Optimization and kinetics of phosphoric acid doping of

Journal of Membrane Science 446 (2013) 422–432

Contents lists available at ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

Optimization and kinetics of phosphoric acid doping of poly(1-vinylimidazole)-graft-poly(ethylene-co-tetrafluorethylene) proton conducting membrane precursors Mohamed Mahmoud Nasef a,b,n, Hamdani Saidi a, Arshad Ahmad a, Amgad Ahmad Ali a a b

Institute of Hydrogen Economy, International Campus, Universiti Teknologi Malaysia, 54100 Kuala Lumpur, Malaysia Malaysia-Japan International Institute of Technology, International Campus, Universiti Teknologi Malaysia, 54100 Kuala Lumpur, Malaysia

art ic l e i nf o

a b s t r a c t

Article history: Received 5 February 2013 Received in revised form 21 May 2013 Accepted 24 May 2013 Available online 20 June 2013

Optimization of the reaction parameters affecting phosphoric acid (PA) doping behavior of poly(1vinylimidazole), P(VIm), grafted poly(ethylene-co-tetrafluoroethylene) (ETFE) proton conducting membrane precursors obtained by radiation induced grafting was studied using the Taguchi method. The reaction parameters such as degree of grafting (G%) in the precursors, PA concentration, temperature and doping time were selected as independent parameters while doping level was the response. The optimum parameters for achieving the maximum doping level (7.45 mmol repeat polymer unit−1) were: G% of 54%, PA concentration of 60%, temperature of 100 1C and reaction time of 5 days. The kinetics of the acid doping reaction was also studied and the doping rate was found to be a function of reaction parameters and followed a first order reaction. The PA doping was verified by Fourier transform infrared (FTIR) and X-ray photoelectron spectroscopy (XPS) analysis of the membranes. The proton conductivity and thermal stability of the membranes were also evaluated. It can be concluded that the Taguchi method provides an effective tool for prediction of acid doping level and optimization of reaction parameters. The kinetics of acid doping is also suggested to be a diffusion-driven reaction following a multi-layers adsorption model. & 2013 Elsevier B.V. All rights reserved.

Keywords: Radiation grafted poly(1-vinylimidazole)/ ETFE precursor Phosphoric acid doping Proton conducting membrane Optimization Kinetics

1. Introduction Proton exchange membrane fuel cell (PEMFC) is one of the most promising environmentally friendly energy systems, which provide a suitable primary power source for transportation and stationary applications [1]. Current state of the art membranes for PEMFC such as Nafion depend on water in their proton conduction. This led to installation of a humidification unit that adds to the increase in the overall cost and complexity of the system [2]. Operation at higher temperatures is an appealing approach to overcome the shortcomings of low temperature PEMFC systems and offers major advantages such as elimination of the humidification unit, increment in the tolerance towards reformed fuel impurities, enhancement of the ionic conductivity and possible utilization of excess heat for cogeneration [3]. Thus, the interest in the development of proton exchange membranes for PEMFC operation above 100 1C is receiving fast growing attention [4].

n Corresponding author at: Institute of Hydrogen Economy, International Campus, Universiti Teknologi Malaysia, 54100 Kuala Lumpur, Malaysia. Tel.: +6 03 2203 1229. E-mail addresses: [email protected], [email protected] (M. Mahmoud Nasef).

0376-7388/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.memsci.2013.05.053

However, high temperature (HT) operation of PEMFC requires new generation of less-water dependent membranes. Basic composite membranes loaded with inorganic acids such as phosphoric acid (PA) doped polybenzimadazole (PBI) membranes have been considered one of the most important candidates for HT-PEMFC operation [5]. These membranes showed reasonable performances upto temperatures close to 190 1C without any additional humidification [6]. However, such membranes are subjected to a degradation that is partially caused by leaching of the loaded liquid electrolyte [7]. PA doping of radiation grafted membrane precursors (films) has been recently proposed for the development of alternative proton conducting membranes for HT-PEMFC [8,9]. In these membranes, heterocyclic monomers such as 1-vinylimidazole (1-VIm) was grafted onto poly(ethylene-co-tetrafluoroethylene) (ETFE) films using the radiation induced grafting method. The obtained poly (1-vinylimidazole), P(1-VIm) grafted films provide the ETFE matrix with-N-groups capable of forming of an acid/base complex when doped with PA under controlled conditions in a post grafting reaction [10]. A plausible mechanism for preparation of proton conducting membranes by radiation induced grafting of 1-vinylimidazole followed by phosphoric acid doping is shown in Fig. 1.

M. Mahmoud Nasef et al. / Journal of Membrane Science 446 (2013) 422–432

423

Fig. 1. Mechanism of preparation of proton conducting membranes by radiation induced grafting of 1-vinylimidazole followed by phosphoric acid doping.

In our recent studies, the reaction parameters of radiation induced grafting of 1-VIm onto ETFE films and its impact on the degree of grafting (G%) were modeled and optimized [10]. The obtained membrane with G% of 54% and 6.6 mmol PA per P(1-VIm) repeating unit achieved an ionic conductivity of 140 mScm−1 at 120 1C and 20% relative humidity [11]. However, there is no detailed study for optimization of PA doping reaction and its kinetics behavior during the preparation of such important membranes. Also, optimization of similar doping reaction for the wellestablished acid doped membranes based on polybenzimidazole was not reported. A study of such kind is very essential for establishing the membrane preparation procedure and controlling their combinational physical and chemical properties particularly when scaling up of preparation is sought. Taguchi is a powerful design of experiments tool developed by Taguchi [12], which provides a simple, efficient and systematic approach to optimize experimental designs for high performance, cost reduction and considerable minimization of experimentation time [13]. Applying this method for optimization of acid doping during preparation of proton conducting membranes has not been reported in literature yet. The objective of the present study is to investigate the effect of acid doping conditions including G%, PA concentration, temperature of solution and doping time on the level of acid doping in the membrane precursors obtained by radiation induced grafting of 1-VIm onto ETFE films. The reaction parameters are optimized and PA doping level is predicted using the Taguchi method and verified with experimental data. The results were subjected to analysis of variance (ANOVA) by means of design expert software. The kinetic behavior of the reaction was also studied and the reaction rates together with the parameters affecting them were investigated.

2. Design of experiment The Taguchi method was developed to overcome the limitations associated with the factorial design method including high cost and prolonged time. Moreover, the contribution of each parameter is not possible to be found and sometimes two different designs can be obtained for the same experiment. Employing the Taguchi method can overcome these problems. In this method an orthogonal arrays is used in the format of Li in which i is the number of necessary trials for the given experiments. The number of levels and factors are included in a parenthesis next to the abbreviation of the orthogonal array. For example, L5(23) is a fivetrial orthogonal array with three factors at two levels. The Taguchi method is a three-stage process. The first stage involves system design and deals with working levels of design factors. The development and testing of a system is performed

Table 1 Parameters of optimization of PA doping study and their levels. Parameters

Code

Level 1

Level 2

Level 3

Degree of grafting (%) Acid concentration (vol%) Reaction temperature (1C) Doping time (day)

A B C D

20 35 5 1

38 50 30 3

54 65 100 5

based on the knowledge, experience and physical vision on the process. The second stage is the parametric design, in which the optimum conditions are determined at specific factor levels with or without the presence of uncontrolled factors. The last stage is design tolerance in which the optimum factor levels obtained in the second stage are tuned. The Taguchi experimental design was implemented to optimize the parameters of PA doping process for ETFE-g-P(1-VIm) films. Initially, Taguchi's orthogonal array L9(34) was chosen to represent four independent doping parameters: G%, acid concentration, temperature and reaction time were varied in three levels as shown in Table 1. The levels of investigated parameters were selected based some preliminary experiments and from the literature review. Design Experts (version 6) software was used for the automatic design and analysis of Taguchi experiments and performed the analysis of the results and optimization of the doping conditions. Fig. 2 shows the flow chart of developing Taguchi optimization for the experiments. The PA doping parameters and the corresponding response (doping level) were summarized in Table 2 in which the mean of three runs is chosen to represent the experimental results. The reliability of the response data were examined by ANOVA performed by the same software. An automatic backward reduction of the insignificant parameters was met at a significance level of p≤0.05.

3. Materials and methods 3.1. Materials ETFE film of 125 mm thickness and 1.69 g cm−3 density was supplied by Good fellow (Cambridge, UK). 1-VIm of 99% purity (Fluka, Switzerland) was used without further purification. 37% hydrochloric acid and 85% PA were obtained from JT Bakers (NJ, USA). Deionized water of resistance of 18 MΩ was produced using a purifier (NANOpure, DIamond™, UK) and used in grafting and doping experiments.

424


Voltage, EPS 3000, Cockroft Walton type, Japan) operated at an acceleration voltage of 2 MeV, 3 mA and 10 kGy per pass. The film was irradiated to a total dose of 100 kGy. After irradiation, the film was kept in a low temperature freezer at −65 1C for a day. Grafting reaction was performed by introducing an aqueous mixture of 1-VIm of known concentration to the irradiated ETFE film in a special ampoule, which was then flushed for 30 min with a purified N2 to remove air and then sealed tightly. The ampoule was transferred to a water bath at a temperature of 60 1C to start the reaction. After 16 h, the grafted films were extracted in 0.1 M HCl solution under sonication to remove the P(1-VIm) homopolymer and unreacted 1-VIm. The degree of grafting (G%) in the grafted films (membrane precursors) was calculated as follows: Gð%Þ ¼

W g −W o 100 Wg

ð1Þ

where, Wg and Wo are the weights of the grafted and original ETFE films, respectively. Five grafted samples with different G% were obtained using an aqueous 1-VIm having concentrations in the range of 20–60 vol%. 3.3. Acid doping reaction The grafted films samples of known weight were doped by placing them in a double-wall glass reactor equipped with stirrer, thermometer, and dip-in capillary to bubble the reaction solution with purified N2. This was followed by introducing PA solution of known concentration, which was varied in the range of 30–60%. The reaction temperature was controlled to the desired values using a thermal liquid circulated with a thermostatic pump. The reaction time was varied in the range of 1–5 days. After the desired time, the membranes were removed, washed with deionized water several times and dried in a vacuum oven at 60 1C for 16 h. The dried membranes were weighted and the acid doping level per repeat unit of polymer (XPA) (mmol repeat unit−1) was calculated using equation 2: X PA ¼

Fig. 2. Flowchart of the Taguchi method.

Table 2 Combinations runs according to R L9(34) array. Runs

1 2 3 4 5 6 7 8 9

Parameters

Response/doping Level (mmol repeat unit−1)

A

B

C

D

38 38 20 38 20 20 54 54 54

50 35 35 65 65 50 65 50 35

100 30 5 5 100 30 30 5 100

1 5 1 3 5 3 1 5 3

6.20 4.45 2.80 5.25 6.36 4.15 6.34 5.45 6.40

3.2. Preparation of membrane precursors/grafted films The ETFE film was cut into samples of known area, washed with ethanol and vacuum dried. The film samples were kept in a thin polyethylene bag sealed under vacuum prior to irradiation by a universal electron beam (EB) accelerator (NHV-Nissin High

ðwd =100ÞM p ð1 þ wi Þ M d ½1−ðwd =100Þwi

ð2Þ

where, wd is the mass fraction of the dopant, Mp is the molar mass of polymer repeat unit (g mol−1), wi is the percentage of weight increase of the grafted films and Md is the molar mass of PA (g mol−1). A summary of various steps involving radiation induced grafting of 1-VIm and its subsequent PA acid doping with the pictures of utilized facilities are illustrated in the schematic representation shown in Fig. 3. 3.4. FTIR spectral analysis Fourier transform infrared (FTIR) measurements of samples were recorded on a Nicolet 5700 spectrophotometer in a transmittance mode at a frequency range of 4000–500 cm−1 with a resolution of 4 cm−1. The spectra of the samples were detected and analyzed using commercial software (Essential FTIR™). 3.5. XPS measurements XPS measurements were conducted on vacuum dried samples using a Kratos XSAM-HS surface micro-analyzer having a Mg Kα X-ray source (300 w, 15 KV and 1253.6 eV) in a fixed analyzer transmission mode under a vacuum of 4.0–10−9 Torr. Wide scans in the range of 50–1150 eV were recorded at pass energy of 160 eV with 1.0 eV and dwell time of 0.1 s per step, respectively. High resolution spectra was recorded at 451 take-off angle by a concentric hemispherical energy electron analyzer operating in


425

Fig. 3. Schematic representation and pictures of facilities used for preparation of radiation grafted and PA doped membranes.

the constant pass energy mode at 29.35 eV, using a 720-mm diameter analysis. Binding energies of all photoelectron effects were corrected by deducing the charging effect values based on C1s at 284.5 eV for terminal hydrocarbon (–CH). The Gaussian peak fitting parameter with a straight baseline was applied for peak analysis using ‘Vision’ software supplied by Kratos. 3.6. Conductivity measurements The ionic conductivity of the membrane was measured at a temperature range of 30–120 1C using a 4-probe conductivity cell (BT-175) from BeckkTech. (Loveland, CO, USA) equipped with thermocouples and temperature controller as shown in Fig. 1. The cell was connected to a DC source meter (Keithley 2400, Cleveland, OH, USA) and a humidity control system and the whole set-up was controlled by Lab view software. Fig. 2 shows the setup for conductivity measurement in this study. The membrane was cut into strips of 3 mm 350 mm size, which was clamped between the platinum electrodes and placed in an externally insulated conductivity cell. The relative humidity (RH%) inside the conductivity cell was maintained in the range of 20–100% by means of pre-conditioned humidified N2. The gas conditioning was performed using the humidification system equipped with humidifier, temperature controller, thermocouple and gas flow meter. The relative humidity was detected by a humidity sensor

attached to the conductivity cell and interfaced with the computer. The ionic conductivity (s, S/cm) was calculated according to the following equation 3: sðS=cmÞ ¼ L=ðRWTÞ

ð3Þ

where, W is width of the membrane sample (cm), T is the thickness (cm), R is the resistance (Ω) and L is the distance between probes (cm). 3.7. Stability of the membranes The stability tested was performed by heating the membrane samples in an oven at 120 1C for prescribed period of time. The stability was monitored in terms of variation of ionic conductivity in correlation acid doping levels.

4. Results and discussion 4.1. Statistical analyses of the response The average of three acid doping runs of the independent parameters in correlation with the responses was recorded. The obtained results were introduced to the Design Expert software (version 6), which was used to fit the responses to a coded

426


Table 3 ANOVA analysis. Source

Sum of squares

F-value

Prob4 F

Significance

Model A B C

12.39 3.97 3.08 5.34

29.29 28.17155 21.85 37.84

0.0334 0.0343 0.0438 0.0257

Significant Significant Significant Significant

equation with respect to the independent parameters of PA doping to determine the optimum doping parameters. The coded equation (ideal function of quality characteristics) was given as follows: X PA ¼ 5:27−0:83A⌊1⌋ þ 0:033A⌊2⌋−0:72B⌊1⌋ þ0:00B⌊2⌋−0:77C⌊1⌋−0:29C⌊2⌋

ð4Þ

where, XPA is the response (doping level), ⌊1⌋ and ⌊2⌋ stand for the lower and upper limits for independent parameters' matrices, respectively. The significant factors in Eq. (4) were maintained whereas the insignificant ones were automatically eliminated using a backward reduction method. Only, three independent parameters: G%, acid concentration and doping temperature showed to have the most significant effects on doping level. The reliability of the ideal function of quality characteristics was examined by means of ANOVA, which is performed by the software and the obtained data is presented in Table 3. It can be seen that the calculated F-value for the model after fixing the doping time was 29.29 indicating the significance of the model. Thus, only 2.12% probability of a “Model F-Value” of such large value could occur due to noise. It can be also seen that the values of “Prob4F” for all the model terms are less than 0.05 and this suggests that all terms are significant and have a significant effect on the response. The doping time was found to have a less significant effect on the optimization and is suggested to be fixed to a maximum value of 5 days without detecting any impact on the model reliability. This was supported by the finding that longer doping time increases PA doping level until saturation (i.e. a complete complexion between available imidazol rings and PA) is reached after 5 days. Similar trend of time dependent doping level and doping saturation attainment after certain time was reported for PA doping of radiation grafted poly(4-vinlypyridine) grafted ETFE membrane precursors [14]. To test the capability of the model, the residual distribution was analyzed by detecting whether the residuals followed a normal distribution or not. Residuals are defined as the differences between actual and predicted values obeying normal distributions as long as the experimental errors are random. The normalization of residuals was primarily performed according to their standard deviations or studentized residual. Fig. 4 shows the variation of normal % probability with the predicted studentized residuals. A linear relationship can be clearly observed and such trend suggests that the data is normally distributed. This means that the model does not exhibit any violation or skewing from the normal distribution of the data. The results of the acid doping (response) analysis were plotted versus an investigated parameter (G%) as shown in Fig. 5 (at 50% acid concentration, 30 1C doping temperature and 5 days reaction time). It can be clearly seen that the doping level increases linearly with the increase in G%. This trend can be attributed to the increase in the number of imidazole rings incorporated in the ETFE matrix with the increase in G%, which provide more positive nitrogen centers for complexion with PA molecules. Fig. 6 shows the relation between the doping level and the acid concentration in the doping solution (at 30 1C doping temperature, 5 days reaction time and various G%). The doping level is found to

Fig. 4. Normal plot of residuals presents the relation between the normal % probability and the studentized residuals of responses.

Fig. 5. Variation of the doping level with G%. Reaction conditions are: 50% acid concentration, 30 1C doping temperature and 5 days reaction time.

increase linearly with the increase in the acid concentration at constant G% and time. The increase in the acid concentration provides more anionic species that occupy more cationic centers (imidazole rings) in the membrane matrix. Also, the doping level increases with the increase in G% at constant acid concentration. The highest doping level of 6.10 (mmol repeat unit−1) was achieved at highest G of 54% under the stated conditions. Fig. 7 shows the variation in the acid doping level with the doping temperature for 50% acid concentration, 5 days reaction time and various G%. It can be clearly seen that the relation between the doping level and the temperature has an increasing exponential trend. Thus, the increase in the temperature leads to an increase in the doping level regardless G%. The highest doping level of 6.62 (mmol repeat unit−1) was achieved at the highest G%¼ 54% and a temperature of 100 1C. Such response emphasizes


427

Fig. 8. Cube graph for the space of the numerical optimization. Fig. 6. Relation between the doping level and the acid concentration in the doping solution. Reaction conditions are: 30 1C doping temperature, 5 days reaction time and various degrees of grafting; (a) 20%, (b) 38% and (c) 54%.

XPA(mmole/repeat unit)

6.5

(d) (c)

6

(b) (a)

5.5

5

4.5

4

0

1

2

3

4

5

6

Time (days) Fig. 9. Variation of acid doping level of membranes with time at various acid concentrations: (a) 30%, (b) 40%, (c) 50% and (d) 60% at a constant temperature of 30 1C.

Fig. 7. Variation in acid doping level with respect to the change in doping temperature for 50% acid concentration, 5 days reaction time and various degrees of grafting; (a) 20%, (b) 38% and (c) 54%.

membrane precursors. Such optimization can be also extended to cover acid doping reactions of various basic polymer matrices during the preparation of widely investigated acid/base complex proton conducting membranes. 4.2. Acid doping kinetics

the significance of the effect of the doping temperature in enhancing the doping level in these membranes. The results of optimization of the four reaction parameters were plotted in Fig. 8, which shows a cube graph for the space of the numerical optimization depicting one of the solutions to maximize the doping level at desirability limit equals to 1. It can be seen that the optimum parameters for achieving the maximum doping level of 7.83 (mmol repeat unit−1) is G% of 54%, acid concentration of 60%, temperature of 100 1C and doping time of 5 days. The corresponding experimental doping level value was found to be 7.45 (mmol repeat unit−1), which deviates by a 5% from the predicted value. This suggests that the adopted Taguchi method is highly effective in optimization the acid doping reaction and predicting the doping level of radiation grafted ETFE-g-P(1-VIm)

The variation of acid doping level of membranes with time at different acid concentrations is presented in Fig. 9. It can be seen that the doping level increases drastically in the first 3 days beyond which it tends to slow down as the time increased except at acid concentration of 50%, which shows continuous progressive increasing trend. The rise in the acid concentration from 30% to 60% also led to an increase in the doping level at all-time intervals. However, such increase was drastic when the acid concentration increased from 30% to 50% and reduced with the further increase upto 60%. These results suggest that the grafted films swell in the dopant solution, which facilitates the doping of PA with-N- of the imidazole rings of the grafted P(1-VIm). The reaction then proceeds progressively from the surface layers towards the film core

428


1.2

Table 5 Variation of doping rate with temperature of the doping medium at an acid concentration of 60%.

dXPA/dt

(d) 1

(c)

0.8

(b) (a)

0.6

Temperature (1C)

Doping rate (mmol repeat unit−1 day−1)

Doping rate (mmol repeat unit−1 h−1)

5 30 100

0.620 0.643 0.657

0.0258 0.0268 0.0274

0.4 0.2 0

0

1

2

3

4

5

6

Time (days) Fig. 10. Acid doping rate of membranes versus time at various acid concentrations: (a) 30%, (b) 40%, (c) 50% and (d) 60% at a temperature of 30 1C.

Table 4 Variation of doping rate with concentration of PA at various acid concentrations at a temperature of 30 1C. PA conc. (%)

Doping rate (mmol repeat unit−1 day−1)

Doping rate (mmol repeat unit−1 h−1)

30 40 50 60

0.327 0.453 0.550 0.657

0.0136 0.0189 0.0229 0.0274

Fig. 12. Arrhenius plot of log acid doping level of membranes versus reciprocal of doping temperature. The reaction time was 5 days at a dopant concentration of 50%.

Fig. 11. Variation of acid doping level of membranes with time at various temperatures: (Δ) 0 1C, (♦) 30 1C and (◊) 100 1C at constant acid concentration of 60%.

Fig. 13. Schematic representation for the multilayer acid adsorption at acid concentration of 60%.

by the diffusion of the acid molecules through the grafted layers following a multilayer diffusion model. Furthermore, the increase in time does not bring about any variation in the doping level. This is due to the complete consumption of the available imidazole rings in the films upon achieving equilibrium. To determine the acid doping rate (dXPA/dt) at various acid concentrations, the function of every curve presented in Fig. 9 was differentiated with respect to doping time (t). The obtained dXPA/dt

was plotted against time (t) as depicted in Fig. 10. As can be clearly shown, the values of dXPA/dt decreased drastically with the increase in time at all acid concentrations. This was coincided with an increase in dXPA/dt with the rise in the PA concentration. This means that the amount of acid adsorbed in the membrane per time unit decreases with the time increase while increasing with the rise in the dopant concentration and tends to slow down upon approaching equilibrium at 5 days.


The doping rates obtained graphically at various PA concentrations are presented in Table 4. As can be seen the rate is doubled by 2 upon increasing the concentration of doping acid from 30% to 60% (two folds). The order of the reaction is given by: −

dX PA ¼ k½X PA a dt

429

of the grafted poly(1-VIm) layers distributed within the membrane precursor in the PA doping solution. It can be suggested that kinetics of PA acid doping in the present membranes follow a progressive (multilayer) diffusion which can be visualized as schematically shown in Fig. 13.

ð5Þ 4.3. Evidence of grafting and acid doping

where, k is the reaction constant at doping time t. Using data in Table 4 and performing regression, the values of k and a are 0.0126 (day−1) and 1, respectively. This suggests that acid doping in this study is a first order reaction. Moreover, the rate of the reaction is a function of acid concentration. The variation of acid doping level of the membranes with time at different temperatures is presented in Fig. 11. The doping rate was obtained graphically from the slope of each curve of doping temperatures and presented in Table 5. The doping rate showed a slight increase with the variation of the temperature from 0 to 100 1C. This suggests that performing the doping reaction at 30 1C is a reasonable option in this work from practical view point. To further understand the effect of temperature on the acid doping behavior, the Arrhenius plot was established as presented in Fig. 12, which depicts a plot of the log of the doping level versus the reciprocate of temperature. The doping level was found to follow the Arrhenius equation. The activation energy (Ea) of the acid doping that was calculated from the slope according to: −Ea/RT, found to be 5.52 kJ mol−1. Such low value of Ea reveals that PA doping with -N- of the imidazole ring takes place through diffusion that encounter a low resistance caused by the swelling

FTIR analysis was performed to provide an evidence for PA acid doping the membrane. Original and grafted ETFE films were used as references as shown in the FTIR spectra presented in Fig. 14. The spectrum of ETFE film displayed a number of strong bands in the range of 1000–1400 cm−1 resembling C-F of CF2 groups, in addition to, a small band at 2915 cm−1 representing stretching vibrations of CH2 groups. The grafted films prevailed additional characteristic bands representing –CN groups in the range of 1525–1575 cm−1. The bands at 2750–3150 and 3200–3650 cm−1 were respectively assigned for N–H and C–H of the grafted imidazole rings, which are mostly involved in a network of H-bonding between the imidazole rings and the moisture as expected from their amphoteric behavior. The spectrum of PA doped membrane showed broad bands in the range of 2100–3600 cm−1. This trend agrees completely with the behavior of other PA membranes reported in the literature and can be attributed to the formation of H-bonding and hydrogen phosphate groups [8,15]. Fig. 15 shows the wide survey spectra of PA acid doped membrane together with the original and grafted ETFE films. The presence of carbon, oxygen and fluorine in the original ETFE film

Fig. 14. FTIR spectra of: (a) original ETFE film, (b) 54% grafted ETFE film and (c) corresponding PA doped membrane.

430


can be seen in the spectrum A in forms of C1s, O1s and F1s at 285.0, 531.8 and 688.2 eV, respectively. The poly(1-VIm) grafted film spectrum (b) contains an additional nitrogen related peak (N1s) at 400 eV. The PA doped membrane showed other additional bands as depicted in the spectrum C resembling P (P1s and P2p) at 134.3 and 101.8 eV, respectively. Such results are in a good agreement with of XPS analysis of PA activated membranes reported in literature [16,17] and provide another evidence for

the grafting of VIm and subsequent incorporation of PA acid by doping of the membranes. 4.4. Ionic conductivity Fig. 16(a–d) shows the variation of proton conductivity with the acid doping level at various temperatures and RH%. It can be observed that the conductivity of all membranes gradually

Fig. 15. XPS wide scan spectra of: (a) original ETFE film, (b) 54% grafted ETFE film and (c) corresponding PA doped membrane.


431

Fig. 16. Variation of proton conductivity with acid doping level at various temperatures: (a) 30 1C, (b) 80 1C, 100 1C, 120 1C and relative humidity: 100%, 47%, 35% and 17% for PA doped poly(VIm) grafted membranes.

increases with the increase in the acid doping level at all temperatures and RH% values. For instance, the proton conductivity increases with the increase in the temperature from 30– 120 1C and the reduction in RH% from 100% to 17%. To explain this trend, it is necessary to understand the way proton conductivity takes place in acid doped membranes. The doping of poly(VIm) grafted film with PA led the formation of NH+ proton donor sites in the membranes causing H+ hopping to take place between NH+ sites and PA anions leading to continuous proton transfer. The increase in acid doping level increases the level of water molecules associated with PA by H-bonding. This allows other PA ionic species such as H2PO4− to contribute to the proton transfer leading to a remarkable increase in conductivity. As the temperature increases, the proton hopping increases leading to higher proton conductivity. The increase in the proton conductivity with the decrease in RH%, suggests that PA doped membrane obtained in this study is considered to be less-water dependent material. These results are going very well with the observation made by He et al. (2008) who reported that the acid molecules are bridging the phosphate and imidazole moieties allowing proton hopping even at low RH% levels [6]. Moreover, PA as a dopant exerts effective proton conductivity even in an anhydrous form, due to its unique proton conduction mechanism by self-ionization and self-dehydration. The effect of temperature, G%, and acid doping level reported for the present membranes are going along with those reported for PA doped PBI [14].

Fig. 17. Variation of ionic conductivity in relation with time for PA doped poly(VIm) grafted membranes at various acid doping level (a) 6.54, (b) 3.18 and (c) 1.29 mmol repeat unit−1 at 120 1C.

4.5. Thermal stability in terms of ionic conductivity loss Finally, the stability of the membranes was monitored in terms of variation of proton conductivity in correlation with time at

432


120 1C as shown in Fig. 17. It can be seen that all the membranes with various doping levels show almost no loss in proton conductivity over a time span of 8 h. This indicates that membrane PA loaded sites remain intact at the test conditions. This result is supported by TGA analysis for a single membrane sample which has G% of 54% reported earlier [11]. From this observation, it can be suggested that the membrane has strong potential for testing in PEM fuel cell operated above 100 1C.

wi wd t W T R

5. Conclusions

ETFE poly(ethylene-co-tetrafluroethylene) FTIR Fourier transform infrared HT high temperature PEMFC proton conducting membrane fuel cell PA phosphoric acid P(1-VIm) poly(1-vinylimidazole) 1-VIm 1-Vinylimidazole XPS X-ray photoelectron spectroscopy ANOVA analysis of variance

Optimization of PA doping of ETFE-g-P(1-VIm) membrane precursors obtained by radiation induced grafting of 1-VIm onto ETFE films was successfully performed using the Taguchi method. The highest doping level reached a value of 7.83 mmol repeat unit−1 at the optimum conditions of G% of 54%, acid concentration of 60%, doping temperature of 100 1C and doping time of 5 days. The predicted value was experimentally verified and found to be 7.45 (mmol repeat unit−1), making a reasonable deviation of 5%. Investigation of the kinetics of PA doping of ETFE-g-P(1-VIm) films revealed that the rate of the doping reaction was in the range of 0.327– 0.657 mmol imidazole ring −1 day−1 in a concentration range of 30– 60%. The doping reaction was found to be first order with a reaction constant of 0.0126 day−1. The activation energy of the doping was obtained from Arrhenius equation and found to be 5.52 kJ mol−1. The kinetics of acid doping was suggested to be a diffusion controlled process. The proton conductivity of the membranes was found to be dependent on doping level, temperature and RH%. Proton conductivities in the range of 56–140 mScm−1 at a temperature of 120 1C and RH% of 17% were achieved for membranes having G% ranging from 8% to 54%, respectively. The high proton conductivity at low RH% suggests that the obtained membranes are less-water dependent material. Acid doping levels suitable for HT-PEMFC application could be achieved. A reasonable thermal stability was also reported. Finally, it can be suggested that the Taguchi method is an effective tool for the optimization of the reaction conditions and predicting the level of acid doping during preparation of proton conduction membranes. This reduces consumption of chemicals, saves time and improves of the overall economy of the process. Acknowledgment The authors wish to thank Malaysian Ministry of Higher Education and Universiti Teknologi Malaysia for sponsoring this work under Research University Fund Scheme with a Grant no. 2544.02H98. A. Aly also acknowledges the financial support from the same research grant.

Nomenclature Ea dXPA/dt G L Li k Wo Wg Mp Md XPA

activation energy, kJ mol−1 rate of doping, mmol repeat unit−1 day−1 degree of grafting, % distance between probes, cm orthogonal arrays of i number of necessary trials in Taguchi, dimensionless reaction constant, day−1 weight of original ETFE film, g weight of grafted ETFE film, g molar mass of repeat unit, g mol−1 molar mass of PA, g mol−1 doping level mmol repeat unit−1

percentage of weight increase of the grafted films mass fraction of the dopant doping time, day width of the membrane sample, cm thickness, cm resistance, Ω

Abbreviations

References [1] M. Winter, R.J. Brodd, What are batteries, fuel cells, and supercapacitors? Chem. Rev. 105 (2005) 1021 –1021. [2] H. Zhang, P.K. Shen, Recent development of polymer electrolyte membranes for fuel cells, Chem. Rev. 112 (2012) 2780–2832. [3] S. Bose, T. Kuila, T.X.H. Nguyen, N.H. Kim, K. Lau, J.H. Lee, Polymer membranes for high temperature proton exchange membrane fuel cell: recent advances and challenges, Prog. Polym. Sci 36 (2011) 813–843. [4] R.K. Nagarale, W. Shin, P.K. Singh, Progress in ionic organic–inorganic composite membranes for fuel cell applications, Polym. Chem. 1 (2010) 388–408. [5] M.A. Hickner, H. Ghassemi, Y.S. Kim, B.R. Einsla, J.E. McGrath, Alternative polymer systems for proton exchange membranes (PEMs), Chem. Rev. 104 (2004) 4587–4612. [6] R. He, Q. Che, B. Sun, The acid doping behavior of polybenzimidazole membranes in phosphoric acid for proton exchange membrane fuel cells, Fibers Polym. 9 (2008) 679–684. [7] A.C. Dupuis, Proton exchange membranes for fuel cells operated at medium temperatures: materials and experimental techniques, Prog. Mater. Sci. 56 (2011) 289–327. [8] C. Schmidt, G. Schmidt-Naake, Proton conducting membranes obtained by doping radiation-grafted basic membrane matrices with phosphoric acid, Macromol. Mater. Eng. 292 (2007) 1164–1175. [9] M.M. Nasef, E. Samsaei, P. Ghassemi, A.A. Aly, A.H. Yahaya, Optimization strategies for radiation induced grating of 4-vinylpyridine onto poly (ethyleneco-tetrafluoroethylene) film using box-Behnken design, Radiat. Phys. Chem. 81 (2012) 437–444. [10] M.M. Nasef, A.A. Aly, H. Saidi, A. Ahmed, Optimization of reaction parameters of radiation induced grafting of 1-vinylimidazole onto poly(ethylene-cotetraflouroethene) using response surface method, Radiat. Phys. Chem. 80 (2011) 1222–1227. [11] M.M. Nasef, A.A. Aly, H. Saidi, Composite proton conducting membrane by radiation induced grafting of 1-vinylimidazole onto poly(ethylene-co-tetrafluoroethylene) and phosphoric acid doping, High Perform. Polym. 25 (2) (2013) 196–202. [12] A. Garcia-Diaz, D.T. Philips, Principles of Experimental Design and Analysis, Chapman & Hall, London, 1995. [13] R.S. Rama, G. Padmanabhan, Application of Taguchi methods and ANOVA in optimization of process parameters for metal removal rate in electrochemical machining of Al/5% SiC composites, Inter. J. Eng. Res. Appl. 2 (2012) 192–197. [14] M.M. Nasef, E. Shamsaei, H. Saidi, A. Ahmad, K.Z.M. Dahlan, Preparation and characterization of phosphoric acid composite membrane by radiation induced grafting of 4-vinylpyridine onto poly(ethylene-co-tetrafluoroethylene) followed by phosphoric acid doping, J. Appl. Polym. Sci. 128 (2013) 549–557. [15] X. Glipa, B. Bonnet, B.J. Mula, D. Jones, J. Roziere, Investigation of the conduction properties of phosphoric and sulfuric acid doped polybenzimidazole, J. Mater. Chem. 9 (1999) 3045–3049. [16] M.J. Ariza, E. Rodríguez-Castellón, R. Rico, J. Benavente, M. Muńoz, M. Oleinikova, X-ray photoelectron spectroscopy analysis of di-(2-ethylhexyl) phosphoric acid activated membranes, J. Colloid Interface Sci. 226 (2000) 151–158. [17] M.J. Ariza, E. Rodríguez-Castellón, M. Muńoz, J. Benavente, Surface chemical and electrokinetic characterizations of membranes containing different carriers by x-ray photoelectron spectroscopy and streaming potential measurements: study of the effect of pH. Surf, Interface Analyt. 34 (2002) 637–641.