CN103364704B - A kind of Forecasting Methodology of polysilicon chip open circuit voltage - Google Patents

Abstract

The invention discloses a kind of Forecasting Methodology of polysilicon chip open circuit voltage.Present invention utilizes CCD imaging technique and luminescence generated by light (PL) imaging technique, from polysilicon chip ccd image, extract the photogenerated current I of expection _l, from PL figure, extract reverse saturation current I ₀, and predict the open circuit voltage V of polysilicon chip on this basis _oc.The invention also discloses predicts the outcome based on open circuit voltage sorts the technology of polysilicon chip.

Description

A kind of Forecasting Methodology of polysilicon chip open circuit voltage

Technical field

The present invention relates to the battery performance prediction field of polysilicon chip, particularly relate to a kind of polysilicon chip open circuit voltage (V _oc) Forecasting Methodology.

Background technology

In polysilicon solar cell field, battery production business and silicon manufacturer all need a kind of method of judgement Si wafer quality of fast and reliable.Wherein, the open circuit voltage V of polysilicon chip _oca kind of important parameter reflecting photovoltaic generation performance, therefore, can based on V _ocpredict the outcome judge Si wafer quality.

Luminescence generated by light imaging (PL Imaging) technology is a kind of fast optical detection method.By process PL image, more existing algorithms at present, predict that the battery performance of polysilicon chip is (comprising prediction open circuit voltage V _oc).But these methods are all confined to the information of luminescence generated by light imaging technique itself, have departing from of can not ignore between different batches experimental result.

Summary of the invention

The present invention is intended to the deficiency overcoming routine techniques means, proposes one polysilicon chip open circuit voltage V more accurately _ocforecasting Methodology.

In general, for achieving the above object, present invention utilizes CCD imaging technique and luminescence generated by light (PL) imaging technique, from polysilicon chip ccd image, extract the photogenerated current I of expection _l, from PL figure, extract reverse saturation current I ₀, and predict the open circuit voltage V of polysilicon chip on this basis _oc.

According to a kind of polysilicon chip open circuit voltage Forecasting Methodology of the present invention, comprising: the luminescence generated by light PL image and the ccd image that a) gather polysilicon chip; B) the PL image that gathers is processed therefrom to extract permanent defects information; C) calculate based on treated PL image wherein β _ito represent in treated PL image the brightness value of a bit; D) calculate (k2-α) based on gathered ccd image, wherein α is the average brightness value of ccd image; And e) will (k2-α) substitutes into open circuit voltage V _ocpredictor formula 1:

$V_{\mathrm{oc}}=\frac{\mathrm{kT}}{q}\ln (k1\frac{k2-\mathrm{\α}}{{\mathrm{\Σ}}_1^m\frac{1}{\sqrt{k3+{\mathrm{\β}}_i}}}+1)$

Wherein k1, k2, k3 are constant, and kT/q is thermal voltage.

According to an aspect of the present invention, described constant k3 by value for making be proportional to total reverse saturation current I of described polysilicon chip ₀.

According to an aspect of the present invention, described constant k3=95.2.

According to an aspect of the present invention, described constant k2 is by the photogenerated current I of value for making (k2-α) be proportional to described polysilicon chip _l.

According to an aspect of the present invention, the least square method based on the batch silicon wafer of known actual open circuit voltage demarcates described constant k1 and k2.

According to an aspect of the present invention, described constant k2=2130, k3=95.2.

According to an aspect of the present invention, described step b) in, maximum filtering is carried out to gathered PL image, and subtracts each other with original image and filtering image.

According to a kind of polysilicon chip open circuit voltage prognoses system of the present invention, comprising: luminescence generated by light imaging device; CCD optical imaging apparatus; And computing equipment, for realizing open circuit voltage Forecasting Methodology of the present invention.

According to test and the method for sorting of a kind of polysilicon chip of the present invention, comprising: by the open circuit voltage V of Forecasting Methodology prediction batch polysilicon chip of the present invention _oc; And based on predicted open circuit voltage V _ocbatch polysilicon chip is tested and sorted.

According to test and the sorting system of a kind of polysilicon chip of the present invention, comprising: aforesaid open circuit voltage prognoses system, for predicting the open circuit voltage V of batch polysilicon chip _oc; And sorting mechanism, based on predicted open circuit voltage V _ocbatch polysilicon chip is sorted.

Accompanying drawing explanation

Comprising accompanying drawing is further understand the present invention for providing, and they are included and form a application's part, and accompanying drawing shows embodiments of the invention, and plays the effect explaining the principle of the invention together with this specification.In accompanying drawing:

Fig. 1 is the polysilicon chip open circuit voltage V according to the embodiment of the present invention _octhe flow chart of Forecasting Methodology.

Fig. 2 a is polysilicon chip luminescence generated by light (PL) image collected according to the embodiment of the present invention.

Fig. 2 b is the polysilicon chip ccd image collected according to the embodiment of the present invention.

Fig. 3 is according to the image after the processing PL image of the embodiment of the present invention.

Fig. 4 is the prediction V obtained according to experimental example of the present invention _ocwith actual V _occomparison diagram.

Embodiment

An exemplary polysilicon chip open circuit voltage Forecasting Methodology is described as follows, it should be noted that, for ease of understanding, following examples give a lot of implementation detail and technical parameter, but these ins and outs and technical parameter are only the object of example, are not construed as limiting the invention.Meanwhile, should be understood that also not all ins and outs and technical parameter are all that enforcement the present invention is necessary.

Fig. 1 illustrates the method for prediction polysilicon chip open circuit voltage according to an embodiment of the invention, comprises following basic step:

1) IMAQ (step 101 in Fig. 1) of polysilicon chip.Comprise and photoluminescence image collection is carried out to polysilicon chip, obtain the PL image such as shown in Fig. 2 a, and ccd image collection is carried out to polysilicon, obtain such as ccd image shown in Fig. 2 b.The resolution of PL image is 16 dark or more Gao Weijia.

2) PL image is processed (step 102 in Fig. 1).This step mainly comprises the information extraction to permanent defects in PL image.The Defect shown in PL image some can remove in cell manufacturing process (as the diffusion of ingot casting process sidewall of crucible causes, the black surround on right side in Fig. 2 a), other then can not (part as cluster-shaped and wire in Fig. 2 a).In addition, in PL image pixel brightness reflection be the size of its position excess carrier concentration Δ n.And when little injection, Δ n is proportional to the minority carrier life time of this point.In order to estimate the impact of PL image permanent defects, can be handled as follows PL image: such as 40 × 40 maximum filtering are carried out to image, and subtracts each other with it with original image, obtain image as shown in Figure 3.In Fig. 3, the brightness value of every bit (is set to β _i) meaning be the difference of maximum in corresponding points brightness and its neighborhood in Fig. 2 a.Brightness is lower, then minority carrier life time is lower.The method of this maximum Neighborhood Filtering can be forgone the information of removable defect in PL figure effectively.

3) reverse saturation current I is estimated ₀(step 103 in Fig. 1)

Reverse saturation current I is estimated based on treated PL image ₀.In prediction algorithm of the present invention, total reverse saturation current I of whole silicon chip ₀computing formula be

$I_0=\underset{1}{\overset{m}{\mathrm{\Σ}}}{I}_{0i},$

Wherein I _0ifor the reverse saturation current of pixel position each in treated PL image, its computing formula is again

$I_{0i}=\frac{\mathrm{Aq}\sqrt{D_e}{n_i}^2}{\sqrt{{\mathrm{\τ}}_{\mathrm{ei}}}{N}_A},$

In above formula: A represents sectional area; Q representation element electric charge; n _iit is the intrinsic carrier concentration of material; N _ait is the p side acceptor concentration of pn knot; D _eit is less sub-diffusion coefficient; τ _eiit is minority carrier life time.Except τ _eiother parameters outer are all constant, therefore I ₀with be directly proportional, k3 is undetermined constant.

Therefore, can calculate as with I ₀the estimated value be directly proportional.

4) photogenerated current I is estimated _l(step 104 in Fig. 1)

Photogenerated current I is estimated based on silicon chip ccd image reflective information _l.Suppose that ccd image mean flow rate is α, then its reflectivity becomes a positive correlation with α.Prediction algorithm of the present invention hypothesis cell reflective rate and silicon chip reflectivity are also a positive correlation, then cell reflective rate also becomes a positive correlation with α, and with photogenerated current I _lbecome a negative correlation.It can thus be appreciated that I _lbe directly proportional to (k2-α), k2 is another undetermined constant.

Therefore, can calculate (k2-α) as with I _lthe estimated value be directly proportional.

5) V is calculated _oc(step 105 in Fig. 1)

Will (k2-α) substitutes into V _ocpredictor formula calculates.Empirical equation after substitution is:

$V_{\mathrm{oc}}=\frac{\mathrm{kT}}{q}\ln (k1\frac{k2-\mathrm{\α}}{{\mathrm{\Σ}}_1^m\frac{1}{\sqrt{k3+{\mathrm{\β}}_i}}}+1)$ (formula 1)

Wherein kT/q is thermal voltage, and parameter k3 is compared by the filtered PL figure of minority carrier lifetime scintigram after polysilicon chip passivation and maximum and obtains, and such as value is 95.2.More particularly, for the silicon chip after abundant passivation, on life test scintigram certain point life value should after filtering corresponding to it on PL figure corresponding points become once relation, k3 can by the intercept decision of this once relation.K1, k2 actual V of known corresponding battery _ocbatch silicon wafer least square method demarcate, such as k1=9.82 × 10 ⁹, k2=2130.

The numbering of said method step be only description just, and do not mean that each step must order successively be implemented according to this.Such as, step 3) and step 4) in fact can carry out simultaneously.

Corresponding polysilicon chip open circuit voltage prognoses system is contained in the present invention simultaneously.According to one embodiment of present invention, polysilicon chip open circuit voltage detection system comprises luminescence generated by light imager and CCD optical imaging instrument, also comprise calculation element, for receiving the silicon chip image that luminescence generated by light imager and CCD optical imaging instrument obtain, and implement open circuit voltage prediction algorithm of the present invention.Described calculation element can be any device realizing special algorithm logic, and it can comprise the general purpose computer of software program for execution, also can comprise the hardware of custom-made, such as FPGA, ASIC, DSP etc.

The present invention also should contain the method and corresponding system that the result based on the prediction of polysilicon chip open circuit voltage sorts batch polysilicon chip.

Experimental example:

An experimental example according to the present invention is as follows:

Get constant k1=9.82 × 10 ⁹, k2=2130, k3=95.2;

Silicon chip is put into test macro, obtains silicon chip photoluminescence image and ccd image respectively;

Process PL image is calculated process ccd image is calculated (k2-α); And

The above results is substituted into formula 1, calculate V _ocpredicted value.

According to above-mentioned experimental example, open circuit voltage test has been carried out to many group batch silicon wafers, and and actual be prepared into battery after the V that records _occompare, as shown in Figure 4, abscissa is actual V _oc, ordinate is prediction V _oc, it is near the straight line of 1 that data point is distributed in slope substantially, surperficial actual V _ocwith prediction V _ocbetween have good degree of conformity.Meanwhile, in experimental example, the test of silicon chip and processing speed are all less than 1s, are highly suitable for online Si wafer quality and judge and sorting.

Technique effect

The present invention utilizes the physical meaning of two kinds of quick silicon wafer test methods, considers the information of silicon chip ccd image and PL image simultaneously, establishes the forecast model of its open circuit voltage.Be proven, open circuit voltage Forecasting Methodology of the present invention can provide the open circuit voltage being coincident with actual value well to predict the outcome, and test and processing speed satisfactory.

Claims (8)

1. a polysilicon chip open circuit voltage Forecasting Methodology, comprising:

A) luminescence generated by light PL image and the ccd image of polysilicon chip is gathered;

B) the PL image that gathers is processed therefrom to extract permanent defects information;

C) calculate based on treated PL image wherein β _ito represent in treated PL image the brightness value of a bit;

D) calculate (k2-α) based on gathered ccd image, wherein α is the average brightness value of ccd image; And

E) will (k2-α) substitutes into open circuit voltage V _ocpredictor formula 1:

$V_{\mathrm{oc}}=\frac{\mathrm{kT}}{q}\ln (k1\frac{k2-\mathrm{\α}}{{\mathrm{\Σ}}_1^m\frac{1}{\sqrt{k3+{\mathrm{\β}}_i}}}+1)$

Wherein k1, k2, k3 are constant, and kT/q is thermal voltage,

Described constant k3 by value for making be proportional to total reverse saturation current I of described polysilicon chip ₀,

Described constant k2 is by the photogenerated current I of value for making (k2-α) be proportional to described polysilicon chip _l.

2. the method for claim 1, is characterized in that, described constant k3=95.2.

3. the method for claim 1, is characterized in that, the least square method based on the batch silicon wafer of known actual open circuit voltage demarcates described constant k1 and k2.

4. the method for claim 1, is characterized in that, described constant k2=2130, k3=95.2.

5. the method for claim 1, is characterized in that, described step b) in, maximum filtering is carried out to gathered PL image, and subtracts each other with original image and filtering image.

6. a polysilicon chip open circuit voltage prognoses system, comprising:

Luminescence generated by light imaging device;

CCD optical imaging apparatus; And

Computing equipment, for realizing the method according to any one of claim 1-5.

7. the test of polysilicon chip and a method for sorting, comprising:

By the open circuit voltage V of the method prediction batch polysilicon chip any one of claim 1-5 _oc; And

Based on predicted open circuit voltage V _ocbatch polysilicon chip is tested and sorted.

8. the test of polysilicon chip and a sorting system, comprising:

Open circuit voltage prognoses system as claimed in claim 6, for predicting the open circuit voltage V of batch polysilicon chip _oc; And

Sorting mechanism, based on predicted open circuit voltage V _ocbatch polysilicon chip is sorted.