Impact of Separator Thickness on Temperature Distribution in Single Cell of Polymer Electrolyte Fuel Cell Operated at Higher Temperature of 90 °C and 100 °C

Abstract

The New Energy and Industry Technology Development Organization (NEDO) road map (Japan, 2017) has proposed that a polymer electrolyte fuel cell (PEFC) system, which operates at a temperature of 90 °C and 100 °C, be applied for stationary and mobility usage, respectively. This study suggests using a thin polymer electrolyte membrane (PEM) and a thin gas diffusion layer (GDL), at the same time, to achieve better power-generation performance, at a higher temperature than usual. The focus of this paper is to clarify the effect of separator thickness on the distribution of temperature at the reaction surface (T_react), with the relative humidity (RH) of the supply gasses and initial operation temperature (T_ini), quantitatively. In this study, separator thickness is investigated in a system using a thin PEM and a thin GDL. Moreover, this study investigates the difference between the maximum temperature and the minimum temperature obtained from the distribution of T_react as well as the relation between the standard deviation of T_react − T_ini and total voltage, to clarify the effect of separator thickness. The impact of the flow rates of the supply gases on the distribution of T_react is not large, among the investigated conditions. It is noticed that the temperature distribution is wider when a separator thickness of 2.0 mm is selected. On the other hand, it is observed that the temperature increases along with the gas flow through the gas channel, by approximately 2 °C, when using a separator thickness between 1.5 mm and 1.0 mm. The impact of the RH on the distributions of T_react − T_ini is larger at T_ini = 100 °C, when a separator thickness of 1.0 mm is selected. It is revealed that the wider temperature distribution provides a reduction in power-generation performance. This study proposes that the thin separators, i.e., with a thickness of 1.5 mm and 1.0 mm, are not suitable for higher temperature operation than usual.

1. Introduction

It is thought that renewable H₂ can be considered to be one procedure to solve global warming. Polymer electrolyte fuel cell (PEFC) is a popular application, to use H₂ as a fuel, which can generate power and heat. In Japan, the New Energy and Industrial Technology Development Organization (NEDO)’s road map (Japan, 2017) [1] has declared that a polymer electrolyte fuel cell (PEFC) should be operated at 90 °C and 100 °C for stationary and mobility application, respectively, which is the target from 2020 to 2025. However, the normal PEFC, using a polymer electrolyte membrane (PEM), such as Nafion, is usually operated within the temperature range between 60 °C and 80 °C [2,3,4]. If we operate the PEFC at a relatively higher temperature than usual (i.e., 90 °C and 100 °C, as mentioned above), we can obtain the following merits: (i) promotion of the electro-chemical kinetics of both electrodes; (ii) reduction in the cooling system because of the larger temperature difference between the PEFC stack and coolant; and (iii) improvement of the endurance to CO, which is available for reformed H₂, including CO [5]. However, we have to consider the following problems, if we operate PEFC at a higher temperature than usual: (i) degradation of the PEM; (ii) erosion of the catalyst; and (iii) uneven distributions of gas flows, pressure, temperature, voltage, and current in the PEFC. We have to solve these problems, in order to commercialize a PEFC system that is operated at a relatively high temperature [6]. In addition, we think temperature distribution at high temperature, also, influences the phase change of water, the performance of PEM, the fuel and oxidant flow characteristics in the gas diffusion layer (GDL), and the catalyst layer. Consequently, it is necessary to analyze the temperature distribution in a single PEFC, to improve the power-generation performance, for achieving a longer operational life span for the PEFC.

Recently, several studies have investigated the characteristics of a higher operating temperature PEFC (HTPEFC), over 100 °C [7,8,9,10,11,12,13,14,15,16]. However, the targets for most of them are the R&D of new materials, which can work at a higher temperature, e.g., the membrane and catalyst [7,8,9,10,11,12,13,14,15,16]. Several numerical studies have investigated the temperature-driven water transport in a cathode GDL [17]; the optimization of catalyst-layer thickness, to obtain high performance without high cost [18]; assessment of mass transport impact on the performance by 3D multi-physics modeling [19]; optimization of GDL, focusing the thickness of porosity by a non-isothermal 3D model [20]; and dynamic simulation for the start-up process by a non-isothermal 3D model [4]. In addition, several numerical studies [21,22,23] have investigated the analysis of aseparator, to optimize the gas channel, considering the widths of the top and bottom, e.g., the cross-sectional area with a trapezium shape [21],the clarification of power-generation performance, reactant and water-saturation distribution, using oriented-type flow channels with porous-blocked baffles [22], and the impact of the channel width/rib width ratio on the performance of and mass distribution in the cell [23]. A few research studies [4,24,25] have reported the numerical simulation on temperature distribution in a single PEFC, which is operated at a relatively higher temperature than usual. However, only a few papers [4,24] have investigated the temperature distribution near the interface between the PEM and catalyst layer at the cathode, which is defined as a reaction surface in this study, excluding other studies by the authors [26,27,28,29]. The authors’ studies [26,27,28,29] have investigated the effect of PEM’s thickness and GDL’s thickness on the distribution of the temperature at the reaction surface (T_react), in a cell of the PEFC at higher temperature, i.e., 90 °C or 100 °C, by a model using the experimental temperature-distribution data obtained by means of a thermograph. However, the effect of separator thickness, including the saddle thickness and channel height, on the distribution of T_react has not been investigated yet. The separator thickness influences the weight, volume, and cost of the PEFC stack, since the commercial PEFC stack is composed of several hundreds of cells. Therefore, it is necessary to understand the effect of separator thickness on the distribution of T_react, in order to commercialize the PEFC stack operated at a higher temperature than usual.

The aim of this study is to analyze the effect of separator thickness on the distribution of T_react in a single PEFC, with the changing flow rates and relative humidity (RH) of the supply gases as well as the initial operation temperature (T_ini), from the general operation temperature of 80 °C to a higher temperature, i.e., 90 °C and 100 °C. The combination of thin PEM and thin GDL is adopted for analysis, since previous studies by the authors [26,27,28,29] have clarified that using thin PEM and thin GDL at the same time is effective to attain higher power-generation performance of a PEFC operated at a higher temperature than usual. The temperature distributions on the separator’s back, of a cell of the PEFC, have been measured by means of a thermograph and used as the boundary-condition data for the heat-transfer model developed by the authors [30,31]. Since this procedure can measure the temperature distribution, with no disturbance of the heat and mass transfer phenomena as well as the power-generation performance, due to sensor installation, it can be said that the temperature distribution under a power-generation condition, with loaded current, is measured accurately. Using the measured data, the previous studies [29,32,33] have developed to build an empirical model, in order to predict the distribution of T_react. As the authors have reviewed the literature, it has been confirmed that there is no previous study estimating the distribution of T_react from the measured temperature data at the separator’s back. The heat-transfer model can predict the distribution of T_react, using the measured separator’s back temperature, and it is preferred to be analyzed by the proposed model. The distribution of T_react can be easily estimated, without difficult and complex temperature measurement, due to the presented heat-transfer model.

According to the previous studies carried out by the authors [26,27,28,29], a 1D multi-plate heat-transfer model, which uses the temperature data of the separator’s back, measured by the thermograph under a power-generation condition, has been presented. A single PEFC is composed of the PEM, catalyst layer, MPL, GDL, and separator. The heat-transfer model developed by the authors [26,27,28,29] has assumed that the heat flows through multi-plates of these cell components. The temperature at the interface between the PEM and the catalyst layer on the cathode, i.e., T_react, is calculated by the heat-transfer model [26,27,28,29]. This is a new procedure to clarify the heat-transfer mechanism in single PEFC, using measured data by the thermograph for the developed model.

2. Heat-Transfer-Analysis Procedure of Single Cell of PEFC

2.1. 1D Multi-Plate Heat-Transfer Model Developed by This Study

Figure 1 shows the multi-plate single PEFC. The separator’s back indicates the opposite side of the surface is contacting with the GDL. The separator’s back surface temperatures, T_{surf, c} and T_{surf, a}, are measured by means of a thermograph. This model assumes that the heat transfer across the module can be thought to be in a 1D direction only. In this module, the cell is divided into a gas channel and a rib part. In addition, the upper and lower part indicate the rib part and channel part, respectively. This study assumes that the heat transfers toward the through-plane direction, for both parts.

The heat generated on the reaction surface is thought to be transferred to the cathode and the anode sides, separately. As the gas flowing through with the gas channel, from the inlet to the outlet of the cell, carries away the heat, the amount of heat taken by the gas flow is less than 1% of the calculated reaction heat, approximately 20 W [34]. Consequently, this study assumes that the heat carried away by the gas flow can be neglected in this model. In addition, the mass flow rate of gas flowing through the gas channel is very small, which ranges from 10⁻⁸ to 10⁻⁶ kg/s, resulting in this study being able to assume the thermal conduction of gas is occurring only in the gas channel.

2.2. Heat-Generation Rate by Electrochemical Reaction in the Model Proposed by This Study

The heat generation rate H_react, according to electrochemical reaction, is calculated as follows:

H_react= E_i − W_E

(1)

where E_i is the ideal energy-generation rate from the water formation by H₂ and O₂, based on the higher heating value (HHV), excluding T_ini = 100 °C. This study adopts the lower heating value (LHV) in case of T_ini = 100 °C. W_E is the electric power generated by a single PEFC. E_i and W_E are expressed as follows:

E_i = m_H2 × q_HHV or q_LHV

(2)

W_E = I × V

(3)

where I is the load current obtained by the power-generation experiment. I is kept at 20 A (=0.80 A/cm²). V is the voltage obtained by the power-generation experiment. m_H2 is the molar flow rate of supplied H₂, which equals the ideal reaction consumption rate of H₂, needed for the generation of 20 A, i.e., at the stoichiometric ratio (s.r.) of 1.0. Here, s.r. is the ratio of the feed amount of H₂ and O₂ needed to generate the current of 20 A. The consumption rate of supplied H₂ at s.r. of 1.0 can be defined as follows:

m_H2 = I/nF

(4)

where m_H2 is the molar consumption rate of supplied H₂ with s.r. of 1.0 [mol/s], n is the valence ion (=2 for H₂), F is the Faraday constant (=96,500 C/mol), and m_O2 is the molar consumption rate of supplied O₂ with s.r. of 1.0 [mol/s], which is calculated as follows:

H₂ + 1/2 O₂ = H₂O

(5)

where m_H2 and m_O2 at 20 A are 1.04 × 10⁻⁴ mol/s and 0.52 × 10⁻⁴ mol/s, respectively.

The actual s.r. for the consumption gas is confirmed, by means of the mass flow controller, which is installed at the inlet of single PEFC, and the mass flow meter, which is installed at the outlet of the cell, in the power-generation experiment [30,31]. The consumption rates of supplied H₂ and O₂ are both s.r. of 1.0, which is confirmed by the experiment.

2.3. Heat-Balance Formulas to Calculate the Reaction Surface Temperature

The reaction heats at the rib and channel are expressed as follows:

H_{rib, c} = K_{rib, c}A(T_{react, rib} − T_{surf, c})/2

(6)

H_{chan, c} = K_{chan, c}A(T_{react, chan} − T_{surf, c})/2

(7)

H_{rib, a} = K_{rib, a}A(T_{react, rib} − T_{surf, a})/2

(8)

H_{chan, a} = K_{chan, a}A(T_{react, chan} − T_{surf, a})/2

(9)

H_react = H_{rib, c} + H_{chan, c} + H_{rib, a} + H_{chan, a}

(10)

where H means the heat flux [K]; K means the overall heat-transfer coefficient [W/(m²·K)]; A means the heat-transfer area, which is the active area of MEA, i.e., the power-generation area (=0.0025 m²); and T means the temperature [K or °C]. As for the subscripts, rib, chan, react, surf, c, and a mean under rib, under chan, reaction surface, separator’s back surface, cathode side, and anode side, respectively. K_{rib, c}, K_{chan, c}, K_{rib, a}, and K_{chan, a} are defined as follows:

1/K_{rib, c} = δ_cat/k_cat + δ_MPL/k_MPL + δ_GDL/k_GDL + δ_rib/k_rib + δ_sep/k_sep

(11)

1/K_{chan, c} = δ_cat/k_cat + δ_MPL/k_MPL + δ_GDL/k_GDL + δ_chan/k_{chan, c} + δ_sep/k_sep

(12)

1/K_{rib, a} = δ_PEM/k_PEM + δ_cat/k_cat + δ_MPL/k_MPL + δ_GDL/k_GDL + δ_rib/k_rib + δ_sep/k_sep

(13)

1/K_{chan, a} = δ_PEM/k_PEM + δ_cat/k_cat + δ_MPL/k_MPL + δ_GDL/k_GDL + δ_chan/k_{chan, a} + δ_sep/k_sep

(14)

where δ_cat means the thickness of the catalyst layer [m], k_cat means the thermal conductivity of the catalyst layer [W/(m·K)], δ_MPL means the thickness of MPL [m], k_MPL means the thermal conductivity of MPL [W/(m·K)], δ_GDL means the thickness of GDL [m], k_GDL means the thermal conductivity of GDL [W/(m·K)], δ_rib means the thickness of the separator rib [m], k_rib means the thermal conductivity of the separator rib [W/(m·K)], δ_sep means the thickness of the separator except for the separator rib [m], k_sep means the thermal conductivity of the separator except for the separator rib [W/(m·K)], δ_chan means the thickness of the separator channel [m], k_chan means the thermal conductivity of the mixture gas in the separator channel [W/(m·K)], δ_PEM means the thickness of PEM [m], and k_PEM means the thermal conductivity of PEM [W/(m·K)].

Table 1. Specification list of polymer electrolyte fuel cell components, according to the manufacturer’s catalog and previous studies [26,27,28,29,30,31].

Components	Dimension	Characteristics	Porosity [-]	Effective Thermal Conductivity [W/(m·K)]
Polymer electrolyte membrane (PEM)	50.0 mm × 50.0 mm × 0.025 mm	NRE-211 (produced by Du Pont Corp.)	0.28	0.195
Catalyst layer	50.0 mm × 50.0 mm × 0.01 mm	Pt/C (20 wt% Pt loading)	0.78	0.27
Micro-porous layer (MPL)	50.0 mm × 50.0 mm × 0.003 mm (attached with PEM)	Carbon black + PTFE	0.60	1.0
Gas-diffusion layer (GDL)	50.0 mm × 50.0 mm × 0.11 mm	Carbon paper (TGP-H-030 produced by Toray Corp.)	0.78	1.7
Separator	75.4 mm × 75.4 mm × 2.0 mm or 1.5 mm or 1.0 mm (2.0 mm = saddle thickness of 1.0 mm and channel height of 1.0 mm; 1.5 mm = saddle thickness of 0.5 mm and channel height of 1.0 mm; 1.0 mm = saddle thickness of 0.5 mm and channel height of 0.5 mm) (gas supply area: 50.0 mm × 50.0 mm)	Carbon graphite, serpentine	0.15	25

lists the specifications of the cell components applied in the heat-transfer model. Nafion NRE-212 (manufactured by Du Pont Corp.), with a thickness of 25 μm, is investigated in this study. In addition, TGP-H-030 (manufactured by Toray Corp.), with a thickness is 110 μm, is investigated. The proposed model is composed of the PEM, catalyst layer, MPL, GDL, and separator. The values shown in Table 1 are the same as those of the components applied in the previous studies [26,27,28,29].

Table 1 lists the effective thermal conductivities of porous media k too. They are inferred from the values of the cell components applied in the power-generation experiment carried out by the authors [31]. In this table, the effective thermal conductivities are listed, considering that the cell-components’ pores are filled with air at room temperature. This study has calculated the corrected effective thermal conductivities, assuming that the cell components’ pores are filled with H₂ or O₂ at 80 °C, 90 °C, and 100 °C. As for the power-generation experiment, with data that were applied in the numerical analysis in this study [31], the single PEFC was pre-heated by an electric heater at T_ini, for temperature measurement by a thermograph. Additionally, the temperature of the supply gas at the inlet can be controlled by an electric heater at T_ini. Thermal conductivities of each of the supply gases are referred to by the “The Japan Society of Mechanical Engineers” [35], in this calculation. The temperatures are measured by means of a thermograph and substituted into the formulas as T_{surf, c} and T_{surf, a}, to solve Equations (6)–(9).

Table 2. Power-generation-operation conditions of polymer electrolyte fuel cell to measure temperature, by means of a thermograph.

Initial temperature of cell (Tini) [°C]	80, 90, 100
Load current [A] (current density [A/cm2])	20 (0.80)
Condition of supply gas
	Anode	Cathode
Gas type	H2	O2
Temperature of supply gas at inelt [°C]	80, 90, 100	80, 90, 100
Relative humidity of supply gas [% RH]	40, 80	40, 80
Pressure of supply gas at inlet (absolute) [MPa]	0.4	0.4
Flow rate of supply gas at inlet [NL/min] (Stoichiometric ratio [-])	0.210 (1.5), 0.280 (2.0), 0.420 (3.0)	0.105 (1.5), 0.140 (2.0), 0.210 (3.0)

lists the power-generation operation conditions for the single PEFC, to measure the temperature by means of the thermograph. For 1D multi-plate heat-transfer analysis, the data obtained under these experimental conditions are used. This study keeps the current density at 0.80 A/cm² in the PEFC power-generation experiment, to obtain the temperature data by means of the thermograph. We can obtain the temperature-distribution data at the separator back caused by the reaction heat. The authors’ previous studies [30,31] have explained the experimental procedure to measure the temperature during power generation. It is seen from Figure 2 that this study divides the in-plane temperature distribution into segments of 10 mm × 10 mm each, in order to use the temperature data measured by means of the thermograph for the 1D multi-plate heat-transfer model. Although the power-generation area is 50 mm × 50 mm, this study sets the observation area at 40 mm × 50 mm, in order to prevent a gas leak, via an observation window in the experiments. Figure 3 illustrates the procedure to measure the temperature distribution by the thermograph, experimentally [30,31], for the readers’ understanding. In addition, Figure 4, also, illustrates the schematic drawing of the experimental setup for the power-generation experiment, to measure the temperature distribution by the thermograph [30,31], also for the readers’ understanding. The gas channel width and rib width of the separator are 1.0 mm. The number of gas channels is 5. Each segment is composed of five parts of the rib and gas channels. The average temperature in each segment at the anode and cathode is applied to the separator’s back temperature, in the 1D multi-plate heat-transfer model. Figure 2 shows the segments, named from A to T, along with the gas-flow direction.

As can be seen in Figure 2, the insulators that cover the gas pipes interfere with the temperature measurement by the thermograph in some areas of the segments, e.g., segments A and T. The effective temperatures of segments A and T, by removing the temperature data, which interfered with the insulator for the total temperature data in the segment, are estimated in this study. For the heat-transfer analysis, we assume that T_{surf, c} on the rib side equals to T_{surf, c} on the channel side as well as T_{surf, a}, since the difference between them could not be confirmed by the data measured by the thermograph [30,31]. According to the above assumptions and Equations (6)–(14), the reaction surface temperature T_react can be expressed as follows:

T_react = T_{react, rib} = T_{react, chan} = {2H_react/A + (K_{rib, c} + K_{chan, c})T_{surf, c} + (K_{rib, a} + K_{chan, a})T_{surf, a}}/(K_{rib, c} + K_{chanc, c} + K_{rib, a} + K_{chan, a})

(15)

where T_react is calculated by H_react, without estimation of the local heat flux for each segment. Here, H_react is assumed as a constant, irrespective of the segment. Additionally, i indicates the segment.

2.4. Validation

Compared with the other previous heat-transfer models [36,37,38], considering the heat-transfer conditions, the model developed by the authors [26,27,28,29,39] has some differences. However, it can be observed that the temperature gradients, for the targeted regions under similar operation conditions, show the same tendency [39]. In addition, the authors [40] have already studied the numerical simulation using a 3D model, by means of commercial CFD software, in order to predict the distributions of T_react. This numerical simulation, with a 3D model, considers the governing equations that consist of conservation equations of mass, momentum, and energy in the porous region as well as the electrochemical reaction. Comparing the results of the numerical simulation using the 3D model with that of the 1D model proposed by the authors under the various operation conditions, the differences in T_react between the two models were from approximately 0.1 °C to 1.5 °C. Therefore, the 1D model proposed by the authors, which is, also, used in this study, has been validated by the 3D model. Consequently, we think that the heat-transfer model proposed by this study is reasonable.

3. Results and Discussion

3.1. Effect of Flow Rate of Supply Gas on Distribution of T_react − T_ini

It is important to clarify the effect of the flow rates of the supply gases at the inlet on not only heat and mass transfer characteristics but also power-generation performance, to manage the operation conditions for the PEFC. Figure 5 shows the effect of the s.r. of the supply gases at the inlet on the distribution of T_react − T_ini, which is simulated by the heat-transfer model proposed by this study. The RH of the supply gases is 80% RH at the anode and 80% RH at the cathode, i.e., A 80% RH and C 80% RH. The s.r. of the supply gas is 1.5, 2.0, and 3.0, while the molar flow rate of the supplied H₂ at the s.r. of 1.5, 2.0, and 3.0 is 1.56 × 10⁻⁴ mol/s, 2.08 × 10⁻⁴ mol/s, and 3.12 × 10⁻⁴ mol/s, respectively. In addition, the molar flow rate of the supplied O₂ at the s.r. of 1.5, 2.0, and 3.0 is 0.78 × 10⁻⁴ mol/s, 1.04 × 10⁻⁴ mol/s, and 1.56 × 10⁻⁴ mol/s, respectively. T _ini is set at 80 °C. The separator thickness is 2.0 mm. Figure 6, also, illustrates the color counter, to clarify the effect of the s.r. of the supply gases at the inlet on the distribution of T_react − T_ini, visually_, which is shown in Figure 5.

It is found from Figure 5 that the effect of the flow rate of the supply gases on the distribution of T_react − T_ini is not significant. The gas supply for the power generation at an s.r. = 1.5 is sufficient for power generation [26,28]. In addition, it is revealed that the effect of the flow rate of the supply gases on the distribution of T_react − T_ini is not significant, regardless of the RH condition, T_ini, and separator thickness investigated in this study. It indicates, from the power-generation experiments in this study [30,31], that the power output exhibits almost the same amount for each different s.r. In the following section, the results with an s.r. = 1.5 are analyzed as the typical case.

In addition, it is seen from Figure 5 that T_react − T_ini declines at positions D, L, and T. Regarding position L, the water droplets are thought to remain there easily; resulting from that, it is positioned at the corner of the serpentine separator. As a result, the electro-chemical reaction is not conducted well [26,41]. Therefore, T_react − T_ini drops in this position at L. As for position D, it is the inlet of the anode side; resulting from that, the cell is cooled by the supply gas, which is cooler than the single cell that is heated, due to the reaction heat [26,41]. Regarding position T, it is thought that water droplets are easily accumulated, since the water with gas flow concentrates at the outlet at the cell [26,27]. Consequently, the gas diffusion toward the catalyst layer is not conducted well; resulting from that, the electro-chemical reaction is not conducted well. Finally, T_react − T_ini drops in this position at T.

3.2. Impact of Separator Size with RH and T_ini on Distribution of T_react − T_ini

Figure 7, Figure 8 and Figure 9 show a comparison of the distribution of T_react − T_ini among different RH conditions and T_ini, respectively. In this analysis, the separator size is changed by 2.0 mm (saddle thickness = 1.0 mm, channel height = 1.0 mm), 1.5 mm (saddle thickness = 0.5 mm, channel height = 1.0 mm), and 1.0 mm (saddle thickness = 0.5 mm, channel height = 0.5 mm). Figure 10, Figure 11 and Figure 12 show a comparison of the polarization curves, which are obtained by the power-generation experiment among different RHs and T_ini, with changes of the separator thickness of 2.0 mm, 1.5 mm, and 1.0 mm, respectively.

From the reported results in Figure 7, Figure 8 and Figure 9, it is seen that the temperature declines at positions D, L, and T in case of a separator thickness of 2.0 mm, which are larger than those of the other considered separator thicknesses. In experiment, to obtain the temperature at separator’s back by means of the thermograph, the temperature is measured after confirming a steady state time of 30 min [31]. Since the heat capacity of a separator thickness of 2.0 mm is larger, and it is believed that the whole cell temperature after balancing with the atmospheric air, is lower compared to the other considered separator thicknesses [42]. If the whole cell temperature decreases, it is easy for the vapor water to be liquid. As a result, the water droplets are thought to remain easily at position L, as described above. As for position T, the water droplets are thought to accumulate easily, resulting from the fact that the water with the gas flow concentrates at the outlet of the cell [26,27] as well as due to hydration. Therefore, the temperature drops at positions L and T are remarkable, in the case of a separator thickness of 2.0 mm, compared to the other separator cases. As for position D, it is the inlet of the anode side, causing the cell to be cooled by the gas, which is cooler than the cell heated, due to reaction heat [26,41], and is discussed earlier.

According to Figure 8 and Figure 9, it is seen that the temperature increases, along with the gas flow through the gas channel, by approximately 2 °C. In addition, it is noticed that the impact of the RH on the distributions of T_react − T_ini is larger, with the increase in T_ini. Especially, this impact is larger at T_ini = 100 °C, using a separator thickness of 1.0 mm. It is concluded that the heat capacity decreases with the decrease in separator thickness. Therefore, it is thought that the whole cell temperature increases [42], resulting in it being easy to dehydrate the PEM and catalyst. It is found from Figure 10, Figure 11 and Figure 12 that the effect of the RH on the power-generation performance is more significant when the separator thickness decreases. Especially, the power-generation performance with A40% RH and C40% RH declines rapidly, with the reduction in separator thickness. Since the heat capacity of the separator decreases with the reduction in separator thickness, the temperature of a cell using the thinner separator becomes higher, after attaining the steady state via heat balance, with the surrounding temperature of 293 K being controlled by the air conditioner [42]. A40% RH and C40% RH is the dry condition, resulting in the PEM and catalyst layer being dehydrated easily [42]. When the power generation is conducted at a higher temperature, the accumulation of liquid water decreases, due to the exponential increase in the saturation of the water vapor, with the increase in temperature [43]. Therefore, the impact of the RH on distributions of T_react − T_ini is larger, with the increase in T_ini as well as the decrease in separator thickness.

3.3. Evaluation on Temperature Difference and Standard Deviation of Distribution of T_react − T_ini

Figure 13, Figure 14 and Figure 15 show the effect of the RH and T_ini on T_{react, max} − T_{react, min} changing the separator thickness. T_{react, max} and T_{react, min} mean the temperature difference between the maximum and the minimum temperature in the distribution of T_react, respectively.

According to Figure 13, Figure 14 and Figure 15, it is seen that T_{react, max} − T_{react, min} decreases with increase in T_ini, irrespective of the RH and separator thickness. However, in the case of A 80% RH and C 80% RH, using a separator thickness of 2.0 mm shows a different tendency. When T_ini increases, the accumulation of liquid water decreases, due to the exponential increase in the saturation of the water vapor with the increase in temperature [43]. Therefore, the power-generation performance declines, since the ohmic losses caused by the reduced proton conductivity of the PEM, due to dehydration, increase [44].

Table 3. Experimental data on power-generation performance, at load current of 20 A.

Separator thickness: 2.0 mm
Tini = 80 °C
A 80% RH, C 80% RH	A 80% RH, C 40% RH	A 40% RH, C 80% RH	A 40% RH, C 40% RH
0.565 V	0.535 V	0.540 V	0.520 V
Tini = 90 °C
A 80% RH, C 80% RH	A 80% RH, C 40% RH	A 40% RH, C 80% RH	A 40% RH, C 40% RH
0.545 V	0.525 V	0.530 V	0.495 V
Tini = 100 °C
A 80% RH, C 80% RH	A 80% RH, C 40% RH	A 40% RH, C 80% RH	A 40% RH, C 40% RH
0.465 V	0.460 V	0.435 V	0.415 V
Separator thickness: 1.5 mm
Tini = 80 °C
A 80% RH, C 80% RH	A 80% RH, C 40% RH	A 40% RH, C 80% RH	A 40% RH, C 40% RH
0.520 V	0.475 V	0.485 V	0.395 V
Tini = 90 °C
A 80% RH, C 80% RH	A 80% RH, C 40% RH	A 40% RH, C 80% RH	A 40% RH, C 40% RH
0.460 V	0.450 V	0.435 V	0.355 V
Tini = 100 °C
A 80% RH, C 80% RH	A 80% RH, C 40% RH	A 40% RH, C 80% RH	A 40% RH, C 40% RH
0.345 V	0.340 V	0.325 V	0.275 V
Separator thickness: 1.0 mm
Tini = 80 °C
A 80% RH, C 80% RH	A 80% RH, C 40% RH	A 40% RH, C 80% RH	A 40% RH, C 40% RH
0.535 V	0.500 V	0.490 V	0.365 V
Tini = 90 °C
A 80% RH, C 80% RH	A 80% RH, C 40% RH	A 40% RH, C 80% RH	A 40% RH, C 40% RH
0.475 V	0.455 V	0.425 V	0.315 V
Tini = 100 °C
A 80% RH, C 80% RH	A 80% RH, C 40% RH	A 40% RH, C 80% RH	A 40% RH, C 40% RH
0.365 V	0.330 V	0.275 V	0.215 V

lists the experimental power-generation performance data, which has been conducted to measure the temperature at the separator’s back by a thermograph. The voltages, at a current of 20 A, are shown in Table 3. According to Table 3, we can see that the power-generation performance declines with the increase in T_ini, irrespective of the RH and separator thickness. For instance, since T_ini = 100 °C is a higher operation temperature, the PEM is easy to be dehydrated. Additionally, it is thought that the effect of not only the phase change of water but also the remaining liquid water on the gas diffusion in the cell is small, at T_ini = 100 °C, regardless of the RH condition [45], providing a uniform temperature distribution [26]. Regarding the case of A 80% RH and C 80% RH, using the separator thickness of 2.0 mm, it is a relatively humidified condition, since the well humidified gases are supplied at the anode and cathode and the thickest separator is used, even at T_ini = 90 °C. Therefore, the temperature decreases at positions D, L, and T are larger, and the temperature increases from the inlet to the outlet, since power generation is improved by the increase in the proton conductivity of the PEM.

Figure 16, Figure 17 and Figure 18 show the relation between the standard deviation of the distribution of T_react − T_ini and the total voltage obtained in the experiment, when changing the T_ini and separator thickness. The data under the different s.r. and RH conditions are shown in each figure. The approximate line is, also, shown in each figure.

It is seen from Figure 16, Figure 17 and Figure 18 that the slope of the approximate line, for the relation between the standard deviation of the distribution of T_react − T_ini and total voltage, becomes negative and larger with the increase in T_ini, irrespective of separator thickness. From this result, the wider temperature distribution provides a reduction in power-generation performance, when the approximate line has a large negative slope. In addition, it can be seen from Figure 16, Figure 17 and Figure 18 that the distribution range, regarding the standard deviation of the distribution of T_react − T_ini at T_ini = 100 °C, is narrower than the other T_ini. As discussed earlier, the effect of not only the phase change of water but also the remaining liquid water on the gas diffusion in the cell is small, at T_ini = 100 °C, regardless of the RH condition [45], which provides uniform temperature distribution [26]. It is difficult to manage the humidification of the PEM and catalyst under a high temperature operation condition. As a result, the power-generation performance drops, even for a narrow temperature distribution.

Summarizing the results and discussion, this study suggests that thin separators (such as a thickness of 1.5 mm and 1.0 mm) are not suitable for a higher temperature operation than usual. Since control of the humidification as well as the in-plane temperature distribution of T_react at a higher temperature is stricter than the usual operation temperature, the hydration of the supply gas and the thermal properties of the cell components of the PEFC should be considered, for controlling the electro-chemical reaction. According to the state-of-art review on the research and development of the separator [46,47,48], the material type, composite, and surface-coating process have been investigated to improve the conductivity and water behavior. However, the thermal properties of the separator have not been investigated yet. Therefore, this study suggests that the thermal properties of the separator, such as heat capacity and thermal conductivity, should be improved to attain better power-generation performance. Additionally, this study adopts a thin PEM (Nafion membrane), which is usually used at a temperature lower than 70 °C. The degradation in a Nafion membrane, when operated at a higher temperature, should be considered. This study conducted its experimental investigation using a thin Nafion membrane at a higher temperature, such as 90 °C and 100 °C [31]. In the experimental analysis, the thin Nafion membrane was operated for 200 h, to evaluate the performance. However, it can be recommended to study the characteristics of the thin Nafion membrane for a longer operation time, e.g., 90,000 h (≒ 10 years), which is the target operation time referred to by the NEDO road map (2017, Japan) [1], for the practical application of the PEFC system.

4. Conclusions

The effect of the separator thickness on the distribution of T_react − T_ini in a single PEFC has been investigated, changing the flow rates and RH of the supply gases as well as T_ini. Especially, this study has focused on the analysis at a high temperature, of 90 °C and 100 °C. The distribution of T_react − T_ini has been calculated by the presented methodology, using the separator’s back temperature measured by a thermograph, experimentally. The following conclusions are obtained:

(i)

The effect of the flow rate of the supply gases on the distribution of T_react − T_ini is not significant, among the investigated conditions.

(ii)

The temperature declines at positions D, L, and T, in the case of a separator thickness of 2.0 mm, which is the thickest separator, though the temperature increases along with the gas flows through the gas channel, by approximately 2 °C, in the case of a thinner separator, compared to 2.0 mm.

(iii)

In the case of a separator thickness of 1.5 mm and 1.0 mm, the impact of the RH on the distributions of T_react − T_ini is larger with the increase in T_ini. Especially, this impact is larger at T_ini = 100 °C, at a separator thickness of 1.0 mm. The heat capacity decreases with the decrease in separator thickness. Therefore, the whole cell temperature increases, resulting in it being easy to dehydrate the PEM and catalyst.

(iv)

T_{react, max} − T_{react, min} decreases with the increase in T_ini, irrespective of the RH and separator thickness. However, in the case of A 80% RH and C 80% RH at a separator thickness of 2.0 mm, it has a different tendency.

(v)

It is revealed that the slope of the approximate line for the relation between the standard deviation of the distribution of T_react − T_ini and the total voltage becomes negative and larger with the increase in T_ini, irrespective of the separator’s thickness, indicating that a wider temperature distribution provides a reduction in power-generation performance. The distribution range regarding the standard deviation of the distribution of T_react − T_ini, at T_ini = 100 °C, is narrower than the other T_ini. Since T_ini = 100 °C is a higher temperature, it is difficult to manage the humidification of the PEM and catalyst, and the power-generation performance declines, even for a narrow temperature distribution. This study proposes that thin separators, such as a thickness of 1.5 mm and 1.0 mm, are not suitable for higher temperature operation than usual.