Patent Application
20250065373
The invention relates to a method for sorting silicon chunks.
Polycrystalline silicon (polysilicon) is conventionally produced by the Siemens method (chemical vapour deposition process). In this case, in a reactor, filament rods (thin rods) of silicon are heated by a direct flow of current through them and a reaction gas containing a silicon-containing component (for example monosilane or halosilane) and hydrogen is introduced. The surface temperature of the filament rods is conventionally more than 1000° C. At these temperatures, the silicon-containing component of the reaction gas is decomposed and elemental silicon is deposited from the gas phase as polysilicon on the surface of the rod, so as to increase the diameter of the rod. After a predetermined diameter has been reached, the deposition is stopped and the polysilicon rods obtained are removed.
Polysilicon is the starting material for the production of monocrystalline silicon, which is produced for example by means of the Czochralski method. It is furthermore required for the production of multicrystalline silicon, for example by means of ingot casting. For both methods, it is necessary to comminute the polysilicon rods into chunks. The chunks are conventionally classified in separating devices according to their sizes.
The separating devices may be multistage screening machines which classify the fragmented polysilicon mechanically in different size classes. For example, U.S. Pat. No. 6,375,011 B1 discloses a vibratory conveyor which allows classification in three size classes.
An improvement of the separation, or even separation according to optical criteria, may be achieved by optopneumatic sorting facilities. US 2007/0235574 A1 discloses such a facility, which is arranged downstream of a comminuting device for polysilicon. The shadow area, projected into a plane, of the comminuted polysilicon chunks is in this case used for the size separation.
Furthermore, U.S. Pat. No. 6,265,683 B1 describes an optopneumatic device for classifying semiconductor materials, the size separation being performed with camera assistance by means of recording a projected area of materials to be sorted. Optionally, sorting of the materials according to their surface condition may also be carried out here.
In the known methods which are based on a two-dimensional (2D) projection of chunks using the transmitted-light method, a disadvantage is that the chunks are recorded only from one side. Chunks may differ greatly in respect of their shape. It may for instance happen that a chunk is oriented unfavourably at the instant when its projected area is recorded, in such a way that its longest extent lies in front of or behind the projected area (i.e. perpendicularly with respect to the projected area). The consequence is an incorrect estimation of the chunk size (sorting error).
This problem has given rise to the object of the invention, namely to provide an improved sorting method in which the orientation of the chunks to be sorted plays only a secondary role.
This object is achieved by a method for sorting chunks, in particular silicon chunks, comprising the following steps.
By the further measuring device, it is possible to supplement the projected area of the chunk, acquired with the aid of the first measuring device, with height information. The measuring devices are in this case preferably arranged at different positions around the chunk and thus inspect the latter from different viewing angles. Information items relating to the extent of the chunk above and/or below the profile plane (in the direction of the applicate axis (z axis)) may thereby be obtained so as to prevent a sorting error. In this way, the quality of the separation is improved.
The chunks are preferably comminuted polysilicon, for example comminuted polysilicon rods from the Siemens method.
The singulation of the chunks is intended, in particular, to mean that the chunks are distanced from one another and are thus no longer arranged above one another or partially overlapping. This may, for example, be carried out by a shaking movement a conveyor belt. It is not absolutely necessary for the chunks to be arranged in a row (behind or next to one another) for the recording by the measuring device.
Preferentially, the first measuring device and at least one of the further measuring devices are photoelectric transmitted-light or reflected-light measuring systems having a detection region through which the chunk passes.
Preferably the silicon chunk passes through the detection region in freefall.
The reflected-light measuring system preferentially comprises a light section sensor and/or at least one camera system.
The light section method carried out with the light section sensor is based on optical triangulation and requires a relative movement of the sensor and the chunk. The chunk is in this case illuminated linearly with the aid of a suitable light source and the resulting light stripe is recorded by means of an area scan camera (a constituent part of the sensor). The surface normals of the light source and of the camera are in this case tilted relative to one another by a triangulation angle.
The camera system may in principle also be only a camera which records the projected area of the chunk with the aid of ambient light as a light source. Preferably, however, the camera is supplemented with an external light source.
The camera system may furthermore be a camera system for photometric stereo analysis. This is a method for the analysis of projected areas and reflections of a surface of an object (chunk) in three-dimensional (3D) space. Conventionally, an external light source is directed onto the object and is moved in order to acquire a plurality of images of the resulting light scenarios. For moving objects such as a chunk in freefall, instead of the moving light source use is made of a plurality of cameras having different viewing angles. Alternatively, it is possible to use a plurality of light sources respectively having a different illumination direction and one camera, or a plurality of light sources and a plurality of cameras in combination.
The transmitted-light measuring system preferably comprises a photoelectric barrier, a light curtain or a light grid. In this case, light beams (for example infrared beams) are emitted from an optical emitter (photoelectric barrier) or a plurality of emitters spaced apart from one another (light grid/light curtain) to the corresponding receivers. If one or more of the beams are interrupted, a signal may be sent to a device and, for example, a deflecting device may consequently be triggered. The resolution of light grids and light curtains may be determined by the distance between the beams. Typical examples of such measuring systems are optical micrometers, light band micrometers, profile projectors, CCD laser micrometers and laser sensors with a photoelectric barrier function.
According to one preferred embodiment, the first measuring device is a camera system, in particular comprising a light source and a camera for recording the projected area.
The further measuring device is preferentially a light curtain or a light grid, particularly when a camera system has been selected as the first measuring device.
Preferably, the recording of the projected area and the recording of the height information item take place with a time spacing of from 0 to 100 ms, preferentially from 0 to 50 ms, particularly preferentially simultaneously. A time spacing that is as short as possible ensures that the position of the chunk does not change substantially, particularly in freefall.
The deflecting device, which is responsible for the sorting of the chunks, may be a pneumatic and/or mechanical deflecting device.
The pneumatic deflecting device preferably comprises at least one nozzle from which gases (for example air, inert gas) or liquids (for example high-purity water) are ejected at pressures of from 3 to 20 bar. In this regard, reference may be made to U.S. Pat. No. 6,265,683 B1.
The method according to the invention is preferably an optopneumatic classification method.
The calculation of the size of the chunk may be performed using an evaluation device (for example a software-aided process control station, for example MATLAB (from the company Math Works)) which is connected both to the measuring devices and to the deflecting device.
The basic measurement information which is obtained with the first measuring device is the projected area of the chunk, in which an information item relating to the shape of the chunk is contained, even if only in 2D as a contour. From the projected area, the evaluation device can obtain various length specifications, in diameters, which allow conclusions about the size of the chunk. In general, numerous methods are known for determining the length specifications or diameters. For example, determination of the diameter derived from an equivalent circle (perimeter equivalent diameter) or determination of a Feret diameter, which involves a whole group of characteristic quantities, all of which are defined by the distance between two tangents to the contour of the projected area in a fixed measurement direction.
The height information item obtained by means of the further measuring device, for example a photoelectric barrier, may in particular be a height value (unit of length). The combination of the acquired measurement information items is then, for example, a 3D point cloud consisting of the contour acquired with the first measuring device and a contour displaced by the height value.
A calculation of the greatest extent may be carried out from the length of a vector between the points lying furthest away from one another:
Point cloud as vector X={[x1,y1,z1], [x2,y2,z2], . . . , [xn,yn, zn]}
Combination of all points with one another and calculation of the distance in 3D:
Determination of the maximum value of L(i,j).
The more detailed the height information items of the further measuring device is, the smaller the measurement error is.
A further aspect of the invention relates to a device for sorting chunks, in particular silicon chunks, comprising
Preferably, the device is a device for carrying out the described method.
The singulating region preferably comprises at least one vibrating conveyor trough and/or conveyor belt. Optionally, the singulating region may also contain a screening plate or a shaking screen for removing fine fragments. In general, it may also be the segment of a conveyor belt, onto which the chunks are applied individually. Suction for dust particles may likewise be provided. The singulating region is preferentially a singulating region as described in EP 0 983 804 A1.
Preferentially, the device comprises a multiplicity of first and/or further measuring devices in order to allow high-throughput sorting. By a multiplicity of measuring devices in combination with a high computing power of the evaluation device, the distance between singulated chunks may be reduced to a minimum.
Preferably, the deflecting device is a pneumatic or mechanical deflecting device, in particular a pneumatic deflecting device which comprises a row or a matrix of individual nozzles.
In respect of the measuring devices and the evaluation device, reference may be made to the comments above and to EP 0 983 804 A1.
As a first measuring device, a camera 40 having an external light source 42 is arranged underneath the sliding surface 20. The camera 40 is, for example, a CCD camera having an optical resolution of from 0.05 to 2.0 mm. The light source 42 is, for example, an LED having diffuse area illumination. A detection region 44 of the first measuring device, in relation to the chunk 30, is indicated as a star. As a further measuring device, a light grid 50 is fitted at a lower end 22 of the sliding surface 20. It consists of an emitter strip 52 having a total of five infrared light sources (laser or LED light sources or light dots in the visible range may also be envisaged), the radiation of which is respectively indicated by a dashed line 54, and a receiver strip 56 which correspondingly has five sensors. Underneath the first measuring device there is a pneumatic deflecting device 70, and underneath the latter there are a first and a second collection container 80, 82. The collection containers 80, 82 are connected to one another by a separating element 81 which is triangular in cross section. Furthermore, both the sensor strip 56 of the first measuring device, the camera 40 and the light source 42 of the second measuring device, as well as the deflecting device 70 are connected to an evaluation device 90. The evaluation device is a computer with image processing software, for example MATLAB.
When a chunk 30, for example a pyramidal chunk, singulated by the shaking movement of the vibrating conveyor belt 10, now reaches the oblique sliding surface 20, it becomes oriented in such a way that its centre of gravity lies as low as possible. This may generally be adapted to the chunk size of the chunks by a sliding surface 20 that is adjustable in its angle. After the end of the sliding surface 22, the chunk 30 passes through the light grid 50 in such a way that its elongated side faces in the z direction, and it is thus recorded in its full length by the light grid 50. The chunk 30 subsequently passes in freefall through the detection region 44 of the camera 40, the latter only recording a projected area 32 corresponding to the base face of the chunk 30. From the two information items, i.e. the projected area 32 and the height information item acquired by the light grid 50, the evaluation device 90 calculates the size of the chunk 30 and forwards this information to the pneumatic deflecting device 70, undeflected chunks 30 being collected in the second collection container 82 and the chunks 30 deflected by a pneumatic pulse being collected in the first collection container 82. The separating element 81 facilitates this separation.
Classification of comminuted (crushed) polysilicon having a chunk size (CS) 2.
The size class of polysilicon chunks is defined as the longest distance between two points on the surface of a silicon chunk (corresponding to the maximum length):
The polysilicon sample material used for the test was produced from a mixture of 9000 chunks in the length range of from 10 to 40 mm (CS2) and 1000 chunks in the length range >40 to 65 mm, namely the fraction to be separated. In order to prepare the sample material, a mechanical screening method (analysis screen according to DIN ISO 3310-2 with a normal hole width W=4 mm (square perforation)) was used to remove the chunk fractions of from 0 to 10 mm. The maximum length both of the chunks in the length range of from 10 to 40 mm and of the chunks of the fraction to be separated in the length range of from 40 to 65 mm was determined manually (vernier calliper) and the polysilicon sample material was then mixed.
This polysilicon sample material was subjected to conventional optopneumatic sorting of the fraction >40 mm.
The optopneumatic sorting device used was equipped with a first 2D measuring device (CCD camera and light source according to
At the end of the sorting, the slippage of the separating method was determined by manual analysis (vernier calliper) in conjunction with manual counting of the chunks. It was 0.1%.
The polysilicon sample material described in the comparative example was separated with an optopneumatic sorting device having substantially the same design (with a separating interval of 40 mm). In contrast to Comparative Example 1, however, the device had a second 2D measuring device. This was a light grid as described with reference to
One possible way of evaluating the information items acquired by the measuring devices is presented below, this having been carried out with MATLAB (from the company Math Works).
The slippage after this sorting was manually established as described above, and was 0.0%.
Classification of comminuted polysilicon in CS3.
First, polysilicon sample material consisting of 9000 chunks in a length range of from 20 to 60 mm (CS3) and 1000 chunks in a length range >60 to 85 mm (the fraction to be separated) was produced.
The verification of the chunk lengths was carried out manually. Following this, the polysilicon sample material was made up by mixing.
The polysilicon sample material was sorted with the optopneumatic sorting device described in Comparative Example 1, but with a separating interval of 60 mm.
The slippage after this sorting was manually established as described above, and was 1.0%.
The polysilicon sample material described in Comparative Example 2 was in this case sorted with an optopneumatic sorting device according to
The slippage after this sorting was manually established as described above, and was 0.0%.
It was possible to show that the sorting outcome can be significantly improved, particularly in the case of pyramidal objects, by an additional height information item determined by means of a second measuring device. The aim of any sorting is essentially a clean separating interval without slippage in respect of a chunk size.
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2022/079552 | 10/24/2022 | WO |
