Hyperbranched Polyimide-Silica Hybrid Materials: Synthesis, Structure, Dynamics, and Gas Transport Properties

This article was downloaded by: [Monash University Library] On: 29 August 2013, At: 03:00 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Journal of Macromolecular Science, Part B: Physics Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/lmsb20

Hyperbranched Polyimide-Silica Hybrid Materials: Synthesis, Structure, Dynamics, and Gas Transport Properties a

b

a

Evgenia Minko , Vladimir Bershtein , Petr Sysel , Larisa Egorova b

b

c

c

, Pavel Yakushev , Vladimir Hynek , Ondrej Vopicka , Karel c

c

c

Friess , Miroslav Zgazar , Krystof Pilnacek & Milan Sipek

c

a

Department of Polymers, Institute of Chemical Technology, Prague, Czech Republic b

Ioffe Physical-Technical Institute, St. Petersburg, Russia

c

Department of Physical Chemistry, Institute of Chemical Technology, Prague, Czech Republic Accepted author version posted online: 21 Aug 2012.Published online: 15 Jan 2013.

To cite this article: Evgenia Minko , Vladimir Bershtein , Petr Sysel , Larisa Egorova , Pavel Yakushev , Vladimir Hynek , Ondrej Vopicka , Karel Friess , Miroslav Zgazar , Krystof Pilnacek & Milan Sipek (2013) Hyperbranched Polyimide-Silica Hybrid Materials: Synthesis, Structure, Dynamics, and Gas Transport Properties, Journal of Macromolecular Science, Part B: Physics, 52:4, 632-649, DOI: 10.1080/00222348.2012.718958 To link to this article: http://dx.doi.org/10.1080/00222348.2012.718958

PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content.

Downloaded by [Monash University Library] at 03:00 29 August 2013

This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/termsand-conditions

R Journal of Macromolecular Science, Part B: Physics, 52:632–649, 2013 Copyright © Taylor & Francis Group, LLC ISSN: 0022-2348 print / 1525-609X online DOI: 10.1080/00222348.2012.718958

Downloaded by [Monash University Library] at 03:00 29 August 2013

Hyperbranched Polyimide-Silica Hybrid Materials: Synthesis, Structure, Dynamics, and Gas Transport Properties EVGENIA MINKO,1 VLADIMIR BERSHTEIN,2 PETR SYSEL,1 LARISA EGOROVA,2 PAVEL YAKUSHEV,2 VLADIMIR HYNEK,3 ONDREJ VOPICKA,3 KAREL FRIESS,3 MIROSLAV ZGAZAR,3 KRYSTOF PILNACEK,3 AND MILAN SIPEK3 1

Department of Polymers, Institute of Chemical Technology, Prague, Czech Republic 2 Ioffe Physical-Technical Institute, St. Petersburg, Russia 3 Department of Physical Chemistry, Institute of Chemical Technology, Prague, Czech Republic A series of organic–inorganic hybrid materials were prepared from a hyperbranched polyimide precursor (hyperbranched polyamic acid), tetramethoxysilane, and/or 3glycidyloxypropyl-trimethoxysilane via a sol-gel process. The hyperbranched polyimidesilica hybrids, whose polyimide moieties were based on commercially available  monomers 4,4 ,4 -triaminotriphenylmethane and 4,4 -oxydiphthalic anhydride taken in molar ratio 1:1, contained from 10 to 30 wt% silica. Their morphology and dynamics were characterized by using scanning electron microscopy, differential scanning calorimetry, dynamic mechanical analysis, laser-interferometric creep rate spectroscopy, and wide-angle X-ray diffraction. Attention was also focused on the relation between morphology/dynamics and gas transport properties of these materials. Keywords gas permeability, glass transition, hybrid materials, hyperbranched polyimide, silica

Introduction Aromatic polyimides (PIs) are a class of high performance polymers with very good thermal, dielectric, and mechanical properties. They are applied in many areas, such as microelectronics, aircraft industry and as membranes for separation technologies.[1,2] In recent years, an increasing attention has been devoted to hyperbranched polyimides (HBPIs) due to a potential possible connection of the known advantages of linear or cross-linked PIs[2] with those of hyperbranched polymers.[3,4] Their important structural features (which distinguish them from linear analogs) are a highly branched structure, absence of chain entanglements, and a large number of terminal functional groups. According to the results of computer simulation, there are many accessible cavities of atomic and slightly larger size Received 20 May 2012; accepted 26 July 2012. Address correspondence to Vladimir Bershtein, Ioffe Physical-Technical Institute, 194021 St. Petersburg, Russia. E-mail: [email protected].

632

Downloaded by [Monash University Library] at 03:00 29 August 2013

Hyperbranched Polyimide-Silica Hybrid Materials

633

in the rigid hyperbranched polymers,[5] thus also HBPIs, contributing to the free volume and properties based thereon, for example, diffusivity. The permeability in a polymer depends on the diffusivity and also on the solubility of the permeating species in that polymer. In agreement with this, we have found[6] that the permeability coefficients of gases [hydrogen (H2 ), carbon dioxide (CO2 ), oxygen (O2 ), nitrogen (N2 )] in the membrane prepared from the HBPI based on 4,4 -oxydiphthalic anhydride (ODPA) and 4,4 ,4-triaminotriphenylmethane (MTA) [HBPI(ODPA-MTA)], as used here, were about three times higher than those in the membrane prepared from linear PI (LPI) based on ODPA and 4,4 -diaminodiphenylmethane (MDA) [LPI(ODPA-MDA)] at the comparable selectivities. Increased gas permeability was observed also for HBPIs based on ODPA and 2,4,6-triaminopyrimidine, cross-linked with ethylene glycol diglycidyl ether, compared with a linear analog.[7] To improve the principal requirement of the membrane technologies on increasing permeability at sufficiently high selectivity, Suzuki and Yamada[8] combined the HBPI based on 1,3,5-tris(4-aminophenoxy)benzene with the silica generated in situ via a sol-gel process. The increasing permeability coefficients observed for CO2 , O2 , and N2 with the silica content in a membrane were explained by the increase in gas solubility only. On the other hand, Cornelius et al.,[9] who dealt with the hybrids based on linear polymers, stated that additives, such as silica, could inhibit the packing efficiency of the polymer chains effectively due to polymer chains packing restriction and/or polymer segmental motion restriction. Additionally, it is well known in general that the interfacial area in composites creates a significant space with properties different from those of both neat polymer and inorganic phases.[8–13] To judge the importance of these phenomena for hyperbranched matrices, a series of the HBPI(ODPA-MTA)-silica hybrid materials (HBPIS) differing in their composition were prepared by using a sol-gel procedure in this work. The interfacial interactions between the organic moieties and the inorganic network (clusters, inclusions) were enhanced in some of these materials by using 3-glycidyloxypropyltrimethoxysilane (GPTMS) as a coupling agent. The structure and molecular dynamics of these materials were evaluated by using scanning electron microscopy (SEM), IR spectroscopy, wide-angle X-ray diffraction (WAXD), dynamic mechanical analysis (DMA), differential scanning calorimetry (DSC), and creep rate spectroscopy (CRS). Simultaneously, the gas transport properties of the films studied were estimated. In addition, a comparison with the data obtained for the LPI(ODPA-MDA) was also made.

Experimental Synthesis of Materials 4,4 -Oxydiphthalic anhydride (ODPA; Aldrich, Czech Republic) was heated to 170◦ C for 5 h in a vacuum before use. 4,4 ,4-Triaminotriphenylmethane (MTA; Dayang Chemicals, China), 4,4 -diaminodiphenylmethane (methylenedianiline, MDA; Aldrich), 3-glycidyloxypropyltrimethoxysilane (GPTMS; Aldrich), and tetramethoxysilane (TMOS; Aldrich) were used as received. 1-Methyl-2-pyrrolidone (NMP; Merck, Czech Republic) was distilled under vacuum over phosphorus pentoxide and stored in an inert atmosphere. Gases in the gas cylinders (Siad, Czech Republic) were used as received (N2 99.99 vol%, O2 99.5 vol%, CO2 99.0 vol%, H2 99.90 vol.%). An amine terminated HBPI precursor (hyperbranched polyamic acid) was prepared from ODPA and MTA [HBPAA(ODPA-MTA)] taken in molar ratio 1:1 at room temperature

Downloaded by [Monash University Library] at 03:00 29 August 2013

634

E. Minko et al.

(24 h, inert atmosphere). The concentration of solids in NMP, used as a solvent, was 4 wt%.[6,14] A portion of this precursor was end-capped with GPTMS. A desired amount of GPTMS was added directly to a 4% solution of HBPAA(ODPA-MTA) in NMP and the reaction was allowed to proceed at room temperature for 24 h. A mixture of the HBPAA(ODPA-MTA) (modified or unmodified) with a calculated amount of the silica precursor (TMOS plus water, 1 mole per 1 mole of methoxy groups) was stirred for 24 h at room temperature. Then it was spread onto a substrate and heated at gradually increasing temperature (60/12, 100/1, 150/1, 200/2, and 250◦ C/1 h). The thickness of the films obtained was about 50 μm. LPI based on ODPA and MDA [LPI(ODPA-MDA)] was also prepared by using the described procedure (monomer molar ratio 1:1, solution concentration 4 wt%). The HBPAA(ODPA-MTA) intrinsic viscosity was 16 ml g−1; the value for the LPAA(ODPA-MDA) was markedly higher (36 ml g−1)[6] due to the entanglements of its chains. Two series of HBPIS films varying in their composition were prepared. The list of the materials prepared is summarized in Table 1. Their properties were compared with those for the pure HBPI(ODPA-MTA) and the linear analog LPI(ODPA-MDA). Series I included HBPIS films synthesized from unmodified HBPAA and TMOS and theoretically contained 10, 16, 20, or 30 wt% silica, without the formation of chemical bonds between the organic and inorganic phases. All HBPIS films of Series II contained, theoretically, 16 wt% silica, and their phases were chemically bounded to different extents via GPTMS; the general chemical structure of the final material is shown in Fig. 1. Since the molar ratio of the trifunctional MTA and bifunctional ODPA was 1:1 in the preparation of HBPAA(ODPA-MTA), the amine end-capped HBPAA was formed. In case all hydrogen atoms of end-capping –NH2 groups were reacted with the coupling agent GPTMS, then 16 wt% of silica was theoretically

Table 1 Samples under study Composition Sample Pure polyimides LPI HBPI HBPIS—Series I HBPIS10 HBPIS16 HBPIS20 HBPIS30 HBPIS—Series II HBPIS16/1 HBPIS16/2 HBPIS16/3 HBPIS16/4 HBPIS16/5 a

SiO2 wt%

NH2 /GPTMSa

— —

— —

10 16 20 30

— [only TMOS] — [only TMOS] — [only TMOS] — [only TMOS]

16 16 16 16 16

1:2 1:1 [+TMOS] 2:1 [+TMOS] 4:1 [+TMOS] 8:1 [+TMOS]

Molar ratio of end-capping –NH2 groups to GPTMS.

Hyperbranched Polyimide-Silica Hybrid Materials

635

O

O O N O

O

O N H C

O

N H

O OH

Si O

Si O

O O

Si O

O O

Si

O O

Downloaded by [Monash University Library] at 03:00 29 August 2013

Figure 1. Chemical structure of HBPI(ODPA-MTA)-silica hybrid material (HBPIS) with the use of GPTMS as a coupling agent.

formed (HBPIS16/1 composition, Table 1). Thus, GPTMS molecules could serve both as a coupling agent linking up PI and silica phases and, similarly as TMOS, as the silica precursor. By using a defined amount of GPTMS (molar ratio NH2 /GPTMS varied from 1:2 to 8:1), the number of the hydrogen atoms of terminating amino groups taken into the reaction with epoxy groups of GPTMS was changed. During thermal exposition, the PI precursor was transformed into the HBPI, and hydrolyzed alkoxy groups of TMOS and GPTMS reacted mutually to form silica nanoinclusions or microinclusions. By using thermogravimetric analysis in an air atmosphere between 25◦ C and 800◦ C, it was found that the real silica contents in the materials were a few percent lower than the theoretical ones; this discrepancy was lower when the higher silica amount was coming from GPTMS. Therefore, this difference was probably given by a TMOS loss during heating in the TGA experiments. Characterization Infrared spectra were recorded on a Nicolet 740 FTIR spectrometer (Nicolet Instruments Corp., USA) using a transmission mode (for HBPAA) or attenuated total reflection mode (ZnSe crystal, incidence angle 45◦ , 64 reflections) for LPI, HBPI, and HBPIS. SEM measurements were performed using a JEOL JSM-5500LV microscope (Japan). The film crosssectional areas, sectioned at ambient temperature, were covered by a deposited platinum layer prior to measurements. WAXD measurements were made using a X’Pert PRO diffractometer (PANalytical, Netherlands) using cooper radiation with wave length λ = 0.154 nm. Dynamic mechanical analysis was performed using a DMA DX 04T apparatus (RMI Bohdanec, Czech Republic), basically at 1 Hz, over the temperature range from 25◦ C to 450◦ C, at the heating rate 3◦ C min−1. The apparent activation energy, Q, of segmental motion within the glass transition could be estimated, with an accuracy of ±20%–30%, from the values of the relaxation (tanδ) peak temperatures, T max , as measured at the frequencies f of 0.1, 1, and 10 Hz and using the following formula: Q=

R d(log f ) · , 0.4567 d(log 1/T )

(1)

where R = 8.31431 J mol−1 K−1 is the gas constant and T = T max (K). Differential scanning calorimetry (Perkin-Elmer DSC-2 apparatus, USA) was used for the characterization of the glass transition and water release from LPI, HBPI, and HBPIS. The measurements were performed under N2 atmosphere over the temperature range from 20◦ C to 370◦ C; amorphous quartz was used as the reference sample. The glass transition

636

E. Minko et al.

Downloaded by [Monash University Library] at 03:00 29 August 2013

temperature, T g , at the half-height of a heat capacity step, Cp , was determined for LPI,  whereas the temperatures of the glass transition onset, T g , and completion, T g , only, were determined for the anomalously broad transition ranges typical of HBPI and HBPIS (see below); the Cp values were also estimated. The second scan was used in order to exclude the side endothermic effect of water desorption in a DSC curve and to deal with waterless samples. Meantime, this endothermic effect was estimated separately in the first scan: the values of dehydration enthalpy, Qd , were determined from the areas of endothermic peaks located between 50◦ C and 250◦ C. Typically, the heating rate V = 20◦ C min−1 and cooling rate 320◦ C min−1 were used. The apparent activation energy, Q, for segmental motion within the glass transition could be estimated by DSC only for LPI (because of the very wide glass transitions in the hybrids), by varying V from 2.5◦ C min−1 to 40◦ C min−1 and using lnV versus T −1(K−1) linear dependencies, by the following formula[15]: Q = −R

d1nV , d(T −1 )

(2)

where T = T g . The accuracy of Q determination was ±20%. Creep rate spectroscopy, as the high-resolution method of relaxation spectrometry and thermal analysis, allows to characterize in a discrete way the dynamics, dynamic heterogeneity and creep resistance of materials over a broad temperature region; the CRS setups and experimental technique used have been described in detail elsewhere.[16] This technique consists in precisely measuring creep rates at a constant low stress, much less than the yield stress, as a function of temperature (creep rate spectrum). For this aim, a laser interferometer based on the Doppler effect is used. The time evolution of deformation is registered as a sequence of low-frequency beats in an interferogram whose beat frequency, ν, yields a creep rate ε˙ =

λν . 2I0

(3)

Here λ = 630 nm is the laser wavelength, and I 0 is an initial length of the working part of a sample. In this work, the creep rate spectra were measured in the temperature range from 20 to 400◦ C at a tensile stress of 1 MPa. This stress was chosen in the preliminary experiments as capable of inducing sufficient creep rates to be measured, while maintaining also a high spectral resolution, without smoothing out and distortion of a spectral contour and preventing premature rupture of a sample. During heating with the rate V = 1◦ C min−1, every 5◦ a sample was loaded, and an interferogram was recorded at 10 s after loading; then the sample was unloaded with the heating continued. The correlative frequency of the CRS experiments was ν corr = 10−3 to 10−2 Hz. CRS and DMA analyses were performed with the film samples of ∼50-micron thickness, 8-mm width, and 2-cm length. A home-built integral permeameter[6,17] was used to determine the permeability and diffusion coefficients of gases in the membranes. The accuracy of the measurements was 15%.

Results and Discussion Structure Figures 2(a), 2(b), and 2(c) show typical IR spectra obtained for the prepared HBPAA, HBPI, and one of the HBPIS (HBPIS16/3), respectively. A few features of these spectra may be

637

Downloaded by [Monash University Library] at 03:00 29 August 2013

Hyperbranched Polyimide-Silica Hybrid Materials

Figure 2. IR spectra of (a) HBPAA, (b) HBPI, and (c) HBPIS16/3.

noted. First, the characteristic absorption bands[18,19] of the imide group were observed at 1777 and 1712 cm−1 (symmetric and asymmetric stretching vibrations of the ring carbonyl groups) and 1364 cm−1 (stretching of the ring C–N bond) in the HBPI and HBPIS16/3 spectra, whereas the absorption band 1677 cm−1, corresponding to the amide groups of

Downloaded by [Monash University Library] at 03:00 29 August 2013

638

E. Minko et al.

the HBPAA, disappeared in these spectra. These results indicate the high imidization level. Only a very weak peak at about 1550 cm−1, of the unreacted end-capping amino groups, was distinguishable in the hybrid spectrum. The band in the HBPIS16/3 spectrum corresponding to the methylene groups coming from GPTMS is probably included in the band with the maximum at 1511 cm−1. The major difference between the HBPI and HBPIS16/3 spectra is the existence in the latter of the absorption at 2900–3700 cm−1 and the broad, complicated absorption band within the range 1300–900 cm−1, with a few maxima characterizing the presence of silica and some intermediates formed during the sol-gel process. Thus, the latter broad contour includes the maxima ∼1200 cm−1 (Si–O–Si network fragments), 1063 cm−1 (Si–O–Si linear fragments), 957 cm−1 (Si–OR), etc. Absorption bands of unreacted Si–OH groups are located at 810–820 cm−1. These data support our previous conclusion based on a solid state 29Si NMR analysis that a condensation degree of Si–OH groups was not complete under these conditions (only by about 80%),[20] due to using of an insufficiently high temperature of thermal treatment (250◦ C) which was limited by the HBPI thermal stability (see Experimental section). The broad peak 3700–2900 cm−1 in the hybrid spectrum was mainly given by –OH groups, including absorbed water. Scanning electron microscopy photos [see Figs. 3(a) and (b)] show the microhomogeneous structure in HBPI and in HBPIS16/1 where the nanosilica regions were formed from GPTMS only. The presence of GPTMS in HBPIS samples whose silica moieties came from both TMOS and GPTMS made also (as in HBPIS16/1) the growth of compact silica microclusters more difficult, although some silica microinclusions apparently developed [see Fig. 3(c), see arrows]. However, the formation of silica microparticles of 1–10 μm in size was the typical feature of the structure in the HBPIS samples prepared with the participation of TMOS only [see Fig. 3(d)]. Dynamics The glass transition dynamics in the LPI, HBPI, and HBPIS under study were characterized using the combined DSC/DMA/CRS approach. These three techniques provided somewhat different, complementary experimental information on the glass transition dynamics, dynamic heterogeneity, and thermal/elastic properties of the HBPIS films compared with these characteristics for LPI and neat HBPI. Figure 4 shows the DSC curves obtained in the first heating scan. The endothermic peaks characterize release of absorbed water from LPI, HBPI, and five HBPIS films; the values of dehydration enthalpy, Qd , were determined from the peak areas; thus, the water uptake degrees could be compared. The HBPI showed a three-fold increase in water uptake compared with that in LPI; this is attributed to the large number of terminal amine groups and increased free volume. Incorporating silica inclusions into HBPI resulted in an additional rise of water uptake (Qd value) in the hybrids; the HBPIS16/3 composition was characterized with the maximal ability to absorb water and, presumably, with the largest free volume: Qd for this hybrid increased to about 200 J g−1 compared with 30 J g−1 for LPI. Figure 5 shows the DSC curves obtained in the second heating scan, which was used to characterize the glass transition in LPI, HBPI, and HBPIS films. Unlike the distinct, narrow glass transition in LPI (T g = 247◦ C, T g = 17◦ C), extraordinarily broad transition ranges were recorded for both neat HBPI (T g = 100◦ C) and the hybrids; in the latter, the transition range was strongly broadened toward higher temperatures. Transition characteristics changed most significantly (increasing T g and heat capacity step Cp ) in

Downloaded by [Monash University Library] at 03:00 29 August 2013

Hyperbranched Polyimide-Silica Hybrid Materials

639

Figure 3. SEM photos of (a) HBPI, (b) HBPIS16/1, (c) HBPIS16/4, and (d) HBPIS30.

the following order: HBPIS16/1 – HBPIS16/2 – HBPIS16/3; the broadest transition range, extending from 90◦ C to 367◦ C, and the maximal, five-fold Cp increase were observed, again, for the HBPIS16/3 composition where the silica regions were formed from the commensurable quantities of GPTMS and TMOS. Then, with increasing contribution of TMOS in the hybrid preparation, the opposite tendency, a narrowing of the glass transition, and decreasing Cp occurred. The glass transition in the hybrids obtained with TMOS only was suppressed to a large extent and could not be discerned at all by DSC (see Fig. 5). The extraordinary width of the glass transition revealed by DSC for HBPI and HBPIS films suggests a pronounced dynamic heterogeneity within their glass transition ranges, presumably caused by the complicated molecular structure, nanostructure, and microstructure of these materials. The DMA and CRS measurements, described below, validated the pronounced dynamic heterogeneity in the studied hybrids. Figure 6 shows absolutely different DMA characteristics obtained for LPI and neat HBPI. Unlike a single, relatively narrow glass transition peak with T max = 276◦ C and tanδ = 0.62 observed for LPI, the HBPI transition region extended from ∼50◦ C to 370◦ C with tanδ ≈ 0.15–0.20. The latter complicated spectrum consisted of four relaxation peaks including two strongly overlapping peaks with T max ≈ 70◦ C and 130◦ C (lower-temperature relaxations I and II, respectively) and two slightly overlapping peaks with maxima at

Downloaded by [Monash University Library] at 03:00 29 August 2013

640

E. Minko et al.

Figure 4. The endotherms of release of absorbed water from LPI, HBPI, and different HBPIS films (DSC curves, the first scan with the heating rate of 20◦ C min−1). (a) LPI (1), HBPI (2), HBPIS16/1 (3), HBPIS16/2 (4), HBPIS16/3 (5), HBPIS16/4 (6), and HBPIS16/5 (7); (b) the enthalpy of dehydration versus sample composition.

∼250◦ C and 370◦ C (mid- and high-temperature relaxations III and IV). The highest peak was observed at 370◦ C. It is clear that only relaxations I and II constituted, in fact, the DSC glass transition of HBPI, whereas relaxations III and IV could not be seen in the DSC thermogram, obviously, due to too slight heat capacity changes. The extraodinarily broad and complicated HBPI glass transition may be understood as a consequence of its complicated molecular structure. The architecture of hyperbranched polymers includes many branches that are distributed randomly throughout the polymers. The random branching theory (RBT)[21] describes their structure as a “collection” of units incorporating the “spherical” nanoregions around branching points, linear polymer sections of different size, and many end groups (i.e., end segments). Additionally, the spatial variation of polymer density is also considered, that is, the presence of the dense nanoregions and nanoregions with loosened packing. Such diversity of structural elements and density variation predetermine, we suggest, the pronounced dynamic heterogeneity and large width of the HBPI glass transition.

Downloaded by [Monash University Library] at 03:00 29 August 2013

Hyperbranched Polyimide-Silica Hybrid Materials

641

Figure 5. DSC curves obtained over the temperature region of glass transition for (1) LPI, (2) HBPI, (3) HBPIS16/1, (4) HBPIS16/2, (5) HBPIS16/3, (6) HBPIS16/4, (7) HBPIS16/5, and (8) HBPIS16–HBPIS30. The second scan with the heating rate of 20◦ C min−1. The transition intervals and heat capacity steps are indicated.

One of the dynamic modes within the HBPI glass transition, relaxation III, which coincides by its temperature location with and may be assigned, presumably, to the glass transition identical to that in LPI, may be assigned to intermolecularly cooperative motion of neighboring segments in the nanoregions with densely packed linear sections of HBPI. The activation energy Q value as determined for the glass transition in LPI by DSC, that is, QIII for relaxation III, turned out to be of high (non-Arrhenius, cooperative) value of about 700 kJ mol−1. At the same time, the DMA experiments, performed at the frequencies of 0.1, 1, and 10 Hz, allowed us to estimate the values of the activation energy Q for the additional relaxations II and IV observed at the much lower and higher temperatures. It was found that relaxation II corresponded to Arrhenius-like, noncooperative motion with the activation energy QII = 120 ± 30 kJ mol−1, that is, to quasi-independent segmental motion within loosely packed nanoregions. On the contrary, relaxation IV characterized a highly cooperative motion with the apparent activation energy QIV ≈ 1200 kJ mol−1. This type of motion, we suggest, may be assigned to “unfreezing” dynamics around branching points. At last, the lower-temperature relaxation I may be associated with quasi-independent motion of end segments in HBPI. Figures 7(a)– (e), 8(a), and (b) present the DMA data obtained for seven HBPIS with different compositions. In HBPIS16/1, silica cross-links could be formed from attached molecules of silica precursor GPTMS only. Different contents of nanosilica inclusions were formed from TMOS only in the HBPIS10 and HBPIS30. At the same time, both silica precursors in different quantities, providing 16 wt% SiO2 in the hybrids, were used for preparing HBPIS16/2, HBPIS16/3, HBPIS16/4, and HBPIS16/5. Figures 7 and 8 show that high-temperature peak of relaxation IV was the most intense in all cases since densely

E. Minko et al.

Downloaded by [Monash University Library] at 03:00 29 August 2013

642

Figure 6. DMA spectra measured at 1 Hz for (a) LPI and (b) HBPI.

packed nanoregions around branching points prevail in both neat HBPI and the hybrids structures. Meanwhile, the DMA spectra of the hybrids manifested a few peculiarities as compared with the spectrum of neat HBPI shown in Fig. 6(b). First, all hybrids’ spectra exhibit a distinctive contour with two to four overlapping peaks but introducing nanosilica and microsilica inclusions resulted in modifying significantly the HBPI spectral contour over the whole 50◦ C–400◦ C range depending on the sample composition. And, secondly, all hybrid spectra showed about twofold or threefold decrease in their intensities. It means

643

Downloaded by [Monash University Library] at 03:00 29 August 2013

Hyperbranched Polyimide-Silica Hybrid Materials

Figure 7. DMA spectra measured at 1 Hz for (a) HBPIS16/1, (b) HBPIS16/2, (c) HBPIS16/3, (d) HBPIS16/4, and (e) HBPIS16/5.

Downloaded by [Monash University Library] at 03:00 29 August 2013

644

E. Minko et al.

Figure 8. DMA spectra measured at 1 Hz for (a) HBPIS10 and (b) HBPIS30.

that the silica phases suppressed segmental mobility within a part of the HBPI regions attached to these nanophases or microphases. At the same time, they did not suppress the dynamics (unchangeable T g ) or even accelerate it (decreasing T g by ∼5◦ C–15◦ C) in the other polymer regions (see Table 2). For HBPIS16/1 and HBPIS16/2, with high contents of GPTMS as cross-linking silica precursor and rather homogeneous structure [as shown for HBPIS16/1 in Fig. 3(b)], some “depletion” of spectral contours occurred since only two peaks were observed distinctly: the lower-temperature one with a maximum at ∼100◦ C and the broadened high-temperature peak in the region from 250◦ C to ∼ 400◦ C; the peak at ∼250◦ C, peculiar to relaxation III, almost disappeared [see Figs. 7(a) and (b)]. Additionally, in these cases the minimal tanδ = 0.06–0.07 for relaxation IV was attained. These changes may be due to grafting GPTMS molecules to the majority of free NH2 functionalities; this results in moving apart the neighboring PI segments in both the same and different molecules that prevents to some extent their cooperative motion. The most complicated spectrum, with strongly overlapping peaks with the maxima at ∼100◦ C, 200◦ C, 270◦ C, and 380◦ C, was observed for HBPIS16/3 [see Fig. 7(c)]. It is noteworthy again that just in this case, the most broad temperature region of the DSC glass transition was observed which practically coincided with the temperature range of the DMA glass transition. With decreasing GPTMS content and increasing TMOS precursor in the hybrid composition, the contribution of relaxation IV to the mechanical loss spectrum increased [e.g., tanδ = 0.18 at 370◦ C for HBPIS16/5, Fig. 7(e)]; additionally, relaxation peak III was restored and observed more distinctly. In fact, a separate formation of silica nanoparticles/microparticles and of dense HBPI regions occurred in this case due to the deficit or absence of cross-linking GPTMS agent. The most broad high-temperature relaxation peak, extending from 200◦ C to ∼400◦ C, was observed for HBPIS30 with 30% TMOS [see Fig. 8(b)]. For the hybrids HBPIS10–HBPIS30, containing only TMOS in their compositions, DSC failed in deciphering a glass transition [see Fig. 5].

645

1350 192 66 11

863 7 45

1.56 27.4 1.47

6.0 17 123 0.55 276

H2 CO2 O2 N2

H2 CO2 O2

H2 CO2 O2

α (O2 /N2 )a α (CO2 /N2 ) α (H2 /N2 ) d (nm)b T g (◦ C)c

6.2 30 113 0.56 377

3.65 119 3.92

740 6 38

2700 714 149 24

HBPI

8.7 34 153 0.56 361

6.64 81.0 2.70

277 5 39

1840 405 104 12

HBPIS10

7.7 36 138 0.55 378

4.60 14.4 4.40

478 4 28

2200 575 123 16

HBPIS16

HBPIS16/1

5.9 27 99 0.56 372 52 171 0.47 350

10

D [m2 s−1] × 1014 910 522 9 4 30 23 S [mol m−3 Pa−1] × 104 2.60 2.26 72.2 88.5 4.70 3.00

P [mol m−1 s−1 Pa−1] × 1018 2380 1180 650 354 141 69 24 6.9

HBPIS30

6.1 31 111 0.55 372

1.65 124 4.03

1339 5 30

2210 620 121 20

HBPIS16/2

b

Oxygen/nitrogen selectivity. Spacing d (from the Bragg equation λ = 2d sin, where λ is wavelengh and 2 is X-ray scattering angle). c T g glass transition temperatures taken from the main tanδ relaxation peak (DMA).

a

LPI

Gas

3.7 19 56 0.56 363

2.80 142 5.03

1039 7 38

2900 994 191 52

HBPIS16/3

8.1 45 157 0.57 372

3.30 178 5.20

769 4 25

2510 712 129 16

HBPIS16/4

Table 2 Gas transport properties, d-spacing, and glass transition temperatures of LPI, HBPI, and HBPIS

Downloaded by [Monash University Library] at 03:00 29 August 2013

7.4 35 135 0.57 378

2.90 177 4.60

928 4 32

2690 709 147 20

HBPIS16/5

Downloaded by [Monash University Library] at 03:00 29 August 2013

646

E. Minko et al.

Figure 9. Creep rate spectra obtained at tensile stress of 5 MPa for HBPIS16/1 (black circles) and HBPIS16/3 (white circles).

Figures 6–8 show, also, the different, peculiar changes of the storage modulus (dynamic modulus) E with temperature for HBPI and different HBPIS hybrids. For HBPI, E decreased from ∼700 MPa at 25◦ C to 230 MPa at 400◦ C. In the HBPIS hybrids, E increased up to ∼1100–1800 MPa at 25◦ C and, moreover, a relatively slight temperature dependence at 20◦ C–400◦ C was observed: E remained basically at the level of 700–1000 MPa at 400◦ C. Figure 9 shows the creep rate spectra measured for HBPIS16/1 and HBPIS16/3. These discrete spectra with the complicated contours consisted of a series of strongly overlapping peaks, including weak peaks at moderate temperatures and more intense ones (maximal at ∼300◦ C) within the temperature range from 200◦ C to 400◦ C. These spectra indicated the presence of a few dynamic modes in the glass transitions of both hybrids confirming, additionally, the pronounced dynamic heterogeneity caused by their complex, heterogeneous molecular, nanostructure and microstructure. The spectra of HBPIS16/1 and HBPIS16/3 differed in a shape: the latter spectrum confirms the more complicated structure of the HBPIS16/3 hybrid, and is in accordance with the widest DSC glass transition range (see Fig. 5) and the most complicated mechanical loss spectrum [see Fig. 7(c)]. Gas Transport Properties Gas transport through a nonporous (dense) membrane can be described by the solution– diffusion mechanism. The permeability (quantified by a permeability coefficient PA ) is proportional to the diffusivity (quantified by a diffusion coefficient DA ) and sorption (quantified by a sorption coefficient SA ) of a medium A. The theoretically valid equation PA = DA × SA was employed earlier, in particular by Suzuki and Yamada[8] and Cong et al.,[22] for a calculation of SA from the experimental values of PA and DA in nanocomposites. The values of P for H2 , CO2 , N2 , and O2 , and D and S values for H2 , CO2 , and O2 , as estimated for the LPI, HBPI, and HBPIS membranes studied, are summarized in Table 2 (D and S values are not shown for N2 due to low accuracy of their estimation). The glass transition temperatures (T g DMA) and d-spacing (d) of these materials are also given in this table (the

Downloaded by [Monash University Library] at 03:00 29 August 2013

Hyperbranched Polyimide-Silica Hybrid Materials

647

relation of d to transport properties was discussed, e.g., in ref.[23]). The transport properties of LPI and HBPI were compared previously in ref.[6]. One can see that the permeability of each material in the HBPIS Series I and II decreased in the order PH2 > PCO2 > PO2 > PN2 , in harmony with the kinetic diameters of these gases: H2 (0.29 nm), CO2 (0.33 nm), O2 (0.35 nm), and N2 (0.36 nm).[8] In Series I of HBPIS films, with unbonded phases (without GPTMS), the permeabilities of all gases increased with increasing silica content and were higher than those for LPI; however, they did not reach the values of the neat HBPI permeabilities. This fact suggests a fairly good interfacial adhesion without increasing free volume holes contributing to the gas fluxes. The permeability coefficients growing with the silica content is attributed to an increase in diffusion coefficients for H2 and (partly) both diffusion and sorption coefficients for CO2 and O2 (see Table 2). It seems that the addition of 10 wt% silica had a modest influence on loosening of the polymer chain packing (quantified by a T g drop from 377◦ C for HBPI to 361◦ C for HBPIS10, Table 2). But the T g values of HBPIS16 (378◦ C) and HBPIS30 (372◦ C) were very close to that of HBPI, probably as a consequence of the reinforcing effect of the higher silica concentration. The modest influence was also supported with a nearly constant d-spacing of around 0.55 nm for HPIS10, HPIS16, and HPIS30. As a consequence of filling with silica, the permeabilities of HBPIS with a lower silica content were smaller than those of pure HBPI. On the contrary, the selectivities, defined as the ratio of permeability coefficients of the media A and B (α = PA /PB ), show for the couples H2 /N2 , CO2 /N2 , and O2 /N2 maximal values for the HBPIS containing 10–16 wt% silica. In Series II of HBPIS films with covalently bonded phases, the theoretical silica content 16 wt% was formed from both TMOS and GPTMS where the latter served also as a coupling agent creating chemical bonds between organic and inorganic phases. Again, the values of P of all gases did not reach those of the pure HBPI; however, the HBPIS16/3 films manifested an anomalous behavior, and the permeability versus GPTMS/TMOS ratio dependencies exhibited maximal values for HBPI16/3, which were higher than those for LPI and HBPI, for example, PN2 = 52, 11, and 24 [mol m−1 s−1 Pa−1] × 1018, respectively, for HBPI16/3, LPI, and HBPI. This result was not accidental since the morphology/dynamics of the HBPIS16/3 hybrid was very complicated [see Figs. 7(c) and 8], and it was characterized with the most broad glass transition range (see Fig. 5) and the highest water absorption (see Fig. 4), that is, with the largest free volume. Nevertheless, the values of d-spacing (0.56 ± 0.01 nm) were practically identical for the hybrids in this series, with an exception being the value d = 0.47 nm for HBPIS16/1 only (see Table 2). At the same time, the highest selectivities were observed just for this HBPIS16/1 hybrid whose silica nanophases were formed from coupling agent GPTMS only: α (O2 /N2 ) = 10, α (CO2 /N2 ) = 51, and α (H2 /N2 ) = 171, that is, higher than for LPI and HBPI. It should be noted that this hybrid was characterized with very good structural homogeneity [see Fig. 3(b)]. Thus, the most interesting hybrids for membrane technologies, HBPIS16/1 and HBPIS16/3, were characterized with loosened molecular packing: their T g s shifted from 377◦ C for HBPI to 350◦ C and 363◦ C, respectively (see Table 2).

Conclusion Two series of HBPIS hybrid materials based on HBPI(ODPA-MTA) and silica were prepared. The silica phase was generated in situ and it was unbounded (Series I) or bounded by chemical bonds to the polymer matrix by using GPTMS as a coupling agent (Series II). Varying the silica content from 10 to 30 wt% and introducing GPTMS influenced the

648

E. Minko et al.

HBPIS morphology, the complicated molecular (glass transition) dynamics, and gas transport properties of membranes made of them. An essential suppression of HBPI dynamics and strong increasing modulus values, especially at high temperatures, due to silica inclusions was shown. The membranes from the studied HBPIS hybrids showed the traditional permeability–selectivity dependence (the higher the permeability, the lower the selectivity[24]); however, some of these hybrids manifested better gas permeabilities or selectivities than linear and HBPIs.

Downloaded by [Monash University Library] at 03:00 29 August 2013

Acknowledgments This work was supported by the grants GA CR 104//09/1357 and MSM 6046137302. Some financial support from specific university research grants [MSMT no. 21/2011 (A1 FCHT 2011 006, and A2 FCHT 2011 046)] is also acknowledged. The authors would like to thank L. Brabec (Heyrovsky Institute of Physical Chemistry, Academy of Science of the Czech Republic, Prague) for his assistance with the SEM analysis.

References 1. Ohya, H.; Kudryavtsev, V.V.; Semenova, S.I. Polyimide Membranes-Applications, Fabrication and Properties; Gordon and Breach Publishers: Tokyo, 1996. 2. Hergenrother, P.M. The use, design, synthesis, and properties of high performance/high temperature polymers: An overview. High Perform. Polym. 2003, 15, 3. 3. Gao, C.; Yan, D. Hyperbranched polymers: From synthesis to applications. Prog. Polym. Sci. 2004, 29, 183. 4. Jikei, M.; Kakimoto, M. Dendritic aromatic polyamides and polyimides. J. Polym. Sci., Part A: Polym. Chem. 2004, 42, 1293. 5. Fang, J.; Kita, H.; Okamoto, K. Gas permeation properties of hyperbranched polyimide membranes. J. Membr. Sci. 2001, 182, 245. 6. Friess, K.; Sysel, P.; Minko, E.; Hauf, M.; Vopicka, O.; Hynek, V.; Pilnacek, K.; Sipek, M. Comparison of transport properties of hyperbranched and linear polyimides. Desal. Water Treat. 2010, 14, 165. 7. Bershtein, V.A.; Egorova, L.M.; Yakushev, P.N.; Sysel, P.; Hobzova, R.; Kotek, J.; Pissis, P.; Kripotou, S.; Maroulas, P. Hyperbranched polyimides crosslinked with ethylene glycol diglycidyl ether: Glass transition dynamics and permeability. Polymer 2006, 47, 6765. 8. Suzuki, T.; Yamada, Y. Physical and gas transport properties of novel hyperbranched polyimidesilica hybrid membranes. Polym. Bull. 2005, 53, 139. 9. Cornelius, C.; Hibshman, C.; Marand, E. Hybrid organic-inorganic membranes. Separ. Purif. Techn. 2001, 25, 181. 10. Joly, C.; Smaihi, M.; Porcar, L.; Noble, R.D. Polyimide-silica composite materials: How does silica influence their microstructure and gas permeation properties? Chem. Mater. 1999, 11, 2331. 11. Zou, H.; Wu, S.; Shen, J. Polymer/silica nanocomposites: Preparation, characterization, properties, and applications. Chem. Rev. 2008, 108, 3893. 12. Chung, T.-S.; Jiang, L. Y.; Li, Y.; Kulprathipanja, S. Mixed matrix membranes comprising organic polymers with dispersed inorganic fillers for gas separation. Progr. Polym. Sci. 2007, 32, 483. 13. Bershtein, V. A.; Egorova, L. M.; Yakushev, P. N.; Georgoussis, G.; Kyritsis, A.; Pissis, P.; Sysel, P.; Brozova, L. Molecular dynamics in nanostructured polyimide-silica hybrid materials and their thermal stability. J. Polym. Sci., Part B: Polym. Phys. 2002, 40, 1056. 14. Sysel, P.; Minko, E.; Cechova, R. Preparation and characterization of hyperbranched polyimides based on 4,4 ,4-triaminophenylmethane. e-Polymers 2009, No. 081, 6. 15. Bershtein, V.A.; Egorov, V. M. Differential Scanning Calorimetry of Polymers. Physics, Chemistry, Analysis, Technology; Ellis Horwood: New York, 1994.

Downloaded by [Monash University Library] at 03:00 29 August 2013

Hyperbranched Polyimide-Silica Hybrid Materials

649

16. Bershtein, V.A.; Yakushev, P.N. Laser-interferometric creep rate spectroscopy of polymers. Adv. Polym. Sci. 2010, 230, 73. 17. Nguyen, X.Q.; Broz, Z.; Vasak, F.; Nguyen, Q. T. Manometric techniques for determination of gas transport parameters in membranes. Application to the study of dense and asymmetric poly(vinyltrimethylsilane) membranes. J. Membr. Sci. 1994, 91, 65. 18. Zhu, Z.; Yang, Y.; Yin, J.; Qi, Z. Preparation and properties of organosoluble polyimide/silica hybrid materials by sol-gel process. J. Appl. Polym. Sci. 1999, 73, 2977. 19. Sysel, P.; Konecna, V.; Volka, K. Structure-curing relation for polyamic acids. Eur. Polym. J. 1996, 32, 317. 20. Minko, E.; Sysel, P.; Hauf, M.; Brus, J.; Kobera, L. Synthesis and properties of hyperbranched polyimides combined with silica. Macromol. Symp. 2010, 295, 88. 21. Konkolewicz, D.; Grey-Weale, A.; Perrier, S. Describing the structure of a randomly hyperbranched polymer. Macromol. Theory Simul. 2010, 19, 219. 22. Cong, H.; Radosz, M.; Towler, B. F.; Shen, Y. Polymer-inorganic nanocomposite membranes for gas separation. Separ. Purif. Techn. 2007, 55, 281. 23. Shimazu, A.; Miazaki, T.; Ikeda, K. Interpretation of d-spacing determined by wide-angle X-ray scattering in 6DA-based polyimide by molecular modeling. J. Membr. Sci. 2000, 166, 113. 24. Robeson, L. M. The upper bound revisited. J. Membr. Sci. 2008, 320, 390.