METHOD FOR PRODUCING POLYCRYSTALLINE SILICON

Abstract

A method for producing polycrystalline silicon includes introducing a reaction gas, which in addition to hydrogen contains silane and/or at least one halosilane, into a reaction space of a gas phase deposition reactor. The reaction space includes at least one heated filament rod upon which by deposition silicon is deposited to form a polycrystalline silicon rod. During the deposition, the the morphology of the silicon rod is determined.

Claims

1-9. (canceled)

10. A method for producing polycrystalline silicon comprising introducing a reaction gas, which in addition to hydrogen contains silane and/or at least one halosilane, into a reaction space of a gas phase deposition reactor, wherein the reaction space comprises at least one heated filament rod upon which by deposition silicon is deposited to form a polycrystalline silicon rod, wherein during the deposition, to determine the morphology of the silicon rod, at least one thermographic image of the surface of said rod comprising a measurement area A.sub.max is generated, by image processing a segmentation of the measurement area A.sub.max into a first and a second area fraction is performed, wherein the first area fraction A.sub.t corresponds to a relatively high temperature T.sub.t compared to local average temperature values and the second area fraction A.sub.p corresponds to a relatively low temperature T.sub.p compared to local average temperature values, and a morphology index M is determined according to $\begin{array}{cc}M=\left(T_t-T_p\right)*\frac{\left(A_t+A_p\right)}{A_{\max }}*\frac{A_t}{\left(A_t+A_p\right)},& \left(\mathrm{formula}I\right)\end{array}$ wherein through variation of at least one parameter selected from the group comprising U, I, surface temperature T.sub.OF, reaction gas composition and volume flow the deposition is controlled such that M has a value of 0 to 4, wherein U is in a range from 50 to 500 V, I is in a range from 500 to 4500 A, T.sub.OF is in a range from 950° C. to 1200° C., the volume flow is in a range from 1500 to 9000 m.sup.3/h and the reaction gas before entry into the reactor contains hydrogen in a proportion of 50% to 90%.

11. The method as claimed in claim 10, wherein the index M has a value from 0.1 to 3, preferably from 0.1 to 2.

12. The method as claimed in claim 10, wherein the index M is kept constant during the deposition.

13. The method as claimed in claim 10, wherein the determination of the index M is carried out continuously during the entire deposition or discontinuously at various points in time during the deposition.

14. The method as claimed in claim 10, wherein the determination of the index M is carried out discretely in a time interval preferably corresponding to a specified growth in the diameter of the silicon rod.

15. The method as claimed in claim 10, wherein at least two thermographic images of the same silicon rod or of different silicon rods are generated to determine M.

16. The method as claimed in claim 10, wherein the segmentation is carried out with a rank filter.

17. The method as claimed in claim 10, wherein the measurement area A.sub.max has a size of 10 to 300 cm.sup.2.

18. The method as claimed in claim 16, wherein the rank filter is a median filter.

19. The method as claimed in claim 17, wherein the measurement area A.sub.max has a size of 30 to 200 cm.sup.2.

20. The method as claimed in claim 19, wherein the measurement area A.sub.max has a size of 50 to 150 cm.sup.2.

Description

[0048] FIG. 1 shows the segmentation of a thermographic image.

[0049] FIG. 2 shows the profile of the morphology index M as a function of diameter for two types of polysilicon.

EXAMPLE 1

[0050] FIG. 1 shows an exemplary thermographic image A. It was recorded with an infrared camera through an inspection window from a silicon rod in a Siemens reactor at a height about halfway between the bridge and the electrode. The silicon rod was in close proximity to the inspection window. Recording was carried out after a deposition time of about 90 h. The Siemens reactor was fitted with 24 rod pairs, wherein the filament rods had a length of 2.5 m (length between bridge and electrode). Type C polysilicon was to be deposited. Accordingly, M was to have a value of 1 to 3. The measurement area A.sub.max corresponds to the area inside the dashed line.

[0051] The images B and C show the result of segmentation of a thermographic image. The software LabVIEW (Fa. National Instruments) and a median filter (30*30 pixel) were used to carry out the segmentation into the area fraction A.sub.p (image B, elevations in white within the dashed line, temperature T.sub.p=1027° C.) and the area fraction A.sub.t (image C, trenches in white within the dashed line, temperature T.sub.t=1033° C., A.sub.t=20 cm.sup.2). The measurement area A.sub.max was 57 cm.sup.2. According to formula I at this point in time of the deposition M was 2.1 which is within the target value range for polysilicon type C.

EXAMPLE 2

[0052] FIG. 2 plots the profile of M against the silicon rod diameter d [mm] for two different deposition processes, i.e. two different polysilicon qualities. The upper curve relates to the production of type D. The lower curve relates to the production of type C. Type C is more compact than type D and is used for more sensitive applications. Type C should have a value for M of 1 to 3 while type D should have a value of from 3 to 5. Both processes were performed in the same Siemens reactor but with different settings for at least one parameter from the group comprising U, I, T.sub.OF, reaction gas composition and volume flow rate. Determination of M was carried out continuously during the entire deposition time. The rod diameter was determined on two rods with a digital camera and image processing.

[0053] Both processes begin with compactly deposited polysilicon having values for M close to 0 which is especially due to the filament rods made of very compact silicon. For production of type D a relatively steep profile was selected for M already shortly after commencement of the deposition. The target level of M of about 3.5 was already to be achieved at a rod diameter of about 90 mm. The steep profile toward a rather porous polysilicon was achieved especially by altering the surface temperature, gas composition and/or volume flow. M was subsequently adjusted to a value between 3.5 and 3.9 (average of about 3.7).

[0054] For production of type C too the target value of about 1.5 was to be achieved at about 90 mm. Control of the above-described parameters was adapted accordingly. For the remaining deposition time M was kept constant at an average of 1.6.

[0055] It it is apparent from the example how conveniently the deposition may be controlled for production of a very wide variety of polysilicon types using the index M.