- Email: [email protected]
The sorption of water vapor by an amorphous polyamide
187
Journal of Membrane Scrence, 65 (1992) 187-199 Elsevrer Scrence Publishers B V , Amsterdam
The sorption of water vapor by an amorphous polyamide Ruben J. Hernandeza, Jack R. Giacin” and Eric A. Grulkeb “School of Packagnzg, MLchzgan State Unrvers@, East Lansmg, MI 48824 (USA) bChemlcal Engzneerzng Department, Mlchtgan State Unwerstty, East Lansmg, MI 48824 (USA) (Received December 23,1988, accepted m revised form July 17,199l)
Abstract A modified dual-mode sorption model represents the sorptron of water vapor by an amorphous polyamide at 23 ‘C The Langmurr equation 1s used to calculate the volume fraction of chemrsorbed solute and the Flory-Huggms equation 1s used instead of Henry’s law to calculate the volume fraction of water which 1s not chemrsorbed This model describes the data over a range of water actrvrtres from zero to one and predicts clusterrng of the sorbant Fourier Transform Infrared (FTIR) spectroscopy data and drelectrrc measurements of the gamma relaxation temperature suggest that the water assocrates with amrde groups at low water activities Keywords barrier membranes, diffusion
dual-sorptron
model, gas and vapor permeatron,
water sorption
and
Introduction
moisture content m Nylon 66 but mcreases with moisture content m the amorphous material
An amorphous polyamide recently has been synthesized and characterized [ 11. This polymer is made from hexamethylenediamme and a mixture of lsophthahc and terephthahc acid. The random placement of these isomers m the polymer chain prevents crystallization of the material. Preliminary studies on the effect of water sorption on the barrier properties of this amorphous polymer show atypical behavior compared to semmrystalhne polyamides. For example, oxygen permeability mcreases as a function of moisture content m Nylon 66 (a semlcrystallme polyamide) but decreases as a function of moisture content in amorphous material The tensile modulus decreases with
[ll.
The mtnguing behavior of the amorphous polyamide system m the presence of water demonstrates the need for equlhbrium models which describe sorption phenomena well over a wide solute activity range Improved models are needed for binary systems and hopefully can be extended to describe ternary systems (gaswater-polymer) m order to interpret and control barrier properties This study models the sorption phenomena of water vapor mto amorphous polyamide as the first step m describing the effects of water on the barrier and mechanical properties of this polymer
Correspondence to Eric A Grulke, Chemical Engmeermg Department, Michigan State Unmemrty, East Lansing, MI 48824 (USA)
Model for gas sorption In glassy polymers Sorption of gases, such as helium, nitrogen,
0376-7388/92/$05
Dual mode sorption models
00 0 1992 Elsevrer Scrence Publishers B V All rights reserved
R.J HERNANDFZ ET AL
188
oxygen, carbon dioxide, and methane, above their critical temperature into glassy polymers has been studied by many researchers. The dual mode sorption model proposed by Barrer et al. [ 21 is the most widely used model for analyzing these data. This model assumes that the solute molecules in the glassy polymer consist of nonspecifically absorbed and specifically adsorbed species which are in dynamic equilibrium in the medium. Equation (1) was proposed to describe Langmuir type curvature observed for gas sorption in glassy polymers at low gas activities. The solubihty of the absorbed species is represented by Henry’s law and the solubility of the adsorbed species are described by a Langmuir type adsorption isotherm. The Langmuir sorption is believed to occur at specific sites, usually considered to be holes in the amorphous polymer structure. Local equilibrium between the absorbed and adsorbed populations is mamtained throughout the polymer matrix [ 31. The total amount of solute sorbed by both mechanisms is:
c=c,+c,=iz,p+
-
CHb
1+bp
where C is the total concentration of sorbed solute in the polymer, Cn and Cn are the solubillties due to absorption (Henry’s law) and adsorption (Langmmr), k,-, is the Henry’s law constant, b is the hole affinity constant, Cn is the hole saturation constant and p is the pressure. The dual model sorption model has been used to describe the solubility of gases in glassy polymers, and in glassy, polar polymers as well. Uragami et al. [4] studied the sorption and equilibrium for CO, m a polyamide film for pressures up to 0.78 atm. Chern et al. [5] applied eqn. (1) to the permeabihty of CO, through Kapton polyamide for pressures up to 16.3 atm Koros and Sanders [6] described a bicomponent sorption process by the dual mode sorption model [CO,/C,H, and CO,/N,O in poly(methy1 methacrylate) 1. In most of the
examples cited above, the gas phase can be considered dilute because the solute activity is much less than one. The solute activity can be approximated by the gas vapor pressure divided by the saturated hquid vapor pressure at the experimental temperature. When the gas acts ideally, simple models such as Henry’s law represent phase equilibria well. Models forpolar systems at hgh solute aciwrtLes Solutes sorbed at temperatures below their boihng point (such as organic solvents and water at room temperature) may not obey simple, ideal phase equilibria relations such as Henry’s law. For these cases, eqn. (1) does not describe the data well over the activity range, 0
(2)
where al is the solute activity, x IS the mteraction parameter, VI is the volume fraction of the solute and V, is the volume fraction of the polymer. Although eqn. (2) does not provide the most accurate description of the thermody-
189
SORPTION OF WATER VAPOR BY AMORPHOUS POLYAMIDE TABLE
1
Models for polar systems at high solute actmltles Solutlon factor
comment
Ref
H,O/Polyacrylomtnle
kd exp (UC)p
Data of [7]
PI
Vmyl chloride/ poly (vmyl chlonde
Flory-Huggms’
Fits for high solute actmltles
PI
k, exp (04 pu=2(l+~)A
Data of [9]
[lOI
)
[Ill
H,O/Kaptons
kip
H,O/Nylon
Flory-Huggms’ ’
Fit by authors
[I21
H,O/EPO~~S
hp
Hysteresis
[I31
H,O/Epoxy’
k,+=exp(l+X)
Two Langmuir factors
[ 141
6s
‘Langmulr sorption observed at low actlvltles but not modeled *RegIstered trademark, E I Du Pont de Nemours and Company, for an aromatic polyether dnmlde 3Trademark of E I De Pont de Nemours, Inc 4Best fit model as determined by authors of this work ‘TGDDM/DDS epoxy resins (MY720, Cuba-Geigy, Ltd ) ‘DBEBA/TETA (Eplkote 828, Shell, Italy) and TBDDM/DDS (Araldlte MY720, Cuba-Geigy, Ltd )
namic performance of polymer solutions, it does contam most of the essential features which distmguish such solutions [ 181 Equation (2) usually fits equihbrla data over a wider range than Henry’s law and reduces to Henry’s law at low gas activities For example, sorption data of toluene m poly (vinyl chloride) is well fit by the Flory-Huggms equation especially for 0.5 < a, < 10 [ 191. Although no model for water sorption mto Nylon 6 was proposed by Sfirakls and Rogers [ 121, this data set fit eqn. (2) with ax value of 188 The use of the Flory-Huggms equation to describe non-specific solute absorption m the place of Henry’s law m eqn. ( 1) is an extension of the model summarized in Table 1 Group contribution or lattice models are alternative choices to eqn. (2 ) , but these are more complex than the Flory-Huggms equation. An issue associated with any model choice is determnnng the value of the model coefficients The con-
stant descrlbmg absorption usually is determined at high solute pressures. The methods for parameter estimation are described in the Results and Discussion section Clusteruzg of solute molecules Self-association of the solute to form clusters has been used to explain a number of transport phenomena occurrmg m polar systems For example, the clustermg of water molecules m polyacrylomtrile coincides with a maximum m the diffusion coefficient [ 201 Clustermg could reduce the effective mobility of water either by mcreasmg the size of the diffusing group or by mcreasmg the tortuosity of diffusion paths due to the presence of clusters A similar effect has been reported for Kapton by Yang et al. [ 111. They calculated the clustermg of water by dlfferentiatlon of a fourth order polynomial fit Puffr and Sabenda [ 211 reported the clustering of water m several polyamides. Sklrrow and Young [22] described the clustering of water, methanol and propanol in Nylon 6 by plotting a,/ V, versus a, In Nylon 6, clustermg has been associated with a change m the diffusion coefficient [ 121 Aronhime et al. [23] reported clustering of water in epoxy resins If clustering is important to solute diffusion, then the equilibrium model should be able to predict its onset The clustermg function developed by Zimm and Lundberg [24,25] has been used with blnary systems m eqmhbrmm to give a measure of the tendency of like molecules to cluster Their clustering function, G,,/V,, is:
Gl VI
-=-
(3(dV1)
v
2
f3a1
_1
(3)
When G,,/ VI is greater than - 1, the solute is expected to cluster Equation (1) is not written m a convenient form for the analytical determmation of the derivative m eqn. (3 ) Common methods for evaluating the clustermg function are either fitting activity versus volume fraction data with pol-
190
R J HERNANDEZ ET AL
ynomials and taking their derivatives or doing a numerical differential of the data. Equation (1) can be rearranged to express the concentration, C, as volume fraction and to express the Henry’s law and Langmuir contributions m terms of activity: v,=V~+V+K,a,+& 1
where Vy and Vk are the Henry’s law and Langmuir volume fraction contributions, KD = kDpS, K= CH f b pS, B = b pS, pS 1s the saturation vapor pressure of the solute and f 1s a conversion factor. Applying eqn. (3) to eqn (4) gives
Gl -V* - -V,
VI - [K,, a, +KaJ(l+B
al)“]
V? (5)
Since the term in brackets is always less than VI, G,,/V, 1s always less than - 1 for the range O
G -= VI
2x 1-2x
v,
(6)
for a constant value of x Equation (6) always predicts clustermg m the range, 0 < V, < l/2 x. The equation is not defined at the upper limit of the range ( VI = 1/2x), which is the spinodal point mdicatmg phase separation Neither the conventional dual mode sorption model or the Flory-Huggms model with a constant value for
x correctly describe the apparent the systems mentioned above.
clustermg
in
Modtfted dual mode sorptron model In this study, eqn. (1) has been modified by using the Flory-Huggms equation to describe non-specific solution rather than Henry’s law. This modification should allow the model to fit over the activity range 0
(7)
where VT” refers to the Flory-Huggms contnbution to the solute volume fraction and Vk refers to the Langmuir contribution. Smce eqn (2) is nonlinear, it is convenient to determme the value for Vy” by numerical methods, such as the Newton-Raphson technique and eqn (7)
(8) The clustering Cl, -=V, +
function
of eqn. (8) 1s
VZ Ko, V: (l+Bc# 1-v:n
VFH (1:2x
1 v;“)--v,
(9)
Equation (9) predicts clustering for some values of the constants but is undefined when* 1-v:n
(1+2x
VCH)=O
(10)
This occurs when V, 1s 1.0 (pure solvent) or l/2 x (phase separation). The replacement of the Henry’s law model by the Flory-Huggms
191
SORPTION OF WATER VAPOR BY AMORPHOUS POLYAMIDE
equation for non-specific absorption gives a model which pre&cts clustering m the activity range O
(ZnS) crystal from a 2% solution with 1,1,1,3,3,3:hexafluoro-2-propanol (Kerr Company, NOW, MI). After evaporation of the solvent at room temperature, the sample was dried m a vacuum oven at 100°C to remove residual solvent and water. The sample was then equlhbrated with water vapor at selected water actlvlty values After attaining equlhbrium, the sample was lmmedlately covered with another crystal of ZnS and transferred to the instrument Dlfferentral scanntng calortmetry and dtelectrtc experrments Film samples were prepared for these analyses by vacuum drying at 100” C and eqmhbrating with the vapor over salt solutions to sve selected water actlvitles. Results and discussion
Dens@ expertments Film densltles were determined m a density gradient column with the gradient made of toluene and carbon tetrachlorlde
Eqwltbrtum sorptton tsotherm Sorption equlhbrmm values of water weight fraction and sample density of the amorphous polyamide at 23°C are presented m Table 2 Isotherm data were obtamed over a wide actlvlty range (0.046
Fourier transform mfrared spectroscopy The sample for infrared analysis was prepared by casting a thin film onto a zmc sulfide
Parameter estlmatlon for equatron (8) Parameter values estimated for eqn (8) were sensitive to the method used to determine them
R J HERNANDEZ ET AL
192
TABLE 2 Water uptake (23°C) Water actlvlty
0 00 0 046 0 056 0 012 0 080 0 090 0 110 0 155 0 189 0 252 0 269 0 308 0 410 0 440 0 565 0 580 0 585 0 635 0 735 0 790 0 860 0 880 0 963
vs water
Water weight fraction 0 00
activity
for amorphous
Density (g/cc) Data
Model
1194
1194 1194 1 194 1 194 1 194 1 194 1 194 1 195 1 196 1 196 1917 1 197 1 199 1 199 1201 1201 1201 1201 1202 1202 1202 1202 1202
0 00468 0 00653 0 00746
-
0 00781 0 00829 0 00929 0 0120 0 0127
1 194 -
0 0164 0 0157
-
0 0215 0 0295 0 0344 0 0401 0 0438 0 0416
1 198 1 199 -
0 0481 0 0583 0 0664 0 0730 0 0758 0 0826
1201 1202
-
Nylon
Volume fraction
000 0 0056 0 0078 0 0089 0 0093 0 0099 0 0111 0 0143 0 0153 0 0197 0 0188 0 0258 0 0354 0 0414 0 0483 0 0527 0 0501 0 0579 0 0702 0 0800 0 0879 0 0913 0 0995
Fig 1 Water volume fraction m amorphous nylon as a function of water actwlty Flory-Huggms model with x=1 63
A nonlinear regressron method was used to determine the values of the three parameters simultaneously. This resulted in Langmuir coefficients with a hrgh amount of error and a fit for eqn. (8) which was not better than the fit for eqn. (2). The least squares technique was not sensltrveto deviations between the data and the model at low activities. However, the devlatlon of eqn. (2) from the data 1s systematic as shown in Fig 1 An alternative parameter estimation method gave coefficients with lower error. The data for a, > 0.4 were used to determine a range of x values which could represent this portion of the curve. x should be between 1.6 and 185 to mmlmize the least square errors for this data set The data for a, < 0 4 were used to determine the Langmuir coefficients. The nonlinear regression method gave the coefficients and the R2 parameter for a set of x values in this range. The R2 value was plotted as a function of x to find the best estimates of the parameters. Figure 2 shows eqn (8) as the solid curve (x= 1.7, K= 0.395 and B = 95.2 ) and the complete data set. At low activities, the Langmuu contrlbutlon 1s high compared to the Flory-Huggins contrlbutlon and at a, = 0 4, rt ISJust over 10% of the total volume fraction of solute. The Langmuu contrlbutlon 1snearly constant (90% of its maximum value) at actlvlties greater than 0.11 and the curve at high water activities should be insensltlve to the values of the Langmmr coefficients. Therefore, the estimation method is consistent with the parameter values. The dashed curve on Fig. 2 1sthe dual-mode sorption model (eqn 4) with a set of coefficients with a low least squares error. The dualmode sorption model fits data of low water actlvltles but shows systematic deviations at high water activities. Sensltmty coeffwents Sensltivlty coefficients mdmate the magmtude of the change of a function due to pertur-
193
SORPTION OF WATER VAPOR BY AMORPHOUS POLYAMIDE
01
-
009
.-
006
--
007 -009 -Vl
005 -001 .-
003 -002 --
I 0
01
02
03
04
06
05 81
07
09
1
09
Fig 2 Comparison of the dual-mode and modified dual-mode sorption models Equation (8) (x= 17, K=O 395, B=95 2), equation (1) (x= 163)
bation in the values of its parameters [%I. Sensitivity coefficients for eqn. (8) are given by the first derivative of the volume fractron with respect to x, I3 and K
(12) (13) where X,, Xn and Xx are the sensltlvlty coefficients of K, B and x. These coefficients were evaluated as a function of activity, a,, at the optimum values of K, B and x and the results are plotted m Fig. 3. For a, > 0.1, X, changes with actlvrty whrle X, and Xn are essentially constant. For high actlvlty values, rt will be difficult to distinguish between K and B parameters At low actlvlty values, they are not linearly dependent and better estimates for them can be obtained The
-151 0
’
’ 2
’
’ 1 4 9 WaterActivity
’
’ 9
1
Fig 3 Sensltlvlty coefficients as a function of water ackvlty
results shown in Fig. 3 suggest that a slmultaneous search for the three parameters will be &fficult over the entire activity range, that high activity data will be msensltive to K and B, and that the best estrmates for K and B will be obtamed at low actrvlty values Equations ( ll)(13 ) give a protocol for evaluating the method of determining the constants and are an improvement over the suggestion that the con-
R J HERNANDEZ ET AL
194
stant (s) for the solution phenomenon should be determined when the solute partial pressure is much larger than the Langmuir affinity constant, b [lo]. Clusterrng analym Equation (9) with the coefficients determined above the empirical polynomial fits of the activity-volume fraction data were used to compute the clustering function for the water sorption data. Thud and fourth order polynomial expressions tit the data with similar least square errors Both expressions approached but did not pass through the zero activity-zero volume fraction point, and neither polynomial replicated the data as well as eqn. (8) Figure 4 shows the clustering function estimated from two polynomial fits (thud and fourth order) and the modified dual-mode sorption model All three equations predict clustering (G,,/ V, > - 1) but &ffer in the solute activity at which this should occur. The fourth order polynomial predicts clustering m the activity range, 0.34~ a, < 0 96 The thud order polynomial predicts clustering for a, > 0 27. Both polynomial fits give clustering functions which go through maxima. There is
001
-31
1
1 -21
’
’ -11
Cluster
’
’
-1
4
B
Function
Fig 4 Clustering function vs water activity (-) eqn (8)) (-------) thud order polynomial, (- - -) fourth order polynomial
no obvious physical explanation for why clustering should decrease, or disappear, at high water activities. Equation (9) predicts clustermg at a, = 0.38 and the function is increasing monotomcally over the whole activity range. Smce the coefficients in eqn. (8) affect the clustering prediction, coefficient accuracy is important to the results Phymal evgdence for solute brndtng The Langmmr isotherm was derived based on chemisorption of a chemical species to a specific bmdmg site characterized by a single energy of sorption Solute sorption described by this isotherm for the water-polyamide solution might result m observable changes m the polymer, particularly d the water associated with a specific group on the polymer backbone One obvious possibihty is water mteraction with the amide group FTIR spectroscopy studies were done to detect possible changes in hydrogen bondmg between N-H and carbonyl groups by observing the vibrational modes of the amide group (Amide I and Amide II bands). The Amide I mode includes contributions from C=O stretching, C-N stretching and CC-N deformation vibrations. The Amide II band includes N-H plane bendmg, C-N stretching and C-C stretching vibrations Figure 5 shows a typical FTIR spectrum taken at room temperature over the range, 800-4000 cm-l for a cast polyamide sample The Amide I and Amide II modes occur at 1640 and 1541 cm-‘, respectively, for the dry sample and are the most intense bands. The frequency scale and the relative mtensity of the Amide I and Amide II modes were internally calibrated with reference to the CH, stretching band at 2858 cm-’ Table 3 lists the absorption frequencies and Table 4 lists the mtensity ratios for the Amide I and Amide II bands of the sample recorded as a function of water volume fraction and activity The frequency of the CH2 stretchmg band, which served as the internal calibra-
195
SORPTION OF WATER VAPOR BY AMORPHOUS POLYAMIDE
40000 L 3000
2000 I
I
1BOO 1
1200 I
1
cm-’
Fig 5 FTIR spectrum of amorphous polyamide at room temperature TABLE 3 Amide I, Amide II and CH absorption frequencies Actlvlty
CH peak
Amide I
Amide II
0 0 08 0 308 0 56 0 88
2858 2858 2858 2858 2858
1640 1640 1641 1641 1641
1541 1545 1545 1545 1546
Units cm-’ TABLE 4 Intensity ratios of Amide I and Amide II with respect to CH bands Actlvlty
Amide I/C-H
Amide II/C-H
0
171 178 187 180 184
5 45 5 44 5 11 5 43 5 14
0 08 0 308 0 56 0 88
tlon band, is presented for reference. Figure 6 shows the absorption region from 1500 to 1600 cm-’ (Amide II mode) as a function of a,. The spectra are displayed on an absolute absorbance scale. The Amide I band remained essentially con-
oo1600
1560
1520
cm-’ Fig 6 Amide II spectrum as a function of water actlvlty
stant at 1640 cm-’ as did its intensity. The peak maxima of the Amide II band shifted to higher frequency, 1541 to 1546 cm-‘, over the range 0%-9.1%, water volume fraction. The intensity ratio of the Amide II band remained essentially constant The shift of the Amide II band to higher frequency with increasing water content is consistent with an increase m the average hydrogen bond strength [ 291 These data suggest that the water does not change m hydrogen-bonding strength of the carbonyl group but does change the hydrogen bonding strength of the N-H group. The shift in the Amide II band is completed at the water activity at which 90% of the Langmuir sites would be filled (Table 3 ) In linear, ahphatlc homopolyamldes there is essentially 100% hydrogen bonding, as evldented by the absence of bands in the infrared spectra above 3300 cm-’ [ 301. In structurally irregular copolymers, such as amorphous polyamide, both the N-H stretchmg region and the Amide I region can be resolved mto “free”
196
and hydrogen-bonded stretching mode. Skranovek et al [29] resolved the N-H stretching region into three components. a “free” N-H mode, a hydrogen bonded N-H stretchmg mode and an Amide B mode The Amide I band was resolved mto a “free” and hydrogen-bonded carbonyl mode Most of the first water molecules sorbed into the polymer associate with the “free” N-H groups of the amorphous polyamide This would be an exothermic process and should have a single activation energy associated with it. The formation of these new hydrogen bonds should have a neghgible enthalpic effect [ 21,311. The extent of disruption of the self-association of polyamide neighbormg groups by water molecules is not know Both the model and the FTIR data suggest that chemisorption has been completed at activities well below those at which clustering occur At a water activity of 0 08 (where the Langmuir sites are 89% saturated), the total amount of chemrsorbed water is only 4% of the total water volume fraction This amount corresponds to 3 34x 10m3 g of water per gram of polymer, or one molecule of water per 22 repeating umts of the polymer Each repeat unit contains 2 amide groups so only 2.3% of the amide groups available form hydrogen bonds with water Either hole filhng or chemisorption might be occurring in the polyamide for the Langmuir contribution as Kapton is known to hydrogen bond with water Water IS known to hydrogen bond with a variety of epoxies and the dualmode sorption model describes some data well [ 131 Apicella et al. [ 141 tested the hole-fillmg hypothesis with an epoxy system by sorbmg water between 0 and 60 atm It was claimed that Langmun factors were needed to model these data one for chemisorption (probably to hydroxyl or secondary amme groups) and one for sorption into holes One conclusion of previous workers [ 33,341
R J HERNANDEZ ET AL
is that the Langmuir component of sorption disappears at the Tg of the polymer. This suggests that sorption is not related to chemisorption with a specific chemical group, since chemisorptlon can occur above and below the system T, The observed association between water and the amide group could be interpreted as showmg that holes preferentially occur near this chemical structure. The physical basis for site-specific holes is not clear. However, data for the sorption of low molecular weight solutes m polyethylene above Tg have been fitted with the Langmuir mode [35]. Whether such sites are the same as Langmuu sites below Tg 1s an open question Relaxatton temperatures Several relaxation temperatures have been described for semicrystalline polyamides Papir et al [31] related the gamma relaxation peak (140K) of Nylon 6 to the movement of methylene and polar groups. Sfirakis and Rogers [ 121 reported that the gamma peak for the Nylon B-water system was affected by water concentration. The change in mtensity of the gamma peak with changmg water concentration leveled off at a concentration correspondmg to the apparent onset of water clustering The gamma relaxation temperature was measured via the dielectric constant at 1 kH. The gamma peak of the amorphous polyamide was 173°K m the dry state. The peak was not observed for samples equilibrated at a, > 0.08 The difference could be due to the mteraction of water with the amide group decreasing the umts available to participate in the relaxation process. Data for the alpha relaxation temperature (glass transition temperature ) were obtained by DSC measurements The effect of sorbed water on Tg 1s shown in Fig 7, where Tg values are plotted with water volume fraction. There is an apparent change m slope of T, with activity near a, = 0.40. A similar trend was observed
197
SORPTION OF WATER VAPOR BY AMORPHOUS POLYAMIDE
r
1:21
.__
1
D
llll-
0 go-
H
D
u
E 1 20 ::
f 79c d
50
1 L
3oo
I
1
0
I
I
3
I
t
5
I
I
7
I
L
I
9
0
Uolun~riraction[x101
11l9k
1 1
I
I
3
I
I
5
I
I
7
I
I
9
10
Water Activity
Fig 7 Glass transhon temperature vs water actlvlty
Fig 8 Density as a function of water actlvlty
on Tg versus water volume fraction obtained from the dielectric experiments. The slope change occurs near the water activity at which clustermg is predicted for the polymer system at 23°C Water solubihty data were not taken at other temperatures so that the effect of temperature on clustering is not known. The data suggest a relationship between water clustering and a change m the slope of Z’* with water activity However, this interpretation depends on temperature having a negligible effect on water clustering in this temperature range
flection point occurs m the density curve near the water activity range (al=0.27 at the maximum of the derivative of density with respect to activity, for these data) over which clustering is predicted to begin The water activity at this inflection point is similar to that near the apparent change m Tg vs. a, curve (Fig. 7). The increase in sample density with mcreasmg water volume fraction occurs as the Tg of the system is decreasing. There are few comparisons of density and Tg data for partially soluble solutes, so it is not clear whether these results are specific to water/polyamides or occur m other systems as well. In general, small solutes plasticize polymers. Water appears to be plasticizing the amorphous polyamide at high temperatures even though, at room temperatureo, adding water mcreases the sample density and the solute could be clustermg
Dens&y data Figure 8 shows the change in total density with water activity. The amorphous nylon solution becomes more dense as water is added This is the opposite effect expected from a mixing rule depending on additive molar volume fractions. A similar result has been reported for an epoxy system [ 321. Notice that there is httle change m density below water activities of 0.1, where the Langmuir coefficients suggest that chemisorption (or “hole filhng”) is essentially completed It is interesting to point out that the permeabihty of this amorphous nylon to oxygen decreases by a factor of 2 for films at water activities greater than 0 10 compared to a sample from which water has been removed An m-
Interpretatwn of the effects of sorpfon on transport and physical properttes of amorphous polyamzde At water activities less than 0.10, water preferentially associates with amide sites, although a low mole fraction of the total amide sites on the polymer is occupied. The association can be detected by observing the wave shift in the amide group absorbance. Langmmr sorption of
R J HERNANDEZ ET AL
198
water is essentially completed at al = 0 1 (about 1 vol.% water). There is no measurable change in system density at this water activity. There is a reduction of the polymer Tgat a water activity of 0.1, probably due to the plastlcizmg effect of water As with other dual-mode sorption system, polymer history and relaxations probably affect the Langmuir sorption capacity, C;, , although such data have not been taken m this study At higher water activities, water solubility is controlled by the Flory-Huggms component and water clustering is predicted by Eqn. (9). System density increases nonlinearly as water activity increases. A change in the Tgof the system with respect to water activity coincides with the clustering prediction as does the inflection point in the density-activity curve Clustering may be associated with the increase in density and the decrease in system specific volume. Acknowledgments The authors acknowledge the support of this work by the grant received from E.1 Du Pont de Nemours and Company, Inc.
6
7
8
9
10
11
12
13
14
15 16 17
References P Blatz, PropertIes of films from blends of amorphous and crystalline nylons, presented at the AIChE Nat Conf , Apnl2-6,1989 R M Barrer, J A Barrier and J Slater, Sorption and dlffuslon m ethyl cellulose Part III Comparison between ethyl cellulose and rubber, J Polymer Scl , 27 (1958) 177 W R Vleth, J M Howell and J H Haeh, Dual sorption theory, J Membrane Scl , 1 (1976) 177 T Uragaml, H B Hopfenberg, W J Koros, D K Yang, V T Stannett and R T Chern, Dual-mode analysis of subatmospheric-pressure CO2 sorption and transport m Kapton H polyamide films, J Appl Polym Scl ,24 (1986) 779 R T Chern, W J Koros, E S Sanders and R Yul, Second component effect m sorption and permeation of gases m glassy polymers, J Membrane Scl , 15 (1983) 157
18
19
20
21
22
W J. Koros and E S Sanders, Multicomponent gas sorption m glassy polymers, J Polym Scl ,72 (1985) 141 V Stannett, M Halder, W J Koros and H B Hopfenberg, Sorption and transport of water vapor m glassy poly (acrylomtnle), Polym Eng Scl ,20 (1980) 300 G R Mauze and S A Stern, The solution and transport of water vapor in poly (acrylonitrile) a reexamination, J Membrane Scl ,12 (1982) 51 A R Berens, The solublhty of vinyl chloride m poly (vmyl chlorlde),Angew Makromol Chem ,47 (1975) 97 G R Mauze and S A Stern, The dual-mode solution of vinyl chlonde monomer m poly (vinyl chloride ) , J Membrane Scl ,18 (1984) 99 D K Yang, W J Koros, H B Hopfenberg and V T Stannett, Sorption and transport studies m water m kapton polyamide I, J Appl Polym Scl , 30 (1985) 1035 A Sflrakls and C E Rogers, Effects of sorption modes on transport and physical properties of Nylon 6, Polym Eng Scl ,20(4) (1980) 294 J A Barrer, P S Sag00 and P Johncock, Sorption and diffusion of water m epoxy resins, J Membrane Scl , 18(1984)197 A Aplcella, R Tesslen and C De Cataldls, Sorption modes of water m glassy epoxies, J Membrane Scl , 18 (1984)211 P J Flory, Thermodynamics of high polymer solutions, J Chem Phys , 10 (1942) 51 P J Flory, Prmclples of Polymer Chemistry, Cornell Umverslty Press, Ithaca, NY, 1953 M L Huggms, Thermodynamic properties of solution of long-chain compounds, Ann N Y Acad Scl ,43 (1942) 1 J M Prausmtz, R N Llchtenhaler and E Gomez de Azevedo, Molecular Thermodynamics of Fluid-Phase Equlhbna, 2nd edn , Prentice-Hall, Inc , Englewood Chffs, NJ, 1986 A R Berens, Predlctlon of organic chemical permeation through PVC pipe, J Am Water Works Assoc , Nov (1985) 57 V T Stannett, G R Ranade and W J Koros, Characterization of water vapor transport in glassy polyacrylomtrlle by combined permeation and sorption techniques, J Membrane Scl , 10 (1982) 219 R Puffr and J Sabenda, On the structure and properties of polyamides XXVII The mechanism of water sorption m polyamides, J Polym Scl , Part C, 16 (1967) 79 G Sklrrow and K R Young, Sorption diffusion and conduction m polyamide-penetrant systems I Sorption phenomena, Polymer, 15 (1974) 771
199
SORPTION OF WATER VAPOR BY AMORPHOUS POLYAMIDE
23
24 25
26
27
28
29
M T Aronhlme, X Peng and J K Gllhan, Effect of time-temperature path of cure on the water absorption of high T, epoxy resins, J Appl Polym Scl , 32 (1986) 3589 B H Zlmm and J L Lundberg, Sorption of vapors by high polymers, J Phys Chem ,60 (1965) 425 J L Lundberg, Clustering theory and vapor sorption by high polymers, J Macromol Scl , Phys Ed, 60 (1965) 425 T A Orofmo, H B Hopfenberg and VT Stannett, Characterlzatlon of penetrant clustering m polymers, J Macromol Scl , Phys Ed, 4 (1969) 777 R J Hernandez, L A Baner and J R Glacm, The evaluation of the aroma barner properties of polymer films, J Plast Films & Sheeting, 2 (1986) 187 J V Beck and K J Arnold, Parameter Estimation m Engineering and Science, J Wiley and Sons, New York, NY, 1977 D J Skronovak, S E Howe, P C Pamter and M M Coleman, Hydrogen bonding m polymers infrared temperature studies of an amorphous polyamide, Macromolecules, 18 (9) (1985) 2218
30
31
32
33
34
35
D S Trlfan and J F Terenzl, Extends of hydrogen bonding m polyamides and polyurethanes, J Polym Scl ,28 (1958) 443 Y S Paper, S Kapur and C E Rogers, Effect of onentatlon, amsotropy, and water on the relaxation behavior of Nylon 6, from 4 2 to 300 K, J Polym Scl , A-2,10 (1972) 1305 EL McKague, J D Reynolds and J E Halklas, Swelling and glass transltlon relations for epoxy matrix material m humid environments, J Appl Polym Scl ,22 (1978) 1643 W J Koros and D R Paul, COz sorption m poly (ethylene terephthalate) above and below the glass transition, J Polym Scl , Polym Phys Ed, 16 (1978) 1947 W J Koros and D R Paul, Observations concernmg the temperature dependence of the Langmmr sorption capacity of glassy polymers, J Polym SCI, Polym Phys Ed, 19 (1981) 1655 G B Gedraltlte, A P Ma’rm and Y A Shlyapmkov, The size dlstnbutlon of additive sorption centers m polyethylene, Eur Polym J ,25 (1) (1989) 39
Journal of Membrane Scrence, 65 (1992) 187-199 Elsevrer Scrence Publishers B V , Amsterdam
The sorption of water vapor by an amorphous polyamide Ruben J. Hernandeza, Jack R. Giacin” and Eric A. Grulkeb “School of Packagnzg, MLchzgan State Unrvers@, East Lansmg, MI 48824 (USA) bChemlcal Engzneerzng Department, Mlchtgan State Unwerstty, East Lansmg, MI 48824 (USA) (Received December 23,1988, accepted m revised form July 17,199l)
Abstract A modified dual-mode sorption model represents the sorptron of water vapor by an amorphous polyamide at 23 ‘C The Langmurr equation 1s used to calculate the volume fraction of chemrsorbed solute and the Flory-Huggms equation 1s used instead of Henry’s law to calculate the volume fraction of water which 1s not chemrsorbed This model describes the data over a range of water actrvrtres from zero to one and predicts clusterrng of the sorbant Fourier Transform Infrared (FTIR) spectroscopy data and drelectrrc measurements of the gamma relaxation temperature suggest that the water assocrates with amrde groups at low water activities Keywords barrier membranes, diffusion
dual-sorptron
model, gas and vapor permeatron,
water sorption
and
Introduction
moisture content m Nylon 66 but mcreases with moisture content m the amorphous material
An amorphous polyamide recently has been synthesized and characterized [ 11. This polymer is made from hexamethylenediamme and a mixture of lsophthahc and terephthahc acid. The random placement of these isomers m the polymer chain prevents crystallization of the material. Preliminary studies on the effect of water sorption on the barrier properties of this amorphous polymer show atypical behavior compared to semmrystalhne polyamides. For example, oxygen permeability mcreases as a function of moisture content m Nylon 66 (a semlcrystallme polyamide) but decreases as a function of moisture content in amorphous material The tensile modulus decreases with
[ll.
The mtnguing behavior of the amorphous polyamide system m the presence of water demonstrates the need for equlhbrium models which describe sorption phenomena well over a wide solute activity range Improved models are needed for binary systems and hopefully can be extended to describe ternary systems (gaswater-polymer) m order to interpret and control barrier properties This study models the sorption phenomena of water vapor mto amorphous polyamide as the first step m describing the effects of water on the barrier and mechanical properties of this polymer
Correspondence to Eric A Grulke, Chemical Engmeermg Department, Michigan State Unmemrty, East Lansing, MI 48824 (USA)
Model for gas sorption In glassy polymers Sorption of gases, such as helium, nitrogen,
0376-7388/92/$05
Dual mode sorption models
00 0 1992 Elsevrer Scrence Publishers B V All rights reserved
R.J HERNANDFZ ET AL
188
oxygen, carbon dioxide, and methane, above their critical temperature into glassy polymers has been studied by many researchers. The dual mode sorption model proposed by Barrer et al. [ 21 is the most widely used model for analyzing these data. This model assumes that the solute molecules in the glassy polymer consist of nonspecifically absorbed and specifically adsorbed species which are in dynamic equilibrium in the medium. Equation (1) was proposed to describe Langmuir type curvature observed for gas sorption in glassy polymers at low gas activities. The solubihty of the absorbed species is represented by Henry’s law and the solubility of the adsorbed species are described by a Langmuir type adsorption isotherm. The Langmuir sorption is believed to occur at specific sites, usually considered to be holes in the amorphous polymer structure. Local equilibrium between the absorbed and adsorbed populations is mamtained throughout the polymer matrix [ 31. The total amount of solute sorbed by both mechanisms is:
c=c,+c,=iz,p+
-
CHb
1+bp
where C is the total concentration of sorbed solute in the polymer, Cn and Cn are the solubillties due to absorption (Henry’s law) and adsorption (Langmmr), k,-, is the Henry’s law constant, b is the hole affinity constant, Cn is the hole saturation constant and p is the pressure. The dual model sorption model has been used to describe the solubility of gases in glassy polymers, and in glassy, polar polymers as well. Uragami et al. [4] studied the sorption and equilibrium for CO, m a polyamide film for pressures up to 0.78 atm. Chern et al. [5] applied eqn. (1) to the permeabihty of CO, through Kapton polyamide for pressures up to 16.3 atm Koros and Sanders [6] described a bicomponent sorption process by the dual mode sorption model [CO,/C,H, and CO,/N,O in poly(methy1 methacrylate) 1. In most of the
examples cited above, the gas phase can be considered dilute because the solute activity is much less than one. The solute activity can be approximated by the gas vapor pressure divided by the saturated hquid vapor pressure at the experimental temperature. When the gas acts ideally, simple models such as Henry’s law represent phase equilibria well. Models forpolar systems at hgh solute aciwrtLes Solutes sorbed at temperatures below their boihng point (such as organic solvents and water at room temperature) may not obey simple, ideal phase equilibria relations such as Henry’s law. For these cases, eqn. (1) does not describe the data well over the activity range, 0
(2)
where al is the solute activity, x IS the mteraction parameter, VI is the volume fraction of the solute and V, is the volume fraction of the polymer. Although eqn. (2) does not provide the most accurate description of the thermody-
189
SORPTION OF WATER VAPOR BY AMORPHOUS POLYAMIDE TABLE
1
Models for polar systems at high solute actmltles Solutlon factor
comment
Ref
H,O/Polyacrylomtnle
kd exp (UC)p
Data of [7]
PI
Vmyl chloride/ poly (vmyl chlonde
Flory-Huggms’
Fits for high solute actmltles
PI
k, exp (04 pu=2(l+~)A
Data of [9]
[lOI
)
[Ill
H,O/Kaptons
kip
H,O/Nylon
Flory-Huggms’ ’
Fit by authors
[I21
H,O/EPO~~S
hp
Hysteresis
[I31
H,O/Epoxy’
k,+=exp(l+X)
Two Langmuir factors
[ 141
6s
‘Langmulr sorption observed at low actlvltles but not modeled *RegIstered trademark, E I Du Pont de Nemours and Company, for an aromatic polyether dnmlde 3Trademark of E I De Pont de Nemours, Inc 4Best fit model as determined by authors of this work ‘TGDDM/DDS epoxy resins (MY720, Cuba-Geigy, Ltd ) ‘DBEBA/TETA (Eplkote 828, Shell, Italy) and TBDDM/DDS (Araldlte MY720, Cuba-Geigy, Ltd )
namic performance of polymer solutions, it does contam most of the essential features which distmguish such solutions [ 181 Equation (2) usually fits equihbrla data over a wider range than Henry’s law and reduces to Henry’s law at low gas activities For example, sorption data of toluene m poly (vinyl chloride) is well fit by the Flory-Huggms equation especially for 0.5 < a, < 10 [ 191. Although no model for water sorption mto Nylon 6 was proposed by Sfirakls and Rogers [ 121, this data set fit eqn. (2) with ax value of 188 The use of the Flory-Huggms equation to describe non-specific solute absorption m the place of Henry’s law m eqn. ( 1) is an extension of the model summarized in Table 1 Group contribution or lattice models are alternative choices to eqn. (2 ) , but these are more complex than the Flory-Huggms equation. An issue associated with any model choice is determnnng the value of the model coefficients The con-
stant descrlbmg absorption usually is determined at high solute pressures. The methods for parameter estimation are described in the Results and Discussion section Clusteruzg of solute molecules Self-association of the solute to form clusters has been used to explain a number of transport phenomena occurrmg m polar systems For example, the clustermg of water molecules m polyacrylomtrile coincides with a maximum m the diffusion coefficient [ 201 Clustermg could reduce the effective mobility of water either by mcreasmg the size of the diffusing group or by mcreasmg the tortuosity of diffusion paths due to the presence of clusters A similar effect has been reported for Kapton by Yang et al. [ 111. They calculated the clustermg of water by dlfferentiatlon of a fourth order polynomial fit Puffr and Sabenda [ 211 reported the clustering of water m several polyamides. Sklrrow and Young [22] described the clustering of water, methanol and propanol in Nylon 6 by plotting a,/ V, versus a, In Nylon 6, clustermg has been associated with a change m the diffusion coefficient [ 121 Aronhime et al. [23] reported clustering of water in epoxy resins If clustering is important to solute diffusion, then the equilibrium model should be able to predict its onset The clustermg function developed by Zimm and Lundberg [24,25] has been used with blnary systems m eqmhbrmm to give a measure of the tendency of like molecules to cluster Their clustering function, G,,/V,, is:
Gl VI
-=-
(3(dV1)
v
2
f3a1
_1
(3)
When G,,/ VI is greater than - 1, the solute is expected to cluster Equation (1) is not written m a convenient form for the analytical determmation of the derivative m eqn. (3 ) Common methods for evaluating the clustermg function are either fitting activity versus volume fraction data with pol-
190
R J HERNANDEZ ET AL
ynomials and taking their derivatives or doing a numerical differential of the data. Equation (1) can be rearranged to express the concentration, C, as volume fraction and to express the Henry’s law and Langmuir contributions m terms of activity: v,=V~+V+K,a,+& 1
where Vy and Vk are the Henry’s law and Langmuir volume fraction contributions, KD = kDpS, K= CH f b pS, B = b pS, pS 1s the saturation vapor pressure of the solute and f 1s a conversion factor. Applying eqn. (3) to eqn (4) gives
Gl -V* - -V,
VI - [K,, a, +KaJ(l+B
al)“]
V? (5)
Since the term in brackets is always less than VI, G,,/V, 1s always less than - 1 for the range O
G -= VI
2x 1-2x
v,
(6)
for a constant value of x Equation (6) always predicts clustermg m the range, 0 < V, < l/2 x. The equation is not defined at the upper limit of the range ( VI = 1/2x), which is the spinodal point mdicatmg phase separation Neither the conventional dual mode sorption model or the Flory-Huggms model with a constant value for
x correctly describe the apparent the systems mentioned above.
clustermg
in
Modtfted dual mode sorptron model In this study, eqn. (1) has been modified by using the Flory-Huggms equation to describe non-specific solution rather than Henry’s law. This modification should allow the model to fit over the activity range 0
(7)
where VT” refers to the Flory-Huggms contnbution to the solute volume fraction and Vk refers to the Langmuir contribution. Smce eqn (2) is nonlinear, it is convenient to determme the value for Vy” by numerical methods, such as the Newton-Raphson technique and eqn (7)
(8) The clustering Cl, -=V, +
function
of eqn. (8) 1s
VZ Ko, V: (l+Bc# 1-v:n
VFH (1:2x
1 v;“)--v,
(9)
Equation (9) predicts clustering for some values of the constants but is undefined when* 1-v:n
(1+2x
VCH)=O
(10)
This occurs when V, 1s 1.0 (pure solvent) or l/2 x (phase separation). The replacement of the Henry’s law model by the Flory-Huggms
191
SORPTION OF WATER VAPOR BY AMORPHOUS POLYAMIDE
equation for non-specific absorption gives a model which pre&cts clustering m the activity range O
(ZnS) crystal from a 2% solution with 1,1,1,3,3,3:hexafluoro-2-propanol (Kerr Company, NOW, MI). After evaporation of the solvent at room temperature, the sample was dried m a vacuum oven at 100°C to remove residual solvent and water. The sample was then equlhbrated with water vapor at selected water actlvlty values After attaining equlhbrium, the sample was lmmedlately covered with another crystal of ZnS and transferred to the instrument Dlfferentral scanntng calortmetry and dtelectrtc experrments Film samples were prepared for these analyses by vacuum drying at 100” C and eqmhbrating with the vapor over salt solutions to sve selected water actlvitles. Results and discussion
Dens@ expertments Film densltles were determined m a density gradient column with the gradient made of toluene and carbon tetrachlorlde
Eqwltbrtum sorptton tsotherm Sorption equlhbrmm values of water weight fraction and sample density of the amorphous polyamide at 23°C are presented m Table 2 Isotherm data were obtamed over a wide actlvlty range (0.046
Fourier transform mfrared spectroscopy The sample for infrared analysis was prepared by casting a thin film onto a zmc sulfide
Parameter estlmatlon for equatron (8) Parameter values estimated for eqn (8) were sensitive to the method used to determine them
R J HERNANDEZ ET AL
192
TABLE 2 Water uptake (23°C) Water actlvlty
0 00 0 046 0 056 0 012 0 080 0 090 0 110 0 155 0 189 0 252 0 269 0 308 0 410 0 440 0 565 0 580 0 585 0 635 0 735 0 790 0 860 0 880 0 963
vs water
Water weight fraction 0 00
activity
for amorphous
Density (g/cc) Data
Model
1194
1194 1194 1 194 1 194 1 194 1 194 1 194 1 195 1 196 1 196 1917 1 197 1 199 1 199 1201 1201 1201 1201 1202 1202 1202 1202 1202
0 00468 0 00653 0 00746
-
0 00781 0 00829 0 00929 0 0120 0 0127
1 194 -
0 0164 0 0157
-
0 0215 0 0295 0 0344 0 0401 0 0438 0 0416
1 198 1 199 -
0 0481 0 0583 0 0664 0 0730 0 0758 0 0826
1201 1202
-
Nylon
Volume fraction
000 0 0056 0 0078 0 0089 0 0093 0 0099 0 0111 0 0143 0 0153 0 0197 0 0188 0 0258 0 0354 0 0414 0 0483 0 0527 0 0501 0 0579 0 0702 0 0800 0 0879 0 0913 0 0995
Fig 1 Water volume fraction m amorphous nylon as a function of water actwlty Flory-Huggms model with x=1 63
A nonlinear regressron method was used to determine the values of the three parameters simultaneously. This resulted in Langmuir coefficients with a hrgh amount of error and a fit for eqn. (8) which was not better than the fit for eqn. (2). The least squares technique was not sensltrveto deviations between the data and the model at low activities. However, the devlatlon of eqn. (2) from the data 1s systematic as shown in Fig 1 An alternative parameter estimation method gave coefficients with lower error. The data for a, > 0.4 were used to determine a range of x values which could represent this portion of the curve. x should be between 1.6 and 185 to mmlmize the least square errors for this data set The data for a, < 0 4 were used to determine the Langmuir coefficients. The nonlinear regression method gave the coefficients and the R2 parameter for a set of x values in this range. The R2 value was plotted as a function of x to find the best estimates of the parameters. Figure 2 shows eqn (8) as the solid curve (x= 1.7, K= 0.395 and B = 95.2 ) and the complete data set. At low activities, the Langmuu contrlbutlon 1s high compared to the Flory-Huggins contrlbutlon and at a, = 0 4, rt ISJust over 10% of the total volume fraction of solute. The Langmuu contrlbutlon 1snearly constant (90% of its maximum value) at actlvlties greater than 0.11 and the curve at high water activities should be insensltlve to the values of the Langmmr coefficients. Therefore, the estimation method is consistent with the parameter values. The dashed curve on Fig. 2 1sthe dual-mode sorption model (eqn 4) with a set of coefficients with a low least squares error. The dualmode sorption model fits data of low water actlvltles but shows systematic deviations at high water activities. Sensltmty coeffwents Sensltivlty coefficients mdmate the magmtude of the change of a function due to pertur-
193
SORPTION OF WATER VAPOR BY AMORPHOUS POLYAMIDE
01
-
009
.-
006
--
007 -009 -Vl
005 -001 .-
003 -002 --
I 0
01
02
03
04
06
05 81
07
09
1
09
Fig 2 Comparison of the dual-mode and modified dual-mode sorption models Equation (8) (x= 17, K=O 395, B=95 2), equation (1) (x= 163)
bation in the values of its parameters [%I. Sensitivity coefficients for eqn. (8) are given by the first derivative of the volume fractron with respect to x, I3 and K
(12) (13) where X,, Xn and Xx are the sensltlvlty coefficients of K, B and x. These coefficients were evaluated as a function of activity, a,, at the optimum values of K, B and x and the results are plotted m Fig. 3. For a, > 0.1, X, changes with actlvrty whrle X, and Xn are essentially constant. For high actlvlty values, rt will be difficult to distinguish between K and B parameters At low actlvlty values, they are not linearly dependent and better estimates for them can be obtained The
-151 0
’
’ 2
’
’ 1 4 9 WaterActivity
’
’ 9
1
Fig 3 Sensltlvlty coefficients as a function of water ackvlty
results shown in Fig. 3 suggest that a slmultaneous search for the three parameters will be &fficult over the entire activity range, that high activity data will be msensltive to K and B, and that the best estrmates for K and B will be obtamed at low actrvlty values Equations ( ll)(13 ) give a protocol for evaluating the method of determining the constants and are an improvement over the suggestion that the con-
R J HERNANDEZ ET AL
194
stant (s) for the solution phenomenon should be determined when the solute partial pressure is much larger than the Langmuir affinity constant, b [lo]. Clusterrng analym Equation (9) with the coefficients determined above the empirical polynomial fits of the activity-volume fraction data were used to compute the clustering function for the water sorption data. Thud and fourth order polynomial expressions tit the data with similar least square errors Both expressions approached but did not pass through the zero activity-zero volume fraction point, and neither polynomial replicated the data as well as eqn. (8) Figure 4 shows the clustering function estimated from two polynomial fits (thud and fourth order) and the modified dual-mode sorption model All three equations predict clustering (G,,/ V, > - 1) but &ffer in the solute activity at which this should occur. The fourth order polynomial predicts clustering m the activity range, 0.34~ a, < 0 96 The thud order polynomial predicts clustering for a, > 0 27. Both polynomial fits give clustering functions which go through maxima. There is
001
-31
1
1 -21
’
’ -11
Cluster
’
’
-1
4
B
Function
Fig 4 Clustering function vs water activity (-) eqn (8)) (-------) thud order polynomial, (- - -) fourth order polynomial
no obvious physical explanation for why clustering should decrease, or disappear, at high water activities. Equation (9) predicts clustermg at a, = 0.38 and the function is increasing monotomcally over the whole activity range. Smce the coefficients in eqn. (8) affect the clustering prediction, coefficient accuracy is important to the results Phymal evgdence for solute brndtng The Langmmr isotherm was derived based on chemisorption of a chemical species to a specific bmdmg site characterized by a single energy of sorption Solute sorption described by this isotherm for the water-polyamide solution might result m observable changes m the polymer, particularly d the water associated with a specific group on the polymer backbone One obvious possibihty is water mteraction with the amide group FTIR spectroscopy studies were done to detect possible changes in hydrogen bondmg between N-H and carbonyl groups by observing the vibrational modes of the amide group (Amide I and Amide II bands). The Amide I mode includes contributions from C=O stretching, C-N stretching and CC-N deformation vibrations. The Amide II band includes N-H plane bendmg, C-N stretching and C-C stretching vibrations Figure 5 shows a typical FTIR spectrum taken at room temperature over the range, 800-4000 cm-l for a cast polyamide sample The Amide I and Amide II modes occur at 1640 and 1541 cm-‘, respectively, for the dry sample and are the most intense bands. The frequency scale and the relative mtensity of the Amide I and Amide II modes were internally calibrated with reference to the CH, stretching band at 2858 cm-’ Table 3 lists the absorption frequencies and Table 4 lists the mtensity ratios for the Amide I and Amide II bands of the sample recorded as a function of water volume fraction and activity The frequency of the CH2 stretchmg band, which served as the internal calibra-
195
SORPTION OF WATER VAPOR BY AMORPHOUS POLYAMIDE
40000 L 3000
2000 I
I
1BOO 1
1200 I
1
cm-’
Fig 5 FTIR spectrum of amorphous polyamide at room temperature TABLE 3 Amide I, Amide II and CH absorption frequencies Actlvlty
CH peak
Amide I
Amide II
0 0 08 0 308 0 56 0 88
2858 2858 2858 2858 2858
1640 1640 1641 1641 1641
1541 1545 1545 1545 1546
Units cm-’ TABLE 4 Intensity ratios of Amide I and Amide II with respect to CH bands Actlvlty
Amide I/C-H
Amide II/C-H
0
171 178 187 180 184
5 45 5 44 5 11 5 43 5 14
0 08 0 308 0 56 0 88
tlon band, is presented for reference. Figure 6 shows the absorption region from 1500 to 1600 cm-’ (Amide II mode) as a function of a,. The spectra are displayed on an absolute absorbance scale. The Amide I band remained essentially con-
oo1600
1560
1520
cm-’ Fig 6 Amide II spectrum as a function of water actlvlty
stant at 1640 cm-’ as did its intensity. The peak maxima of the Amide II band shifted to higher frequency, 1541 to 1546 cm-‘, over the range 0%-9.1%, water volume fraction. The intensity ratio of the Amide II band remained essentially constant The shift of the Amide II band to higher frequency with increasing water content is consistent with an increase m the average hydrogen bond strength [ 291 These data suggest that the water does not change m hydrogen-bonding strength of the carbonyl group but does change the hydrogen bonding strength of the N-H group. The shift in the Amide II band is completed at the water activity at which 90% of the Langmuir sites would be filled (Table 3 ) In linear, ahphatlc homopolyamldes there is essentially 100% hydrogen bonding, as evldented by the absence of bands in the infrared spectra above 3300 cm-’ [ 301. In structurally irregular copolymers, such as amorphous polyamide, both the N-H stretchmg region and the Amide I region can be resolved mto “free”
196
and hydrogen-bonded stretching mode. Skranovek et al [29] resolved the N-H stretching region into three components. a “free” N-H mode, a hydrogen bonded N-H stretchmg mode and an Amide B mode The Amide I band was resolved mto a “free” and hydrogen-bonded carbonyl mode Most of the first water molecules sorbed into the polymer associate with the “free” N-H groups of the amorphous polyamide This would be an exothermic process and should have a single activation energy associated with it. The formation of these new hydrogen bonds should have a neghgible enthalpic effect [ 21,311. The extent of disruption of the self-association of polyamide neighbormg groups by water molecules is not know Both the model and the FTIR data suggest that chemisorption has been completed at activities well below those at which clustering occur At a water activity of 0 08 (where the Langmuir sites are 89% saturated), the total amount of chemrsorbed water is only 4% of the total water volume fraction This amount corresponds to 3 34x 10m3 g of water per gram of polymer, or one molecule of water per 22 repeating umts of the polymer Each repeat unit contains 2 amide groups so only 2.3% of the amide groups available form hydrogen bonds with water Either hole filhng or chemisorption might be occurring in the polyamide for the Langmuir contribution as Kapton is known to hydrogen bond with water Water IS known to hydrogen bond with a variety of epoxies and the dualmode sorption model describes some data well [ 131 Apicella et al. [ 141 tested the hole-fillmg hypothesis with an epoxy system by sorbmg water between 0 and 60 atm It was claimed that Langmun factors were needed to model these data one for chemisorption (probably to hydroxyl or secondary amme groups) and one for sorption into holes One conclusion of previous workers [ 33,341
R J HERNANDEZ ET AL
is that the Langmuir component of sorption disappears at the Tg of the polymer. This suggests that sorption is not related to chemisorption with a specific chemical group, since chemisorptlon can occur above and below the system T, The observed association between water and the amide group could be interpreted as showmg that holes preferentially occur near this chemical structure. The physical basis for site-specific holes is not clear. However, data for the sorption of low molecular weight solutes m polyethylene above Tg have been fitted with the Langmuir mode [35]. Whether such sites are the same as Langmuu sites below Tg 1s an open question Relaxatton temperatures Several relaxation temperatures have been described for semicrystalline polyamides Papir et al [31] related the gamma relaxation peak (140K) of Nylon 6 to the movement of methylene and polar groups. Sfirakis and Rogers [ 121 reported that the gamma peak for the Nylon B-water system was affected by water concentration. The change in mtensity of the gamma peak with changmg water concentration leveled off at a concentration correspondmg to the apparent onset of water clustering The gamma relaxation temperature was measured via the dielectric constant at 1 kH. The gamma peak of the amorphous polyamide was 173°K m the dry state. The peak was not observed for samples equilibrated at a, > 0.08 The difference could be due to the mteraction of water with the amide group decreasing the umts available to participate in the relaxation process. Data for the alpha relaxation temperature (glass transition temperature ) were obtained by DSC measurements The effect of sorbed water on Tg 1s shown in Fig 7, where Tg values are plotted with water volume fraction. There is an apparent change m slope of T, with activity near a, = 0.40. A similar trend was observed
197
SORPTION OF WATER VAPOR BY AMORPHOUS POLYAMIDE
r
1:21
.__
1
D
llll-
0 go-
H
D
u
E 1 20 ::
f 79c d
50
1 L
3oo
I
1
0
I
I
3
I
t
5
I
I
7
I
L
I
9
0
Uolun~riraction[x101
11l9k
1 1
I
I
3
I
I
5
I
I
7
I
I
9
10
Water Activity
Fig 7 Glass transhon temperature vs water actlvlty
Fig 8 Density as a function of water actlvlty
on Tg versus water volume fraction obtained from the dielectric experiments. The slope change occurs near the water activity at which clustermg is predicted for the polymer system at 23°C Water solubihty data were not taken at other temperatures so that the effect of temperature on clustering is not known. The data suggest a relationship between water clustering and a change m the slope of Z’* with water activity However, this interpretation depends on temperature having a negligible effect on water clustering in this temperature range
flection point occurs m the density curve near the water activity range (al=0.27 at the maximum of the derivative of density with respect to activity, for these data) over which clustering is predicted to begin The water activity at this inflection point is similar to that near the apparent change m Tg vs. a, curve (Fig. 7). The increase in sample density with mcreasmg water volume fraction occurs as the Tg of the system is decreasing. There are few comparisons of density and Tg data for partially soluble solutes, so it is not clear whether these results are specific to water/polyamides or occur m other systems as well. In general, small solutes plasticize polymers. Water appears to be plasticizing the amorphous polyamide at high temperatures even though, at room temperatureo, adding water mcreases the sample density and the solute could be clustermg
Dens&y data Figure 8 shows the change in total density with water activity. The amorphous nylon solution becomes more dense as water is added This is the opposite effect expected from a mixing rule depending on additive molar volume fractions. A similar result has been reported for an epoxy system [ 321. Notice that there is httle change m density below water activities of 0.1, where the Langmuir coefficients suggest that chemisorption (or “hole filhng”) is essentially completed It is interesting to point out that the permeabihty of this amorphous nylon to oxygen decreases by a factor of 2 for films at water activities greater than 0 10 compared to a sample from which water has been removed An m-
Interpretatwn of the effects of sorpfon on transport and physical properttes of amorphous polyamzde At water activities less than 0.10, water preferentially associates with amide sites, although a low mole fraction of the total amide sites on the polymer is occupied. The association can be detected by observing the wave shift in the amide group absorbance. Langmmr sorption of
R J HERNANDEZ ET AL
198
water is essentially completed at al = 0 1 (about 1 vol.% water). There is no measurable change in system density at this water activity. There is a reduction of the polymer Tgat a water activity of 0.1, probably due to the plastlcizmg effect of water As with other dual-mode sorption system, polymer history and relaxations probably affect the Langmuir sorption capacity, C;, , although such data have not been taken m this study At higher water activities, water solubility is controlled by the Flory-Huggms component and water clustering is predicted by Eqn. (9). System density increases nonlinearly as water activity increases. A change in the Tgof the system with respect to water activity coincides with the clustering prediction as does the inflection point in the density-activity curve Clustering may be associated with the increase in density and the decrease in system specific volume. Acknowledgments The authors acknowledge the support of this work by the grant received from E.1 Du Pont de Nemours and Company, Inc.
6
7
8
9
10
11
12
13
14
15 16 17
References P Blatz, PropertIes of films from blends of amorphous and crystalline nylons, presented at the AIChE Nat Conf , Apnl2-6,1989 R M Barrer, J A Barrier and J Slater, Sorption and dlffuslon m ethyl cellulose Part III Comparison between ethyl cellulose and rubber, J Polymer Scl , 27 (1958) 177 W R Vleth, J M Howell and J H Haeh, Dual sorption theory, J Membrane Scl , 1 (1976) 177 T Uragaml, H B Hopfenberg, W J Koros, D K Yang, V T Stannett and R T Chern, Dual-mode analysis of subatmospheric-pressure CO2 sorption and transport m Kapton H polyamide films, J Appl Polym Scl ,24 (1986) 779 R T Chern, W J Koros, E S Sanders and R Yul, Second component effect m sorption and permeation of gases m glassy polymers, J Membrane Scl , 15 (1983) 157
18
19
20
21
22
W J. Koros and E S Sanders, Multicomponent gas sorption m glassy polymers, J Polym Scl ,72 (1985) 141 V Stannett, M Halder, W J Koros and H B Hopfenberg, Sorption and transport of water vapor m glassy poly (acrylomtnle), Polym Eng Scl ,20 (1980) 300 G R Mauze and S A Stern, The solution and transport of water vapor in poly (acrylonitrile) a reexamination, J Membrane Scl ,12 (1982) 51 A R Berens, The solublhty of vinyl chloride m poly (vmyl chlorlde),Angew Makromol Chem ,47 (1975) 97 G R Mauze and S A Stern, The dual-mode solution of vinyl chlonde monomer m poly (vinyl chloride ) , J Membrane Scl ,18 (1984) 99 D K Yang, W J Koros, H B Hopfenberg and V T Stannett, Sorption and transport studies m water m kapton polyamide I, J Appl Polym Scl , 30 (1985) 1035 A Sflrakls and C E Rogers, Effects of sorption modes on transport and physical properties of Nylon 6, Polym Eng Scl ,20(4) (1980) 294 J A Barrer, P S Sag00 and P Johncock, Sorption and diffusion of water m epoxy resins, J Membrane Scl , 18(1984)197 A Aplcella, R Tesslen and C De Cataldls, Sorption modes of water m glassy epoxies, J Membrane Scl , 18 (1984)211 P J Flory, Thermodynamics of high polymer solutions, J Chem Phys , 10 (1942) 51 P J Flory, Prmclples of Polymer Chemistry, Cornell Umverslty Press, Ithaca, NY, 1953 M L Huggms, Thermodynamic properties of solution of long-chain compounds, Ann N Y Acad Scl ,43 (1942) 1 J M Prausmtz, R N Llchtenhaler and E Gomez de Azevedo, Molecular Thermodynamics of Fluid-Phase Equlhbna, 2nd edn , Prentice-Hall, Inc , Englewood Chffs, NJ, 1986 A R Berens, Predlctlon of organic chemical permeation through PVC pipe, J Am Water Works Assoc , Nov (1985) 57 V T Stannett, G R Ranade and W J Koros, Characterization of water vapor transport in glassy polyacrylomtrlle by combined permeation and sorption techniques, J Membrane Scl , 10 (1982) 219 R Puffr and J Sabenda, On the structure and properties of polyamides XXVII The mechanism of water sorption m polyamides, J Polym Scl , Part C, 16 (1967) 79 G Sklrrow and K R Young, Sorption diffusion and conduction m polyamide-penetrant systems I Sorption phenomena, Polymer, 15 (1974) 771
199
SORPTION OF WATER VAPOR BY AMORPHOUS POLYAMIDE
23
24 25
26
27
28
29
M T Aronhlme, X Peng and J K Gllhan, Effect of time-temperature path of cure on the water absorption of high T, epoxy resins, J Appl Polym Scl , 32 (1986) 3589 B H Zlmm and J L Lundberg, Sorption of vapors by high polymers, J Phys Chem ,60 (1965) 425 J L Lundberg, Clustering theory and vapor sorption by high polymers, J Macromol Scl , Phys Ed, 60 (1965) 425 T A Orofmo, H B Hopfenberg and VT Stannett, Characterlzatlon of penetrant clustering m polymers, J Macromol Scl , Phys Ed, 4 (1969) 777 R J Hernandez, L A Baner and J R Glacm, The evaluation of the aroma barner properties of polymer films, J Plast Films & Sheeting, 2 (1986) 187 J V Beck and K J Arnold, Parameter Estimation m Engineering and Science, J Wiley and Sons, New York, NY, 1977 D J Skronovak, S E Howe, P C Pamter and M M Coleman, Hydrogen bonding m polymers infrared temperature studies of an amorphous polyamide, Macromolecules, 18 (9) (1985) 2218
30
31
32
33
34
35
D S Trlfan and J F Terenzl, Extends of hydrogen bonding m polyamides and polyurethanes, J Polym Scl ,28 (1958) 443 Y S Paper, S Kapur and C E Rogers, Effect of onentatlon, amsotropy, and water on the relaxation behavior of Nylon 6, from 4 2 to 300 K, J Polym Scl , A-2,10 (1972) 1305 EL McKague, J D Reynolds and J E Halklas, Swelling and glass transltlon relations for epoxy matrix material m humid environments, J Appl Polym Scl ,22 (1978) 1643 W J Koros and D R Paul, COz sorption m poly (ethylene terephthalate) above and below the glass transition, J Polym Scl , Polym Phys Ed, 16 (1978) 1947 W J Koros and D R Paul, Observations concernmg the temperature dependence of the Langmmr sorption capacity of glassy polymers, J Polym SCI, Polym Phys Ed, 19 (1981) 1655 G B Gedraltlte, A P Ma’rm and Y A Shlyapmkov, The size dlstnbutlon of additive sorption centers m polyethylene, Eur Polym J ,25 (1) (1989) 39