- Email: [email protected]
Differential scanning calorimetry and rheological experiments to study membrane formation via thermally-induced phase-separation
New Insights into Membrane Science and Technology: Polymeric and Biofunctional Membranes D. Bhattacharyya and D.A. Butterfield (Editors) 9 2003 Elsevier Science B.V. All rights reserved.
Chapter 3
Differential scanning calorimetry and rheological experiments to study membrane formation via thermallyinduced phase-separation P.C. van der Heijden, M.H.V. Mulder, M. Wessling University of Twente, PO Box 217, 7500 AE Enschede, The Netherlands, Fax: xx31-53-4894611 1.
INTRODUCTION
1.1
Thermally-Induced Phase-Separation Porous polymer membranes can be prepared by different techniques: track-etching, stretching and phase-separation [ 1]. The phase-separation method is very common to prepare polymeric membranes and much attention has been paid to prepare membranes from various polymers and diluents [1,2]. The preparation of a porous structure consists of two steps. First, the homogeneous polymer solution has to undergo liquid-liquid demixing to obtain a polymer-rich continuous matrix and a dispersed polymer-lean phase of almost pure solvent. The second step is a fixation step to give the structure mechanical stability. Liquid-liquid demixing takes place when the solvent power is not sufficient anymore to keep the polymer dissolved. Two major methods can be distinguished to decrease the solvent power, either by adding a non-solvent which is called diffusion-induced phase-separation (DIPS) or by changing the temperature, defined as thermally-induced phase-separation (TIPS). The latter method is studied in this paper. Fig. 1 shows a schematic representation of the formation process of a porous structure with the TIPS method. The dashed arrow represents a typical cooling trajectory. When passing the binodal, the homogenous polymer solution liquid-liquid demixes in a polymer-rich phase with a polymer concentration given by the binodal and a polymer-lean phase of almost pure diluent. After liquid-liquid demixing of the binary polymer-diluent system, the structure has to be fixed. This can be achieved by crystallization, vitrification (glass transition), or gelation of the polymer-rich phase. After the structure fixation, it is possible to remove the polymer-lean phase by evaporation or extraction. An advantage of the TIPS method over the DIPS method is the ability to form porous structures with polymers which do not form a solution at room temperature, for example polyethylene and polypropylene.
45
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
H
500 I
binodal
V v
400
t,....
:3
",1 1
(!1 13. E 300
F-
200
1
0.0
S
'
I 0.2
'
I 0.4
'
I 0.6
'
I 0.8
'
1.0
Volume fraction polymer (-) Figure 1. Example of a phase diagram of a binary polymer- diluent solution. The black arrow represents a cooling trajectory to obtain a porous structure with the TIPS method. It goes from the homogeneous solution (H) via the liquid-liquid demixing gap (L-L) into the structure fixation area (S). ~c is the critical concentration of the solution. The porous structures obtained with TIPS can be used for more applications than membranes, for example: separators m electrochemical cells, synthetic leather, 'breathable' rainwear, diapers, surgical dressings and bandages [3], inertial confmed fusion targets [4,5], biodegradable implants for cell transplantation [6], and scaffolds for tissue engineering [7]. Typical cell sizes obtained with this method are m the order of micrometers.
1.2
Historical Background The formation of porous structures via TIPS was firstly reported by
Castro [8]. In this patent, he investigated hundreds of polymer-diluent systems, both amorphous and crystalline polymers, on their ability to form a porous structure. Later on, Shipman [9] described a procedure to prepare porous structures with a certain shape, and Josefiak et al. [10] introduced a second liquid prior to the cooling step to adjust the morphology of the porous structure. The first papers on the TIPS method were published in 1984 [11 ]. Since then, the interest in the formation of porous structures of various polymerdiluent systems with the TIPS method has increased rapidly. Table 1 gives a
46
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijdcn
summary of crystalline polymer-diluent systems used in the TIPS method and Table 2 gives an overview of the formation of amorphous porous structures via TIPS. Besides the studies on new polymer-diluent systems, studies were performed on variations of the TIPS method by introducing, for example, an evaporation step to obtain an asymmetric structm'e [ 12-16]. Table 1. Crystalline polymer-diluent systems which have been used to form a porous structure with TIPS.
Polymer
Diluent
Diphenylether [ 17-19] n,n-Bis(2-hydroxylethyl)tallowamme [11,20-231 Polyvinylidene fluoride Cyclohexanone [ 11] Butrolactone (PVDF) Propylenecarbonate Carbitolacetate Nitrobenzene [24] Isotactic polystyrene (iPS) Cyclohexane [25] Dioxane / isopropanol Poly(tetrafluorethylene-co-per- Chlorotrifluorethylene [26] fluoro-(propyl vinylether)) (Teflon| Neoflona~crPFA) Poly(2,6-dimethyl- 1,4Cyclohexanol [ 12] phenylene ether) (PPE) Nylon 12 Poly(ethyleneglycol) [27] Dioxane / water [28] Poly(~,-benzyl-l-glutamate) Benzene (PBIG) Diehloroethane Polyethylene (HDPE) Diphenylether [29] 1,2-Ditridecylphtalate / hexadecane[30] Polysilastyrene (PSS) Cyclohexane [31 ] Benzene 4-Methyl- 1-pentene (PMP) Diisopropylbenzene [5] Polylactide 1,4-Dioxane [6]
Isotactic polypropylene (iPP)
As mentioned before, the formation of a porous structure with the TIPS method consists of two steps: liquid-liquid demixing to obtain a porous structure and the fixation step to obtain mechanical stability. Both steps have been studied separately for quite a time. The first papers on liquid-liquid demixing of polymer solutions were already published in 1950's (for references see [43,44]).
47
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
The existence of crystalline polymers has been known from the mid 1940's (a historic overview can be found in [45]) and the vitrification temperature depression (or glass transition temperature depression), for example, was already quantified in 1961 [46]. Table 2. Amorphous polymer-diluent systems which have been used to form a porous structure with TIPS. Polymer Diluent Atactic poly(methylmethacrylate) Cyclohexanol [32,33] Tetramethylenesulfone (sulfolane) (aPMMA) [34,35] 1-Butanol [32] Tert-butylalcohol [36] Cyclohexane [37] Atactic polystyrene (aPS) Diethylmalonate Cyclohexanol [38-41 ] Trans- decahy dronap htal ene (decalin) [42]
Liquid-liquid demixing of polymer solutions is extensively studied since tb.e last half of the twentieth century (for references see [43,44]). Phase diagrams of many polymer-diluent systems have been determined visually or with other optical techniques such as optical microscopy and light scattering [21,23,27,28,35,38,43]. These techniques are also used to follow the time dependency of the demixing process to study the kinetics of demixing [17,27,33,47-50]. Also other experimental techniques have been occasionally used to compose phase diagrams or to follow the demixing process like viscometry [51 ], NMR [34], X-ray scattering [52] and DSC [32,42,53,54]. 1.3
Aim of this Paper In this paper, DSC is used to study liquid-liquid demixing and the structure fixation step of an amorphous polymer solution. In addition, rheological experiments are carried out to study liquid-liquid demixing as well. Using DSC and rheology, new insights may be obtained in the membrane formation process. However, practical reasons can be mentioned to use these techniques instead of the frequently used light based techniques. For example, with light based techniques the refractive index between both the diluent and polymer has to differ significantly, and transparency of the experimental setup is required to observe a signal. This is not a problem with DSC and rheological experiments. Furthermore, with DSC and rheology it is possible to detect both the liquid-liquid demixing step and the structure fixation step in one experiment. 48
Dtfferential Scanning Calorimetry And Rheological Experiments To Study Membrane Formauon Via Thermally-Induced Phase-Separation Van Der Heijden
With optical microscopy for example, liquid-liquid demixing can be observed but it is impossible to observe the glass transition. Finally, high temperatures are involved in the TIPS process and to exclude evaporation of diluent, closed sample holders are necessary to use, and this is very easy to achieve with DSC. 2.
DIFFERENTIAL SCANNING CALORIMETRY [55]
2.1
Introduction
With DSC, the heat flow is measured as a function of time or temperature to observe physical transitions. After the introduction of the first differential scanning calorimetry (DSC) apparatus polymers have been studied extensively. See Refs. [56,57] for a historical overview about the development of thermal analysis devices in general and for polymers in particular. DSC is a well-known technique to study solid-liquid demixing (crystallization) [20,23,24,29] and vitrification [32,38,54] of liquid-liquid demixed polymer solutions. Berghmans and co-workers [32,42,53,54] used DSC as well to determine the liquid-liquid demixing temperature. They carried out DSC experiments for the systems atactic polystyrene / decalin and atactic polymethylmethacrylate / 1-butanol and cyclohexanol respectively. Upon cooling (cooling rate 5 K.min-1), an exothermic heat flow shift was observed and the onset was taken as the liquid-liquid demixing temperature. This signal agreed very well with optical observations. With one DSC run they could determine both the liquid-liquid demixing temperature and the glass transition temperature of the polymer-rich phase. However, DSC is hardly used for the determination of liquid-liquid demixing because the heat effect is very small and disappears easily in the base line drift. Recently, a rather new technique has been developed: temperature modulated differential scanning calorimetry (TMDSC) [58,59]. In spite of some discussion about the interpretation of the measured signals [60-67], TMDSC is a very useful device to measure small heat effects and to separate overlapping thermal events. In the next sections, TMDSC experiments are used to examine the cooling trajectory of binary polymer solutions in the concentrated region (at the right side of the critical point in Fig.l). The experimental work described in this work is carried out with the polymer-diluent system atactic polystyrene in 1dodecanol. A number of reasons can be mentioned to choose this system. This system forms a porous polymer structure with the TIPS method at room temperature [8], 1-dodecanol is a non-toxic solvent, polystyrene has solvents at room temperature and therefore easy to use, and many physical data are available for both components.
49
Differential Scanning Calorimetry And Rheologicai Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
2.2
Temperature Modulated Differential Scanning Calorimetry
An extensive description of this technique can be found in Refs. [58,62,68]. Here, only a short description is given. With TMDSC, a second function (for example a sine wave) is superimposed onto the conventional linear or isothermal temperature ramps (see Fig. 2).
200
--
o o 195-(9 Q.
E
(9 I--
190 -I
0
'
I
1
'
I
'
2
Time
I
3
'
I
4
'
5
(rain)
Figure 2" Linear cooling trajectory (thick black line) with superimposed temperature modulations. The temperature ramp can then be described as: (1)
T = To + bt + A sin cot
where To is the initial temperature, b the underlying scanning rate, A the temperature amplitude, co the angular frequency, and t the time. The TAinstruments user manual [68] suggests a value of the underlying scanning rate (b) below 5 K.min -1 and typical values for A and co are 1 K and 60 s. The resulting heat flow consists of two contributions. The first part is caused by rapid processes and is proportional to the scanning rate. The second part is caused by kinetically hindered or irreversible processes and hence independent of the scanning rate: dQ dt
= cp
dT -~
+ f(t,T)
(2)
"
50
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
dQ/dt is the heat flow, Cp the heat capacity, and fit, T) a contribution to kinetic effects. The resulting heat flow is a wave function as well. With the help of discrete Fourier transformation software which resides in the DSC module, the heat flow signal is deconvoluted in a cyclic signal and an underlying signal (which is equivalent to the conventional DSC). In this work the modulus of the complex heat capacity is used to present the TMDSC results. This quantity (]cp*]) is calculated from only the amplitude of both the temperature and the heat flow modulation. It is thus completely derived from the cyclic signal. Physical processes, which take place at time scales smaller or comparable to the modulation period, contribute to this complex heat capacity. Examples of such physical processes are vibrations and rotations of atoms. Slow physical processes, like the mobility of vitrified polymers with time scales much larger than a modulation period does not contribute to the complex heat capacity.
(Icp*l)
2.3
Experimental
2.3.1 Materials Atactic polystyrene (aPS) (Styron* 686E) was kindly supplied by Dow Benelux NV (Mw and M~/Mn" 2.3.105 g.mo1-1 and 2.1 respectively, determined with GPC). The diluent, 1-dodecanol (purity > 98%, Merck-Schuchardt), was used without further purification.
2.3.2 Sample preparation A homogeneous solution of aPS and 1-dodecanol was prepared in a threeneck bottle under nitrogen at 200~ 1-Dodecanol vapor was allowed to evaporate during stirring with a mechanical stirrer. Small amounts of various polymer concentrations were poured in Petri-dishes and cooled in air. The compositions of the samples were determined by thermogravimetric analysis. About 20 mg of the sample was inserted in a platinum sample pan of a TGA 2950 Thermogravic Analyzer of TA Instruments and heated up to 200~ with a heating rate of 10 K.min 1. Afterwards the temperature was kept constant at 200~ for maximum 2 h to evaporate all the 1-dodecanol. From the ultimate weight loss the polymer concentration was determined.
2.3.3 Temperature
Modulated
Differential
Scanning
Calorimetry
(TMDSC) The TMDSC used is a DSC 2920 of TA Instruments. Calibration with indium and high density polyethylene (HDPE) (for calibration of the heat capacity) was carried out. About 5 mg of the sample was put in the aluminum closed sample pan. The TMDSC was heated to 200~ and kept isothermally for
51
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
30 min to ensure homogeneity. The cooling rate was set to 2 K.min -1 to 0~ and after an isothermal step of 5 min the sample was heated again with 2 K.min -1. The amplitude of the superimposed sine wave was 1 K with a period of 60 s [68]. The glass transition temperature Tg and the liquid-liquid demixing temperature TL-Las well as the heat capacity shift at TL_Lwas determined with the TA Universal Analysis software.
2.3.4 Optical Microscopy (O51) To compare the TMDSC results with a well-known technique for studying liquid-liquid demixing, optical microscopy experiments are carried out. The polymer sample was placed on an object glass within an aluminum ring (thickness 0.1 mm, inner diameter 5 mm) and covered by a second glass. To prevent diluent loss caused by evaporation, laboratory grease was used to stick the aluminum spacer to the object glasses [23]. The sample was placed in a hot stage (Linkam THMS 600) which was controlled by the Linkam TMS92 hot stage controller. The sample was heated and cooled with a rate of 2 K.min -1 and demixing was observed visually with an Olympus BH2 microscope (magnification 200x). 2.4
Results Cooling and subsequent heating curves of aPS - 1-dodecanol are plotted in Fig. 3 for two polymer concentrations (weight fractions of 0.38 and 0.69). The modulus of the complex heat capacity (fCp*f) is plotted at the y-axis. Two transitions can be observed: the glass transition and a small baseline shift at higher temperatures, which is assumed to be the liquid-liquid demixing temperature. In the following, the onset of this signal upon cooling is defined as the liquid-liquid demixing temperature (TL_L) comparable with the observations of Amauts et al. and Vandeweerdt et al. with the conventional DSC [32,42,53,54]. The glass transition temperature (Tg) is chosen as the onset upon cooling because below this temperature influences can be expected of vitrification on the liquid-liquid demixing behavior. The difference in heat capacities (Icp'l) between the results for the two polymer concentrations can be explained by the heat capacity difference between both single components. Literature values of the heat capacity for pure polystyrene and 1-dodecanol at T = 180~ are 2.1 J.gl.Kl [69] and 3.1 J.gl.K-1 [70], respectively. Therefore, a polymer solution that contains the higher content of polystyrene should have a smaller heat capacity than a diluted polymer solution. At least, when it is assumed that the heat capacity of a solution of aPS in 1-dodecanol is between the values of both pure components.
52
Differential Scanning Calorimetry And Rheological Expenments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
3.0
,7 ,7
2.5--
o o
r
--3 "
~
2.0
0
==...=.=.
1.5
--
A
169'wt/~176 I !T'LL I
40
80
120
160
' 200
Temperature (~ Figure 3. Cooling and subsequent heating curves (weight fractions of polymer: 0.38 and 0.69). Gray lines" heating curves, black lines: cooling curves.
With conventional DSC experiments, an optimum has to be found between a high scanning rate which results in a high intensity and a low scanning rate which results in a high resolution of a thermal signal. Consequently, a small thermal transition such as liquid-liquid demixing is very difficult to observe with conventional DSC. With TMDSC, both a low scanning rate (the underlying scanning rate) and higher scanning rates (the temperature modulations) are present within one experiment and therefore it is very suitable to detect liquid-liquid demixing. In the cooling curves the L-L phase transitions at Tz.-L are represented by a steep heat capacity shift. The heating curves have the same slopes as the cooling curves only at the liquid-liquid demixing temperatures the transition is not as distinct. This difference is discussed later in this section. In the following, the details of the cooling curves are further used and discussed only. Performing such TMDSC cooling experiments over a large concentration range allows the construction of the phase diagram of the polymer-diluent system. To support the assumption of the base line shift to stem from the L-L demixing, the TMDSC results are compared with optical microscopy (OM) data indicating visually the phenomenon of L-L demixing. In Fig. 4, the TL_Land Tg determined with TMDSC and OM are plotted. The open circles represent the TMDSC liquid-liquid demixing data whereas the filled black squares are the OM data. The closed circles are the glass-transition temperature data points. The liquid-liquid demixing data obtained from TMDSC
53
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
and OM agree well and therefore it can be concluded that the observed TMDSC signals are indeed caused by liquid-liquid demixing. 200
160 0 o ~
120
Q.
E I-.
so
40 0.0
0.2 Weight
0.4 fraction
0.6
0.8
polymer
1.0 (-)
Figure 4. Phase diagram aPS- 1-dodecanol. Open circles: TMDSC data L-L demixing. Closed squares: OM data L-L demixing (cloud points). Closed circles: TMDSC data glass transition. Lines are drawn to guide the eye. The shift of the TMDSC curve at the liquid-liquid temperature is assumed to be completely caused by the enthalpy of demixing. The difference in values in the heat capacity shift (Acp* at TL-L, defined in Fig. 3) between the different concentrations is caused by the interaction between polymer and diluent. This can be quantified by calculating the enthalpy of mixing with the help of the Flory-Huggins theory [54]. To minimize the error in the calculation of the modulus of the complex heat capacity with the TMDSC software, it is recommended that at least 4 complete superimposed cycles fit within a phase transition [68]. This requirement is satisfied for the glass transition because this transition covers a temperature range of at least 10 K. However, in case of liquid-liquid demixing the heat capacity shift only covers a temperature interval of 2 K, so only one modulated cycle fits within this transition. By lowering the underlying cooling rate, the number of cycles within the transition can be increased; the resulting TMDSC curves are shown in Fig. 5. From Fig. 5 it can be concluded that cooling rates of 2 K.min -1 and lower have no significant influence on the measured modulus of the complex heat capacity. As already mentioned the measured complex heat capacity is only measured from the amplitudes of the modulated heat flow and modulated temperature, hence it is not influenced by the cooling rate. However, the shape of the curve can be influenced by time-dependent effects like the growth of the
54
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separauon Van Der Heijden
demixing domains and the increasing viscosity of the polymer solution (see section 3) upon cooling. The growth of the demixed domains can be the reason for the difference between the cooling and heating curve (see Fig. 3 for a 38 wt.% polymer solution). In the demixed polymer solution regions of almost pure diluent grow when the temperature of the solution is in between the TL-Land the Tg. The TMDSC heating curve was measured after the cooling trajectory, so, the demixed domains already had a long time to grow. Therefore, it is more difficult for the polymer solution to follow the temperature modulation because of larger distances the diluent molecules must diffuse. 0.2 K.min 1 2.8
2 K.min ~
~
"7 ~ 2.4 -v
-1
O 2.0
1.6
~"
~
5 K-min ~
I 50
'
I 100
~
I
'
150
I 200
T e m p e r a t u r e (~C) Figure 5. Influence of cooling rate on modulus of the complex heat capacity. Weight fraction of polymer is 0.48.
The shape of the curves at low cooling rates (between 0.2 K.min 1 and 2 K.min 1) are comparable, this means that influence of time depending effects play a negligible role in the cooling curves at these cooling rates. 2.5
Conclusions
and Outlook
With Temperature Modulated DSC, liquid-liquid demixing of polymerdiluent systems can be determined as well as the glass transition of the polymerrich phase in one run at a relatively low scanning rate. Liquid-liquid demixing observed with TMDSC agrees well with visually observed cloud points.
55
Differential Scannmg Calonmetry And Rheolog~cal Expenments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
Upon quantifying the heat capacity shift at the liquid-liquid demixing temperature additional information with respect to the thermodynamics of the polymer solutions can be obtained. Furthermore, it offers the possibility to follow the heat capacity shift as a function of temperature both theoretically and experimentally to study the kinetics of liquid-liquid demixing. Especially this is interesting in the region of the glass transition temperature in which the structure fixation step can be studied. 3.
VISCOMETRY AND RHEOLOGY
3.1
Introduction
During the formation of porous structures with the TIPS process, the homogenous solution is demixed in two liquid phases and afterwards the polymer-rich phase is vitrified. It is expected that these phase transitions can be observed with rheological experiments. The possibility of studying liquid-liquid demixing with viscometry and rheology experiments is briefly discussed in this section. The experiments are carried out for the polymer-diluent system aPS diisodecylphthalate (DIDP). The reason to study this system is the following. The high vapor pressure of many common diluents in relation to the liquidliquid demixing temperature is often a problem studying the thermally-induced phase-separation method experimentally. Special care has to be taken to prevent evaporation of the diluent during an experiment. The diluent DIDP shows a cloud point with the polymer of about 50~ and has a very low vapor pressure. (For comparison, the vapor pressure of water at T = 20~ is 2.103 Pa and the vapor pressure of DIDP at T = 100~ is 0.1 Pa [70].) Therefore, for this system no special care has to be taken to avoid evaporation of diluent.
3.2
Experimental Homogeneous solutions of aPS in DIDP (Merck-Schuchardt, purity>99%) were prepared in the same way as described in section 2.3.2. The viscosity of the polymer diluent system aPS-DIDP was measured with a Brabender Viscotron, Mod. Nr 8024. Starting at high temperatures (in the homogeneous solution), the viscosity was measured for different shear rates at a fixed temperature. Subsequently the temperature was lowered and the complete procedure was repeated. Furthermore, polymer solutions of aPS in DIDP were studied in a Bohlin Rheometer CS50. Upon cooling with a cooling rate of 2 K.min -1, the complex viscosity was measured for different strains at a constant frequency of 0.1 Hz.
56
Differential Scanning Calorimetry And Rheological Expenments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
3.3
Results 200
/
150 - (/) t~ fl_
~3
S"1
II
>'100 (/) o o (/)
>
50
~'~11 I
20
w
40
I
60
s-~
J
80
Temperature (~ Figure 6. Viscosity of a 30 wt.% solution of aPS in DIDP as a function of the temperature upon cooling for different shear rates (indicated in the figure). Results of the viscosity measurements are visualized in Fig. 6 for a polymer concentration of 30 wt.%. Experiments in the concentration range of polymer between 20 and 40 wt.% show a comparable trend'as plotted in Fig. 6. The observation of an increasing viscosity upon cooling as the temperature approaches the liquid-liquid demixing temperature is in agreement with the work of Wolf et aL [51 ]. In fact, the location of the sharp decrease in viscosity (between T = 45 and 50~ is in good agreement with cloud point experiments of this polymer-diluent system (T = 49~ The observation of the sudden decrease at lower temperatures inside the liquid-liquid demixing gap is not in agreement with the expectation. Upon cooling a polymer solution, the viscosity increases. At the liquid-liquid demixing temperature, a polymer-rich matrix is formed enclosing the polymerlean phase. The viscosity is expected to be dominated by the viscosity of the continuous, polymer-rich, phase of the system. Therefore, it should be expected that upon liquid-liquid demixing the viscosity increases further. An explanation of the sharp decrease in viscosity could be the loss of contact between the polymer rich matrix and the rotating cone of the viscometer. Upon storage below the cloud point temperature the demixed polymer solution displays synereses; DIDP is expelled out of the continuous phase. This process is probably enhanced by the applied shear rate resulting into the formation of a slip
57
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
layer of DIDP between the polymer rich matrix and the rotating cone. To apply less mechanical force on the polymer solution, an oscillating rheometer was used instead of the rotating viscometer. 2 O.
S9
1
0.5
8 20
30
40
50
60
70
Temperature (~ Figure 7. Complex viscosity for a 30 wt.% solution of aPS in DIDP. The frequency is 0.l Hz. The strain varies from 0.1 to 0.001. The cooling rate is 2 K.min-1. In Fig. 7, a small plateau can be observed upon cooling at the liquid-liquid demixing temperature in the complex viscosity at a low strain (0.001). However, as expected, the complex viscosity increases upon further cooling. With increasing strains the plateau becomes longer and the complex viscosity even shows a depression at a strain of 0.1. The reason of this depression is probably the same as in the viscosity experiments. Diluent is pushed out of the demixed solution and forms a slip layer. When using rheology or viscometry experiments to determine the liquidliquid demixing temperature, sufficient mechanical force should be applied to achieve immediate synereses of the demixed solution. The synereses results into an apparent viscosity drop that could be used as a marker for the onset of demixing. However, for studying the mechanical properties of the demixed solution, very small strains are recommended in order to avoid shear enhanced synereses. Otherwise, the measurement of the mechanical properties of the demixed solution is unreliable.
58
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-lnduced Phase-Separation Van Der Heijden
3.4
Conclusions and Outlook Liquid-liquid demixing temperatures can be observed with rheology and viscometry by using high shear rates. To study the mechanical behavior of demixing polymer solutions, care has to be taken to avoid altering of the liquid structure due to shear forces. It should be of much interest when rheological experiments can be carried out with a polymer solution at different frequencies. These results can be compared with results of emulsions to obtain information about the structure belonging to the demixed solution. Furthermore, rheological experiments can be very suitable to study the mechanical behavior of the polymer-diluent system at a temperature in the region of the structure fixation step. For example, the system aPS - DIDP forms a gel at room temperature because of the high viscosity of the polymer solution. After the gelation of the solution, no growth of the demixed domains is observed. Hence, gelation is sufficient to stop the liquid-liquid demixing process. The polymer-diluent system aPS - 1-dodecanol, which was studied in section 2, shows an onset of the glass transition temperature at about 65~ The glass transition temperature is known to be sufficient to stop the liquid-liquid demixing process. However, it is very interesting to study whether at higher temperatures than the glass transition temperature, the liquid-liquid demixing process is influenced by or even stopped by gelation. Rheological experiments have the potential to clarify the structure fixation step of a polymer solution.
4.
ACKNOWLEDGEMENTS
E. Schomaker, J. B o s e n R. Lammers of Akzo Nobel in Arnhem are acknowledged for discussions with respect to the TMDSC work, and for giving us the possibility to carry out the TMDSC experiments. B. Reuvers (Akzo Nobel in Amhem) and M. van Egmond (University of Twente) are acknowledged for carrying out the rheological experiments and for the discussions. REFERENCES [1]
[2] [3] [4] [5] [6]
M. Mulder, Basic principles of membrane technology, Kluwer Acedemic Publishers, Dordrecht, (1996). L.J. Zeman and A.L. Zydney, Microfiltration and Ultrafiltration, Marcel Dekker, Inc., New York, (1996). D.R. Lloyd, K.E. Kinzer and H.S. Tseng, J. Membrane Sci., 52 (1990) 239. A.T. Young, J. Vac. Sci. Techn., A 4 (1985) 1128. J.M. Williams and J.E. Moore, Polymer, 28 (1987) 1950. C. Schugens, V. Maquet, C. Grandfils, R. Jerome and P. Teyssie, Polymer, 37 (1996) 1027.
59
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
[7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]
[24] [25] [26] [27]
[28] [29]
[301 [31] [32] [33] [34]
[35] [36] [37]
[38] [39]
[40] [41] [42] [43] [44] [451
Y.S. Nam and T.G. Park, J. Biomed. Mat. Res., 47 (1999) 8. A.J. Castro, US-patent 4247498 (1981). G.H. Shipman, US-patent 4539256 (1985). C. Josefiak and F. Wechs, GB-patent 2115425B (1985). W.C. Hiatt, G.H. Vitzthum, W.K.B., K. Gerlaeh and C. Josefiak, in Material science of synthetic membranes, Lloyd, D.R.(Ed.) American Chemical Society, Missouri (1984) 229. S. Berghmans, H. Berghmans and H.E.H. Meijer, J. Membrane Sci., 116 (1996) 171. H. Matsuyama, S. Berghmans and D.R. Lloyd, J. Membrane Sci., 142 (1998) 213. H. Matsuyama, S. Berghmans and D.R. Lloyd, Polymer, 40 (1999) 2289. P.M. Atkinson and D.R. Lloyd, J. Membrane Sci., 175 (2000) 225. P.M. Atkinson and D.R. Lloyd, J. Membrane Sci., 171 (2000) 1. A. Laxminarayan, K.S. McGuire, S.S. Kim and D.R. Lloyd, Polymer, 35 (1994) 3060. K.S. McGuire, A. Laxminarayan and D.R. Lloyd, Polymer, 35 (1994) 4404. K.S. McGuire, K.W. Lawson and D.R. Lloyd, J. Membrane Sci., 99 (1995) 127. K.E. Kinzer and D.R. Lloyd, Polym. Mater. Sci. Eng., 61 (1989) 794. D.R. Lloyd, S.S. Kim and K.E. Kinzer, J. Membrane Sci., 64 (1991) 1. S.S. Kim and D.R. Lloyd, J. Membrane Sci., 64 (1991) 13. S.S. Kim and D.R. Lloyd, Polymer, 33 (1992) 1047. J.H. Aubert, Macromolecules, 21 (1988) 3468. J.H. Aubert and C.R.L., Polymer, 26 (1985) 2047. M.R. Caplan, C.-Y. Chiang, D.R. Lloyd and L.Y. Yen, J. Membrane Sci., 130 (1997) 219. B.J. Cha, K. Char, J.-J. Kim, S.S. Kim and C.K. Kim, J. Membrane Sci., 108 (1995) 219. C.L. Jackson and M.T. Shaw, Polymer, 31 (1990) 1070. L. Aerts, M. Kunz, H. Berghmans and R. Koningsveld, Makromolek. Chem., 194 (1993) 2697. H.C. Vadalia, H.K. Lee, A.S. Myerson and K. Levon, J. Membrane Sci., 89 (1994) 37. L.L. Whinnery, W.R. Even, J.V. Beach and D.A. Loy, J. Polym. Sci. Polym. Chem., 34 (1996) 1623. P. Vandeweerdt, H. Berghmans and Y. Tervoort, Macromolecules, 24 (1991) 3547. P.D. Graham, A.J. Pervan and A.J. McHugh, Macromolecules, 30 (1997) 1651. G.T. Caneba and D.S. Soong, Macromolecules, 18 (1985) 2538. F.J. Tsai and J.M. Torkelson, Macromolecules, 23 (1989) 775. F.J. Tsai and J.M. Torkelson, Macromolecules, 23 (1990) 4983. S.-W. Song and J.M. Torkelson, Macromolecules, 27 (1994) 6389. R.M. Hikmet, S. Callister and A. Keller, Polymer, 29 (1988) 1378. J.H. Aubert, Macromolecules, 23 (1990) 1446. S.-W. Song and J.M. Torkelson, Polym. preprints, 34 (1993) 496. S.-W. Song and J.M. Torkelson, J. Membrane Sci., 98 (1995) 209. J. Arnauts, H. Berghmans and R. Koningsveld, Makromolek. Chem., 194 (1993) 77. R. Koningsveld and A.J. Staverman, J. Polym. Sci., Part A-2, 6 (1968) 349. R. Koningsveld, W.H. Stockmayer and E. Nies, Polymer phase diagrams, Oxford University Press, Oxford, (2001). F. Khoury and E. Passaglia, in Treatise on solid state chemistry, Plenum Press, Hannay, N.B. (Ed.), New York-London, (1976) 335.
60
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
[46] [47]
[48] [49]
[50] [51] [52] [53] [54]
[55] [56] [57]
[58] [59]
[60] [61] [62] [63] [64] [65] [66] [67]
[68] [69]
[70]
F.N. Kelley and F. Bueche, J. Polym. Sci., 50 (1961) 549. S. Nojima, K. Shiroshita and T. Nose, Polym. J., 14 (1982) 289. J. Lal and R. Bansil, Macromolecules, 24 (1991) 290. K.S. McGuire, A. Laxminarayan and D.R. Lloyd, Polymer, 36 (1995) 4951. J. Szydlowski and W.A. Van Hook, Macromolecules, 31 (1998) 3255. B.A. Wolf and M.C. Sezen, Macromolecules, 10 (1977) 1010. Y. Xie, K.F.J. Ludwig, R. Basnil, P.D. Gallagher, C. Kon~ik and G. Morales, Macromolecules, 29 (1996) 6150. J. Arnauts and H. Berghmans, Polym. Comm., 28 (1987) 66. J. Arnauts, R. De Cooman, P. Vandeweerdt, R. Koningsveld and H. Berghmans, Thermochim. Acta, 238 (1994) 1. P.C. van der Heijden, M.H.V. Mulder and W. M., Thermochim. Acta, 378 (2001) 27. B. Wunderlich, Thermal Analysis, Academix Press, Inc., London, (1990). V.B.F. Mathot, Calorimetry and thermal analysis of polymers, Hanser Publishers, Munich, (1994). M. Reading, Trends Polym. Sci., 1 (1993) 248. M. Reading, B.K. Hahn and B.S. Crowe, US-patent 5224775 (1993). J. Schawe, Thermochim. Acta, 260 (1995) 1. T. Ozawa and K. Kanari, Thermochim. Acta, 253 (1995) 183. K.J. Jones, I. Kinshott, M. Reading, A.A. Lacey, C. Nikolopoulos and H.M. Pollock, Thermochim. Acta, 304/305 (1997) 187. G.W.H. HOhne, Thermochim. Acta, 304/305 (1997) 121. J.E.K. Schawe, Thermochim. Acta, 304/305 (1997. B. Wunderlich, A. Boiler, I. Okazaki and K. Ishikiriyama, Thermochim. Acta, 304/305 (1997) 125. R. Scherrenberg, V. Mathot and A. Van Hemelrijck, Thermochim. Acta, 330 (1999) 3. R. Scherrenberg, V. Mathot and P. Steeman, J. Therm. Anal., 54 (1998) 477. TA Instruments, Modulated DSC (MDSC) How does it work ? (1998). B. Wunderlich, ATHAS Databank, http://web.utk.edu/-~athas/databank/intro.html (2000). T.E. Daubert, R.P. Danner, H.M. Sibul and C.C. Stebbins, Physical and thermodynamic properties of pure chemicals. Data compilation, Taylor&Francis, Pennsylvania, (1989).
61
Chapter 3
Differential scanning calorimetry and rheological experiments to study membrane formation via thermallyinduced phase-separation P.C. van der Heijden, M.H.V. Mulder, M. Wessling University of Twente, PO Box 217, 7500 AE Enschede, The Netherlands, Fax: xx31-53-4894611 1.
INTRODUCTION
1.1
Thermally-Induced Phase-Separation Porous polymer membranes can be prepared by different techniques: track-etching, stretching and phase-separation [ 1]. The phase-separation method is very common to prepare polymeric membranes and much attention has been paid to prepare membranes from various polymers and diluents [1,2]. The preparation of a porous structure consists of two steps. First, the homogeneous polymer solution has to undergo liquid-liquid demixing to obtain a polymer-rich continuous matrix and a dispersed polymer-lean phase of almost pure solvent. The second step is a fixation step to give the structure mechanical stability. Liquid-liquid demixing takes place when the solvent power is not sufficient anymore to keep the polymer dissolved. Two major methods can be distinguished to decrease the solvent power, either by adding a non-solvent which is called diffusion-induced phase-separation (DIPS) or by changing the temperature, defined as thermally-induced phase-separation (TIPS). The latter method is studied in this paper. Fig. 1 shows a schematic representation of the formation process of a porous structure with the TIPS method. The dashed arrow represents a typical cooling trajectory. When passing the binodal, the homogenous polymer solution liquid-liquid demixes in a polymer-rich phase with a polymer concentration given by the binodal and a polymer-lean phase of almost pure diluent. After liquid-liquid demixing of the binary polymer-diluent system, the structure has to be fixed. This can be achieved by crystallization, vitrification (glass transition), or gelation of the polymer-rich phase. After the structure fixation, it is possible to remove the polymer-lean phase by evaporation or extraction. An advantage of the TIPS method over the DIPS method is the ability to form porous structures with polymers which do not form a solution at room temperature, for example polyethylene and polypropylene.
45
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
H
500 I
binodal
V v
400
t,....
:3
",1 1
(!1 13. E 300
F-
200
1
0.0
S
'
I 0.2
'
I 0.4
'
I 0.6
'
I 0.8
'
1.0
Volume fraction polymer (-) Figure 1. Example of a phase diagram of a binary polymer- diluent solution. The black arrow represents a cooling trajectory to obtain a porous structure with the TIPS method. It goes from the homogeneous solution (H) via the liquid-liquid demixing gap (L-L) into the structure fixation area (S). ~c is the critical concentration of the solution. The porous structures obtained with TIPS can be used for more applications than membranes, for example: separators m electrochemical cells, synthetic leather, 'breathable' rainwear, diapers, surgical dressings and bandages [3], inertial confmed fusion targets [4,5], biodegradable implants for cell transplantation [6], and scaffolds for tissue engineering [7]. Typical cell sizes obtained with this method are m the order of micrometers.
1.2
Historical Background The formation of porous structures via TIPS was firstly reported by
Castro [8]. In this patent, he investigated hundreds of polymer-diluent systems, both amorphous and crystalline polymers, on their ability to form a porous structure. Later on, Shipman [9] described a procedure to prepare porous structures with a certain shape, and Josefiak et al. [10] introduced a second liquid prior to the cooling step to adjust the morphology of the porous structure. The first papers on the TIPS method were published in 1984 [11 ]. Since then, the interest in the formation of porous structures of various polymerdiluent systems with the TIPS method has increased rapidly. Table 1 gives a
46
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijdcn
summary of crystalline polymer-diluent systems used in the TIPS method and Table 2 gives an overview of the formation of amorphous porous structures via TIPS. Besides the studies on new polymer-diluent systems, studies were performed on variations of the TIPS method by introducing, for example, an evaporation step to obtain an asymmetric structm'e [ 12-16]. Table 1. Crystalline polymer-diluent systems which have been used to form a porous structure with TIPS.
Polymer
Diluent
Diphenylether [ 17-19] n,n-Bis(2-hydroxylethyl)tallowamme [11,20-231 Polyvinylidene fluoride Cyclohexanone [ 11] Butrolactone (PVDF) Propylenecarbonate Carbitolacetate Nitrobenzene [24] Isotactic polystyrene (iPS) Cyclohexane [25] Dioxane / isopropanol Poly(tetrafluorethylene-co-per- Chlorotrifluorethylene [26] fluoro-(propyl vinylether)) (Teflon| Neoflona~crPFA) Poly(2,6-dimethyl- 1,4Cyclohexanol [ 12] phenylene ether) (PPE) Nylon 12 Poly(ethyleneglycol) [27] Dioxane / water [28] Poly(~,-benzyl-l-glutamate) Benzene (PBIG) Diehloroethane Polyethylene (HDPE) Diphenylether [29] 1,2-Ditridecylphtalate / hexadecane[30] Polysilastyrene (PSS) Cyclohexane [31 ] Benzene 4-Methyl- 1-pentene (PMP) Diisopropylbenzene [5] Polylactide 1,4-Dioxane [6]
Isotactic polypropylene (iPP)
As mentioned before, the formation of a porous structure with the TIPS method consists of two steps: liquid-liquid demixing to obtain a porous structure and the fixation step to obtain mechanical stability. Both steps have been studied separately for quite a time. The first papers on liquid-liquid demixing of polymer solutions were already published in 1950's (for references see [43,44]).
47
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
The existence of crystalline polymers has been known from the mid 1940's (a historic overview can be found in [45]) and the vitrification temperature depression (or glass transition temperature depression), for example, was already quantified in 1961 [46]. Table 2. Amorphous polymer-diluent systems which have been used to form a porous structure with TIPS. Polymer Diluent Atactic poly(methylmethacrylate) Cyclohexanol [32,33] Tetramethylenesulfone (sulfolane) (aPMMA) [34,35] 1-Butanol [32] Tert-butylalcohol [36] Cyclohexane [37] Atactic polystyrene (aPS) Diethylmalonate Cyclohexanol [38-41 ] Trans- decahy dronap htal ene (decalin) [42]
Liquid-liquid demixing of polymer solutions is extensively studied since tb.e last half of the twentieth century (for references see [43,44]). Phase diagrams of many polymer-diluent systems have been determined visually or with other optical techniques such as optical microscopy and light scattering [21,23,27,28,35,38,43]. These techniques are also used to follow the time dependency of the demixing process to study the kinetics of demixing [17,27,33,47-50]. Also other experimental techniques have been occasionally used to compose phase diagrams or to follow the demixing process like viscometry [51 ], NMR [34], X-ray scattering [52] and DSC [32,42,53,54]. 1.3
Aim of this Paper In this paper, DSC is used to study liquid-liquid demixing and the structure fixation step of an amorphous polymer solution. In addition, rheological experiments are carried out to study liquid-liquid demixing as well. Using DSC and rheology, new insights may be obtained in the membrane formation process. However, practical reasons can be mentioned to use these techniques instead of the frequently used light based techniques. For example, with light based techniques the refractive index between both the diluent and polymer has to differ significantly, and transparency of the experimental setup is required to observe a signal. This is not a problem with DSC and rheological experiments. Furthermore, with DSC and rheology it is possible to detect both the liquid-liquid demixing step and the structure fixation step in one experiment. 48
Dtfferential Scanning Calorimetry And Rheological Experiments To Study Membrane Formauon Via Thermally-Induced Phase-Separation Van Der Heijden
With optical microscopy for example, liquid-liquid demixing can be observed but it is impossible to observe the glass transition. Finally, high temperatures are involved in the TIPS process and to exclude evaporation of diluent, closed sample holders are necessary to use, and this is very easy to achieve with DSC. 2.
DIFFERENTIAL SCANNING CALORIMETRY [55]
2.1
Introduction
With DSC, the heat flow is measured as a function of time or temperature to observe physical transitions. After the introduction of the first differential scanning calorimetry (DSC) apparatus polymers have been studied extensively. See Refs. [56,57] for a historical overview about the development of thermal analysis devices in general and for polymers in particular. DSC is a well-known technique to study solid-liquid demixing (crystallization) [20,23,24,29] and vitrification [32,38,54] of liquid-liquid demixed polymer solutions. Berghmans and co-workers [32,42,53,54] used DSC as well to determine the liquid-liquid demixing temperature. They carried out DSC experiments for the systems atactic polystyrene / decalin and atactic polymethylmethacrylate / 1-butanol and cyclohexanol respectively. Upon cooling (cooling rate 5 K.min-1), an exothermic heat flow shift was observed and the onset was taken as the liquid-liquid demixing temperature. This signal agreed very well with optical observations. With one DSC run they could determine both the liquid-liquid demixing temperature and the glass transition temperature of the polymer-rich phase. However, DSC is hardly used for the determination of liquid-liquid demixing because the heat effect is very small and disappears easily in the base line drift. Recently, a rather new technique has been developed: temperature modulated differential scanning calorimetry (TMDSC) [58,59]. In spite of some discussion about the interpretation of the measured signals [60-67], TMDSC is a very useful device to measure small heat effects and to separate overlapping thermal events. In the next sections, TMDSC experiments are used to examine the cooling trajectory of binary polymer solutions in the concentrated region (at the right side of the critical point in Fig.l). The experimental work described in this work is carried out with the polymer-diluent system atactic polystyrene in 1dodecanol. A number of reasons can be mentioned to choose this system. This system forms a porous polymer structure with the TIPS method at room temperature [8], 1-dodecanol is a non-toxic solvent, polystyrene has solvents at room temperature and therefore easy to use, and many physical data are available for both components.
49
Differential Scanning Calorimetry And Rheologicai Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
2.2
Temperature Modulated Differential Scanning Calorimetry
An extensive description of this technique can be found in Refs. [58,62,68]. Here, only a short description is given. With TMDSC, a second function (for example a sine wave) is superimposed onto the conventional linear or isothermal temperature ramps (see Fig. 2).
200
--
o o 195-(9 Q.
E
(9 I--
190 -I
0
'
I
1
'
I
'
2
Time
I
3
'
I
4
'
5
(rain)
Figure 2" Linear cooling trajectory (thick black line) with superimposed temperature modulations. The temperature ramp can then be described as: (1)
T = To + bt + A sin cot
where To is the initial temperature, b the underlying scanning rate, A the temperature amplitude, co the angular frequency, and t the time. The TAinstruments user manual [68] suggests a value of the underlying scanning rate (b) below 5 K.min -1 and typical values for A and co are 1 K and 60 s. The resulting heat flow consists of two contributions. The first part is caused by rapid processes and is proportional to the scanning rate. The second part is caused by kinetically hindered or irreversible processes and hence independent of the scanning rate: dQ dt
= cp
dT -~
+ f(t,T)
(2)
"
50
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
dQ/dt is the heat flow, Cp the heat capacity, and fit, T) a contribution to kinetic effects. The resulting heat flow is a wave function as well. With the help of discrete Fourier transformation software which resides in the DSC module, the heat flow signal is deconvoluted in a cyclic signal and an underlying signal (which is equivalent to the conventional DSC). In this work the modulus of the complex heat capacity is used to present the TMDSC results. This quantity (]cp*]) is calculated from only the amplitude of both the temperature and the heat flow modulation. It is thus completely derived from the cyclic signal. Physical processes, which take place at time scales smaller or comparable to the modulation period, contribute to this complex heat capacity. Examples of such physical processes are vibrations and rotations of atoms. Slow physical processes, like the mobility of vitrified polymers with time scales much larger than a modulation period does not contribute to the complex heat capacity.
(Icp*l)
2.3
Experimental
2.3.1 Materials Atactic polystyrene (aPS) (Styron* 686E) was kindly supplied by Dow Benelux NV (Mw and M~/Mn" 2.3.105 g.mo1-1 and 2.1 respectively, determined with GPC). The diluent, 1-dodecanol (purity > 98%, Merck-Schuchardt), was used without further purification.
2.3.2 Sample preparation A homogeneous solution of aPS and 1-dodecanol was prepared in a threeneck bottle under nitrogen at 200~ 1-Dodecanol vapor was allowed to evaporate during stirring with a mechanical stirrer. Small amounts of various polymer concentrations were poured in Petri-dishes and cooled in air. The compositions of the samples were determined by thermogravimetric analysis. About 20 mg of the sample was inserted in a platinum sample pan of a TGA 2950 Thermogravic Analyzer of TA Instruments and heated up to 200~ with a heating rate of 10 K.min 1. Afterwards the temperature was kept constant at 200~ for maximum 2 h to evaporate all the 1-dodecanol. From the ultimate weight loss the polymer concentration was determined.
2.3.3 Temperature
Modulated
Differential
Scanning
Calorimetry
(TMDSC) The TMDSC used is a DSC 2920 of TA Instruments. Calibration with indium and high density polyethylene (HDPE) (for calibration of the heat capacity) was carried out. About 5 mg of the sample was put in the aluminum closed sample pan. The TMDSC was heated to 200~ and kept isothermally for
51
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
30 min to ensure homogeneity. The cooling rate was set to 2 K.min -1 to 0~ and after an isothermal step of 5 min the sample was heated again with 2 K.min -1. The amplitude of the superimposed sine wave was 1 K with a period of 60 s [68]. The glass transition temperature Tg and the liquid-liquid demixing temperature TL-Las well as the heat capacity shift at TL_Lwas determined with the TA Universal Analysis software.
2.3.4 Optical Microscopy (O51) To compare the TMDSC results with a well-known technique for studying liquid-liquid demixing, optical microscopy experiments are carried out. The polymer sample was placed on an object glass within an aluminum ring (thickness 0.1 mm, inner diameter 5 mm) and covered by a second glass. To prevent diluent loss caused by evaporation, laboratory grease was used to stick the aluminum spacer to the object glasses [23]. The sample was placed in a hot stage (Linkam THMS 600) which was controlled by the Linkam TMS92 hot stage controller. The sample was heated and cooled with a rate of 2 K.min -1 and demixing was observed visually with an Olympus BH2 microscope (magnification 200x). 2.4
Results Cooling and subsequent heating curves of aPS - 1-dodecanol are plotted in Fig. 3 for two polymer concentrations (weight fractions of 0.38 and 0.69). The modulus of the complex heat capacity (fCp*f) is plotted at the y-axis. Two transitions can be observed: the glass transition and a small baseline shift at higher temperatures, which is assumed to be the liquid-liquid demixing temperature. In the following, the onset of this signal upon cooling is defined as the liquid-liquid demixing temperature (TL_L) comparable with the observations of Amauts et al. and Vandeweerdt et al. with the conventional DSC [32,42,53,54]. The glass transition temperature (Tg) is chosen as the onset upon cooling because below this temperature influences can be expected of vitrification on the liquid-liquid demixing behavior. The difference in heat capacities (Icp'l) between the results for the two polymer concentrations can be explained by the heat capacity difference between both single components. Literature values of the heat capacity for pure polystyrene and 1-dodecanol at T = 180~ are 2.1 J.gl.Kl [69] and 3.1 J.gl.K-1 [70], respectively. Therefore, a polymer solution that contains the higher content of polystyrene should have a smaller heat capacity than a diluted polymer solution. At least, when it is assumed that the heat capacity of a solution of aPS in 1-dodecanol is between the values of both pure components.
52
Differential Scanning Calorimetry And Rheological Expenments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
3.0
,7 ,7
2.5--
o o
r
--3 "
~
2.0
0
==...=.=.
1.5
--
A
169'wt/~176 I !T'LL I
40
80
120
160
' 200
Temperature (~ Figure 3. Cooling and subsequent heating curves (weight fractions of polymer: 0.38 and 0.69). Gray lines" heating curves, black lines: cooling curves.
With conventional DSC experiments, an optimum has to be found between a high scanning rate which results in a high intensity and a low scanning rate which results in a high resolution of a thermal signal. Consequently, a small thermal transition such as liquid-liquid demixing is very difficult to observe with conventional DSC. With TMDSC, both a low scanning rate (the underlying scanning rate) and higher scanning rates (the temperature modulations) are present within one experiment and therefore it is very suitable to detect liquid-liquid demixing. In the cooling curves the L-L phase transitions at Tz.-L are represented by a steep heat capacity shift. The heating curves have the same slopes as the cooling curves only at the liquid-liquid demixing temperatures the transition is not as distinct. This difference is discussed later in this section. In the following, the details of the cooling curves are further used and discussed only. Performing such TMDSC cooling experiments over a large concentration range allows the construction of the phase diagram of the polymer-diluent system. To support the assumption of the base line shift to stem from the L-L demixing, the TMDSC results are compared with optical microscopy (OM) data indicating visually the phenomenon of L-L demixing. In Fig. 4, the TL_Land Tg determined with TMDSC and OM are plotted. The open circles represent the TMDSC liquid-liquid demixing data whereas the filled black squares are the OM data. The closed circles are the glass-transition temperature data points. The liquid-liquid demixing data obtained from TMDSC
53
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
and OM agree well and therefore it can be concluded that the observed TMDSC signals are indeed caused by liquid-liquid demixing. 200
160 0 o ~
120
Q.
E I-.
so
40 0.0
0.2 Weight
0.4 fraction
0.6
0.8
polymer
1.0 (-)
Figure 4. Phase diagram aPS- 1-dodecanol. Open circles: TMDSC data L-L demixing. Closed squares: OM data L-L demixing (cloud points). Closed circles: TMDSC data glass transition. Lines are drawn to guide the eye. The shift of the TMDSC curve at the liquid-liquid temperature is assumed to be completely caused by the enthalpy of demixing. The difference in values in the heat capacity shift (Acp* at TL-L, defined in Fig. 3) between the different concentrations is caused by the interaction between polymer and diluent. This can be quantified by calculating the enthalpy of mixing with the help of the Flory-Huggins theory [54]. To minimize the error in the calculation of the modulus of the complex heat capacity with the TMDSC software, it is recommended that at least 4 complete superimposed cycles fit within a phase transition [68]. This requirement is satisfied for the glass transition because this transition covers a temperature range of at least 10 K. However, in case of liquid-liquid demixing the heat capacity shift only covers a temperature interval of 2 K, so only one modulated cycle fits within this transition. By lowering the underlying cooling rate, the number of cycles within the transition can be increased; the resulting TMDSC curves are shown in Fig. 5. From Fig. 5 it can be concluded that cooling rates of 2 K.min -1 and lower have no significant influence on the measured modulus of the complex heat capacity. As already mentioned the measured complex heat capacity is only measured from the amplitudes of the modulated heat flow and modulated temperature, hence it is not influenced by the cooling rate. However, the shape of the curve can be influenced by time-dependent effects like the growth of the
54
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separauon Van Der Heijden
demixing domains and the increasing viscosity of the polymer solution (see section 3) upon cooling. The growth of the demixed domains can be the reason for the difference between the cooling and heating curve (see Fig. 3 for a 38 wt.% polymer solution). In the demixed polymer solution regions of almost pure diluent grow when the temperature of the solution is in between the TL-Land the Tg. The TMDSC heating curve was measured after the cooling trajectory, so, the demixed domains already had a long time to grow. Therefore, it is more difficult for the polymer solution to follow the temperature modulation because of larger distances the diluent molecules must diffuse. 0.2 K.min 1 2.8
2 K.min ~
~
"7 ~ 2.4 -v
-1
O 2.0
1.6
~"
~
5 K-min ~
I 50
'
I 100
~
I
'
150
I 200
T e m p e r a t u r e (~C) Figure 5. Influence of cooling rate on modulus of the complex heat capacity. Weight fraction of polymer is 0.48.
The shape of the curves at low cooling rates (between 0.2 K.min 1 and 2 K.min 1) are comparable, this means that influence of time depending effects play a negligible role in the cooling curves at these cooling rates. 2.5
Conclusions
and Outlook
With Temperature Modulated DSC, liquid-liquid demixing of polymerdiluent systems can be determined as well as the glass transition of the polymerrich phase in one run at a relatively low scanning rate. Liquid-liquid demixing observed with TMDSC agrees well with visually observed cloud points.
55
Differential Scannmg Calonmetry And Rheolog~cal Expenments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
Upon quantifying the heat capacity shift at the liquid-liquid demixing temperature additional information with respect to the thermodynamics of the polymer solutions can be obtained. Furthermore, it offers the possibility to follow the heat capacity shift as a function of temperature both theoretically and experimentally to study the kinetics of liquid-liquid demixing. Especially this is interesting in the region of the glass transition temperature in which the structure fixation step can be studied. 3.
VISCOMETRY AND RHEOLOGY
3.1
Introduction
During the formation of porous structures with the TIPS process, the homogenous solution is demixed in two liquid phases and afterwards the polymer-rich phase is vitrified. It is expected that these phase transitions can be observed with rheological experiments. The possibility of studying liquid-liquid demixing with viscometry and rheology experiments is briefly discussed in this section. The experiments are carried out for the polymer-diluent system aPS diisodecylphthalate (DIDP). The reason to study this system is the following. The high vapor pressure of many common diluents in relation to the liquidliquid demixing temperature is often a problem studying the thermally-induced phase-separation method experimentally. Special care has to be taken to prevent evaporation of the diluent during an experiment. The diluent DIDP shows a cloud point with the polymer of about 50~ and has a very low vapor pressure. (For comparison, the vapor pressure of water at T = 20~ is 2.103 Pa and the vapor pressure of DIDP at T = 100~ is 0.1 Pa [70].) Therefore, for this system no special care has to be taken to avoid evaporation of diluent.
3.2
Experimental Homogeneous solutions of aPS in DIDP (Merck-Schuchardt, purity>99%) were prepared in the same way as described in section 2.3.2. The viscosity of the polymer diluent system aPS-DIDP was measured with a Brabender Viscotron, Mod. Nr 8024. Starting at high temperatures (in the homogeneous solution), the viscosity was measured for different shear rates at a fixed temperature. Subsequently the temperature was lowered and the complete procedure was repeated. Furthermore, polymer solutions of aPS in DIDP were studied in a Bohlin Rheometer CS50. Upon cooling with a cooling rate of 2 K.min -1, the complex viscosity was measured for different strains at a constant frequency of 0.1 Hz.
56
Differential Scanning Calorimetry And Rheological Expenments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
3.3
Results 200
/
150 - (/) t~ fl_
~3
S"1
II
>'100 (/) o o (/)
>
50
~'~11 I
20
w
40
I
60
s-~
J
80
Temperature (~ Figure 6. Viscosity of a 30 wt.% solution of aPS in DIDP as a function of the temperature upon cooling for different shear rates (indicated in the figure). Results of the viscosity measurements are visualized in Fig. 6 for a polymer concentration of 30 wt.%. Experiments in the concentration range of polymer between 20 and 40 wt.% show a comparable trend'as plotted in Fig. 6. The observation of an increasing viscosity upon cooling as the temperature approaches the liquid-liquid demixing temperature is in agreement with the work of Wolf et aL [51 ]. In fact, the location of the sharp decrease in viscosity (between T = 45 and 50~ is in good agreement with cloud point experiments of this polymer-diluent system (T = 49~ The observation of the sudden decrease at lower temperatures inside the liquid-liquid demixing gap is not in agreement with the expectation. Upon cooling a polymer solution, the viscosity increases. At the liquid-liquid demixing temperature, a polymer-rich matrix is formed enclosing the polymerlean phase. The viscosity is expected to be dominated by the viscosity of the continuous, polymer-rich, phase of the system. Therefore, it should be expected that upon liquid-liquid demixing the viscosity increases further. An explanation of the sharp decrease in viscosity could be the loss of contact between the polymer rich matrix and the rotating cone of the viscometer. Upon storage below the cloud point temperature the demixed polymer solution displays synereses; DIDP is expelled out of the continuous phase. This process is probably enhanced by the applied shear rate resulting into the formation of a slip
57
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
layer of DIDP between the polymer rich matrix and the rotating cone. To apply less mechanical force on the polymer solution, an oscillating rheometer was used instead of the rotating viscometer. 2 O.
S9
1
0.5
8 20
30
40
50
60
70
Temperature (~ Figure 7. Complex viscosity for a 30 wt.% solution of aPS in DIDP. The frequency is 0.l Hz. The strain varies from 0.1 to 0.001. The cooling rate is 2 K.min-1. In Fig. 7, a small plateau can be observed upon cooling at the liquid-liquid demixing temperature in the complex viscosity at a low strain (0.001). However, as expected, the complex viscosity increases upon further cooling. With increasing strains the plateau becomes longer and the complex viscosity even shows a depression at a strain of 0.1. The reason of this depression is probably the same as in the viscosity experiments. Diluent is pushed out of the demixed solution and forms a slip layer. When using rheology or viscometry experiments to determine the liquidliquid demixing temperature, sufficient mechanical force should be applied to achieve immediate synereses of the demixed solution. The synereses results into an apparent viscosity drop that could be used as a marker for the onset of demixing. However, for studying the mechanical properties of the demixed solution, very small strains are recommended in order to avoid shear enhanced synereses. Otherwise, the measurement of the mechanical properties of the demixed solution is unreliable.
58
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-lnduced Phase-Separation Van Der Heijden
3.4
Conclusions and Outlook Liquid-liquid demixing temperatures can be observed with rheology and viscometry by using high shear rates. To study the mechanical behavior of demixing polymer solutions, care has to be taken to avoid altering of the liquid structure due to shear forces. It should be of much interest when rheological experiments can be carried out with a polymer solution at different frequencies. These results can be compared with results of emulsions to obtain information about the structure belonging to the demixed solution. Furthermore, rheological experiments can be very suitable to study the mechanical behavior of the polymer-diluent system at a temperature in the region of the structure fixation step. For example, the system aPS - DIDP forms a gel at room temperature because of the high viscosity of the polymer solution. After the gelation of the solution, no growth of the demixed domains is observed. Hence, gelation is sufficient to stop the liquid-liquid demixing process. The polymer-diluent system aPS - 1-dodecanol, which was studied in section 2, shows an onset of the glass transition temperature at about 65~ The glass transition temperature is known to be sufficient to stop the liquid-liquid demixing process. However, it is very interesting to study whether at higher temperatures than the glass transition temperature, the liquid-liquid demixing process is influenced by or even stopped by gelation. Rheological experiments have the potential to clarify the structure fixation step of a polymer solution.
4.
ACKNOWLEDGEMENTS
E. Schomaker, J. B o s e n R. Lammers of Akzo Nobel in Arnhem are acknowledged for discussions with respect to the TMDSC work, and for giving us the possibility to carry out the TMDSC experiments. B. Reuvers (Akzo Nobel in Amhem) and M. van Egmond (University of Twente) are acknowledged for carrying out the rheological experiments and for the discussions. REFERENCES [1]
[2] [3] [4] [5] [6]
M. Mulder, Basic principles of membrane technology, Kluwer Acedemic Publishers, Dordrecht, (1996). L.J. Zeman and A.L. Zydney, Microfiltration and Ultrafiltration, Marcel Dekker, Inc., New York, (1996). D.R. Lloyd, K.E. Kinzer and H.S. Tseng, J. Membrane Sci., 52 (1990) 239. A.T. Young, J. Vac. Sci. Techn., A 4 (1985) 1128. J.M. Williams and J.E. Moore, Polymer, 28 (1987) 1950. C. Schugens, V. Maquet, C. Grandfils, R. Jerome and P. Teyssie, Polymer, 37 (1996) 1027.
59
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
[7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]
[24] [25] [26] [27]
[28] [29]
[301 [31] [32] [33] [34]
[35] [36] [37]
[38] [39]
[40] [41] [42] [43] [44] [451
Y.S. Nam and T.G. Park, J. Biomed. Mat. Res., 47 (1999) 8. A.J. Castro, US-patent 4247498 (1981). G.H. Shipman, US-patent 4539256 (1985). C. Josefiak and F. Wechs, GB-patent 2115425B (1985). W.C. Hiatt, G.H. Vitzthum, W.K.B., K. Gerlaeh and C. Josefiak, in Material science of synthetic membranes, Lloyd, D.R.(Ed.) American Chemical Society, Missouri (1984) 229. S. Berghmans, H. Berghmans and H.E.H. Meijer, J. Membrane Sci., 116 (1996) 171. H. Matsuyama, S. Berghmans and D.R. Lloyd, J. Membrane Sci., 142 (1998) 213. H. Matsuyama, S. Berghmans and D.R. Lloyd, Polymer, 40 (1999) 2289. P.M. Atkinson and D.R. Lloyd, J. Membrane Sci., 175 (2000) 225. P.M. Atkinson and D.R. Lloyd, J. Membrane Sci., 171 (2000) 1. A. Laxminarayan, K.S. McGuire, S.S. Kim and D.R. Lloyd, Polymer, 35 (1994) 3060. K.S. McGuire, A. Laxminarayan and D.R. Lloyd, Polymer, 35 (1994) 4404. K.S. McGuire, K.W. Lawson and D.R. Lloyd, J. Membrane Sci., 99 (1995) 127. K.E. Kinzer and D.R. Lloyd, Polym. Mater. Sci. Eng., 61 (1989) 794. D.R. Lloyd, S.S. Kim and K.E. Kinzer, J. Membrane Sci., 64 (1991) 1. S.S. Kim and D.R. Lloyd, J. Membrane Sci., 64 (1991) 13. S.S. Kim and D.R. Lloyd, Polymer, 33 (1992) 1047. J.H. Aubert, Macromolecules, 21 (1988) 3468. J.H. Aubert and C.R.L., Polymer, 26 (1985) 2047. M.R. Caplan, C.-Y. Chiang, D.R. Lloyd and L.Y. Yen, J. Membrane Sci., 130 (1997) 219. B.J. Cha, K. Char, J.-J. Kim, S.S. Kim and C.K. Kim, J. Membrane Sci., 108 (1995) 219. C.L. Jackson and M.T. Shaw, Polymer, 31 (1990) 1070. L. Aerts, M. Kunz, H. Berghmans and R. Koningsveld, Makromolek. Chem., 194 (1993) 2697. H.C. Vadalia, H.K. Lee, A.S. Myerson and K. Levon, J. Membrane Sci., 89 (1994) 37. L.L. Whinnery, W.R. Even, J.V. Beach and D.A. Loy, J. Polym. Sci. Polym. Chem., 34 (1996) 1623. P. Vandeweerdt, H. Berghmans and Y. Tervoort, Macromolecules, 24 (1991) 3547. P.D. Graham, A.J. Pervan and A.J. McHugh, Macromolecules, 30 (1997) 1651. G.T. Caneba and D.S. Soong, Macromolecules, 18 (1985) 2538. F.J. Tsai and J.M. Torkelson, Macromolecules, 23 (1989) 775. F.J. Tsai and J.M. Torkelson, Macromolecules, 23 (1990) 4983. S.-W. Song and J.M. Torkelson, Macromolecules, 27 (1994) 6389. R.M. Hikmet, S. Callister and A. Keller, Polymer, 29 (1988) 1378. J.H. Aubert, Macromolecules, 23 (1990) 1446. S.-W. Song and J.M. Torkelson, Polym. preprints, 34 (1993) 496. S.-W. Song and J.M. Torkelson, J. Membrane Sci., 98 (1995) 209. J. Arnauts, H. Berghmans and R. Koningsveld, Makromolek. Chem., 194 (1993) 77. R. Koningsveld and A.J. Staverman, J. Polym. Sci., Part A-2, 6 (1968) 349. R. Koningsveld, W.H. Stockmayer and E. Nies, Polymer phase diagrams, Oxford University Press, Oxford, (2001). F. Khoury and E. Passaglia, in Treatise on solid state chemistry, Plenum Press, Hannay, N.B. (Ed.), New York-London, (1976) 335.
60
Differential Scanning Calorimetry And Rheological Experiments To Study Membrane Formation Via Thermally-Induced Phase-Separation Van Der Heijden
[46] [47]
[48] [49]
[50] [51] [52] [53] [54]
[55] [56] [57]
[58] [59]
[60] [61] [62] [63] [64] [65] [66] [67]
[68] [69]
[70]
F.N. Kelley and F. Bueche, J. Polym. Sci., 50 (1961) 549. S. Nojima, K. Shiroshita and T. Nose, Polym. J., 14 (1982) 289. J. Lal and R. Bansil, Macromolecules, 24 (1991) 290. K.S. McGuire, A. Laxminarayan and D.R. Lloyd, Polymer, 36 (1995) 4951. J. Szydlowski and W.A. Van Hook, Macromolecules, 31 (1998) 3255. B.A. Wolf and M.C. Sezen, Macromolecules, 10 (1977) 1010. Y. Xie, K.F.J. Ludwig, R. Basnil, P.D. Gallagher, C. Kon~ik and G. Morales, Macromolecules, 29 (1996) 6150. J. Arnauts and H. Berghmans, Polym. Comm., 28 (1987) 66. J. Arnauts, R. De Cooman, P. Vandeweerdt, R. Koningsveld and H. Berghmans, Thermochim. Acta, 238 (1994) 1. P.C. van der Heijden, M.H.V. Mulder and W. M., Thermochim. Acta, 378 (2001) 27. B. Wunderlich, Thermal Analysis, Academix Press, Inc., London, (1990). V.B.F. Mathot, Calorimetry and thermal analysis of polymers, Hanser Publishers, Munich, (1994). M. Reading, Trends Polym. Sci., 1 (1993) 248. M. Reading, B.K. Hahn and B.S. Crowe, US-patent 5224775 (1993). J. Schawe, Thermochim. Acta, 260 (1995) 1. T. Ozawa and K. Kanari, Thermochim. Acta, 253 (1995) 183. K.J. Jones, I. Kinshott, M. Reading, A.A. Lacey, C. Nikolopoulos and H.M. Pollock, Thermochim. Acta, 304/305 (1997) 187. G.W.H. HOhne, Thermochim. Acta, 304/305 (1997) 121. J.E.K. Schawe, Thermochim. Acta, 304/305 (1997. B. Wunderlich, A. Boiler, I. Okazaki and K. Ishikiriyama, Thermochim. Acta, 304/305 (1997) 125. R. Scherrenberg, V. Mathot and A. Van Hemelrijck, Thermochim. Acta, 330 (1999) 3. R. Scherrenberg, V. Mathot and P. Steeman, J. Therm. Anal., 54 (1998) 477. TA Instruments, Modulated DSC (MDSC) How does it work ? (1998). B. Wunderlich, ATHAS Databank, http://web.utk.edu/-~athas/databank/intro.html (2000). T.E. Daubert, R.P. Danner, H.M. Sibul and C.C. Stebbins, Physical and thermodynamic properties of pure chemicals. Data compilation, Taylor&Francis, Pennsylvania, (1989).
61