initial charging material (42) bed temperature is increased up to 500° C. Then nitrogen is switched to hydrogen mixed with vaporized STC while maintaining the total SGV of the gas mixture over 30 cm/sec. At the same time special metallurgical grade silicon particles, which has manganese contents lower than 10 ppmw, (43) are introduced to the initial charging material (42) bed through the silicon feeding line (33), which reaches a point (34) just below the upper end (35) of the lower reactor section (21) with an angle (36) from a vertical line, which is extended from the wall of the lower reactor section, smaller than 20 degrees.
The MGSI feeding line (33) is connected to an outer carrier gas feeding line (44). Part of hydrogen gas and/or STC needed is introduced through the carrier gas feeding line (44) and disperses and carries the silicon particles (43) into the bed. Major portion of hydrogen and STC is heated up to 400° C. and introduced to the FBR from the bottom of the FBR through the gas distribution plate (24).
As disclosed in many prior arts, they start up the hydrogenation to convert STC to TCS production with un-necessarily excess amount of silicon. Their usual SGV is around 10 cm/sec to meet the long retention time of over 50 seconds to 60 seconds for maximum conversion. However, at this velocity, which is about 2˜3 times of minimum fluidizing velocity, the bed of silicon just start to swell and does not mix the bed. Then the concentration of gas components, reactants and the produced HCl, does not disperse uniformly across the bed. Especially when the reactor is big, it causes many un-desired reaction. To avoid such unnecessary side reaction, the applicant developed couple of methods as follows.
First method is to use inert initial charging material to disperse and mix the reactants, special MGSI, which has manganese concentration less than 35 ppmw, hydrogen and STC, well with high SGV to operate the bed as ‘bubbling bed mode.’ The initial charging material is chemically inert at the reaction environment and the physical properties are same as those of the silicon granule used as reactant. Pure silica (SiO2) granules, quartz, have almost the same physical properties and showed no chemical reaction at the reaction conditions of 550° C. and at 5 atm pressure. The amount of the initial charging material introduced at the start up is the amount that can fill the height (H) of the lower reactor section with a dimension that is one to six times of the internal diameter (D1) of the lower reactor section. Second method is to use a feeder that feeds the special MGSI granules continuously with accuracy of ±5% at 105 Pisa. It is well known in this industry that on-off block valve or some ball valves are used in commercial STC to TCS synthesis process. On-off valves provide pulse feeding and ball valves are easily worn out by the silicon granules. Therefore, both valves provide unstable feeding. Then, pulse or unstable feeding of the silicon granules result in ramping of the reactor temperature, due to sudden introduction of cold MGSI, and loose of temperature control. It is clear that the reaction condition becomes unstable and the products composition distribution also unstable according to the unstable temperature control. The third method is the gas distribution plate design.
In case of exothermal reaction, heat generated by the reaction can not be effectively removed by the gas and as a result a ‘hot spot’ is formed. At this ‘hot spot’ the reaction produces non-desirable result, such as high molecular weight siloxane and viscous particles aggregated together.
In case of endothermic reaction, like STC hydrogenation, the heat supplied from outside of the reactor will not be transferred well and the temperature at this stationary zone (60) will be different from the main bed and the product distribution will be different from main bed. Then the efficiency of the bed is decreased.
By combining the above methods, the applicant can introduce the reactants, gas and solid, stably into the FBR (20).
The initial charging material (42) is inert to the reaction. In addition to that, it helps dispersion of the reactants and products uniformly throughout the fluidizing bed. At the same time the charging material transfers heat between the reactants and to the wall. So, temperature of the ‘fluidizing bed’ becomes uniform too.
With combination of the above methods, the fluidized bed will convert STC to TCS more stably and continuously by maintaining temperature of the reaction bed controlled between 300° C. to 600° C., more specifically at 550° C., within the mean temperature deviation, between thermo couples in the bed, of ±5 C. Reaction pressure is maintained between 3 to 10 bars, more specifically at 5 bars. Contact time between the STC, hydrogen and the special MGSI in the reaction bed is controlled to be shorter than 50 seconds, more specifically less than 30 seconds due to the high SGV and the dilution effect of the initially charging material.
This is a divisional application of U.S. patent application Ser. No. 12/802,320, which is Continuation In Part of the U.S. patent application Ser. No. 12/456,979, which is now abandoned and which was non-provisional application of the Provisional Application No. 61/133,688 which was filed on Jul. 1, 2008.
| Number | Date | Country | |
|---|---|---|---|
| 61133688 | Jul 2008 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 12802320 | Jun 2010 | US |
| Child | 12931581 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 12456979 | Jun 2009 | US |
| Child | 12802320 | US |
