2628
Ind. Eng. Chem. Res. 1992,31, 2628-2631
variable is also much less. Now that the f d steady-state value of the manipulative variable is zero, even with a unit step change in load. This is because the process contains an integrator. Note that if' lese conbewative settings are desired, the specified peak in the phase angle curve can be lowered in eq 10. The result will be a more underdamped closed loop system. Conclullion A design procedure is proposed that permits the cdculation of the tightest controller settings when a PI controller is applied to an integrator/dead time process. The basic criterion is a maximum closed loop log modulus of +2 a. This is mntially equivalent to specifying a closed loop damping coefficient of about 0.4 and then finding the best settings that will minimize the closed loop time constant (maximize the closed loop resonant frequency). Unlike the IMC approach in which a closed loop time constant must be assumed and then the results tested, the proposed procedure involves no trial and error. These settings have been tested on a wide variety of processes and have worked much better than the classical ZieglerNichols settings. Nomenclature B = feedback controller transfer function D = dead time (min) G M = process open loop transfer function K, = controller gain ( % / % ) K, = process gain ( % / m i d
LcL = closed loop log modulus (dB) L c =~maximum closed loop log modulus
(a)
Greek Letters 7cL = closed loop time constant (min)
= reset time (min) = frequency (radfmin) up = frequency at which phase angle peak occurs (rad/min) 71
w
Literature Cited Chien, I. L.; Fruehauf, P. S. Consider IMC Tuning to Improve Performance. Chem. Eng. Prog. 1990 (Oct.), 33-41. Fuentes, C.; Luyben, W. L. Control of High Purity Distillation Columns. Ind.Eng. Chem. Process Des. Dev. 1983,22,361-366. Luyben, W . L. Process Modeling, Simulation and Control for Chemical Engineers, 2nd ed.; McGraw-Hill: New York, 1990. Rivera, D. E.;Morari, M.; Skogestad, S. Intarnal Model Control. 4. PID Controller Design. Znd. Eng. Chem. Process Des. Dev. 1986, 25, 252-265. *Author to whom all correspondence should be addressed.
Bjorn D.Tyreus E . I. d u Pont de Nemours & Company P.O. Box 6090, Newark, Delaware 19714-6090
William L. Luybem* Department of Chemical Engineering Zacocca Hall, Lehigh University 111 Bethlehem, Pennsylvania 18015 Received for review April 20, 1992 Revised manuscript received August 17, 1992 Accepted September 9,1992
Solubility Characteristics of Tetrabromobisphenol-A Polycarbonate in Various Liquids As a guide for preparing permselective membranes useful for gas separations, the relative solubility of the polymer tetrabromobisphenol-A polycarbonate (TBBA-PC) was observed a t temperatures up to 100 "C in numerous organic liquids. Forty-five compounds were identified that dissolved a t least 50 w t % polymer. Some cases of solution instability and limited solubility were observed. The reason for such behaviors is not known,but it is tentatively attributed to the formation of a weak "structure" through polymer-liquid interaction such as solvate formation and/or solvent-induced crystallization. The correlation of solvent ability with Hildebrand total solubility parameters was poor; the correlation with three-component Hansen parameters was much better. On the baais of observations with 33 solvents. the solubility parameter of TBBA-PC was estimated to be 20.8 (10.2 ( ~ a l / c m ~ ) ' / ~ ) . Introduction The polymer poly[oxycarbony~oxy(2,6-dibromo-1,4phenylene) (1-methylethylidene)(3,5-dibromo-1,4phenylene)], commonly known as tetrabromobisphenol-A polycarbonate or TBBA-PC (Chemical Abstrads Service Registry No. 28774-934, is a promising candidate for permselective membrane compositions especially for the separation of gas mixtures, e.g., oxygen and nitrogen (Sanders et al., l988,1990a,b; Beck et al., 1990; Muruganandam et al., 1987). The polymer has a glass temperature in excess of about 260 "C (Muruganandam et al., 1987), and at the temperatures necessary for melt spinning of homogeneous membranes, significant thermal-induced degradation may result. However it is known that both homogeneous and asymmetric membranes can be readily prepared from many polymers, such as cellulose triacetate, polysulfone, and polyether sulfone, at significantly lower temperatures from suitable solutions in solvents or, optionally, in solvent-nonsolvent mixtures (Kesting, 1985).
This has been found to be true also for TBBA-PC (Sanders et al., 1988,19904b; Beck et al., 1990). Proper membrane formation from polymer solutions requires knowledge of the relative solubility of the particular polymer in various liquids. Until recently such information has been lacking for TBBA-PC. This paper will describe the solubility characteristics of TBBA-PC in over 100 organic liquids. Experimental Section The polymer used was prepared from tetrabromobisphenol-A (Chemical Abstracta Service Registry No. 7994-7) and phosgene in methylene chloride in the presence of pyridine. The polymer was end-blocked with p-tertbutylphenol and was precipitated in excess heptane. The resulting dried polymer was a lowdensity-high-bulk "fluff" with a fibrous consistency and a relatively high surface area. The inherent viscosity in methylene chloride at 25 "C was 0.412 dL/g. The liquids examined as potential solvents were from several sources, mainly from Aldrich
0888-5885/92/2631-2628$03.W/0 0 1992 American Chemical Society
Ind. Eng. Chem. Res., Vol. 31, No. 11, 1992 2629 Table I. ref no. 31 70 99 92 50
101 76 I9 42 21 31 30 44 39 86 100 24 20 22 51 91 I1 48 49 96 21 59 45 98 58 23 1 19 25 94 65 67 69 95 80 4 60 85 43 81 5
I 13 103 46 15 91
Relative Solubility 02 TBBA-PC in Variour Liquids compound re1 solubility," 96 acetophenone s > 50.1b 1-acetvbiwridine S > 50.1b bend-dehyde S > 50.1bJ 1-benzylpyrrolidinone-2 25.9 < S < 50.1;b>50.1f bis(2-methoxyethyl) ether S > 50.3b 25.3 < S < 50.0b*'?J 1-bromonaphthalene S > 50.2b y -butyrolactone 25.3 < S < 50.1;b > 50.lf e-caprolactone S > 50.4bqcJJ chlorobenzene S > 50.Bbvc chloroform s > 50.3b cyclohexanone s > 50.5b cyclohexyl acetate cyclopentanone S > 50.3b S > 50.1bscJJ o-dichlorobenzene diethyl ketone S > 50.2b diiodomethnne S > 25.2bJ Nfl-dimethylacetamide S > 50.2b NJV-dimethylformamide S > 55.0b s > 50.lb p-dioxane 25.2 < S < SO.lb*cJJ ethyl benzoate 1-ethylpyrrolidinone-2 S > 50.1b s >50.lb N-ethylmorpholine 1-formylpiperidine S > 50.1b 25.6 < S < 50.2;b >50.2f N-formylmorpholine furaldehyde-2 s > 60.lb 25.3 < S < 50.1;b>50.lf isophorone s > 50.5b'C?J methyl benzoate S > 50.lbnC? methyl ethyl ketone methyl iodide s > 50.26 25.6 < S < 50.1;b >50.lf methyl salicylate s > 50.2b 1-methylpyrrolidinone-2 methylene chloride S > 5l.lbsC S > &).3b.c?J~ nitrobenzene s > 50.1b n-octyl acetate S > 50,lbd.eJ phenyl ether S > 26.3b3df piperidine pyridine s > 50.lb 25.8 < S < 50.1;b >50.lf pyrrolidinone-2 s > 50.0b'C?J styrene oxide tetraethylene glycol dimethyl ether s > 50.36 S > 51.4b tetrahydrofuran 1,1,3,3-tetramethylurea S > 50.3b*cJJ 25.4 < S < 50.1;b*c? >50.lf 1,2,4-trichlorobenzene trichloroethylene s > 50.2bfC triethylene glycol dimethyl ether s > 50.4b dimethyl sulfoxide M Z 14.lbvef I < 4.6b*c acetone I < 4.9b acetonitrile I < 4.9b 4-acetylmorpholine butyl stearate I < 4.9bJ tert-butylcyclohexane I < 4.9bJ I < 4.Bb butyraldehyde
ref no. 40 26 93 33 14 68 29 35 64 72 32 28 38 10 17 53 83 90 34 54 9 12 14 102 2 62 55 18 I1 18 88 81
61 3 16 15 41 63 51 11
66 84
73 89 41 6 56 52 36 82 8
compound carbon tetrachloride cyclohexane 1-cyclohexylpyrrolidinone-2 cyclohexylbenzene decahydronaphthalene (cis and trans) diacetone alcohol dicyclohexyl diethylene glycol dibutyl ether diisopropyl ketone dimethyl malonate dimethyl phthalate dioctyl phthalate dodecane ethanol ethyl acetate ethyl formate ethylbenzene ethylene carbonate hexadecane hexamethylphosphoramide hexane hexanol-1 isopropylcyclohexane malononitrile methanol methyl acetate methyl formate methyl isobutyl ketone methyl laurate methyl myristate methvl stearate N-mlthylacetamide N-methylformamide methylcyclohexane nitromethane perchloroethylene perfluoro(methylcyc1ohexane) poly(dimethylsiloxane), 50 cs polyethylene glycol E400 propanol-2 propionitrile n-propylbenzene styrene sulfolane tetrahydronaphthalene toluene tricresyl phosphate triethyl orthoformate triethyl phosphate triethylamine xylene
re1 solubility," W I < 4.Ib I < 4.Bb I < 4.1;b >4.1f I < 4.BbJ I < 4.4bJ I < 4.9bJ I < 4Abf I < 4.9bJ I < 4.9bJ I < 4BbJ I < 4.9;b >4.4f I < 4.Ibf I < 4.IbJ I < 3.9b I < 4.Ibsc I < 5.0b*' I < 4.9b.CJ I < 5.PJ I < 4BbJ I < 4.9;b >4.4' I < 1.6b I < 4.IbJ I < 4.95bJ I < 4.PJ I < 1.5b I < 4.BbvC I < 5.0b I < 4.lbycJ I < 4.Ibf I < 4.9bJ I < 4.7eJ I < 4.6eJ I < 5.0bJ I < 4.6bf I < 5.0bJ I < 4.9bJ I < 1.4b I < O.BbJ I < 2.2bJ I < 2.gb I < 4.9b3' I < 4.9bICJ I < 4.7b.c I < 4.6;O >4.6' I < 4Abf I < 4.5bpCf I < 5.0;b >5.0' I < 4.5bJ I < 4.8;b*"@ >4.6 I < 4.Ib I < 5.5;b.c >5.5'
"I = insoluble: S5W. M = moderately soluble: 525%. S = soluble: >25%. b A t room temperature. 'Insoluble fraction and/or 'solvated polymer" and/or dvent-induced "order"? dReacta with polymer. eAt 50 "C. 'At 100 "C. #Clear.
Chemical Company, and were used as received without further purification. Weighed amounta of polymer and liquid were placed in 4-dram-capacity glass vials with polyethylene-lined caps. About 2.5 g of liquid was usually used. Initial polymer concentration was about 5 wt 9%. The vials were placed on mechanical rollers for at least 24 h or until complete dissolution was obse~ed.Additional polymer, if indicated, was then added to prepare concentrations of about 10,25, and 50 wt ?%. Insoluble misturea with liquid boiling pointa in excess of about 100 OC were placed in a 100 OC forced-air oven for at least 24 h of observation or until dissolution was completed. The polymer was arbitrarily designated as being "insoluble" in the liquid if about 5 wt 90or less dissolved; "moderately" soluble if 625% dissolved; and "soluble" if more than 25% dissolved. Results and Discussion One hundred three commercially-available compounds were examined for their solvent ability for the polymer (Table I). They are listed alphabetically into essentially
two groups: those designated as solvents and those as nonsolventa. (The "reference numbers" are arbitrary and for ease of reference only.) Twelve classes of compounds are represented: halocarbons (11 compounds), esters (21), heterocyclics (8),ketones (9),aldehydes (5), hydrocarbons (16),ethers (8),alcohols (5), nitriles (3), amines (3), amides (6), and eight miscellaneous compounds. Their boiling points vary from 34 O C to in excess of about 420 "C at 1 atm pressure. Their approximate Hildebrand-Scott solubility parameters (Hildebrand and Scott, 1950)vary from about 11.0to about 32.9 (Barton,1983;Brandrup and Immergut, 1975). Except for a few common lowboiling alcohols, in general, no alcohols, polyols, phenols, or other potentially obviously reactive compounds were examined. The solvent abilities of the liquids were examined at room temperature; six low-melting compounds were examined at 50 OC instead. Compounds having boiling points in excess of about 100 "C and having no appreciable solvent behavior at room temperature were also examined at 100 "C. The relative solubilities were recorded as de-
2630 Ind. Eng. Chem. Res., Vol. 31, No. 11,1992 25 2(
S O L W
mNylllslE
I
0
0
5
10
15
.. \ \
/
/ /
I
20
\
'Pp
25
HYDROGEN BONDING PARAMETER, MPa112
Figure 1. Solubility diagram for TBBA-PC. Hansen polar vs hydrogen-bonding components for solvents and nonsolvents.
100 0
scribed in the Experimental Section (Table I). Forty-five compounds are designated as solvents, 1 compound is designated as having intermediate solvent ability, and 57 compounds are designated as nonsolvents. The ultimate relative solubilities may vary slightly from the recorded ones due to the effeds of dissolution rates within the time frames of observation. Dissolution r a w can vary and are functions of temperature, viscosity, the particular compound or liquid, the amount and type of agitation, and the polymer surface area. The facts that dissolution was determined visually and that refractive index effects may cause erroneous observations must also be considered as sources of error. The experimental procedures used here do not readily distinguish between true solubility,reaction, and/or degradation followed by dissolution unless complete separations and analyses of the resulting mixtures are also carried out. For example, piperidine (no. 65) appears to react with the polymer at room temperature as evidenced by the considerable heat of reaction; however, the behavior of triethyl phosphate (no. 36) can be questioned. The reported solubility parameter of TBBA-PC was calculated to be 20.64 MPa112 using the tables of Hoy (Muruganandam et al., 1987). We obtain an average value of 21.97 f 1.15 MPa1/2using the group additive solubility constants of Fedora (19741,(22.89 MPa1/2),Hoy (19701, (20.68MPa112),and Van Krevelen and Hoftyzer (1976) (22.34MPa1/2). Density for the latter two calculations assumed a reported value of 1.953 g/cm3 (Muruganandam, 1987). Poor or no correlation of solvent ability was found with Hildebrand total solubility parameters. The 45 solvents found (Table I) have solubility parameters between 16.8 and 30.1 MPa112,and poor, medium, and/or strong hydrogen-bondingabilities. Solvents were observed from 9 of the 12 classes of compounds examined; no hydrocarbons, alcohols, or nitriles were found to be Solvents. No compound classes were observed to yield only soluble mixtures with the polymers. Better correlations of compound structure with solvent ability were obtained using the dispersion, polar, and hydrogen-bonding components of Hansen (Barton, 1983; Hansen and Beerbower, 1971). A plot of polar versus hydrogen-bondingcomponents (Chawla and Chang, 1975; Hansen and Beerbower, 1971)is shown in Figure 1 for 71 compounds for which Hansen parameters were available. The enclosed area representing all 33 solvents includes only 12 (32%) nonsolvents. Correlation of solvent behavior with all three Hansen parameters using a graphical representation on triangular coordinates also yields a better corre-
20
40
60
FD
80
100
Figure 2. Solubility diagram for TBBA-PC. Graphical representation of relative Hansen solubility components of various liquids. (See text for definition of coordinates.)
lation, Figure 2. Here FD represents 100 times the Hansen dispersion components divided by the sum of the corresponding dispersion, polar, and hydrogen-bonding components (Lloyd et al., 1984);similar fractional contributions are calculated for the other two components. In this case the enclosed area includes all 33 aolventa but only 11 (29%) of the nonsolvents. The total solubility parameter calculated from the averages of the dispersion, polar, and hydrogen-bondingcomponents for 33 solvents is 20.8 Wall2. This "experimental" value corresponds to the average calculated value of 21.97 MPa112noted above. The behavior of the polymer in 22 compounds is marked by a footnote c in Table I. Such behavior includes (a) partial dissolution followed by opacification and whitening of the clear swollen polymer accompanied by cessation of further dissolution; this behavior is frequently accompanied by a hazy or cloudy supernatant liquid; (b) dissolution to give a clear solution followed by precipitation at the same temperature of a white and opaque solid, mushy gellike formation, or, at the extreme, a solidification to a 'candle wax-like" solid; and (c) dissolution at elevated temperature followed by precipitation of solid, "gelation", and/or a hazy-cloudy formation in the supernatant liquid upon cooling. The reason(s) for these behaviors is (are) not known, but it might include, among others, solventinduced crystallization, polymer-liquid solvate formation, precipitation due to supersaturation, fractionation by polymer molecular weight, gel formation, or reaction/ degradation. The behavior frequently appears to be a function of time; the "structure" appears to build with increasing time and is also most likely a function of temperature, concentration, and viscosity. Large typical polymer birefringent spherulites, indicative of solvent-induced crystallization, have been observed in concentrated solutions (-50%) of some solvents, e.g. acetophenone, after a long time standing at room temperature. Seven particularly noteworthy cases of this behavior were noted. (a) Methylene chloride (no. 1)solutions containing about 51% polymer became hard candle wax-like solids after about 17 days standing at room temperature. (b) Dimethyl sulfoxide (no. 5) solutions containing about 14% polymer were readily formed at room temperature; they changed to a white opaque slush after about 36 h.
Ind. Eng. Chem. Res., Vol. 31, No. 11,1992 2631 Redissolution would not occur at elevated temperature. (c) Chloroform (no. 21) solutions containing about 51% polymer were clear at room temperature but changed into candle wax-like solids after about 14 days standing. (d) Chlorobenzene (no. 42)solutions containing about 50% polymer became clear stiff gels after about 11 days at mom temperature. The gels became clear at 100 "C but became cloudy when cooled. (e) Tetramethylurea (no. 60) containing about 50% polymer was clear and soluble at room temperature but became a rigid gel after about 8 days standing. The gel became clear at 100 "C; the clear solution then became cloudy when cooled to room temperature. (f) A clear solution of about 50% polymer in ethyl benzoate (no. 57) became a rigid opaque gel after 11 days at room temperature. (9) n-Propylbenzene (no. 84) dissolved less than 4.9% polymer at room temperature; solubility was almost complete at 100 "C. This warm solution became a candle wax-like solid when cooled to room temperature. These types of solvent-polymer interaction were not observed with the heterocyclic, aldehyde, alcohol, amine, and amide compounds that were used. Toluene (no. 6) also causes this "gelation/precipitation" behavior. The white insoluble solid resulting from attempted dissolution of the polymer at room temperature was removed by filtration and dried overnight at ambient temperature and at less than 0.1-Torr pressure. The resulting solid and the virgin TBBA-PC were subjected to DSC (at 10 "C/min scanning rate), TGA, and X-ray analyses. A large broad endotherm was observed peaking at about 145 "C in the toluene-treated solid; it was not observed in the virgin polymer or upon rescanning the same sample. The weight loss upon scanning at the same rate was 7.82%. This corresponds to 0.52 mol of toluene per molar unit of polymer. The X-ray patterns of both samples were identical; no peaks due to crystallinity were evident, but two broad peaks indicative of "structure" were observed. Similar behavior was noted earlier by Soviet workers for the neat polymer (Kozlov and Perepelkin, 1967). This observation suggests that the mlvent-polymer behavior is not due to solvent-induced crystallization in this case, but it may be due to a formation of a weak "structure" such as might be expected through a solventtype "cross-link". For example, 1 mol of solvent could "connect" with 2 molar units of polymer thus effectively "doubling" the effective molecular weight, affecting a "cross-link density", reducing the polymer solubility, and increasing the chances of "gelation". Further work is necessary to establish the true reason for this behavior. Conclusions Numerous organic liquids comprising at least 12 diferent chemical classes have been identified as solvents and nonsolventa for tetrabromobisphenol-A polycarbonate between room temperature and 100 "C. The combinations of theee solvents with the nonaolventa offer the membrane specialist literally hundreds of choices of potential spinning dope compositions. Poor correlation of observed solvent ability with Hildebrand-Scott total solubility parameters was noted. Correlation was much better with the three Hansen-type solubility parameter components. The solubility parameter of the polymer was estimated from the three Hansen solubility Components of the identified solvents to be about 20.8 MPa1/2,which compares favorably with values calculated from group additive solubility
constants. Some liquids interact with the polymer in a manner not yet full understood; this results in limited solubility and solution instability. Acknowledgment
I am indebted to James E. Magner and Thomas 0. Jeanes (T.O.J.) of Central Research & Development, The Dow Chemical Company, for samples of polymer. I also wish to acknowledge helpful discussions with T.O.J. regarding thiswork and similar limited solubility studies of his on the same polymer. The DSC, TGA, and X-ray analyses were carried out by Wing Toy of DOW'SWestern Division Analytical Laboratories. Nomenclature DSC = differential scanning calorimetry TGA = thermogravimetric analysis solubility parameters: (megapascals)1/2(MPa1/2);( ~ a l / c m ~ ) l / ~ = (MPa1/2/2.0455) inherent viscosity reported dL/g (deciliters per gram), at a polymer concentration of 0.5 g/dL of methylene chloride Literature Cited Barton, A. F. M. CRC Handbook of Solubility Parameters and Other Cohesion Parameters; CRC Press: Boca Raton, FL, 1983. Beck, H. N.; Sanders, E. S.; Lipscomb, G. G. U.S. 4,962,131, Oct 09, 1990. Brandrup, J., Immergut, E. H., Eds. Polymer Handbook, 2nd ed.; Wiley: New York, 1975. Chawla, A. S.; Chang, T. M. S. Use of Solubility Parameters for the Preparation of Hemodialysis Membranes. J.Appl. Polym. Sci. 1975,19, 1723-30. Fedora, R. F. A Method for Estimating Both the Solubility Parameters and Molar Volumes of Liquids. Polym. Eng. Sci. 1974,14 (2), 147-154. Hansen, C.; Beerbower, A. Solubility Parameters. In Kirk-Othmer Encyclopedia of Chemical Technology, 2nd ed.; Mark, H. F., McKetta, J. J., Jr., Othmer, D. F., Eds.; Wiley-Interscience: New York, 1971; Supplement Volume, pp 889-910. Hildebrand, J. H.; Scott, R. L. The Solubility of Nonelectrolytes, 3rd ed.; Reinhold: New York, 1950. Hoy, K. L. New Values of the Solubility Parameters from Vapor Pressure Data. J. Paint Technol. 1970,42, 76. Kesting, R. E. Synthetic Polymeric Membranes, A Structural Perspectiue, 2nd ed.; Wiley: New York, 1985. Kozlov, P. V.; Perepelkin, A. V. Investigation of the Crystallization of Polycarbonates with Various Chemical Structures. Vysokomol. Soedin. 1967, A9 (2), 370-376. Lloyd, D. R.; Prado, T.; Kinzer, K. E.; Wightman, J. P.; McGrath, J. E. Asymmetric Membrane Preparations from Solventless Casting Systems. Polym. Mater. Sci. Eng. 1984, 50, 152-155. Muruganandam, N.; Koros, W. J.; Paul, D. R. Gas Sorption and Transport in Substituted Polycarbonates. J. Polym. Sci., Part B: Polym. Phys. 1987,25 (9), 1999-2026. Sanders, E. S.; Clark, D. 0.;Jensvold, J. A.; Beck, H. N.; Lipscomb, G.; Coan, F. L. (The Dow Chemical Company) U.S. 4,772,392, Sept 20, 1988. Sanders, E. S.; Jensvold, J.; Clark, D. 0.; Coan, F. L.; Beck, H. N.; Mickols, W. E.; Kim, P. K.; Admassu, W. (The Dow Chemical Company) U.S. 4,955,993, Sept 11, 1990a. Sanders, E. S.; Wan, H. S.; Beck, H. N. (The Dow Chemical Company) U.S. 4,975,228, Dec 4, 1990b. Van Krevelen, D. W.; Hoftyzer, P. J. Properties of Polymers, 2nd ed.; Elsevier: Amsterdam, 1976.
H. Nelson Beck Central Research & Development- Walnut Creek The Dow Chemical Company Walnut Creek, California 94598
Received for review June 23, 1992 Accepted September 9, 1992
