Fabrication and characterization of brominated matrimid^® 5218 membranes for CO₂/CH₄ separation: application of response surface methodology (RSM)

2.1 Materials

Matrimid 5218 with a chemical structure as shown in Figure 1 was received from Huntsman advanced materials American Inc. (Los Angeles, CA, USA). Chloroform (99%) as a solvent and methanol (99.5%) as an extraction agent were purchased from Merck (Darmstadt, Hesse, Germany) and bromine (99.5%) as a reaction agent was provided from Sigma Aldrich (Saint Louis, MO, USA). Solvents and polymers were used as received.

Figure 1:

(A) Two-dimensional (2D) representation of matrimid 5218 and (B) three-dimensional (3D) representation of matrimid 5218.

2.2 Bromination reaction

All of the equipments and polymers were dried at 100°C overnight prior to the reaction. The reaction occurred in a 100 ml round-bottom flask equipped with a condenser and a magnetic stirrer at room temperature. Moreover, the reaction was accompanied by a nitrogen purge. Matrimid 5218 (4.00 g) was dissolved in chloroform (27.00 ml) under stirring for 2 h at ambient temperature. Afterward, bromine (14.00 ml) was added to the solution. The color of the solution changed from yellow to dark red after an injection of bromine. The mentioned solution was stirred for 20 h in order to reassure that good connection between bromine and polymer chains occurred. In the meantime, a dark gum precipitated from the reaction mixture. It means that the reaction was carried out completely. Finally, methanol (100.00 ml) was poured dropwise into the flask under stirring in order to precipitate yellow modified polymers. The color of methanol became dark eventually. Precipitated polymers were stirred in fresh methanol again until no bromine could be extracted. Colorless methanol symbolizes that all of the trapped bromine has been extracted. After vacuum drying the brominated matrimid is ready to use. Reaction of bromine with matrimid is exhibited in Figure 2. Three dimensional (3D) representation of brominated matrimid is demonstrated in Figure 3.

Figure 2:

Bromination reaction representation (17).

Figure 3:

Three-dimensional (3D) representation of brominated matrimid (Br represents, grafted bromine to polymer matrix).

2.3 Pristine and modified membranes preparation

A total of 0.35 g matrimid 5218 was dissolved in 10.00 ml chloroform in a 50 ml round-bottom flask under stirring at ambient temperature for 1 h. The yellow solution was placed in an ultrasonic bath for 1 h in order to obtain a homogeneous polymer solution. After the solution de-bubbled, it was filtered and then poured into a Petri dish. The film (which formed after solvent evaporation at ambient temperature) was separated from the glass substrate and dried in a vacuum oven at 70°C for 1 h. Figure 4 depicts the scanning microscopy image (SEM) image (surface and cross-section) of the pristine membrane. As can be seen, there are no voids on the surface of the membrane which was confirmed by permeation tests. It means that no unusual permeation through (modified or unmodified) membranes were observed.

Figure 4:

SEM images of pristine matrimid membrane [(A) surface and (B) cross section].

Membranes which were prepared from brominated matrimid were too fragile and they ruptured during gas permeation experiments especially when the operating pressure was high, so brominated matrimid was added to the primary matrimid in order to fabricate modified membranes. Brominated matrimid was added from 10% (of membrane weight) up to 50%, however, results of 20% and 50% were reported only as remarkable conclusions. The preparation procedure of modified and unmodified membranes was similar.

2.4 Characterization methods

Membrane morphologies were studied using SEM images. They were obtained using a VEGA II TESCAN and ZEISS instruments. The tensile strength of pristine and modified membranes was measured using samples on a Zwick/Roell instrument. Fourier transform infrared (FTIR) analysis of matrimid and brominated matrimid was recorded employing a Bruker FTIR spectrometer. Thermal stability of primary matrimid and brominated matrimid was investigated using a Mettler-Toled thermogravimetry analyzer (TGA/SDTA851e). The samples were heated from 25 to 1000°C at a heating rate of 10°C/min under a nitrogen purge. Glass transition temperature (Tg) of modified and unmodified polymers was measured employing a DSC7 (Perkin Elmer Instruments DSC 8000). The samples were heated from 30 to 450°C at a heating rate of 10°C/min under a nitrogen purge.

2.5 Permeability measurements

Prior to gas permeation experiments, membranes were placed in a vacuum oven for 24 h at 40°C. Pristine and modified membranes permeabilities were investigated thorough a gas separation membrane unit which is illustrated in Figure 5. The permeability tests were carried out giving priority to CH₄ and the permeability tests of CO₂ gas were performed a day after. After each measurement the membrane was kept in the vacuum oven overnight. During gas permeation through the membrane, the feed stream was divided into two streams: 1. permeate stream which is a part of feed stream selectively passing through the membrane, 2. retentate stream which is a part of feed stream that does not pass through the membrane. Moreover, another stream that can be mentioned for a membrane separation unit is the sweep stream which is a stream to help permeate gas to leave the membrane cell easier (1).

Figure 5:

Schematic of gas separation membrane unit.

All gas permeation experiments were done at constant pressure. Pure gases (CH₄ and CO₂, Roham Company, 99.9%) were used. Gas permeation tests were done at ambient temperature and applied pressure was in the range of 2 up to 12 bars. In order to achieve steady state conditions, after the membranes were placed in the cell, all of membranes were kept at the pressure of 2 bars for 1 h. The permeate gas flow was measured using an AALBORG flow meter (with an accuracy of 0.01 ml/min).

The permeability of a membrane is calculated from equation 1:

where P is the permeability in Barrer [1 Barrer=10⁻¹⁰ cm³ (STP) cm/cm² s cmHg], Q is the volumetric flow rate [cm³ (STP)/s], L is the membrane thickness [cm], Δp is the pressure difference between two sides of the membrane [cmHg] and A is the effective membrane area [cm²] (2).

2.6 Calculations of gas permeations and fractional free volume (FFV)

Permeation of gases through a dense polymeric membrane occurs via the solution-diffusion mechanism (2). Gases are adsorbed by the polymeric membrane at certain pressures (p_feed). Then, adsorbed gases are diffused through effective membrane thickness and finally on the other side of membrane, gases are desorbed at low pressure (p_perm) (p_feed>p_perm). The transport rate (flux, J) is obtained by equation 2:

D_A and S_A are the diffusion and sorption coefficients of component A, respectively. L is the effective membrane thickness. The permeability coefficient of component A (P_A) is obtained by multiplying D_AS_A. Ideal selectivity is obtained by equation 3 (2):

For FFV calculations, densities of the matrimid and brominated matrimid were measured by the displacement method using a Mettler densitometer. The measurements were carried out using anhydrous ethanol as a displacement fluid at 25°C. For FFV calculations, the following equation 4 was used:

where V is the specific volume of a polymer matrix (can be calculated from density), and V₀ is the van der Waals volume that can be calculated by the group contribution method as outlined in Bondi (20).

2.7 Experimental design

Application of RSM as a statistical and mathematical method can lead to optimizing the process. The interactions between operational parameters (Br₂ concentration in modified membranes and operating pressure) were investigated using the RSM. In addition, this method provides the minimum number of tests by taking into account the sensitive points of the design (21), (22), (23), (24). A summary of the design, factors and responses are listed in Table 1. It should be noted that CO₂ permeability (Y₁) and CO₂/CH₄ ideal selectivity (Y₂) were considered as responses.

In order to predict the optimum points in the experiments, the following equation was used which is shown as follows:

where η is response value, β₀ is a constant coefficient, e_i is the experimental error, X_i and X_j are variables and β_j, β_jj and β_ij are linear, square and cross effects, respectively (24). The design matrix for modified membrane preparation is listed in Table 2, which was conducted using the software. In other words, Design-Expert version 7 provided from State Ease (Minneapolis, MN, USA) was used to calculate regression and graphical analysis of data.

3.1 TGA analysis and Tg

Thermal resistance of matrimid and brominated matrimid were studied using TGA analysis. TGA diagrams of pure matrimid and brominated matrimid are indicated in Figure 6. Degradation temperatures of pristine and modified polymers are over about 350°C. There are no significant weight losses at temperatures below about 450°C. Both polymers have weight losses <10% at temperatures below 300°C. The main reason for this observation can be referred to the early decomposition of impurities, water and low molecular weight polymers within the samples. The main weight loss for all of the samples occurs at the temperatures between 450°C and 700°C. According to the results, thermal stability of brominated matrimid is less than pure matrimid.

Figure 6:

Thermal degradation curves of matrimid and brominated matrimid.

Glass transition temperature (Tg) of matrimid and brominated matrimid are tabulated in Table 3 (Figure 7). Tg of both polymers was determined by using DSC analysis. Tg of brominated matrimid is higher than pure matrimid, it can be attributed to the bulky bromine atoms which are joined to polymer chains and they intercepted rotation of polymer chains which were exposed to the heat treatment. Moreover, by employing more precision it can be observed that Tg of brominated matrimid approaches the degradation temperature.

Figure 7:

Differential scanning calorimetry curves of matrimid and brominated matrimid.

3.2 FTIR analysis

The FTIR spectrum of pristine membrane is shown in Figure 8A. Aromatic C-H stretching and bending vibrations are indicated via bands at 3063 cm⁻¹ and 721 cm⁻¹, respectively. Bands at 2850–2970 cm⁻¹ show the presence of aliphatic C-H stretching vibrations. The bands at 1779 cm⁻¹ and 1736 cm⁻¹ represent presence of asymmetric and symmetric C=O stretching vibrations in the imide group, respectively. The band shown at 1382 cm⁻¹ indicates C-N stretching vibration in the imide group. The multiple bands in the region of 1400–1600 cm⁻¹ are stood for C=C stretches (14), (16).

Figure 8:

(A) FTIR spectrum of matrimid 5218 and (B) FTIR spectrum of brominated matrimid.

FTIR spectrum of brominated matrimid is shown in Figure 8B. As can be observed from the IR spectrum, there are peaks which overlap so that it is hard to find the newly appeared peak responsible for aromatic C-Br. For this reason, we take the advantage of the red shifts of the C=C bands that occurred after the bromination reaction in the region of 1400–1600 cm⁻¹ as a proof of bromine addition to the aromatic ring (see Figure 8).

3.3 Tensile strength analysis

Tensile strength measurements of pristine and modified (20% brominated matrimid − 80% matrimid) membranes were examined. For each sample, the average values of tensile measurements were reported. As can be seen from Table 4, modified membranes were ruptured at lower stress than pristine membranes, implying that the tensile strength of pristine membranes is higher than modified ones. In fact, the connection between bromine and the polymer matrix leads to weaken polymer chains. So, modified membranes become fragile by adding brominated matrimid.

3.4 Fractional free volume (FFV) calculations

Grafting bromine to matrimid can lead to the increment in FFV of modified polymers. In fact, when bulky bromine molecules graft to polymer matrix, it leads to the looseness in the interchain packing of the polymer. As a result, some new spaces will form within the polymer chains and thus the permeation of gases through the brominated matrimid will dramatically rise (17). Guiver et al. also calculated FFV of their modified polymers (brominated matrimid), the enhancement in FFV was reported by them as being from 0.11 to 0.14 (17). Xiao et al. demonstrated that derived carbon membranes from brominated matrimid were more permeable than ones derived from pure matrimid. In other words, they attributed this observation to the amount of membranes FFV (18). Additionally, McCaig et al. reported that grafting bromine to the polymer matrix led to the increase in FFV of modified polymers (19).

It is worth noting that, free spaces within brominated matrimid enhanced that even large molecules (CH₄) can easily pass through the modified membrane. Table 5 exhibits the effect of modification on the values of density and calculated FFV. As can be seen, the density of the brominated matrimid increases by introducing Br₂ to the matrimid structure; it leads to the change in density from 1.19 g/cm³ to 1.3 g/cm³. For matrimid, FFV is calculated at about 0.190 while for brominated matrimid, it is obtained at about 0.208. For this reason, permeation through modified membranes has been increased. Gas permeation experiments support the results of this section. As a matter of fact, the increment in permeation of single gases through membranes can be attributed to the rise in fractional free volumes of brominated matrimid.

3.5 Gas permeation evaluations

Table 6 exhibits permeation tests of pristine and modified membranes (25°C). As it can be seen in the table, CO₂ and CH₄ permeation through modified membranes boost in comparison with the pristine membrane. It is worth noting that, by increasing the content of brominated matrimid in modified membranes (from 20% to 50%); permeation of pure gases was enhanced. In fact, combination of bulky bromine atoms with polymers chains can lead to the creation of some novel free volumes across the modified membranes. For this reason, gas permeation through modified membranes rises in comparison with unmodified ones. As regards ideal selectivity as mentioned before, there is often an inverse relation between permeability and ideal selectivity. Ideal selectivity of modified membranes decreased compared to the pristine membranes, so the mentioned relation is true here. As it is evident in Table 6, ideal selectivity of modified membranes which contain 50% of brominated matrimid is less than the other modified membranes. In short, the increase in permeability can be attributed to greater FFV values of the modified membranes.

3.6 Effect of feed pressure on permeability and ideal selectivity

The effects of feed pressure on permeabilities and ideal selectivities of pristine and modified membranes were evaluated (Figure 9). As can be observed from Figure 9A, CO₂ permeability of pristine and modified membranes first decreases to a minimum point and then it increases. In other words, the minimum pressure in each graph is plasticization pressure (Figure 9A). The plasticization pressure occurs because of the interaction between polymer and CO₂ gas. Immediately after sorption of CO₂, the polymer swells and causes an increase in gas permeability. In short, by enhancing feed pressure, below plasticization pressure, CO₂ permeability declines and conversely, CO₂ permeability increases above that pressure (Figure 9A). As can be seen from Figure 9B, CH₄ permeability of pristine and modified membranes is relatively stable despite the rise in ΔP_CH4 which leads to the slight increment in Q_CH4. Modified membranes which contain 50% of brominated matrimid are much more permeable than other modified and pristine membranes. Generally, if the content of brominated matrimid in modified membranes increases, permeability of the modified membrane will improve and it can be attributed to the enhancement in FFVs within modified membranes.

Figure 9:

Effect of feed pressure on (A) CO₂ permeation, (B) CH₄ permeation and (C) CO₂/CH₄ Ideal selectivity.

CO₂/CH₄ ideal selectivity of modified and pristine membranes improves by increasing feed pressure (Figure 9C). As can be seen from the figure, ideal selectivity of the pristine membrane is more than modified membranes. Ideal selectivity of modified membranes decreases by enhancing the content of brominated matrimid in modified membranes. As a matter of fact, by increasing the content of brominated matrimid, permeation of both single gases (CH₄ and CO₂) in modified membranes will increase. However, according to equation 3 the ratio of CO₂/CH₄ will decrease. For this reason, ideal selectivity of modified membranes which contain 50% of brominated matrimid is less than in modified membranes which contain 20% of brominated matrimid.

3.7 Comparative assay

Guiver et al. could improve permeability of CO₂ and CH₄ using brominated matrimid membranes instead of matrimid membranes. In fact, CO₂ and CH₄ permeability improved from 8.7 and 0.24 Barrer to 14 and 0.42 Barrer, respectively (17). It is worth noting that results of Guiver’s study were approximately achieved in the present study (Table 6, modified membranes which were made up of 50% brominated matrimid+50% matrimid). However, the marked difference between two works is that, herein brominated matrimid was added to pure matrimid (from 10% up to 50% of membrane weight) to enhance permeation of single gases thorough membranes. While Guiver et al. prepared modified membranes just by using brominated matrimid. In other words, 100% weight of their membrane was made up of brominated matrimid. As a result, in this work a lower amount of brominated matrimid was employed to increase gas permeation of membranes.

On the other hand, concerning tensile strength analysis (Table 4), combining matrimid with brominated matrimid can improve the mechanical strength of membranes in comparison with ones which were just prepared from brominated matrimid. In other words, membranes prepared here can bear high pressures during experiments.

3.8 Optimization of modified membranes fabrication using RSM

As mentioned in Introduction section, Design-Expert software was aimed to optimize the process. Also, it provides a large number of models which can be selected to improve the efficiency of the process (24). The effects of Br₂ concentration (within modified membranes) and operating pressure on CO₂ permeability and CO₂/CH₄ ideal selectivity were investigated. The coefficients of the model developed for two responses, were estimated using multiple regression analysis on the experimental data. CO₂ permeability and CO₂/CH₄ ideal selectivity as functions of the independence variables (Br₂ concentration and pressure) are considered as follows:

where A, B are two independent variables and Y₁ (CO₂ permeability) and Y₂ (CO₂/CH₄ ideal selectivity) are responses. The results of the quadratic model for CO₂ permeability and the linear model for CO₂/CH₄ ideal selectivity in the shape of analysis of variance (ANOVA) are tabulated in Tables 7 and 8, respectively. With respect to the both tables, the values of R² and adjusted R² are next to 1.0 which is very high and it represents a high correlation between independent variables and responses. In other words, the regression model provides an excellent description of the relationship between observed values and the predicted values. As can be observed in Table 7, the F-value for a quadratic model is 19.36188, this confirms that the quadratic model is valid for experimental data. Precision more than 4 symbolizes accurate calculations; herein precision is reported 14.99847 which is excellent and an acceptable accuracy. p-Values <0.05 indicates that the model is consistent with experimental data however, p-values more than 0.1 represent less consistency between the model and the experimental data. Concerning Table 8, F-value and precision for second response (Y₂) are 722.57 and 78.460, respectively. In accordance with the above results, it is evident that the quadratic model can be used to predict CO₂ permeability and the linear model for anticipating CO₂/CH₄ ideal selectivity is completely applicable (25), (26), (27).

3.9 Optimization of operating parameters

The data were plotted to evaluate the correlation between the actual data (or experimental data) and predicted responses (Figure 10). As can be observed, the points are well distributed adjacent to the straight line with the R² values of about 0.993 (for CO₂/CH₄ ideal selectivity) and 0.932 (for CO₂ permeability), which implies an excellent relationship between the experimental data and predicted responses for CO₂ permeability and CO₂/CH₄ ideal selectivity. The results also symbolize that the quadratic and linear models were appropriately selected for presuming responses for the experimental data.

Figure 10:

Normal % probability per internally studentized residual for (A) CO₂ permeability and (B) CO₂/CH₄ ideal selectivity.

The effects of feed pressure and concentration of brominated matrimid in modified membranes on CO₂ permeability and CO₂/CH₄ ideal selectivity are depicted in Figures 11 and 12, respectively. As can be seen in Figure 11, the increase in feed pressure leads to the decrease in CO₂ permeability and then it leads to the increment in CO₂ permeability. In addition, the rise in content of brominated matrimid in modified membranes leads to the enhancement in CO₂ permeability. Given Figure 12, CO₂/CH₄ ideal selectivity increases steadily by rising feed pressure. Moreover, elevating the content of brominated matrimid in modified membranes leads to the decline in CO₂/CH₄ ideal selectivity. It is worth noting that these results were also explained in more details in Section 3.6. As a consequence, the optimum pressure and concentration (percent of brominated matrimid in modified membranes) are 12 bar and 50%, respectively. In this case, maximum permeation of single gases through modified membranes can be achieved.

Figure 11:

The effects of feed pressure and concentration of brominated matrimid on CO₂ permeability, (A) Surface plot and (B) counter plot.

Figure 12:

The effects of feed pressure and concentration of brominated matrimid on CH₄ permeability, (A) surface plot and (B) counter plot.