Ankit M. Kansaraab,
Vinod K. Aswalc and
Puyam S. Singh*ab
aCSIR-Central Salt & Marine Chemicals Research Institute, RO Membrane Discipline, G. B. Marg, Bhavnagar-364 002, Gujarat, India. E-mail: puyam@csmcri.org; Fax: +91-278-2567562; Tel: +91-278-2566511
bAcademy of Scientific and Innovative Research (AcSIR-CSMCRI), G. B. Marg, Bhavnagar-364002, Gujarat, India
cSolid State Physics Division, Bhabha Atomic Research Centre, Trombay, Mumbai-400085, India
First published on 22nd May 2015
Poly(dimethylsiloxane) (PDMS) membrane, which is generally prepared by cross-linking a hydroxyl-terminated-poly(dimethylsiloxane) liquid with a polymethylhydrosiloxane cross-linker through a condensation reaction between the hydroxyl end groups and hydride groups liberating hydrogen, is one of the most studied polymeric membranes for the separation of gases and liquids. Herein, a new PDMS membrane series prepared by direct cross-linking hydroxyl terminated polydimethylsiloxane pre-polymer liquid with RSiCl3 molecules of different types in n-heptane solvent under a nitrogen environment were studied, wherein the alkyl chain, R, was varied as methyl CH3, octyl C8H17, perfluorooctyl C8H4F13, and octadecyl C18H37. For each membrane series, the amount of cross-linker to pre-polymer was varied as 13:87, 33:67 and 50:50 (w/w) to compare the membranes at different cross-linking densities. The cross-linked network structure of the membrane comprised dimethylsiloxane network structures of two cross-links and alkylsiloxane network structure of two or three cross-links. The changes in the structure and properties of the membranes depending on the concentration and type of trichloro(alkyl)silane used were observed by XRD, SEM, TEM, SANS, TGA, DSC, 29Si NMR, IR, cross-linking density, contact angle (water) and ethanol–water separation performance measurements. The membranes with an optimal amount of trichloro(alkyl)silane cross-linker of approximately 10–30% (w/w) showed better separation performance than the other reported conventional PDMS membranes in terms of the ethanol selectivity and flux from an aqueous feed containing 5% (w/w) ethanol. Among the membranes prepared, the membranes cross-linked with perfluorooctylsiloxane/octadecylsiloxane showed better separation performance than those membranes cross-linked with methylsiloxane/octylsiloxane.
It is generally prepared by cross-linking the hydroxyl-terminated-poly(dimethylsiloxane) liquid pre-polymer with a polymethylhydrosiloxane cross-linker4–12 through a condensation reaction between the hydroxyl end groups and hydride groups in the presence of a reduction catalyst liberating hydrogen. The properties of the PDMS membrane could be influenced greatly by the type of solvent used in the preparation, cross-linking reaction temperature, and curing conditions, including degassing prior to curing and curing at different temperatures9 as well as the membrane thickness.13 The thin membrane film coated on a porous support normally exhibits a less-dense structure with a larger amount of chain aggregates. Alkoxysilane such as tetraethoxysilane Si(OR)4 (ref. 14–22) is also used widely as a cross-linker, in which the hydroxyl end groups of the PDMS pre-polymer liquid are cross-linked multi-dimensionally with the ethoxy groups (–OR) of the monomolecular Si(OR)4 unit. By choosing trialkoxysilane R′Si(OR)3, where R′ is a phenyl23 or vinyl group,24 PDMS membranes with different properties as a result of different distributions of network chain lengths could be obtained. Recently, we prepared improved PDMS membranes via a cross-linking reaction between the hydroxyl terminated polydimethylsiloxane macromolecules and polymethylhydrosiloxane macromolecules along with a small amount of monomolecular cross-linker either of trichloro(n-octadecyl)silane or trichloro(1H,1H,2H,2H-perfluorooctyl)silane in toluene solvent, which showed better separation performance as compared to the typical PDMS membrane in the removal of volatile organics from aqueous streams.25 Some disadvantages of these monomolecular organosilane cross-linkers include the formation of aggregated silica particles by the hydrolysis of the organosilane with moisture present in the environment. By carrying out the reaction under a moisture-free nitrogen environment, the hydrolysis of the organosilane can be almost prevented.
In the present study, a new PDMS membrane series prepared by direct cross-linking hydroxyl terminated polydimethylsiloxane liquid with RSiCl3 molecules of different types in n-heptane solvent under a nitrogen environment were studied, wherein the alkyl chain, R, was varied as methyl CH3, octyl C8H17, perfluorooctyl C8H4F13, octadecyl C18H37. For each membrane series, the amount of cross-linker to pre-polymer was varied as 13:87, 33:67 and 50:50 (w/w) to compare the membranes at different cross-linking densities.
Sample series | Octadecyltrichlorosilane | Perfluorooctyltrichlorosilane | Octyltrichlorosilane | Methyltrichlorosilane | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
PDMS-ODS1 | PDMS-ODS2 | PDMS-ODS3 | PDMS-FOS1 | PDMS-FOS2 | PDMS-FOS3 | PDMS-OS1 | PDMS-OS2 | PDMS-OS3 | PDMS-MS1 | PDMS-MS2 | PDMS-MS3 | |
Cross-linker (alkyltricholorosilane), wt% | 13 | 33 | 50 | 13 | 33 | 50 | 13 | 33 | 50 | 13 | 33 | 50 |
Base polymer liquid (PDMS), wt% | 87 | 67 | 50 | 87 | 67 | 50 | 87 | 67 | 50 | 87 | 67 | 50 |
I(Q) = ϕΔρ2VpP(Q)S(Q) | (1) |
(2) |
The Debye–Anderson–Brumberger (DAB) equation is a Lorentzian-square term that can be used to calculate the scattering from a randomly distributed (non-particulate) two-phase system.27 A measure of the average spacing between two phase regions is given by ξDAB.
(3) |
The scattering function given by the sum of a Lorentzian and a Lorentzian-squared term is the prediction of Random Field Ising Model as
(4) |
The Random Field Ising Model can describe the correlation from the interactions between the polymer chain segmental units, wherein the correlation length varies between tens and hundreds of angstroms.28 Furthermore, the Gaussian characteristics for the polymer chains can be studied using the Kratky plot [Q2I(Q) vs. Q], as described elsewhere.29
The separation performance of the membranes was evaluated through pervaporation experiments for the removal of organics from aqueous solutions. The experiments were carried out with laboratory fabricated unit described elsewhere9 at 70 °C using aqueous feed solution containing ethanol. The effective membrane area was 18.86 cm2. The downstream pressure was maintained at ∼2 mbar generated by a two stage vacuum pump (Alcatel, France) and measured using a MKS Series 925 pressure transducer with micro-pirani readout. The permeate was collected in a cold trap cooled in liquid nitrogen. Before each measurement, the pervaporation system was stabilized for one hour and the initial permeate of one hour was discarded. The feed and permeate concentrations were measured using an offline GC (Perichrom GC, France). The separation factor of the mixture is expressed as,
(5) |
Both the PDMS-MS and PDMS-OS membrane series cross-linked with trichloro(methyl)silane and trichloro(octyl)silane were very transparent at all the compositional (cross-linker:pre-polymer) ratios studied. Relatively, the film transparency was lower for the PDMS-FOS and PDMS-ODS samples cross-linked with larger amounts (33–50 wt%) of trichloro(perfluorooctyl)silane or trichloro(octadecyl)silane. The highest opaque among the membrane films was the sample made with 50 wt% of trichloro(octadecyl)silane. The film opaqueness is expected if aggregates or clusters are formed in the membrane film. This is possible if some of the trichloro(alkyl)silane molecules themselves polymerized to poly(alkylsiloxane) as a result of the hydrolysis reaction of trichloro(alkyl)silane with the moisture present in the system environment. The formation of poly(alkylsiloxane) upon the hydrolysis of trichloro(alkyl)silane, followed by the condensation and polymerization reactions of the hydrolysed species is well-known,30 as shown in reactions 1, 2 and 3 of Scheme 2.
Scheme 2 Formation of poly(alkylsiloxane) upon hydrolysis of trichloro(alkyl)silane, followed by condensation and polymerization. |
XRD patterns of the PDMS membrane samples, as shown in Fig. 2, also indicated a change in membrane structure after cross-linking with trichloro(alkyl)silane. A broad diffraction peak at 2θ value of about 12.1° was observed for the samples prepared from the cross-linking reaction with small amounts of trichloro(alkyl)silane (13 wt%), suggesting the partly crystalline nature of PDMS. The intensity of the peak was lower for the samples obtained from the reaction with a higher amount (33–50 wt%) of the trichloro(methyl)silane, while this peak disappeared for the samples obtained from the reaction with 33–50 wt% of the trichloro(octyl)silane, trichloro(perfluorooctyl)silane, or trichloro(octadecyl)silane.
The XRD results suggested that the PDMS membranes formed by crosslinking with trichloro(methyl)silane had crystalline regions even from the reaction with 50 wt% trichloro(methyl)silane. This is in contrast to the others, which exhibited a completely amorphous nature of PDMS after the reaction with 33–50 wt% trichloro(alkyl)silane. On the other hand, the sample cross-linked with 33–50 wt% of trichloro(octadecyl)silane showed an additional broad peak at about 21.3°, which is referred to as poly(octadecylsiloxane).30 Changes in membrane structure in terms of the microstructure of the samples cross-linked with different amounts of trichloro(octadecyl)silane were clearly distinguishable in the SEM images, as shown in Fig. 3(A1–A3). The PDMS-ODS3 sample cross-linked with 50 wt% of trichloro(octadecyl)silane showed a rugged surface in micrometre-length-scale compared to relatively smooth surface of the PDMS-ODS1 and PDMS-ODS2 samples prepared with 13 and 33 wt% trichloro(octadecyl)silane. All other sample series (PDMS-MS, PDMS-FOS and PDMS-OS) cross-linked with trichloro(methyl)silane trichloro(octyl)silane and trichloro(perfluorooctyl)silane exhibited a relatively smoother surface than the PDMS-ODS samples. The rugged surface morphology of PDMS-ODS3 may be related to the hybrid structure formed by PDMS and poly(octadecylsiloxane) components. The poly(octadecylsiloxane) is crystalline and so it is harder than the PDMS component. Therefore, it is likely that poly (octadecylsiloxane) crystalline components in excess are separated from the PDMS components. Indeed, nanoscale particles presumably of the poly(octadecylsiloxane) material were clearly visible by TEM in the samples prepared by cross-linking with 33–50 wt% of the trichloro(octadecyl)silane.
Fig. 3 A1, 2, 3 are SEM surface images, while B1, 2, 3 are TEM images of PDMS-ODS1, 2, 3. respectively. |
The TEM images of the PDMS-ODS samples are shown in Fig. 3(B1–B3). Large aggregations of these nanoparticles were clearly observed in the TEM image of the PDMS-ODS3 sample cross-linked with 50 wt% of trichloro(octadecyl)silane. Although such large nanoparticle aggregates were not observed in the other membrane series, the formation of nanoparticles in the membrane series cross-linked with excess trichloro(alkyl)silane was evident from the TEM studies of the samples. SANS measurements of the samples were performed to probe membrane nanostructure in detail.
Sample details | Ornstein–Zernike | Debye–Anderson–Brumberger | ||
---|---|---|---|---|
ξOZ, Å | I(0)OZ, cm−1 | ξDAB, Å | I(0)DAB, cm−1 | |
PDMS-ODS1 | 10.1 | 0.7 | 24.2 | 1.4 |
PDMS-ODS2 | 11.9 | 1.1 | 24.3 | 1.8 |
PDMS-ODS3 | 11.4 | 1.1 | 62.9 | 7.6 |
PDMS-FOS1 | 9.4 | 0.4 | 31.8 | 1.0 |
PDMS-FOS2 | 10.1 | 0.5 | 43.8 | 2.1 |
PDMS-FOS3 | 8.9 | 0.4 | 46.5 | 3.2 |
PDMS-OS1 | 7.5 | 0.3 | 19.9 | 0.5 |
PDMS-OS2 | 9.3 | 0.6 | 50.6 | 23.9 |
PDMS-OS3 | 7.0 | 0.5 | 52.7 | 31.8 |
PDMS-MS1 | 7.6 | 0.4 | 44.0 | 1.0 |
PDMS-MS2 | 7.3 | 0.3 | 44.6 | 1.2 |
PDMS-MS3 | 8.7 | 0.6 | 46.4 | 2.1 |
Kratky plots [Q2I(Q) vs. Q] for the membranes (as shown in Fig. 5) were compared to identify the structural differences in Gaussian characteristics of the scattering chains. All the plots were inclined to a horizontal asymptote implying the Gaussian statistics of the polymer chain.
Fig. 5 Kratky plots [Q2I(Q) vs. Q] for the membranes of the lowest and highest alkylsiloxane content. |
Two important observations from the Kratky plots are (1) the Q value at which the plateau value reached was lower for the PDMS-ODS membrane series than the other membrane series and (2) a peak at a low Q value in each of PDMS-OS2 and PDMS-OS3 membrane was clearly observed. This indicated that the Gaussian segmental length of the polymer chain networks was longer for the PDMS-ODS membrane series compared to the other membrane series because the plateau Q value is inversely proportional to the Gaussian segmental length. On the other hand, the low Q peak observed in case of PDMS-OS2 and PDMS-OS3 may correspond to polymer mesh size or aggregated cluster of polymer chains formed by localized cross-linking. Such a low Q peak was also observed in some PDMS membranes.12 Upon closer examination of the Kratky plots, a shoulder peak at low Q was apparently present in the PDMS-ODS3, all the PDMS-FOS series and all the PDMS-MS series, implying some degree of polymer chain aggregation in most of the samples.
Fig. 7 shows DSC plots of all the alkylsiloxane cross-linked PDMS membrane samples. The glass transition temperatures (Tg) for all the membranes were similar at about −100 °C while the melting peaks (Tm) were in the temperature range from −48 °C to −40 °C.
The presence of both the Tg and Tm peaks in the membranes indicated the semi-crystalline nature of the membrane material because the Tg peak corresponds to the amorphous region, while the Tm peaks correspond to the crystalline region. For each membrane series, the intensity of the Tm peak for the membrane decreased by an increase in the alkylsiloxane content. Furthermore, shifts of the Tm peaks towards lower values were observed for the PDMS-ODS (−40 °C to −45 °C), PDMS-FOS (−40 °C to −48 °C) and PDMS-OS (−42 °C to −44 °C) membranes of higher alkylsiloxane contents except the PDMS-MS samples, which had the same Tm at about −42 °C. The decrease in the intensity and shifts to lower value of the Tm peak indicated that the crystallinity of the initial PDMS structure was decreased by cross-linking with alkylsiloxane. For the cross-linked structure, the segmental mobility of the polymer chain is expected to be decreased, resulting in a decrease in heat capacity.
The density of the membrane structure in terms of the cross-linking density (Vc) was examined by the membrane swelling in carbon tetrachloride. From the swelling experiments, the volume fraction of polymer (ϕp) was estimated. The Vc values were calculated from the ϕp values using the Flory–Rehner equation for all the membrane series. The Mc value for the sample was calculated based on the relation, Mc = density × Vc−1, in which the density of the PDMS membrane was taken as the density of pure polydimethylsiloxane material, 0.97 g cm−3.31 The results are given in Table 3.
Membrane | ϕp | Vc × 10−5 mol cm−3 | Mc g mol−1 |
---|---|---|---|
PDMS-ODS1 | 0.139 | 8.3 | 11563 |
PDMS-ODS2 | 0.204 | 19.0 | 5098 |
PDMS-ODS3 | 0.233 | 25.8 | 3761 |
PDMS-FOS1 | 0.176 | 13.8 | 7042 |
PDMS-FOS2 | 0.227 | 24.2 | 4011 |
PDMS-FOS3 | 0.258 | 32.6 | 2972 |
PDMS-OS1 | 0.147 | 9.4 | 10268 |
PDMS-OS2 | 0.225 | 23.8 | 4068 |
PDMS-OS3 | 0.315 | 53.3 | 1820 |
PDMS-MS1 | 0.109 | 5.1 | 18844 |
PDMS-MS2 | 0.127 | 7.0 | 13857 |
PDMS-MS3 | 0.133 | 7.7 | 12570 |
Among all the membranes, the Vc value 5.1 × 10−5 mol cm−3 of the PDMS-MS1 was the lowest. The density of the membrane structure was affected by the type of trichloro(alkyl)silane used as well as the trichloro(alkyl)silane concentration. The membrane structure density in terms of the ϕp, Vc and Mc values was the highest for the PDMS-FOS1 at a low loading (13 wt%) of trichloro(alkyl)silane. In contrast, at a high loading (50 wt%) of trichloro(alkyl)silane, the highest Vc 53.3 × 10−5 mol cm−3 with the corresponding Mc = 1820 g mol−1 was observed for the PDMS-OS3, which was about 10-times that of the corresponding value of PDMS-MS1. PDMS-MS sample series exhibited a relatively loose membrane structure with a Vc value range of 5.1–7.7 × 10−5 mol cm−3.
The infrared spectra of the membrane series are given in Fig. 9. A broad absorption band at about 3230 cm−1 due to O–H stretching vibrations of the absorbed moisture, was observed in all the spectra. The absorption band due to the O–H the stretching vibration can be at a higher wave number range if they are from structural O–H group of the materials, such as silanol alkylsiloxane structures (HO) (R)Si(OSi)2 of the membranes. Depending on the nature of OH groups, such as free OH groups, hydrogen bonded pairs (OH⋯HO), and perturbed OH, various structural OH absorption bands can be present in the spectra.34
This is what has been observed in all the spectra, which showed three weak bands at about 3600 cm−1, 3723 cm−1 and 3820 cm−1 due to the silanols (T2 units) in the network structure of the membranes. The bands in the wavenumber range of 1500–1690 cm−1 were assigned as (Si)O–H⋯O(Si) vibrations, in which the (Si)O–H group forms hydrogen bonds with the oxygen of the siloxanes and bending mode of absorbed water. Such water and hydroxyl-related bands were observed in the IR spectra of silicate glasses.35 The strong absorption band observed at 800 cm−1 is due to the symmetric stretching vibration of Si–O–Si, rocking alkyl group and stretching vibration of Si–C bond of the alkylsiloxane structure of the membranes. The asymmetric stretching vibration of the Si–O–Si (siloxane) bonds for the membranes was observed at 1020 and 1090 cm−1. Long chains of the D siloxane units generally show two distinct bands at 1020 and 1090 cm−1.36 The other absorption band due to the O–Si–C linkage can also interfere at this range.37 The Si-alkyl group from the D and T units was identified by two overlapping absorption bands at about 1240–1260 cm−1. This may be due to the chemical cross-linking of the hydroxyl group of the polymer precursor with trichloro(alkyl)silane.
Furthermore, the membrane prepared with trichloro(alkyl)silane of a longer alkyl chain showed a higher contact angle, which indicated that the long alkyl chain in the structure helped increase the hydrophobicity of the membrane.
On the other hand, a larger amount of cross-linker was not helpful in increasing the hydrophobicity. The PDMS-FOS1 and PDMS-FOS2 had a similar contact angle of 120°, which are slightly higher than the PDMS-FOS3 value of 117°. The same trend was observed for the PDMS-ODS series, in which both the PDMS-ODS1 and PDMS-ODS2 showed a contact angle of 113°, while the PDMS-ODS3 had a value of 110°. In the case of the PDMS-OS and PDMS-MS series, the contact angle decreased from 111° to 105° and 108° to 103°, respectively, with increasing cross-linker amount. This may be due to secondary reaction of trichloro(alkyl)silane hydrolysis when used in excess forming poly(alkyl)siloxane inside the membrane, as discussed above. The separation performance of the membranes in terms of the ethanol selectivity and flux from an aqueous feed containing 5% (w/w) ethanol was examined.
Fig. 11 shows the performance comparison of the present PDMS membranes prepared by trichloro(alkyl)silane cross-linking and the other conventional PDMS membranes prepared by cross-linking with polymethyl(hydro)siloxane. The highest ethanol/water separation factor of 18 and normalized permeate flux of 183 g m−2 h−1 (flux normalized per unit membrane thickness of 100 μm) was observed for the PDMS-FOS1 sample, while a slight decrease in the ethanol/water separation factor 17, but a slight increase in flux, 190 g m−2 h−1, was observed for the PDMS-ODS1 sample. Both the PDMS-OS1 and PDMS-MS1 samples exhibited a higher flux of about 210 g m−2 h−1 but with a moderate separation factor of 14. There was a decrease in the membrane performance for all the membranes series prepared with increasing amount of cross-linker loading, which may be due to poly(alkyl)siloxane particle formation in the membrane film rather than cross-linking as discussed above. The details of the membrane performance studies of all the membrane series in ethanol–water as well as butanol–water separation will be published elsewhere. It can be seen from Fig. 11 that the present PDMS membranes prepared with an optimal loading of trichloro(alkyl)silane cross-linker performed better than the other reported conventional PDMS membranes.
Fig. 11 Comparison of the present PDMS membranes with other reported conventional PDMS membranes. Experimental conditions: Present work: feed = 5% (w/w) of ethanol in water, operating temperature = 70 °C. Ref. 38: feed = 1.5% (w/w) of ethanol in water, operating temperature = 66 °C. Ref. 39: feed = 6% (w/w) of ethanol in water, operating temperature = 65 °C. Ref. 40: feed = 7% (w/w) of ethanol in water, operating temperature = 22 °C. All the membrane fluxes were normalized to a membrane thickness of 100 μm. |
The membranes formed by crosslinking with trichloro(methyl)silane had crystalline PDMS regions even from the reaction with 50 wt% trichloro(methyl)silane. This is in contrast to the others, which exhibited a completely amorphous nature of PDMS after the reaction with 33–50 wt% trichloro(alkyl)silane. The additional presence of poly(alkylsiloxane) groups from the self-condensation of trichloro(alkyl)silane was observed in the membranes formed with 33–50 wt% of trichloro(alkyl)silane. The cross-linked network structure comprising the dimethylsiloxane network structures of two cross-links and alkylsiloxane network structure of two or three cross-links were observed by 29Si NMR and IR spectra. Thus, the PDMS membrane structure comprised two characteristic segments of dimethylsiloxane and alkylsiloxane units. The dimethylsiloxane networks are long chain polymeric networks that may be considered soft segment because it is being derived from the PDMS pre-polymer liquid while the alkylsiloxane networks are short chain networks and a hard structure resulting from the cross-linking reaction directly with molecular trichloro(alkyl)silane or with the species derived from hydrolysis, condensation and polymerization reactions of trichloro(alkyl)silane. SANS models of the Ornstein–Zernike approach and Debye–Anderson–Brumberger equation described the soft and hard segments comprising dimethylsiloxane and alkylsiloxane groups of the cross-linked membrane structure. The hard segment of the membrane was increased in the membrane samples formed with a larger amount of trichloro(alkyl)silane. Changes in the thermal properties and structure density of the membranes depending on concentration and type of trichloro(alkyl)silane used were also observed.
An optimal amount of trichloro(alkyl)silane cross-linker of about 10–30% (w/w) was required to prepare uniformly homogenous membrane and these membranes showed better separation performance than the other reported conventional PDMS membranes in terms of the ethanol selectivity and flux from an aqueous feed containing 5% (w/w) ethanol.
This journal is © The Royal Society of Chemistry 2015 |