Priyanka
Ghosh
a,
Chandan Kumar
Dhole
a,
Saibal
Ganguly
b,
Dipali
Banerjee
c and
Kajari
Kargupta
*a
aChemical Engineering Department, Jadavpur University, Kolkata, India. E-mail: karguptakajari2011@gmail.com
bChemical Engineering Department, BITS PILANI, Goa Campus, India
cDepartment of Physics, Indian Institute of Engineering Science and Technology (I.I.E.S.T), Shibpur, West Bengal, India
First published on 21st June 2018
Portable smart low-cost proton conductive gel paste electrolytes, capable of retaining their properties over a few months and of being stored, are of significance in the context of portable fuel cell power sources. A low-cost, sol gel derived inorganic gel paste electrolyte with an optimum phosphorous to silicon ratio of five, whose proton conductivity exceeds Nafion in the broad temperature range of 50–180 °C, is reported herein. The fuel cell assembled with this gel paste coated on a glass mat, operating at a temperature of 160 °C, produces a maximum current density of >1 A cm−2, a power density of >210 mW cm−2 and offers stable durable operation up to 29 hours under a 50 mA cm−2 load. The significance of these findings is that the gel paste may be used to replace conventional liquid and membrane electrolytes in fuel cells and be applied in new generation electronics with portable fuel cells.
Conductive gels are generally synthesized by incorporating conducting particles such as conducting polymers, graphene, carbon nanotubes (CNT), etc. into the gel matrix.9,10 On the contrary, proton conductive gels can be synthesized by trapping acid into organic or inorganic matrices.11,12 Among the different acids, phosphoric acid contains three protons per molecule and therefore offers the best proton conductivity. The potential application of a proton conductive gel is to replace the conventionally used phosphoric acid liquid electrolyte of phosphoric acid fuel cells (PAFCs) and the Nafion membrane electrolyte. Originally, developed as a liquid electrolyte device, PAFCs use phosphoric acid as an electrolyte, soaked in any proton conducting matrix such as silicon carbide or a glass mat, etc. PAFCs can operate over a higher temperature range than a Nafion fuel cell and are less sensitive to CO poisoning. Fuel cells with liquid electrolytes are susceptible to the crossover of gases, ionic shorts, flooding and drying (of the electrolyte). They are now being refined by trapping the phosphoric acid into solid organic or inorganic electrolyte materials. Recently, the incorporation of phosphoric acid in a polymeric matrix such as polybenzimidazole (PBI)13 has made these types of cells a promising and cost effective alternative to Nafion based fuel cells. Phosphoric acid doped high temperature resistant polymer based fuel cells offer a higher temperature range of operation (up to 160 °C) compared to the low temperature operation of Nafion based fuel cell. Organic polymer based phosphoric acid gels or polymeric gel electrolytes for performance enhancement have been published in the literature. Shmukler et al.14–16 have synthesized proton conductive gel electrolytes based on poly(methyl methacrylate) (PMMA), poly(vinylidene fluoride) (PVdF), and mixtures of PMMA with PVdF or poly(vinyl chloride) and showed the dependence of conductivity on the acid and polymer concentration. H. B. Aiyappa et al.17 have synthesized a proton conducting metallogel based on FNPA [ferric nitrate (FN)–phytic acid (PA)], by immobilizing a protogenic ligand (phytic acid) using iron(III) nitrate in dimethylformamide (DMF). Kargupta et al.18 have reported the performance enhancement of PAFCs using a phosphosilicate gel paste. The advantage of this Si and phosphorous based gel electrolyte is that a large amount of phosphoric acid and water molecules can be trapped into the thermally stable Si–O–Si cavities without crystallization. The phosphoric acid trapped in the 3-dimensional structure creates proton conductive solid pores, filled with water molecules at low humidity, which in turn form pathways for proton transfer. As the acid and water molecules are trapped into the optimum sized pores in thermally stable Si–O–Si 3-dimensional networks, leaching out of the acid can be prevented.19–21
Recently, impressive work has been reported by Ansari et al.22 on a flexible all inorganic fiberglass reinforced SiPOH gel membrane. They reported that the all inorganic material, with very good conductivity, can perform well in a fuel cell up to 150 °C and could be used for up to 24 hours. The SiPOH gel reported in this paper represents a very successful sequestration of phosphoric acid in a flexible nano-permeated, defect-free, but open (amorphous zeolitic), network of pure silica.
Herein, we report a portable, ready-to-use, smart, phosphorous and silica based three dimensional, highly proton conductive inorganic gel paste electrolyte that can retain its physical (gel paste) form and properties over several months, for application in fuel cells. Scheme 1 depicts the schematic illustration of the gel paste formation with its 3D structure and its application as a fuel cell electrolyte. The structural, morphological and electrochemical features of the sol gel derived gel paste can be observed by varying its P/Si ratios. ‘Structure-property-performance’ mapping was used to identify the optimum ratio of P/Si that offers the best values of proton conductivity, acid and water uptake and the maximum power density and stability of the fuel cell. This gel paste has a proton conductivity greater than that of Nafion, is capable of being stored and can be used as a portable soft gel when required. These inorganic gel paste could be a door opener in modern technologies for long term uninterrupted power supply with smart portable handy features. Due to its portable nature, the gel could be used in new generation electronics operating with fuel cells, batteries or supercapacitors.
Scheme 1 Schematic illustration of the gel paste formation with its 3D structure and its application as a fuel cell electrolyte. |
The morphological analysis was performed using scanning electron microscopy (SEM) and the corresponding EDX spectra. The surface morphology was observed using a JEOL JSM-6360 scanning electron microscope. The EDX analysis was performed using an FEI INSPECT F50 instrument. The thermal stability of the gel paste was measured by thermogravimetric analysis (TGA) using a PerkinElmer Pyris Diamond model with a 150 ml min−1 rate nitrogen flow with a heating rate of 10 °C min−1. The water uptake and acid uptake properties of the gel paste were calculated from the TGA analysis.
The temperature dependent proton conductivities of the gel paste coated glass mat membranes were determined by a two probe method using a Metrohm Autolab AUT84999 PGSTAT 302N potentiostat/galvanostat (Ecochemie, The Netherlands) over a frequency range of 1 Hz to 1 MHz with a sine wave amplitude of 10 mV. A constant voltage was supplied through the Autolab potentiostat. Gel pastes with different P/Si ratios were coated on both sides of a glass mat and sandwiched between two 0.785 cm2 area platinum electrodes. The distance between the two electrodes was 0.065 cm, i.e. the thickness of the gel coated glass mat. The proton conductivity of 88% H3PO4 was measured in an electrochemical cell, as described in the literature.23 A schematic illustration of the measurement of the conductivity of the gel paste is shown in the ESI (see Fig. S5 in the ESI†). A sinusoidal voltage signal from the Autolab potentiostat was applied across this platinum plate and gel paste membrane combination. The conductivity was calculated using the following formula:24
σ = (L/A) × (1/R) (S cm−1) |
Fig. 1 PXRD patterns, and FTIR and Raman spectra of the gel paste with different P/Si ratios. (a) The PXRD patterns, (b) the FTIR spectra and (c) the Raman spectra. |
FTIR spectra in the range of 500–4000 cm−1 for gel pastes with different P/Si ratios are shown in Fig. 1(b). The absorption peak at 〈1〉 at around 480 cm−1 is due to Si–O–Si rocking vibrations.30 The intense peak 〈2〉 at 1008 cm−1 is due to the stretching vibration of Si–O, which is assigned to the Si–O–Si linkages.12,31 The characteristic peaks in the range of 950 cm−1 to 1050 cm−1 are assigned to the Si–O–Si and P–O, and P–OH linkages of the gel paste.12,28,30,32–35 The symmetric modes of PO and O–P–O are mainly observed in the range of 950–1200 cm−1, which confirms that the silica network is doped with phosphate units.12,28,36 The peak 〈3〉 at 1693 cm−1 is assigned to –H bonded or adsorbed molecular water and confirms the presence of –OH groups in the cavity of the pores.28,30,36–40 The absorption peak 〈4〉 at 2310 cm−1 is the characteristic vibration of P–O–H, which confirms the presence of phosphate linkages in the gel paste.12 The broad absorption band 〈5〉 at 3360 cm−1 can be mainly assigned to Si–O–H and P–O–H arising from the –H bonding with adsorbed water molecules. In the gel paste, a large amount of water molecules are trapped in the cavity, which form –H bonds with the oxygen atoms of the Si–O and P–O linkages. The water molecules show peaks at around 3950 cm−1. All of the absorption peaks are evident for all three paste compositions (P/Si = 4, 5, 6). The data concludes that the gel is a large network of Si–O–Si, Si–O–H, and P–O–H bonds, where water molecules and orthophosphoric acid are trapped.
The Raman spectra of gel pastes with different compositions are shown in Fig. 1(c). Two peaks at 511 cm−1 and 915 cm−1 are observed in Fig. 1(c). McMillan et al.41 have shown that in the low frequency region, vibration bands in the region of 400–700 cm−1 are due to the presence of bridging oxygen or Si–O–Si linkages and these linkages confirm the presence of more polymerized silicates than orthosilicates. A broad peak 〈a〉 starting at 430 cm−1 and ending at 570 cm−1 can be observed in the Raman spectrum. The peak is at a maximum at 511 cm−1 and is a characteristic peak of silicates with almost all four oxygens bridged. A low intensity hump is obtained rather than an intense peak, mainly due to the defects in the structure and defects arising due to hydrogen bonding and other van der Waals force of attraction arising in the cavities. The other intense peak at around 915 cm−1 is due to the P–OH symmetric stretching of the PO43− and the out of plane –OH bending vibrations.42–44 By varying the P/Si ratio from 4 to 6, the peak intensity increases. As the sample is heated only at 100 °C, no Si–O–P bonds can be formed, and H3PO4 remains in a hydrogen bonded state. The PO43− units are hydrogen bonded in the cavities.
The liquid-like mobility in the solid silicate network was confirmed using three different types of NMR analysis. The 1H NMR, 31P NMR and 29Si NMR spectra of gel paste-5 are depicted in Fig. 2 and 3(a) and (b), respectively. The corresponding 1H NMR, 31P NMR and 29Si NMR spectra of gel paste-4 and gel paste-6 are shown in the ESI (see Fig. S1, S2 and S3,† respectively). The sharp and intense resonance peak referenced to the TMS standard at δ = 7.392 for gel paste-4 (see Fig. S1(a)†), δ = 5.493 for gel paste-5 and δ = 7.055 (see Fig. S1(b)†) for gel paste-6 are assigned to the POH groups involved in the weak hydrogen bonding.45,46 The deviation of the δ value of gel paste-5 (δ = 5.493) from that of gel paste-4 (δ = 7.392) can be attributed to the variation in the extent of the water content. With an increase in the acid content (P/Si 4 to 5), the gel becomes more gelatinous and pasty in nature and absorbs more water leading to upfield shifting. In gel paste-6, the δ value further increases due to an over flooding of the cavities and the presence of external surface water, which is also clearly observable from the SEM images.47 The small intense resonance peaks at δ = 1.06 and δ = 3.8 can be assigned to the methyl (CH3CH2O–) and methylene (CH3CH2O–) parts of the ethoxy group, respectively.45 The small and narrow shape indicates that the ethoxy group undergoes high frequency pseudo-isotropic dynamics. Another sharp and intense peak at δ = 2.46 is due to the use of deuterated dimethyl sulfoxide (C2D6OS) as the solvent. In the 31P NMR spectrum of gel paste-5, shown in Fig. 3(a), the sharp and intense peak observed at 0 ppm (Qo), can be attributed to the isolated orthophosphoric acid.21,48–50 The absence of any other peaks in the 31P NMR spectrum confirms the absence of condensed phosphoric acid, i.e. phosphoric acid with bridging oxygens such as Si–O–P or P–O–P. The absence of any Q1, Q2, Q3 and Q4 peaks eliminates the probability of the presence of singly, doubly, triply and fully coordinated P atoms. We prepared gel pastes from TEOS and H3PO4 at room temperature. The results confirm that for all of the gel pastes, the phosphoric acid remains in a pure and isolated form; there is no solid condensation bonding. It also confirms the absence of Si–O–P and P–O–P bonds. Similar features were also observed for gel paste-4 and 6 (see Fig. S2(a) and (b),† respectively). The solid state 29Si NMR spectrum of gel paste-5 shown in Fig. 3(b) shows one resonance peak at −112 ppm referenced to the standard TMS (using solid tetrakis(trimethylsilyl)silane (TTSS) at −9.8 ppm as a secondary reference) confirming the four-coordination of the silica to the bridging oxygen in the tetrahedral silicate networks. A chemical shift of −115 ppm is the extreme downfield of the chemical shift range reported for [SiO4] groups.22 Only pure zeolitic silica has a chemical shift more downfield than −110 ppm, which is average for Q4 groupings of silicate minerals. This confirms that the Si in the gel paste is in the pure zeolitic form. The 29Si NMR spectra of gel pastes-4 and 6 (Fig. S3(a) and (b)†) show resonance peaks at −111 ppm for gel paste-4 and -110 ppm for gel paste-6, which are characteristic of four coordinated silica. The absence of any other intense resonance peaks in the more downfield region eliminates the chances of the presence of six coordinated silica.
Fig. 3 NMR studies of gel paste-5. (a) The solution state 31P NMR spectrum in d6-DMSO and (b) the solid state 29Si NMR spectrum. |
The presence of four coordinated silica, as evidenced from 29Si NMR, and isolated, non-bridged P, as evidenced from 31P NMR and weakly hydrogen bonded POH, as evidenced from 1H NMR, confirms that in the gel paste, H3PO4 is in a hydrogen bonded state within a Si–O–Si cage.
Fig. 4 (a) Thermogravimetric analysis of gel pastes of different compositions, and (b) water and acid uptake properties evaluated from the TGA plot. |
Interestingly, at any temperature, the maximum weight loss was observed for gel paste-5. The values of water uptake and acid uptake for different P/Si ratios were evaluated from the TG analysis and are shown in Fig. 4(b). Comparative analysis of the TGA of gel paste-4, 5 and 6 reveals that gel paste-5 contains a greater amount of trapped acid and adsorbed water. As a result, gel paste-5 exhibits higher proton conductivity values and better fuel cell performance compared to gel paste-6. This will be discussed below in Sections 3.4 and 3.5. The evaluated water and acid uptake values from the TG analysis have a good correlation with the composition of the gel paste. The calculated water uptake values of gel pastes-4, 5, 6 are 7.19%, 7.56% and 4.79%, respectively, whereas the acid uptake values are 8.31%, 10.05% and 7.28%, respectively.
EDX analysis of the three gel paste samples confirmed the compositions of P and Si in the samples (see Fig. S3†). The mole ratio obtained from the EDX analysis has good correlation with the experimental mole ratios. From the EDX analysis, the calculated values of the P/Si ratios with respect to weight percentages are 3.646, 4.591, and 6.081 for gel pastes-4, 5 and 6, respectively, and with respect to the atomic percentage are 3.307, 4.165 and 5.520, respectively.
Fig. 6 (a) Temperature dependent proton conductivity of different gel pastes compared with 88% phosphoric acid and Nafion 115 at ambient pressure; the Nafion 112 and 100% H3PO4 data have been incorporated from literature.22,56 The solid symbols represent the experimental data from this work, and the open symbols represent the data from literature. (b) The corresponding Nyquist plots at 160 °C of three gel pastes and 88% phosphoric acid and (c) shows the fitting circuit of the Nyquist plot. |
The Randles–Ershler model54 was found to fit perfectly with the experimental data. The circuit is depicted in Fig. 6(c). In this model, RΩ and Rct represent the ohmic resistance and charge transfer resistance, respectively, and Ws represents the finite-length Warburg impedance, arising due to the diffusion of protons through pores.23,54,55 The conventional double layer capacitance is replaced by a constant phase element (CPE) because of the non-uniform distribution of the capacitance caused by double layer charging.23,54
All the conductivity values are compared with two different literatures; one with 100% H3PO4 from crystal22 another with Nafion 112 at fully hydrated state.56 Upon an increase in the temperature, the proton conductivity was observed to increase for the gel pastes. The gel pastes gave the best results at 160 °C and then showed a decrease and can sustained up to 180 °C. Among all the different cases, the gel paste with the P/Si = 5 ratio was observed to exhibit the best proton conductivity values over the entire range of temperatures from 30–180 °C. The proton conductivity values and the variation in the proton conductivity values upon changing the P/Si ratios, both reached a maximum at 160 °C and the increase in the P/Si ratio from 5 to 6 caused a decrease in the proton conductivity due to the presence of less orthophosphoric acid and more pyro- and triphosphoric acids (as evidenced from the TG analysis). Here, the conductivity of the gel paste is compared with the conductivity of the conventional phosphoric acid fuel cell (PAFC) and the commercially purchased Nafion membrane. In conventional PAFCs, an 88% orthophosphoric acid soaked glass mat is used for the conductivity measurements. The gel paste shows better conductivity than the 88% phosphoric acid soaked glass mat used in PAFCs. In PAFCs, upon an increase in the temperature, acid leaches out from the glass mat, whereas in the gel paste, the acid molecules are trapped in the thermally stable Si–O–Si network. The commercially purchased Nafion membranes were soaked in 88% orthophosphoric acid for three days and tested in conductivity measurements. The conductivities of all three of the gel pastes were higher than those of the Nafion 115 and Nafion 112 membranes at 100% relative humidity. The maximum operating temperature of the Nafion membrane is 90 °C, but it showed the best result at 70 °C. At 90 °C, the melting of the Nafion membrane begins to occur.
It is known that the proton conductivity of phosphoric acid depends on the formation of a contact ion pair (H4PO4+/H2PO4−), and the electrostatic interaction of the associated electric dipole moments within the unrelaxed solvent environment with transiently short hydrogen bonds.23,56 The proton transportation follows the jumping of the proton through the Grotthuss mechanism.58–60 In the gel paste, the phosphoric acid remains trapped inside the 3D Si–O–Si network via hydrogen bonding. The solvent environments in normal phosphoric acid solution and in the gel paste are wildly different since the Si–O–Si network provides a facile proton conductive passage and water trapped in the cavities plays a major role in proton conduction.23,57,61
The comparative fuel cell performance (voltage and power versus current density) of the different gel paste electrolytes with P/Si ratios of 4, 5 and 6 and a conventional PAFC at a 160 °C operating temperature is illustrated in Fig. 7(b). Among the three electrolytes, the gel paste with a P/Si ratio of five offers the best performance (maximum power density > 210 mW cm−2). At a cell temperature (TCell) of 160 °C and humidifier temperature (TH) of 70 °C, the variation in the P/Si ratio from 4 to 5 increases the peak power generation by 25.49% and the corresponding current density is also shifted towards a higher value. In comparison with the PAFC, 67.51% peak power enhancement was achieved using the gel paste electrolyte with a P/Si ratio of 5. Also, it may be inferred from the voltage versus current density plots, that among all of the electrolytes, gel paste-5 exhibits the highest value (1.05 A cm−2) of threshold current density (at which the voltage becomes zero). The result follows the same trend as that of the proton conductivity data shown in Fig. 6.
The fuel cell durability plot of the different gel paste (coated glass mat matrices) electrolytes is shown in Fig. 8. In order to address the endurance of electrolyte under load (usually referred to as a degradation rate), the gel paste electrolyte was placed between the two electrodes and the fuel cell was operated at 160 °C under a fixed load current density of 50 mA cm−2 for 29 hours and the variation of voltage over time was recorded and plotted, as shown in Fig. 7. Among the different electrolytes, the gel paste with a P/Si ratio of 5 was observed to offer the best durability.
Fig. 8 Stability plot of gel pastes with different P/Si ratios and the PAFC at a humidifier temperature (TH) of 70 °C and cell temperature (TCell) of 160 °C with a constant current load of 50 mA cm2. |
The voltage efficiency is defined as
The theoretical electromotive force (EMF) is 1.2 volts. Two different values of voltage efficiency are calculated: one based on the open circuit voltage (OCV) and another based on the voltage corresponding to the peak power density. Among all the different electrolytes, gel paste-5 had the maximum values of threshold load current, maximum peak power density and voltage efficiencies. Although the maximum operating temperature of the fuel cell with gel paste-5 electrolyte is 180 °C, the best performance was observed at 160 °C.
Table 1 summarizes the performance of all of the inorganic gel paste electrolytes for application in fuel cells. The quantitative parameters include maximum operating temperature, threshold load current, maximum peak power density, voltage efficiency and durability in hours. The threshold load current is the maximum sustainable current density value of the cell, i.e. the specific current density of the cell at zero voltage. The maximum peak power density is another valuable parameter of a fuel cell. It is the power density that actually determines the strength of the fuel cell.
Matrices | Maximum operating temperature (°C) | Threshold load current (mA cm−2) | Maximum peak power density (mW cm−2) | Voltage efficiency (%) (OCV) | Voltage efficiency at Pmax (%) | Durability (hours) | Ref. |
---|---|---|---|---|---|---|---|
Phosphoric acid 85% (cell area 0.5 cm2) | 226 | 303 | 59 | 85.7 | 32.9 | 23 | 4 |
SiPOH (cell area 0.5 cm2) | 226 | 966 | 184 | 85.4 | 33.8 | 23 | 4 |
Stabilised SiPOH gel | 154 | 1100 | 200 | 85.83 | 33.33 | 24 | 22 |
Fe(III) phytate metallogel | 130 | 1.2 (at 80 °C) | 0.55 | 72.5 | 48.24 | — | 17 |
Gel paste-4 | 180 | 900 | 172 | 67.83 | 31.91 | 29 | This work |
Gel paste-5 | 180 | 1050 | 216 | 74.75 | 36.083 | 29 | This work |
Gel paste-6 | 180 | 950 | 178 | 73.25 | 33.083 | 29 | This work |
PAFC | 160 | 750 | 129 | 66.75 | 26.91 | 22.5 | This work |
In order to reveal the superiority of the performance of gel paste-5, the results of the present work were compared with reported data, as shown in Table 1.4,17,22 The literature on gel electrolytes including fuel cell performance is scant. Some of the literature data on gel electrolytes showing fuel cell performance are compared here. H. B. Aiyappa et al.17 synthesized a proton conducting metallogel based on metal organic materials (MOMs), constituting a phosphate ester based ligand immobilized via gelation with Fe3+ ions in DMF and used them to perform fuel cell operations. A power density of 0.55 mW cm−2 at 80 °C and a high proton conductivity value of 2.4 × 10−2 S cm−1 at 120 °C were achieved with these MOMs. Ansari et al.4 synthesized an anhydrous modified H3PO4 liquid electrolyte embedded with a silica network [Si(PO4H)2] that showed a power density of 184 mW cm−2 at 226 °C. The all inorganic fiberglass reinforced gel membrane made recently by Ansari et al.22 achieved a maximum current of >1 A cm−2, a power density of 200 mW cm−2 at 154 °C and could be used for up to 24 hours with a load of 50 mAcm−2. Gel paste 5 offers a threshold load current of >1 A cm−2 and a peak power density of 216 mW cm−2 at 160 °C. The maximum power density, which actually determines the strength of a fuel cell, achieved in this study was observed for gel paste-5. It also showed the maximum durability of 29 hours at a 50 mA cm−2 current load.
Fig. 9(a) summarizes the effect of the P/Si ratio on the proton conductivity and performance of the fuel cell with the gel paste electrolyte. The results show that the proton conductivity, peak power density and corresponding current density, and threshold load current are maximized at the optimum P/Si ratio value of 5. For commercial, easy and smart use purposes, gel paste-5 was stored in a tube and used as per requirement. To check the performance of the gel paste stored in the tube, the proton conductivity was measured. Fig. 9(b) shows the proton conductivity value of gel paste-5 with respect to time. It also gives a clear idea of how long the gel paste can be used for good performance or in commercial language how long it can suitably be used before expiry. Around 8%, 18.91%, 30.23% and 60.25% decay in the proton conductivity was observed after 6 months, 10 months, 14 months and 20 months, respectively.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c8se00142a |
This journal is © The Royal Society of Chemistry 2018 |