CN102565659A - Solar grade ingoting polycrystalline silicon chip characterization method - Google Patents

Abstract

The invention discloses a solar grade ingoting polycrystalline silicon chip characterization method, which comprises the following steps: obtaining a polycrystalline silicon chip image through a fluorescence photoluminescence imager; drawing a luminance frequency histogram for the polycrystalline silicon chip image; taking the peak value of the luminance frequency histogram as a threshold value, performing object recognition for the polycrystalline silicon chip image, and converting the polycrystalline silicon chip image into a two-value matrix; summarizing the obtained two-value matrix, so as to obtain the pixel points of a clear zone; and realizing the efficiency characterization of the polycrystalline silicon chip of the solar cell through the clean crystal grain zone ratio epsilon as the clean crystal grain zone ratio epsilon is the ratio of the pixel points of a clear zone to the total pixel points. The quantification treatment method based on the threshold valve interception image recognition method and provided by the invention can be used for predicating the efficiency of the cell through the quantitative defect area, and the experiment finds out that the clean crystal grain zone ratio epsilon and the corresponding cell efficiency are in positive correlation, and fast screening of silicon chips can be carried out through the positive correlation, thereby finally improving the quality of the end product.

Description

Solar level ingot casting polysilicon chip characterizing method

Technical field

The present invention relates to solar level ingot casting polysilicon chip characterizing method; Can be used for the shelves of presorting before the ingot casting polysilicon chip cell preparation; The battery sheet that defect concentration is high filters out in advance, and this method has the relation that meets preferably to the quantitative resolution of silicon chip and the efficient of final products battery sheet.

Background technology

The ingot casting polycrystalline solar cell has occupied silicon solar cell 60% above share at present.The fluorescence photoluminescence mainly contain fluorescence photoluminescence spectrometer and fluorescence photoluminescence imager dual mode at present, but there is certain limitation in two kinds of methods in the use because of its characteristics as the conventional means that semiconductor detects:

1. the fluorescence photoluminescence spectroscopy is primarily aimed at the semiconductor microcell and carries out defect detection, and what obtain is the composition information of material 0 dimension.Can't estimate fast whole polysilicon chip.

2. what fluorescence photoluminescence imager obtained is the gray scale contrast of whole silicon chip; As shown in Figure 1, dark part is represented high defect area, and brightness major part is represented clean crystal grain; This method has the potentiality of comprehensive evaluation silicon chip quality; But at present the resulting data of the method are lacked reliable quantification means, particularly and the relevance between the final products battery sheet can't obtain, restricted the application of quick sorting silicon chip.

Summary of the invention

Technical matters to be solved by this invention is: a kind of solar level ingot casting polysilicon chip characterizing method is provided, the silicon chip quality is carried out quantitative analysis, be used for the shelves of presorting before the ingot casting polysilicon chip cell preparation, the battery sheet that defect concentration is high filters out in advance.

The technical solution adopted for the present invention to solve the technical problems is: a kind of solar level ingot casting polysilicon chip characterizing method has following steps:

1) the polysilicon chip image that uses fluorescence photoluminescence imager to obtain;

2) make the brightness frequency histogram of polysilicon chip image;

3) with the peak value of brightness frequency histogram as threshold value, the polysilicon chip image is carried out object identification, be converted into two values matrix;

4) two values matrix that obtains is sued for peace, obtain the clear zone pixel number, its ratio with total pixel number is clean die region area ratio ε, characterizes the efficient of final polysilicon chip solar cell with the clean die region area ratio ε of silicon chip.

The invention has the beneficial effects as follows: the quantizing method that the present invention is based on threshold value intercepting image-recognizing method; Predict the efficient of battery sheet through the quantify defects area; Silicon wafer to manufacture after the test is become the battery sheet, obtain its efficiency data, can find that through Fig. 4 there is positive correlation in the clean die region area ratio ε of silicon chip with corresponding battery efficiency; Utilize this positive correlation to carry out rapid screening, improve the quality of final products silicon chip.

Description of drawings

Below in conjunction with accompanying drawing and embodiment the present invention is further specified.

Fig. 1 is a fluorescence photoluminescence imager image of the present invention;

Fig. 2 is the frequency histogram of silicon chip luminance matrix of the present invention;

Fig. 3 is a binaryzation luminance matrix picture among the present invention;

Fig. 4 be among the present invention the clean chip area of silicon chip fluorescence photoluminescence image than ε value and efficient corresponding relation.

Embodiment

A kind of solar level ingot casting polysilicon chip characterizing method; The polysilicon chip image that uses commercial fluorescence photoluminescence imager to obtain; And with the peak value of brightness frequency histogram as threshold value; It is carried out object identification, be converted into two values matrix and calculate its clean die region area ratio, this numerical value has good corresponding relationship with the conversion efficiency of final battery.

Key step is following:

1) use commercial fluorescence photoluminescence imager to obtain a luminosity matrix of polysilicon chip, matrix size is 1024 * 1024, or other size, also is image as shown in Figure 1.

2) with the GTG of some,, make the interval frequency histogram of different luminosity as 128, as shown in Figure 2, and obtain frequency histogram peak value p, the i.e. abscissa value on histogram summit.Think that brightness is greater than or equal to the clean grained region of regional corresponding silicon chip of p value, and brightness value is less than the zone of the high density of defects in the regional corresponding silicon chip of p value.

3) be threshold value with frequency histogram peak value p, the luminosity matrix is converted into two values matrix (by following formula manipulation), wherein the pixel more than or equal to p is 1, is 0 less than the pixel of p.

$A(i,j)=\left\{\begin{array}{c}0,\mathrm{AO},D<y\\ 1,\mathrm{AO},D\&GreaterEqual;y\end{array}\right\}$

With 2 * 2 following matrixes $\left(\begin{array}{cc}1000& 1000\\ 1300& 1280\end{array}\right)$ Be example: suppose that obtaining threshold value p as histogram is 1200, brightness is made as 1 more than or equal to 1200 pixel point value, the pixel less than 1200 is made as 0, is promptly only comprised 0 and 1 two values matrix.

$\left(\begin{array}{cc}1000& 1000\\ 1300& 1200\end{array}\right)\stackrel{1000}{\&RightArrow;}\left(\begin{array}{cc}0& 0\\ 1& 1\end{array}\right)$

If silicon chip data shown in Figure 1 are carried out above-mentioned processing and picture, then obtain Fig. 3 as a result, the former relatively figure of the form contrast of its bright area and dark areas meets relatively goodly, and this binarization method is traditional monodrome threshold method.

4) two values matrix that obtains is sued for peace, obtain the clear zone pixel number.Its ratio with total pixel number is clean die region area ratio ε.

Embodiment 1:

1. it is some to get a collection of clean ingot casting polysilicon chip, gathers its fluorescence photoluminescence view data.

2. adopt flow processing view data of the present invention, obtain the clean area ratio ε of every silicon chip.

3. be intercepting line with ε=0.8, choose ε greater than 0.8 silicon chip, the corresponding battery efficiency of this part silicon chip is in relatively low level.

Claims (1)

1. solar level ingot casting polysilicon chip characterizing method is characterized in that: have following steps:

1) the polysilicon chip image that uses fluorescence photoluminescence imager to obtain;

2) make the brightness frequency histogram of polysilicon chip image;

3) with the peak value of brightness frequency histogram as threshold value, the polysilicon chip image is carried out object identification, be converted into two values matrix;

4) two values matrix that obtains is sued for peace, obtain the clear zone pixel number, its ratio with total pixel number is clean die region area ratio ε, characterizes the efficient of final polysilicon chip solar cell with the clean die region area ratio ε of silicon chip.

Images