Predicting performance values of solar cells from electrical impedance spectroscopy measurements

Abstract

The invention relates to a computer-implemented method for providing output data, wherein the output data comprises a performance value of a perovskite-based solar cell. The method comprises the steps of (i) receiving input data, wherein the input data comprises data from an electrical impedance spectroscopy performed on the solar cell or a portion thereof, (ii) applying a trained function to the input data, wherein the output data is generated, and (iii) providing the output data. The invention further relates to a computer-implemented method for providing a trained function, the method comprising training a function capable of predicting a performance value of a perovskite-based solar cell based on an electrical impedance spectroscopy performed on the solar cell or a portion thereof. Furthermore, the invention relates to a corresponding providing system and to a corresponding computer-readable medium.

Description

Technical Field

[0001] This invention relates to a computer-implemented method for predicting performance values ​​of perovskite-based solar cells based on electrical impedance spectra performed on a solar cell or a portion thereof. Furthermore, this invention relates to a computer-implemented method for providing a trained function, comprising training a function capable of predicting performance values ​​of perovskite-based solar cells based on electrical impedance spectra performed on a solar cell or a portion thereof; and further, this invention relates to a corresponding providing system and a corresponding computer-readable medium. Background Technology

[0002] Perovskite-based solar cells are a novel and promising photovoltaic technology. In a relatively short period of development, cell efficiencies exceeding 25% have been achieved on a laboratory scale. This is competitive with the record-breaking efficiencies of conventional polycrystalline silicon or monocrystalline silicon solar cells, which are in the range of 23% and 26% respectively on a laboratory scale. A record-breaking light conversion efficiency of 28% has been demonstrated for tandem configurations including perovskite-based and silicon solar cells.

[0003] Nevertheless, scaling up the manufacture of perovskite-based solar cells and corresponding photovoltaic cells presents new challenges. Compared to the widely available and mature manufacturing processes for silicon-based solar cells, the manufacturing processes for perovskite-based solar cells still require optimization. For the manufacture of tandem solar cells, even more manufacturing process steps must be established, some of which have not yet been developed for large-scale industrial production.

[0004] Precise monitoring of perovskite-based solar cells during successive manufacturing process steps, and analysis of performance deviations traced back to specific process steps, will make the process development and production of perovskite-based solar cells cost-effective.

[0005] Today, to quantitatively assess the performance of perovskite-based solar cells, the fully treated cells are characterized in a controlled environment under completely known illumination. Specifically, the cell's I(V) curve (current-voltage curve) is recorded on a solar simulator, a device that provides illumination close to natural sunlight. The key performance values ​​of the tested solar cell are then derived from this.

[0006] However, this process requires significant investment, is time-consuming, and must be performed on fully processed solar cells. In other words, there is no established, rapid, and reliable method to inspect intermediate steps in the perovskite-based solar cell production chain. To assess process quality, all fabrication steps must be followed, and the performance of the resulting cells can only be characterized after the solar cells are completed. Typically, 1-V measurements under full illumination are performed according to established standards such as IEC 60904-9. These measurements are used to evaluate the key performance parameters of perovskite-based solar cells.

[0007] Therefore, there is a desire to develop a concept that addresses how to predict the performance of perovskite-based solar cells in a way that is alternative to existing technologies. Ideally, the results could be used to monitor and optimize the processing of solar cells more quickly and directly. Summary of the Invention

[0008] This objective is achieved through the subject matter of the independent claims. Advantageous embodiments and variations are disclosed in the dependent claims and the specification.

[0009] According to the present invention, a computer-implemented method for providing output data is provided, wherein the output data includes performance values ​​of a solar cell, and the solar cell comprises a perovskite material. The method includes the following steps:

[0010] - Receive input data, wherein the input data includes data from electrical impedance spectroscopy performed on a solar cell or a portion thereof.

[0011] - The trained function is applied to the input data, where output data is generated, and

[0012] - Provides output data.

[0013] A key aspect of this invention is the use of electrical impedance spectroscopy to predict one or more performance values ​​of perovskite-based solar cells. The use of electrical impedance spectroscopy is attractive because it does not need to be performed on fully manufactured (in other words, fully processed) solar cells, but can in principle also be performed on intermediate cell materials, i.e., before the cell processing is complete. In other words, electrical impedance spectroscopy measurements can be performed on fully or partially processed solar cells.

[0014] Electrical impedance spectroscopy (EIS) is a technique used to evaluate the current response to an alternating current voltage application based on frequency. Specifically, EIS measures the resistive and capacitive properties of a material by applying a sinusoidal AC excitation signal, for example, in the range of 2 mV to 10 mV. The impedance spectrum is obtained by varying the frequency within a predefined range. For example, EIS is known for studying the dynamics of electrical processes, including elucidating relevant ionic and electronic processes occurring at different interfaces in various electrical devices.

[0015] However, EIS has never been used to predict the performance values ​​of solar cells (especially perovskite-based solar cells) to date.

[0016] Unlike conventional IV measurements, EIS does not require a solar simulator. Instead, EIS measurements can be performed in low light or even no light at all.

[0017] For example, impedance spectroscopy can be performed at frequencies ranging from 100 millihertz (mHz), or 0.1 Hz, to 1 megahertz (MZh), or 1,000,000 Hz.

[0018] In an advantageous embodiment of the invention, electrical impedance spectroscopy is performed at a finite number of predefined frequencies. This has the advantage of significantly reducing measurement time. Therefore, measurement times can be achieved in the range of a few seconds to less than one second. This has the advantage that EIS measurements can potentially be performed online during the manufacturing process of perovskite-based solar cells. Advantageously, the frequencies are chosen such that they are well-suited for characterizing the material under test and thus can form the basis for performance value predictions. Choosing a suitable frequency is generally convenient in practice because, since the same cell type is typically manufactured along a single production process, the most suitable frequency is known.

[0019] EIS measurements are performed on the solar cell or a portion thereof. The solar cell does not yet need to be fully fabricated. For example, an anti-reflective coating is not yet required. Furthermore, a front contact (e.g., transparent conductive oxide, TCO) is not required. However, some type of contact is necessary because a voltage needs to be applied to the material. If a contact is not yet present in the solar cell, small point contacts can be provided, for example, at the boundaries of the solar cell material (especially at corners). In principle, even a back contact (e.g., achieved by completely covering the back of the solar cell with aluminum) can be replaced by the test structure.

[0020] The purpose of the method according to the present invention is to predict the performance values ​​of perovskite-based solar cells. Perovskite-based solar cells can also be referred to as perovskite solar cells, or simply PSCs. Specifically, the solar cell includes a photosensitive layer, and the photosensitive layer comprises a perovskite material.

[0021] Perovskite-based solar cells are solar cells that include a perovskite-structured compound as the light-collecting active layer, or photosensitive layer, most commonly a hybrid organic-inorganic lead or tin halide-based material. The general chemical structure of perovskite is ABX3, and the common specification is (CH3NH3)PbX3.

[0022] The method of the present invention proposes to use data from electrical impedance spectra performed on a solar cell or a portion thereof as input data.

[0023] In one embodiment of the invention, the input data, namely EIS data, includes the impedance, admittance, modulus function, complex permittivity, and / or dielectric constant of the solar cell or a portion thereof. These data can, in principle, be derived from the Nyquist plot or Bode plot measured by EIS.

[0024] In another embodiment of the invention, input data is derived by modeling one or more electrical equivalent circuits based on the impedance spectrum performed on a solar cell or a portion thereof. Based on the modeled electrical equivalent circuits, parameters such as series resistance, capacitance, or Warburg impedance of the solar cell or a portion thereof can be derived. These parameters can then be used as input data for a trained function used to predict the performance values ​​of the solar cell.

[0025] In another embodiment of the invention, the input data further includes at least one process parameter relating to the treatment of the solar cell. Specifically, the process parameter may include the physical or chemical properties of at least one of the substances used in the treatment of the solar cell, such as the material composition or viscosity. In other words, in addition to the input data obtained from EIS measurements, the process parameters from the treatment of the solar cell under test can also be used to predict one or more performance values ​​of the solar cell.

[0026] Process parameters can also include deposition parameters during the deposition process of the solar cell elements. Here, temperature, pressure, printing, or spin coating parameters appear to be particularly valuable as input data.

[0027] Finally, the physical and chemical properties of the solar cell elements (such as layer thickness) can also be taken into account and added to the input data.

[0028] This provides one or more performance values ​​of the solar cell as output data. Specifically, the performance values ​​are selected from the group consisting of: short-circuit current; open-circuit voltage; voltage of the solar cell at its maximum power point (also called maximum power point voltage); current of the solar cell at its maximum power point (called maximum power point current); maximum power (which is the product of maximum power point voltage and maximum power point current); fill factor (which is the product of short-circuit current and open-circuit voltage divided by maximum power); and cell efficiency, also known as the light conversion efficiency of the solar cell. These performance values ​​are often referred to as the “key cell parameters” of the solar cell. A quantitative prediction of one or more of these parameters (or values) provides a specific and quantitative estimate of the quality and performance of the solar cell.

[0029] According to the present invention, instead of seeking an analytical function between the input and output data, a relationship is established between the input and output data, referred to as a function in this patent application. This function is trained to represent the relationship between the input and output data as best as possible.

[0030] One approach to training a function is to apply multivariate statistics. Another approach is to use neural networks, such as convolutional neural networks.

[0031] The function is trained according to the following method: First, input training data is received, where the input data includes exemplary data from the electrical impedance spectrum applied to a perovskite-based solar cell or a portion thereof. Second, output training data is received, where the output data includes performance values ​​of the solar cell. The output training data is correlated with the input training data, meaning that each output training data is associated with (in other words: correlated) the input training data. In the third step, the function is trained based on the input and output training data. This leads to the fourth and final step: providing the so-called trained function.

[0032] Descriptively speaking, a function is trained by providing a set of input / output training data. Once training is complete, the function is able to predict output data given unknown input data, with the help of the training.

[0033] Furthermore, the present invention also relates to a providing system comprising: a first interface configured to receive input data, wherein the input data includes data from electrical impedance spectroscopy performed on a perovskite-based solar cell or a portion thereof; a second interface configured to provide output data, wherein the output data includes performance values ​​of the solar cell; and a computing unit configured to apply a trained function to the input data, wherein the output data is generated.

[0034] All implementation methods and variations already discussed in the context of computer-implemented methods also apply to the mentioned providing system. For the sake of brevity, they will not be repeated verbatim here.

[0035] Finally, the present invention also relates to a computer-readable medium comprising instructions that, when a program is executed by a providing system, cause the providing system to perform a method according to one of the above embodiments. Attached Figure Description

[0036] The invention will be described below with the aid of the accompanying drawings, in which...

[0037] Figure 1 The Nyquist plots of three perovskite-based solar cells are shown. Detailed Implementation

[0038] Figure 1 The real and imaginary parts of the impedance in frequency relation are shown in the form of a Nyquist plot. The real part of the impedance in ohms is plotted at x-axis 10, and the imaginary part of the impedance in ohms is plotted at y-axis 20. The first Nyquist plot 31 and the second Nyquist plot 32 are generated by impedance spectral measurements performed on a first-type perovskite solar cell. The first-type perovskite solar cell consists of two distinct cells; the first Nyquist plot 31 belongs to the first cell, and the second Nyquist plot 32 belongs to the second cell, but both the first and second cells have been processed using the same technology. The third Nyquist plot 33 is generated by impedance spectral measurements performed on a third perovskite solar cell. The third solar cell is processed using a significantly different technology. Therefore, the third solar cell can be classified as a second-type perovskite solar cell.

[0039] The first solar cell is characterized by high performance, thus resulting in a good selective performance value. This selective performance value can be, for example, the solar cell's light conversion efficiency. The second solar cell is characterized by low performance, resulting in a poor performance value (in this example, a low light conversion efficiency).

[0040] By comparing the first Nyquist plot 31 and the second Nyquist plot, it can be seen that significantly different Nyquist plots were obtained for two solar cells belonging to the same type but exhibiting different performance. In other words, the difference in performance values ​​(cell efficiency in this example) is already visible in the Nyquist plot obtained from EIS measurements. A key aspect of the invention is that the trained function will be able to identify (and thus predict) the poor performance of the second solar cell solely from the Nyquist plot.

[0041] However, Figure 1As shown, caution is required when comparing EIS measurements from different cell types. The third Nyquist plot 33 belongs to a different cell type compared to the first plot 31 and the second plot 32. Therefore, a different function is applied when predicting the performance values ​​of the third solar cell, and this function needs to be trained separately.

[0042] Essentially, this invention allows for the prediction of perovskite-based solar cell performance based on EIS measurements, whereas the prediction of perovskite-based solar cell performance would otherwise only be achieved through complex and expensive IV measurements on a solar simulator. The predicted performance can be valuable in several ways:

[0043] The results obtained can be used to select and classify cells during manufacturing. This classification is useful for constructing photovoltaic modules with matched cells in order to optimize the performance of individual modules.

[0044] The results obtained can also be used to optimize the production process itself. Measurements can be performed directly after intermediate processing steps (i.e., before the battery is fully processed). Compared to many research tools, EIS is non-destructive (requiring only contact with two electrodes). Therefore, on the one hand, the results obtained can be used as feedback to adjust the processing machine to optimize the processing of the next batch. On the other hand, the results can be used as feedforward for subsequent processing machines to compensate for deviations from previous processing steps and thus “correct” the most recent batch. In particular, for series-connected batteries that require even more processing steps (sometimes completed at different manufacturers and shipped between manufacturers), EIS is a valuable tool for inspecting incoming pre-processed parts before the start of processing at the next site.

[0045] Another application of this invention relates to monitoring perovskite-based solar cells during operation. As is known, a challenge of perovskite-based solar cells involves their potential degradation over time. This method is also applicable to assessing cell performance degradation over time independently of illumination conditions. While aging mechanisms are known, achievable stability during operation is not fully understood. For these reasons, perovskite-based solar cells may exhibit varying degrees of degradation over time, leading to a mismatch in cell performance within a module or power plant. This mismatch can be quantified using the described method to estimate the overall loss in production. The anticipated loss can be partially compensated for by adjusting the operation of the solar cells in consideration of EIS measurements, to, for example, avoid complete degradation or destruction of the cells.

[0046] In existing photovoltaic (PV) sites, the performance of cells or modules over time can be measured. When using EIS, these measurements do not require a solar simulator with explicitly defined lighting conditions, but can be performed at any appropriate time (e.g., at night) without disrupting normal daily operations. The acquired data can be analyzed online and presented to stakeholders in real time.

[0047] In summary, an attractive concept for predicting one or more performance values ​​of perovskite-based solar cells has been proposed. This concept can be used not only to monitor and optimize the manufacturing process of solar cells, but also, for example, to monitor the performance of perovskite-based solar cells during their field operation.

[0048] List of reference numerals

[0049] 10 x-axis

[0050] 20 ordinates

[0051] 31 First Nyquist Diagram

[0052] 32 Second Nyquist Diagram

[0053] 33 Third Nyquist Diagram

Claims

1. A computer-implemented method for providing output data, wherein, The output data includes performance values ​​of the solar cell, and the solar cell comprises perovskite material. The method includes the following steps: - Receive input data, wherein the input data includes electrical impedance spectroscopy data performed on the solar cell during the manufacturing process prior to the completion of processing of the perovskite-based solar cell. - The trained function is applied to the input data, wherein the output data is generated, and the trained function is a multivariate statistical model or a neural network model, and - Provide the output data.

2. The method according to claim 1, wherein, Data from the electrical impedance spectrum includes the impedance, admittance, modulus function, complex permittivity, and / or dielectric constant of the solar cell.

3. The method according to any one of claims 1-2, wherein, The data from the impedance spectrum includes data derived from modeling the electrical equivalent circuit based on the impedance spectrum performed on the solar cell.

4. The method according to claim 3, wherein, The resulting data includes the series resistance, capacity, or Warburg impedance of the solar cell.

5. The method according to any one of claims 1-2, wherein, The performance values ​​of the solar cell are selected from the group consisting of: short-circuit current, open-circuit voltage, maximum power point voltage, maximum power point current, maximum power, fill factor, and cell efficiency.

6. The method according to any one of claims 1-2, wherein, The input data also includes at least one process parameter related to the processing of the solar cell.

7. The method according to claim 6, in, The process parameters are selected from the group consisting of: - The physical or chemical properties of at least one of the substances used in the processing of the solar cell. - Deposition parameters during the deposition process of the elements in the solar cell, and - The physical or chemical properties of the elements of the solar cell.

8. The method according to any one of claims 1-2, wherein, The solar cell includes a photosensitive layer, and the photosensitive layer contains the perovskite material.

9. The method according to any one of claims 1-2, wherein, The impedance spectrum is performed at frequencies in the range of 100 mHz to 1 MHz.

10. The method according to any one of claims 1-2, wherein, The impedance spectrum is performed at a finite number of predefined frequencies.

11. A method for providing a computer implementation of a trained function, comprising: - Receive input training data, wherein the input data includes exemplary data from electrical impedance spectroscopy applied to a solar cell, wherein the solar cell comprises a perovskite material, and wherein the electrical impedance spectroscopy is performed during the manufacturing process prior to the completion of processing of the perovskite-based solar cell. - Receive output training data, wherein the output training data is related to the input training data, and wherein the output training data includes the performance values ​​of the solar cell. - Train the function based on the input training data and the output training data, and - Provides a trained function, which is a multivariate statistical model or a neural network model.

12. A system for providing output data, comprising: - A first interface, configured to receive input data, wherein... The input data includes data from electrical impedance spectroscopy performed on a solar cell, wherein the solar cell comprises a perovskite material, and wherein the electrical impedance spectroscopy is performed during the manufacturing process prior to the completion of processing of the perovskite-based solar cell. - A second interface, configured to provide output data, wherein the output data includes performance values ​​of the solar cell, and - A computing unit configured to apply a trained function to the input data, wherein the output data is generated, and the trained function is a multivariate statistical model or a neural network model.

13. A computer-readable medium comprising instructions that, when executed by a providing system, cause the providing system to perform the method according to any one of claims 1 to 10.

Images