Quantitative evaluation of thermal runaway tolerance in space solar cells

To evaluate the critical solar light intensity for the thermal runaway of thin-film multi-junction solar cells on a solar panel of a spacecraft, a coupon panel was irradiated with simulated solar light under vacuum and at low temperature. We prepared a coupon panel that was equipped with InGaP/GaAs/InGaAs IMM tripe-junction (IMM3J) solar cells in cell-interconnect-coverglass (CIC) assemblies mounted on an aluminum honeycomb panel. The thickness of the IMM3J cell is ∼30 µm. The solar cell array was formed by three IMM3J solar cells connected in series. The test configuration is shown in Fig. 1. The coupon panel was placed in a vacuum chamber with a shroud cooled down to −170 °C and illuminated through a lens by an air mass zero (AM0)²³⁾ solar simulator (11 suns at maximum) located outside the chamber.²⁴⁾ The chamber can simulate the dawn environment, which is likely to cause a strong temperature gradient in the cell plane.²⁵⁾ In order to simulate the environment in Earth's orbit, the coupon panel was thermally insulated from the chamber, and was cooled only by radiation. The temperature distribution was observed using a thermo-camera through an infrared-transmitting window, and the absolute temperature of the cell was measured using a thermocouple attached to the coupon panel. The array was kept under open-circuit condition, and the open-circuit voltage (V_oc) was monitored. The light intensity of the solar simulator was adjusted with a power meter in the chamber. The panel temperature before light irradiation was −80 °C, and the irradiation duration was 10 min. The results of this experiment are explained in Sect. 3.1.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The thin-film multijunction solar cells on a solar panel were irradiated with solar light and a laser beam with variable intensity at room temperature under atmosphere and vacuum. Furthermore, to obtain the quantitative ΔT_th for the thermal runaway of the solar cells under space environment, the coupon panel was irradiated with solar light and the laser beam under vacuum and low temperature. Because the laser is located outside the chamber, a window with transparency for the laser light was installed in the chamber as indicated in Fig. 1. The solar cell in the center of the array was excited with a laser beam of 532 nm wavelength to create an artificial shunt spot by heat generation. Prior to the laser beam experiment, the location of the laser spot was confirmed by observing the photoluminescence (PL) emission of an InGaP subcell due to the laser excitation. The array was irradiated simultaneously with both the simulated solar light and the laser beam after the panel temperature reached room temperature or −80 °C. The AM0 solar light intensity was fixed at 1 sun and the other conditions were the same as those outlined in Sect. 2.1. The results of this experiment are explained in Sect. 3.2.

Figure 2 indicates the time evolution of the array's V_oc when it was exposed to the solar light under vacuum and at low temperature. In general, the V_oc decreases as time proceeds, which is assigned to the temperature increase of the solar cells and the panels. The V_oc and panel temperature saturated within about 10 min, and we consider that these values represent the steady-state condition. It is interesting to note that under illumination with 1.6 suns, a sharp voltage drop occurred after continuous illumination for about 30 s. Such a characteristic drop indicates a thermal runaway. The V_oc dropped by about 1.5 V within a time interval of about 10 s. The start and end of the interval are indicated in Fig. 2 with t₁ and t₃, respectively. The time t₂ indicates a point during the fast drop of the voltage. The corresponding temperature distributions on the cells' surface are shown with the thermo-images in the inset of Fig. 2. The dashed lines indicate the position of the solar cell array. The thermal runaway occurred in the vicinity of the n-side electrode pad, and the temperature reached more than 100 °C. After t₃, the shunt current reached its steady-state condition, and it was maintained as long as the intense illumination proceeded. After the cooling of the coupon panel in the dark, the sample recovered and the V_oc of the array returned to its normal value (initial values shown with the curves for 0.7 to 1.3 suns) upon irradiation with the solar simulator. Thus, the cell suffered no significant damage caused by the thermal runaway under this condition. In the case of conventional "thick" triple-junction solar cells (∼160 µm), thermal runaway was reported to occur at a light intensity of 6 suns under ambient environment (standard atmosphere and room temperature).^3,4) In contrast, our experiment verified that the IMM3J solar cells without a substrate have poor thermal runaway tolerance because thermal runaway occurred already at 1.6 suns under vacuum and at low temperature. However, the observed critical excitation intensity depends on the characteristics of the shunt spot. The thermal runaway may occur in other samples from the same batch even if the light intensity is 1 sun.

Zoom In Zoom Out Reset image size

The usual approach for a quantitative evaluation of ΔT_th for the thermal runaway of a cell (under a certain solar intensity) requires clarification of the size, number, resistance, and position of shunt spots. However, it is difficult to investigate these characteristics of shunt spots in advance. Actually, we recorded an electroluminescence (EL) image^26–28) of the coupon panel under ambient environment prior to the test to locate areas that potentially become shunt spots, but no implications such as bright spots were confirmed at the position where thermal runaway occurred. Since EL is observed under external bias, the electric-field distribution is different from that under solar irradiation.

Therefore, in order to quantitatively evaluate ΔT_th for the thermal runaway of a cell, a more sophisticated approach is required. We propose an evaluation method using an arbitrary artificial shunt spot induced by a laser beam, which is discussed in the following section.

Figure 3 presents an EL image (injection current of 400 mA) of the solar cell array in the chamber. A filter was chosen to select the EL emitted from the IMM3J solar cells' InGaP layers. The bright spot surrounded by the red broken line indicates PL emission from the InGaP layer due to the laser excitation. The area of the laser spot was about 6 mm², which is equivalent to the area of the n-side electrode pad where the thermal runaway occurred under solar irradiation with 1.6 suns (shown in Fig. 2). The laser beam was focused at the corner of Cell 2 near the bus bar electrode (see Fig. 3), where heat dissipation should be lesser than in the inside of the cell and current easily concentrates compared with in the region near the finger electrode.

Zoom In Zoom Out Reset image size

Figures 4(a)–4(c) provide the transient V_oc during the simultaneous excitation with simulated AM0 solar light and the laser beam irradiation under atmosphere at room temperature, under vacuum at room temperature, and under vacuum at low temperature, respectively. The laser intensity absorbed by the cell was calculated from the reflectance of the IMM3J CIC and the measured laser power with a power meter in the chamber. Since the external luminescence efficiency of the EL from the InGaP subcell is about 1%, it can be considered that almost all absorbed laser power is converted to thermal energy. In order to investigate the threshold laser power for thermal runaway, the laser intensity was gradually increased. Considering the smooth changes in V_oc obtained for excitations less than 1413 mW under atmosphere at room temperature, 1304 mW under vacuum at room temperature, and 977 mW under vacuum at low temperature, ΔT_th was not reached. However, at the laser intensity of 1849 mW under atmosphere at room temperature, 1413 mW under vacuum at room temperature, and 1195 mW under vacuum at low temperature, thermal runaway was observed after continuous laser excitation at about 4, 10, and 12 s, respectively. The final temperature distributions after 600 s of illumination for the laser intensity of 1195 mW under vacuum at low temperature is shown with a thermo-camera image in the inset of Fig. 4(c). The image shows evidence that thermal runaway occurred at the laser irradiation spot. We confirmed that the solar cell suffered no damage by the thermal runaway under this condition. The laser intensity required to induce thermal runaway under vacuum or at low temperature was lower than that under atmosphere or at room temperature. The reasons for the difference of the laser intensity inducing thermal runaway are considered to be as follows heat dissipation in vacuum is significantly lower than that in the atmosphere, and the thermal resistance of solar cells is higher at low temperature than at room temperature. In the following, analysis using the experimental results under vacuum and low temperature, which is the condition in a simulated space environment, is explained.

Zoom In Zoom Out Reset image size

The temperatures of the cell and the shunt spot are essential for thermal analysis, but it is difficult to measure the temperature directly. If a thermocouple is attached to the laser spot, the thermocouple itself becomes a thermal path and therefore the thermal characteristics of the shunt spot cannot be evaluated correctly. Also, since the thermo-camera cannot detect temperatures below −20 °C, the temperature of the spot cannot to be evaluated for low excitation powers or short light exposure duration at low temperature. Therefore, in the present work, the temperatures of the cell and the laser irradiated area were calculated from the SPICE²⁹⁾ simulation using the measured V_oc, which changes with increasing laser beam irradiation duration.

Figure 5 depicts the equivalent circuit of the array. Cell 2, which was irradiated with the laser beam, is defined with a local shunt spot with an area of 6 mm². The total area of the IMM3J cell is 27.4 cm². In addition, a series resistance (R_s) is introduced in Cell 2, to define the current flow from the entire cell to the shunt spot. We choose R_s = 0.5 Ω, which is a typical value of our IMM3J solar cells. The temperature of the cells [T_cell(t)] is assumed to be independent of the laser power. The temperature dependence of the reverse saturation current of the diode in the area excluding the laser spot was obtained from the V_oc data recorded for a laser intensity of 170 mW under vacuum and low temperature, which was sufficiently small not to affect V_oc. Since the temperature dependence of the optical absorption is known to be negligibly small, the optical generation current was set to be constant at 450 mA. The generation current for the shunt spot was calculated from the area ratio, which results in 1 mA. This current corresponds to that determined by the current-limiting subcell under solar excitation with AM0 1 sun. Using the equivalent circuit, we obtained the temperature [T_spot(t)], inflow current [I_spot(t)] and voltage [V_spot(t)] of the shunt spot by fitting the measured V_oc with the reverse saturation current of the shunt spot.

Zoom In Zoom Out Reset image size

A thermal network model was constructed by using an analogy between transient heat-flow and electrical networks.³⁰⁾ Figure 6 schematically shows three thermal resistances (R_{t_rad}, R_{t_panel_1}, and R_{t_cell}) around the artificial shunt spot. R_{t_rad} is the thermal resistance for the heat dissipation via radiation, R_{t_panel_1} couples the spot to the aluminum honeycomb panel via the adhesive, and R_{t_cell} is the thermal resistance along the cell's in-plane direction. Since the heat dissipation under vacuum was negligibly small compared with the heat generation by the laser beam, we assumed R_{t_rad} to be infinite. The panel temperature under the shunt was unknown, and therefore we defined the thermal network in terms of T_cell. Figure 7 illustrates the thermal network and includes a thermal resistance between the panel and the cell (R_{t_panel_2}), a thermal capacity (C_t) between the shunt spot and the cell, a voltage source representing temperature, and a current source representing thermal flow. The heat can be dissipated to the panel and to the cell itself by the thermal resistances indicated. The thermal resistance between the shunt spot and the cell is unaffected by the laser beam and can be expressed with a combined thermal resistance,

$$\begin{equation} R_{\text{t}} = \frac{R_{\text{t_cell}}(R_{\text{t_panel_1}} + R_{\text{t_panel_2}})}{R_{\text{t_cell}} + R_{\text{t_panel_1}} + R_{\text{t_panel_2}}}. \end{equation} \tag{ 1 }$$

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In our model, the time dependence of the temperature difference between the shunt spot and the cell area that was not affected by the laser beam [ΔT(t)] can be expressed as

$$\begin{align} \Delta T(t) &= T_{\text{spot}}(t) - T_{\text{cell}}(t) \\ &= R_{\text{t}}(Q_{\text{laser}} + Q_{\text{cell}}(t))\left[1 - \exp\left(-\frac{t}{R_{\text{t}}C_{\text{t}}}\right)\right], \end{align} \tag{ 2 }$$

$$\begin{align} Q_{\text{cell}}(t) &= I_{\text{spot}}(t)\cdot V_{\text{spot}}(t). \end{align} \tag{ 3 }$$

Here, Q_laser and Q_cell represent the heat generated by the laser and that generated by the inflow photocurrent from the surrounding cell into the shunt spot, respectively. Q_cell is the product of the inflow photocurrent and the voltage of Cell 2 as expressed in Eq. (3). The limit of Eq. (2) can be expressed as

$$\begin{equation} \lim_{t\to\infty}\Delta T = R_{\text{t}}[Q_{\text{laser}} + Q_{\text{cell}}(t)], \end{equation} \tag{ 4 }$$

which means that ΔT(t) is determined by the product of the thermal resistance and thermal flow.

Figure 8 shows the calculated I_spot and V_spot (= V_Cell2) of Cell 2 under vacuum and low temperature obtained from the experimental results and the equivalent circuit shown in Figs. 4(c) and 5, respectively. For the laser irradiation with 759 and 977 mW, the voltage decrease due to laser irradiation is relatively low, and the inflow current into the beam spot is also low. On the other hand, for a strong laser excitation of 1195 mW, the voltage starts to decrease significantly around t = 7 s, and accordingly, the inflow current increases exponentially. At t = 13 s, the total photogeneration current in Cell 2 is concentrated at the shunt spot, and the inflow current saturated.

Zoom In Zoom Out Reset image size

Figure 9 plots the temperature difference between the shunt spot and the cell area that was not irradiated under vacuum and low temperature. The closed circles are the calculated data obtained from the measured V_oc using the equivalent circuit shown in Fig. 5 and the curves are calculation results based on the thermal network shown in Fig. 7. The solid lines were the calculated results using Eq. (2). From the fitting, we obtained R_t = 22 °C/W and C_t = 0.2 J/°C. Q_cell(t) was calculated from Fig. 7 and Eq. (3). ΔT could be fitted well by the thermal model for laser intensities of 759 and 977 mW. However, although ΔT(t) was also fitted well until t = 12 s under laser excitation with 1195 mW, the observed peculiar phenomenon of the rapid temperature increase between t = 13 and 20 s could not be reproduced. The calculated curve shows saturation at around 50 °C because the inflow current into the shunt spot saturates at t = 13 s as shown in Fig. 8, i.e., 50 °C seems to be the upper limit of ΔT(t) for R_t = 22 °C/W. Since Q_laser and Q_cell (t > 13 s) were constant, we consider that the rapid temperature increase reflects the change in R_t due to thermal runaway. To include the upper and lower limits of R_t, Eq. (2) is rewritten as

$$\begin{align} \Delta T& = R_{\text{t}}Q_{\text{laser}}\left[1 - \exp\left(-\frac{t}{R_{\text{t}}C_{\text{t}}}\right)\right] \\ &\quad + R_{\text{t_TR}}Q_{\text{cell_TR}}\left[1 - \exp\left(-\frac{t - t_{\text{TR}}}{R_{\text{t_TR}}C_{\text{t_TR}}}\right)\right]. \end{align} \tag{ 5 }$$

Zoom In Zoom Out Reset image size

Here, R_{t_TR} and C_{t_TR} indicate the thermal resistance and capacity after the thermal runaway saturated, respectively. t_TR is the continuous illumination duration that is required to initiate thermal runaway. The dashed line in Fig. 9 represents ΔT(t) for R_{t_TR} and C_{t_TR} being equal to 120 °C/W and 0.01 J/°C, respectively. The calculated curve agrees well with ΔT(t) obtained from the experiment, and thus we determine the ΔT_th of about 50 °C under this experimental condition.

The above analysis clarified that before thermal runaway occurs, the temperature of the shunt spot can be modeled by the steady state analysis using Eq. (2), but after transition into the state with strong current inflow and heating, Eq. (5) is required for accurate description. In other words, thermal runaway is a phenomenon in which the thermal resistance of the shunt spot increases when the spot temperature difference reaches the threshold value. The reason why thermal resistance change is considered that the diameter of the spot where photocurrent flows decreases. It is possible to prevent the thermal runaway when the thermal resistance is designed to be less than the ratio between ΔT_th for the shunt spot and the maximum heat generation. Our proposed method helps to identify the ΔT_th quantitatively at the arbitrary shunt spot under the simulated space environment, which is applicable to the design of reliable satellite power supply systems.