Electroluminescence as a Tool to Study the Polarization Characteristics and Generation Mechanism in Silicon PV Panels

Abstract

Electroluminescence is a defect detection method commonly used in photovoltaic industry. However, the current research mainly focuses on qualitative analysis rather quantitative evaluation, since there exists some shortcomings, such as fuzzy edges, unclear texture, etc., in the obtained electroluminescence images. Electroluminescence polarization imagery is a new method for defect detection in photovoltaic modules, which can effectively make up for the aforementioned deficiencies. In this paper, the polarization characteristics and formation mechanism of silicon solar panels was investigated based on the principle of electroluminescence. Firstly, the polarization imaging mechanism of electroluminescence of photovoltaic modules was studied. Then, an electroluminescence polarization experimental platform was built, and the polarized electroluminescence images of photovoltaic panels were obtained and preprocessed with Gaussian filter. Finally, the influence of view angle, bias voltage and other factors on the polarization characteristics was discussed. The results show that the electroluminescence of photovoltaic modules has polarization characteristics, and the degree and angle of polarization are related to the view angle and bias voltage; the degree of polarization of photovoltaic panel electroluminescence increases with the angle of view, and first increases then rapidly decreases with bias voltage.

1. Introduction

Photovoltaic cells and modules are the core parts of the solar power generation system, and the defects such as fragments, broken grids and hidden cracks are prone to occur in the process of production, transportation, or application; these not only cause production waste but also reduce the power generation efficiency, and may even lead to major safety accidents if excessive temperature caused by equipment damage exists inside the panel. Therefore, it is of very important application value and practical significance to study real-time and efficient defect detection methods for photovoltaic cells and modules.

In 2005, Fuyuki et al. captured the light emission of solar cells under positive bias voltage using a short-wave infrared (SWIR) CCD, and found that there is a one-to-one relationship between the emission intensity and the diffusion length of minority carriers. Then, Fuyuki et al. proposed a semi-quantitative analysis method for detecting the diffusion length distribution, diffusion length mapping and degraded areas of minority carriers in silicon-based solar cells with electroluminescence (EL) [1]. In 2007, Fuyuki et al. realized the rapid detection of EL intensity distribution and defects in the region captured by silicon CCD camera within 1 s; they did so by analyzing the impact of the inherent material characteristics, such as diffusion length, recombination velocity at surface and subsurface, and external defects, such as wafer damage and electrode breakdown, on minority carriers. [2]. Based on the above work, the detailed information inside the solar panel, such as minority carrier lifetime, two-dimensional electrical characteristics, series resistance and shunt resistance, has been obtained through electroluminescence, and a defect testing method for crystalline silicon photovoltaic modules based on electroluminescence was proposed [3]. However, the images obtained by traditional electroluminescence detection methods mainly reflect the intensity of shortwave infrared radiation generated by carrier recombination, and the impact of radiation emitted from optical and physical properties, edge characteristics, texture and other aspects of the defect area has not been taken into account, which makes it difficult to quantify the defect area and give accurate judgments on the type and impact of the defects.

Excited by external positive bias voltage, photovoltaic cells usually emit short wave infrared light with a peak wavelength of about 1150 nm which has polarization characteristics [4]. Polarization is the fourth piece of important information in addition to basic attribute strength, wavelength, and phase of electromagnetic waves. Through polarization imaging, not only the intensity but also the physical characteristics, geometric information and surface texture characteristics of the target can be obtained or resolved. In 2002, Bush et al. found that solar panels are the main source of the polarization characteristics of satellite surface scattering when predicting the polarization characteristics of satellites in different Earth orbit scenarios, using the advanced tracking time domain analysis simulation system (TASAT) to generate polarization renderings of satellite models [5]. In 2010, Li et al. found that the polarization characteristics of the scattered light on the surface of the solar panel are related to the composition materials; the angle of incidence and the attitude have a large impact, where the amount of polarization can be used to improve the recognition ability of space targets [6]. Niu et al. studied the infrared radiation polarization effect mechanism and typical characteristics of solar panels through theoretical simulation modeling and outdoor experiments and analyzed the influence of view angle, refractive index, surface state, temperature and other factors on the infrared polarization effect characteristics; they found that the thermal infrared polarization characteristics of solar panels are obvious and strongly correlated with the emission angle [7]. Wang Fangbin et al. proposed a defect detection method for crystalline silicon photovoltaic cells based on electroluminescent polarization image fusion. It was found that polarization imaging can highlight the contour edges and texture details of photovoltaic cell defect images, and the characteristics of photovoltaic cell defect images after the fusion of polarization image and light intensity image are more prominent [8].

The above research mainly focuses on the polarization characteristics of reflection or thermal radiation on the surface of photovoltaic panels. However, the polarization characteristics and generation mechanism in the electroluminescence process of photovoltaic cells has not been studied in-depth thus far. In 2011, Matthew P. Peloso observed that some material structures of silicon solar cells produce strong polarization characteristics, and found the polarization characteristics are related to the anisotropy of the internal charge of defects in silicon solar cells during the optical recombination process, especially for high dislocation density areas, where the partial polarization characteristic of the sub-band gap electroluminescence was highlighted; additionally, the polarization direction is consistent with the dislocation direction [9,10]. In 2013, Radek Stojan found that the relationship between the electroluminescence intensity of solar cells and the rotation angle of the linear polarization analyzer is sinusoidal through experiments and 3D data analysis, indicating that the electroluminescence of solar cells has polarization characteristics which are related to the internal structural and defect of photovoltaic cells [11,12]. The above research shows that the electroluminescence of solar cells has polarization characteristics, but it does not tell us which external objective factors are related to the polarization characteristics of electroluminescence, and it only aims at the research object of polysilicon solar cells, which have limitations. Based on this, this paper studies the polarization mechanism of electroluminescence and explores which external factors are related to its polarization characteristics.

In this paper, the generation mechanism of electroluminescence polarization characteristics was investigated, taking the silicon solar panel as the object; the method for acquiring the polarized electroluminescence images was given by constructing a short-wave infrared polarization imaging platform and conducting experiments. In addition, the effects of view angle and bias voltage on the polarization characteristics of silicon solar panel electroluminescence were analyzed.

2. Generation Mechanism of Electroluminescence Polarization Characteristics

2.1. Electroluminescence

The crystalline silicon photovoltaic cells are generally N+/P homogeneous junction structure, and the surface of a photovoltaic cell usually has a dense pyramid texture structure of micrometer size or is covered with an antireflection film to reduce light reflection energy loss [13].

In the equilibrium state, there is a built-in electric field E in the PN junction of crystalline silicon from the P to N region. The carrier diffusion current and drift current cancel out. When the photovoltaic cell is connected to the forward bias voltage, the carrier balance is broken, so that the electrons and holes can pass through the PN junction and become unbalanced carriers. The unbalanced carriers compound with majority carriers and radiate outward; this is called electroluminescence [14], and is shown in Figure 1.

When the positive bias voltage is applied externally, the crystalline silicon solar cell can be simplified to an ideal PN junction with a thickness of $d$. It can be assumed that the voltage directly acts on the PN junction, as shown in Figure 2.

Under the conditions of small injection and abrupt depletion layer, ignoring the reabsorption of photons generated by radiation recombination [15,16], the electroluminescence intensity generated at different positions $x$ can be represented as:

$$I_L(E,x)=\frac{2\pi }{h^3{c}^2}\alpha (E)E^2\exp (-\frac{E}{kT})\exp (\frac{q{V}_f}{kT})[\frac{\exp (-\frac{x}{\sqrt{L_n}})}{1-\exp (-\frac{2d}{\sqrt{L_n}})}+\frac{\exp (\frac{x}{\sqrt{L_n}})}{1-\exp (\frac{2d}{\sqrt{L_n}})}]$$

(1)

where, $h$ is Planck constant; $c$ is the speed of light in vacuum; $\alpha (E)$ is the absorption coefficient of photons; $E$ is the energy of photons generated by radiation recombination; $q$ is the unit charge; $V_f$ is the positive bias voltage applied on the battery chip; $k$ is Boltzmann constant; $T$ is the absolute temperature; $L_n$ is the electron diffusion length, and $L_n=\sqrt{D_n{\tau }_n}$ ($D_n$ is the diffusion coefficient, ${\tau }_n$ is the minority carrier lifetime).

By integrating Formula (1), the luminous intensity of the photovoltaic cell can be calculated:

$$I_L={\displaystyle \underset{E}{\int }\frac{2\pi }{h^3{c}^2}\alpha (E)E^2\exp (-\frac{E}{kT})dE•{\displaystyle {\int }_0^d[\frac{\exp (-\frac{x}{\sqrt{L_n}})}{1-\exp (-\frac{2d}{\sqrt{L_n}})}+\frac{\exp (\frac{x}{\sqrt{L_n}})}{1-\exp (\frac{2d}{\sqrt{L_n}})}]dx•\exp (\frac{q{V}_f}{kT})}}$$

(2)

Generally, the thickness $d$ of silicon wafer is much greater than the diffusion length $L_n$, so Formula (2) can be simplified as:

$$I_L=C\sqrt{L_n}\exp (\frac{q{V}_f}{kT})$$

(3)

where $C$ is the constant coefficient after simplification.

It can be seen from Formula (3) that the electroluminescence intensity of photovoltaic cells is related to the external positive voltage and minority carrier diffusion length (minority carrier lifetime). At room temperature and constant voltage, the non-equilibrium carrier density at the defect is relatively low, the minority carrier life is reduced, and the diffusion length is shortened; therefore, the radiation intensity is relatively weak. The defect area is darker in the image and shows different morphologies. Therefore, electroluminescence can be used for defect detection of crystalline silicon photovoltaic cells.

2.2. Polarization Characteristic Generation

In the production process of solar cells, silicon wafers usually need to be solidified. During the curing process, residual stress is usually introduced into the wafer and makes the crystal anisotropic; subsequently, plastic deformation will occur at a high temperature, which leads to the formation and movement of crystal defects such as dislocations. As a result, the quantum mechanical effect at the dislocation core gives the electroluminescence polarization characteristics, especially at the regions with high dislocation density, where the polarization direction is consistent with the dislocation direction.

According to the principle of quantum mechanics, the radiation recombination of photovoltaic cells depends on the wave function of the carrier, and the electroluminescence intensity I is positively correlated with the probability R of optical transition at the defect. According to the Bloch theorem, the wave function at the crystal defect can be reduced to the spatially related conduction $|u_c\rangle$ and valence $\langle u_v|$ Bloch bands [10]. Thus, for an anisotropic semiconductor, the probability of optical transition can be formalized as the emission of linearly polarized photons including the polarization vector [10]; that is, the luminous intensity can be expressed as:

$$I\propto K{\left|\langle u_v|\overrightarrow{e}\cdot \widehat{p}|u_c\rangle \right|}^2$$

(4)

where $\overrightarrow{e}$ is photon; $K$ and $\widehat{p}$ is the overlap of the envelope describing the electron and hole; and $\widehat{p}$ is the momentum operator.

Linearly polarized $\overrightarrow{p}$ can be decomposed into a parallel component $\overrightarrow{e_{\parallel }}$ and a perpendicular component $\overrightarrow{e_{\perp }}$ to the incident plane. By time integrating $\overrightarrow{e_{\parallel }}$ and $\overrightarrow{e_{\perp }}$, the parallel $I_{//}$ and vertical components $I_{\perp }$ of the outgoing light can be obtained:

$$\{\begin{array}{l}{I}_{//}={\epsilon }_{//}I\\ I_{\perp }={\epsilon }_{\perp }I\end{array}$$

(5)

where, ${\epsilon }_{//}$ and ${\epsilon }_{\perp }$ are the emissivity of the parallel component and the vertical component of the outgoing light.

The emissivity is mainly affected by the type of target material, surface roughness, physical and chemical nature, and material thickness. Defects and doping are the determining factors of carrier type, concentration, transmission, and non-radiative recombination of photogenerated carriers in semiconductors. There are many kinds of defects, concentrations, and anisotropy in the crystal structure of solar cells, which makes the emissivity of the parallel component and the vertical component of the light emitted from the defect area unequal. Research shows that the electroluminescence from the defect area of solar cells shows a high degree of polarization, especially from a high dislocation area [11]. On the other hand, the etching of silicon-based solar cells makes the surface uneven to form a strong diffuse reflection, which also causes the emissivity of the parallel and vertical components of the outgoing light are unequal [17]. For PV panels, from top to bottom, they comprise glass, EVA, a solar cell, and a backplane and aluminum frame sealed together by lamination. When photons are emitted from the PN junction, they need to pass through two layers of structure (EVA, glass) to refract. The emission conditions are different under different voltages and different emission angles, so the polarization characteristics are related to the forward bias and emission angle. The degree and angle of polarization can be written as:

$$\{\begin{array}{l}DOP=\left|\frac{I_{//} - I_{\perp }}{I_{//} + I_{\perp }}\right|=\left|\frac{{\epsilon }_{//} - {\epsilon }_{\perp }}{{\epsilon }_{//} + {\epsilon }_{\perp }}\right|\\ AOP=\arctan \left(\frac{I_{//}}{I_{\perp }}\right)==\arctan \left(\frac{{\epsilon }_{//}}{{\epsilon }_{\perp }}\right)\end{array}$$

(6)

3. Experiment

3.1. Setup

The experimental setup is mainly composed of a SWIR camera Owl 640S, a high-precision turntable equipped with a SWIR polarizer, a DC power supply and an embedded mini-PC with image acquisition software. The response wavelength of the SWIR camera, using a 640 * 512 InGaAs sensor, is between 900–1700 nm. The exposure time of the camera is 14 $\mathsf{\mu }\mathrm{s}$ and the effective pixel ratio is greater than 99%. The polarizer is a 20LP-NIR precision linear grid polarizer with an extinction ratio of 1000:1. The DC power supply is a Gwinstek 3030DD, which can provide constant current output with a voltage range of 0–30 V and a current range of 0–3 A. The schedule of the electroluminescence polarization experimental platform is shown in Figure 3, and the experimental spot is shown in Figure 4.

3.2. Sample

The experimental samples include polycrystalline silicon and monocrystalline silicon photovoltaic solar panels. The size of the polysilicon solar panel is 135 mm * 125 mm, and its peak power is about 6 W; the size of the monocrystalline silicon photovoltaic panel is 295 mm * 220 mm, and its peak power is about 10 W. There are a positive and a negative electrode on the back of the solar panel connecting the DC power supply. The samples used in this paper are shown in Figure 5.

3.3. Infrared Polarization Image Acquisition

Because the circular polarization is very small in practice and can be ignored, assuming that the Stokes vector of the outgoing electroluminescence from the panel is $S={[I,Q,U]}^T$ (where $I$, $Q$, $U$ denote the intensity, linear polarization component in horizontal direction, and linear polarization component in the diagonal direction of the electroluminescence), the energy received by the camera through the polarizer at the polarization azimuth $\alpha$ can be presented as:

$$I(\alpha )=\frac{1}{2}(I+Q\cos 2\alpha +U\sin 2\alpha )$$

(7)

$\alpha$ is usually selected as 0°, 60°, 120° for the convenience of calculation in engineering. Then $I$, $Q$, $U$ can be calculated as follows:

$$\{\begin{array}{l}I=\frac{2}{3}[I(0^{\circ })+I({60}^{\circ })+I({120}^{\circ })]\\ Q=\frac{2}{3}[2I(0^{\circ })-I({60}^{\circ })-I({120}^{\circ })]\\ U=\frac{2\sqrt{2}}{3}[I(0^{\circ })-I({120}^{\circ })]\end{array}$$

(8)

where I (0°), I (60°), I (120°) denote the SWIR polarization images at the azimuth 0°, 60°, 120°. Next, the degree of polarization DOP and the angle of polarization AOP images can be obtained:

$$\{\begin{array}{l}DOP=\frac{\sqrt{Q^2+U^2}}{I}\\ AOP=\frac{1}{2}\arctan (\frac{U}{Q})\end{array}$$

(9)

3.4. Experimental Procedure

Before the experiment, the crystalline silicon photovoltaic panels to be tested were fixed on an optical platform, and connected the DC power supply. The SWIR camera was fixed on a tripod, and the turntable with the polarizer also was fixed on the optical platform. The optical center of the SWIR camera was tuned in a straight line with that of the polarizer which was perpendicular to the center of the photovoltaic panel to be measured. The distance between the camera lens and the panel was adjusted to obtain a suitable field of view.

During experiments, the panel was first excited with the power supply, and then the turntable was driven to turn the polarizer to the azimuths at a high speed. Then, the camera shot and the SWIR polarization images I (0°), I (60°), I (120°) were obtained.

In order to avoid the interference of environmental noise, the experiments were conducted in a dark room where the indoor temperature was kept at 25 °C.

3.5. Image Preprocess

In the process of generation and transmission, polarized images are prone to be influenced by random noise due to the camera and environmental factors. Therefore, the acquired images were filtered with a Gaussian filter:

$$G(x,y)=\frac{1}{2\pi {\sigma }^2}{e}^{-(\frac{x^2+y^2}{2{\sigma }^2})},$$

(10)

where $\sigma$ is the filter constant.

The EL image and DOP image obtained are grayscale images, and the grayscale value of each pixel point is related to the electroluminescent intensity and polarization. The values under different voltages and angles are different; this is represented by different pixel brightness [18]. In order to quantitatively describe this difference, it is necessary to calculate the electroluminescent intensity and degree of polarization of each pixel. It is meaningless to compare the value of each pixel point, so we selected the ROI region and calculated the average value in the region. The average value represents the electroluminescence intensity or polarization of the entire PV panel, and the DOP and EL curve were drawn by these values.

4. Analysis and Discussion

4.1. Image Quality

I, Q, U, DOP, AOP images of the observed polycrystalline and monocrystalline silicon solar panels were resolved by Equations (7) and (8), as shown in Figure 6 and Figure 7, where I (0°), I (60°), I (120°), I, Q, U, DOP and AOP were listed from left to right, respectively.

From Figure 6 and Figure 7, the edges of the defects are blurry, and several small defects are difficult to distinguish from image I, which is equivalent to traditional electroluminescence. However, the geometric information of defects can be noticed from I (0°), I (60°), I (120°). Q and U images are relatively sensitive to defect edge details, DOP and AOP images were significantly enhanced with textural details, where the internal grid structure can be clearly recognized and the defect area has a higher contrast.

To evaluate the quality of each image objectively, five indicators of information, entropy (IE), average gradient (AG), standard deviation (SD), space frequency (SF), and edge intensity (EI) were adopted, and the values of them were calculated and shown in

Table 1. Objective evaluation index of polycrystalline silicon solar panel.

Image	IE	AG	SD	SF	EI
I (0°)	5.2738	2.5139	22.3112	5.1122	17.2192
I (60°)	5.2769	2.5012	22.3881	5.0838	17.1187
I (120°)	5.2752	2.5131	22.3823	5.1059	17.1947
I	5.7001	3.4125	31.1978	6.8641	23.1996
DOP	4.2414	14.3442	35.2736	31.2222	80.5407
AOP	3.299	71.2911	72.914	134.6397	292.2719

and

Table 2. Objective evaluation index of monocrystalline silicon solar panel.

Image	IE	AG	SD	SF	EI
I (0°)	5.7211	2.8818	17.4242	5.8161	22.1181
I (60°)	5.7366	2.8868	17.3715	5.8409	22.1573
I (120°)	5.7335	2.8862	17.4171	5.8285	22.1839
I	6.34	4.4121	27.0704	8.8893	33.7119
DOP	5.1754	12.2685	30.811	27.7195	64.3701
AOP	3.7124	65.1589	68.8897	118.9183	259.8149

as follows:

From Table 1 and Table 2, the standard deviation, average gradient, spatial frequency, and edge intensity of DOP and AOP images have been greatly improved compared to image I, indicating that DOP and AOP images are rich in more texture details, which is consistent with the subjective observation of human eyes.

4.2. Polycrystalline Silicon Solar Panel

4.2.1. Bias Voltage

At the emission angle of 40°, the forward bias voltage increases from 9.5 V to 15 V with the step of 0.5 V. Image acquisition is carried out at the same time of each voltage change. The images I and DOP of the polycrystalline photovoltaic panel were obtained as shown in Figure 8 and Figure 9, respectively.

It can be seen from Figure 8 that when the bias voltage is below 10 V, the panel is not conductive and the electroluminescence intensity is very small and can almost be ignored; therefore, there is no luminous area shown in the image. With the increase in bias voltage, the intensity of electroluminescence increases gradually. When the bias voltage reaches 11.5 V, the contrast of the image is significantly enhanced, and the difference between the defect and normal area becomes obvious. With a further increase in bias voltage, an overexposure area appears in the image when the bias voltage exceeds 12.5 V. At this moment, it is difficult to distinguish the defect area from other areas. If the bias voltage continued to increase, the solar panel may be damaged heavily. In general, with increases in bias voltage, the electroluminescence intensity is continuously enhanced; this is basically consistent with the mathematical model given in Formula (3).

From DOP image in Figure 9, when the bias voltage is less than 10 V, the degree of polarization is almost equal to 0, as is the intensity. With the increase in bias voltage, DOP gradually increases and reaches a peak value at about 12 V. This is mainly due to the increase in minority carriers involved in radiation recombination; the difference between the parallel and vertical components of the outgoing light increases until the degree of polarization reaches a maximum value. When the bias voltage exceeds 12 V, the DOP value decreases rapidly with the bias voltage. This is mainly because the number of injected carriers increases, and the numbers of photons participating in radiation recombination in the vertical and horizontal directions approaches equal, which makes the parallel and vertical components of outgoing light almost equal. Additionally, the DOP value decreased, meaning a depolarization effect begins to appear.

In short, there exist many overexposed areas in the electroluminescent intensity image if the bias voltage is too large, and these areas cannot appear in DOP images. In addition, it can also be noticed that for a polycrystalline silicon photovoltaic panel, the DOP value of the defective area is usually lower than that of the flawless area, and the voltage corresponding to the peak DOP value is smaller. The curves of the electroluminescence intensity and degree of polarization of the photovoltaic panel with the bias voltage are shown in Figure 10, as follows.

It can be seen from Figure 10 that the electroluminescence intensity increases with bias voltage and gradually tends to saturation when the bias voltage approaches 12 V. The degree of polarization of the photovoltaic panel first increases and reaches a peak at about 12 V, and then decreases rapidly with the bias voltage, which is consistent with the above qualitative analysis.

4.2.2. View Angle

To study the effect of view angle on electroluminescence polarization imaging, Figure 11 shows DOP images at different view angles under 12 V bias voltage, and the curve of DOP value with the view angle is shown in Figure 12.

From Figure 11 and Figure 12, when the view angle is small, the degree of polarization of the panel is low. This moment, the light received by the camera mainly belongs to that emitted vertically from the surface of the panel; therefore, the difference of the vertical and parallel components is very small, depending on the anisotropy of the crystal. With the increase of the view angle, the degree of polarization increases significantly. In other words, the polarization characteristics of the outgoing light not only depend on those of the light incident to the interface after generation of the inside panel, but also on the change of the characteristics caused by interface refraction and multi-directional scattering; this shows a similar characteristic to the infrared spontaneous radiation from the hot metal plate [7].

4.3. Monocrystalline Silicon Solar Panel

4.3.1. Bias Voltage

The view angle was set to 40°, and the bias voltage increased from 22 V to 27.5 V with a step of 0.5 V. The images I and DOP of monocrystalline silicon photovoltaic panel were obtained as shown in Figure 13 and Figure 14, respectively. The curve of DOP value with bias voltage is shown in Figure 15.

It can be seen from Figure 13, Figure 14 and Figure 15 that, with the increase in bias voltage, the intensity kept increasing until saturation, while the degree of polarization increased first and then decreased; this is consistent with the conclusion of polysilicon photovoltaic modules. This is mainly due to the anisotropy of the material and external field composition system, which transmits information about defect structure and charge distribution direction [11].

4.3.2. View Angle

Under the condition of 24 V bias voltage, the view angle was increased from 0° to 80° every 10°. Figure 16 shows DOP images at different view angles under 24 V external positive bias voltage, and the curve of DOP value with view angle is shown in Figure 17.

It can be seen from Figure 16 and Figure 17 that, with the increase in the view angle, the degree of polarization of the monocrystalline silicon solar panel gradually increased; this is consistent with the trend of the degree of polarization of the polycrystalline silicon solar panel, which confirmed the above qualitative analysis.

4.4. DOP Comparison between Monocrystalline Silicon and Polycrystalline Silicon Cells

To further compare the polarization characteristics of monocrystalline and polycrystalline panel, two photovoltaic cells with size of 52 mm * 52 mm were selected and measured, fixed on the same plane, as shown in Figure 18, wherein the monocrystalline silicon photovoltaic cell is above and the other is below.

Considering that the rated parameters of voltage, current, power, and others are different for monocrystalline and polycrystalline silicon photovoltaic cells, the same type of DC power supplies were used to apply a preset bias voltage at the same time. In the experiment, the bias voltage was adjusted and monitored by a SWIR camera so that the outgoing intensity from the two cells stayed almost equal, as shown in Figure 19.

Figure 20 shows DOP images of two monocrystalline silicon and polycrystalline silicon cells, and the curves of DOP value with view angle are shown in Figure 21.

It can be seen from Figure 21 that the degree of polarization of monocrystalline and polycrystalline silicon cells increases with the view angle. Under the same luminous intensity, the degree of polarization value of polycrystalline silicon cell is slightly greater than that of the monocrystalline one. This may be because there are more crystal defects in polycrystalline silicon solar cells, where high density dislocations significantly polarize the EL sub-band gap and affect the anisotropic charge distribution in the recombination process. It should be noted that due to the difference between the structure of the cell and the panel, the polarization degree of the cell is lower than that of the panel, and the data fluctuates, resulting in a DOP value lower than the theoretical value at 20° and 40°.

5. Conclusions

In this paper, polarization imaging technology was introduced based on the principle of photovoltaic panel electroluminescence, and the factors affecting the polarization characteristics of photovoltaic panel electroluminescence were analyzed by building a SWIR polarization imaging platform. The results show that the electroluminescence of photovoltaic panels has polarization characteristics; this is related to the view angle and bias voltage. The degree of polarization of photovoltaic panel electroluminescence increases with the view angle. The intensity increases continuously before saturation, and the polarization degree of electroluminescence increases first and then decreases rapidly with external bias voltage.

The research in this paper is an extension of traditional EL technology, by which the electroluminescence process of silicon photovoltaic panels can be furthered and the defects quantitatively evaluated. It is of great significance to the exploration of new methods for defect detection in photovoltaic cells and modules.