Identification of Polyurethanes by High Resolution Nuclear Magnetic Resonance Spectrometry Edward G . Brame, Jr., Raymond C. Ferguson, and George J. Thomas, Jr.’ E. I. du Pont de Nemours and Co., Experimental Station, Wilmington, Del. 19898 High resolution proton magnetic resonance spectra of polyurethanes are useful for identification and estimation of their compositions. The commonly used monomeric and polymeric components produce highly characteristic patterns which are usually resolved sufficiently for electronic integration. The 60 Mc/sec spectra of eleven different polyurethanes in AsCh solutions at looo C are shown, and the chemical shift data for structures in these and related polymers are summarized.
THEINCREASING IMWIMANCE of polyurethanes as commercial polymers and the proliferation of polyurethane compositions have created the need for a rapid, systematic method for the identification of unknown compositions. Linear polyurethanes are usually synthesized by the following route (1):
R(OH)2
+ R’(NCO)z
.-*
0
0
ll
11
X--(OROCNHR’HNC)n-X
where R is typically art alkyl, a polyether, or a polyester chain, and R’ is usually a low molecular weight aryl or alkyl moiety. Further reaction of the linear polymer with other polyfunctional agents leads to chain-extended or cross-linked polymers which may contain other unique structural groups. Identification of t,he components of polyurethanes by isolation and identification of hydrolysis products is difficult and time consuming. Infrared spectra are useful in distinguishing between polyether and polyester types of polyurethanes, and the spectra aid sometimes in identifying end groups*.g., COCl, NCO, OH, NH, or NHz-but, in general, they are neither conclusive nor quantitative. High resolution proton magnetic resonance spectra prove to be highly characteriistic of the type of polyurethane and also provide useful quantit,ative data (2). Previous NMR studies of polyurethanes have dealt with the identification of various amide groups (3) and the quantitative analysis of various polyether compositions ( 4 ) that are used to make polyurethane foams through reaction with diisocyanates. Related papers on polyester resins ( 5 ) and polyalkalene glycols (6) contain additional NlMR data pertaining to structures not included here. In this, paper we outline the use of NMR for identifying components and estimating their concentrations in polyurethanes, and we summarize pertinent chemical shiftstructure correlation data obtained in our laboratories. 1 Present address, Massachusetts Institute of Technology, Cambridge, Mass.
(1) J. H. Saunders and K. C. Frisch, editors, “High Polymers,” Vol. XVI, Part I, Interscience, New York, 1964. (2) R. C. Ferguson, Kautschuk Gummi, 18 ( l l ) , 723 (1965). (3) N. Surni, Y. Choklci, Y. Nakai, M. Nakabayashi, and T. Kanzawa, Makromol. Chemie, 78, 146 (1964). (4) A. Mathias and N. Idellor, ANAL.CHEM.,38, 472 (1966). (5) D. F. Percival and h4. P. Stevens, Zbid., 36, 1574 (1964). (6) T. F. Page, Jr., and W. E. Bresler, Zbid., p. 1981.
EXPERIMENTAL
The polyurethanes studied in this work were obtained from various commercial sources. They were prepared for the NMR examination by dissolving them in Baker and Adamson reagent grade arsenic trichloride (7)at a concentration of about 15% w/v and by using standard 5-mm 0.d. precision glass tubes. Oven-dried glassware was used, and exposure of either the solvent or solutions to atmospheric moisture was minimized. The spectra of all samples were obtained at about 100’1IO” C with a Varian Associates, Model A-60 NMR spectrometer equipped with a variable temperature probe. The spectra were run at a 1 cpslsec sweep rate and a sweep width of 500 cps. The spectra were calibrated us. tetramethylsilane (TMS) as the internal reference. In some cases, the less volatile compound hexamethyldisiloxane (HMDS) was used as the internal reference. HMDS degrades in hot AsC13 with the appearance of lines at about 0.2 and 0.3 ppm but each measurement was made before noticeable degradation occurred. All data in this paper have been converted to the delta scale (6 = 0.00 for the internal reference, TMS, and positive &values indicate resonances at lower field strengths). RESULTS AND DISCUSSION
Qualitative Applications. Good quality high resolution NMR spectra were obtained from various polyurethanes dissolved in AsC13. Conventional solvents such as CS2, CC14, CDCL, and CeD6generally give poor quality spectra even near the boiling point of the solvent. Trifluoroacetic acid may also be used to obtain reasonably good quality spectra from polyurethanes. Polyurethanes degrade slowly in AsC13, but the spectra of freshly prepared solutions are adequately reproducible. The spectra did not change significantly in less than two or three days even though the samples were degrading slowly. Since chemical shifts are somewhat concentration and temperature dependent, the set of standard conditions described in the Experimental Section were chosen for comparison of data. Typical spectra and integrals of various polyurethanes are shown in Figures 1-4. Figures 1 and 2 show spectra of four polyether-based polyurethanes. The ether moiety contained in these polyurethanes is either poly(tetramethy1ene ether) glycol (PTMEG) or poly(propy1ene) glycol (PPG). Spectra of six different polyester-based polyurethanes are shown in Figures 3 and 4. The ester moiety in each of these polyurethanes is an adipate. Certain characteristic resonances occur frequently throughout the spectra shown. One of these is attributed to the structure X C H ~ C H ~ C H ~ C H which Z X produces a characteristic pair of multiplets. The higher field multiplet is always attributed to the internal methylene groups, and its chemical shift is less sensitive to the identity of X than is the shift of the CHZX multiplet. (Although these two pairs of methylene groups are chemically equivalent, their protons are not (7) H. A. Syzmanski, A. Bluemle, and W. Collins, Appl. Spectry., 19, 137 (1965). VOL. 39, NO.
4, APRIL 1967
517
magnetically equivalent, and the complex multiplet patterns are due to strong coupling.) In the butane diol (BD)-based polyurethanes, X is the oxygen atom of the urethane group, and the internal CH2 and CH2X multiplets appear at 1.68 and 4.13 ppm, respectively. These multiplets appear at essentially the same chemical shifts for polyesters of BD and for polyurethanes based on the polyesters (Figure 3-B). Similarly, the adipic acid (AA) moiety in polyesters and in the corresponding polyurethanes has characteristic CH2 and CH2Xmultiplets at 1.62 and 2.30 ppm, respectively. In this case, Xis the carbonyl group of the ester. A similar XCH2CH2CH2CH2Xpattern was expected for (PTMEG) and for polyurethanes based on PTMEG. However, PTMEG and its polyurethanes produce a more complex, but highly characteristic pattern with the internal CH2 multiplet at 1.62 ppm and the CH2Xresonances falling in the region 3.4 to 4.2 ppm. In the polyurethanes two types of X group are present: ether oxygens and urethane oxygens. However, these alone do not account for the complex CHzX pattern, particularly in PTMEG itself; therefore, the effect must be due in part to the distribution of chain lengths in FTMEG.
Table I. Polyethers
0
Polyesters 0
0
I/
I/
-COCHzCHzOC0 0
/I
-C-OCHCHzOC-
II
I
CHs 0
0
II
/I
-C-OCH~CHZCHZCHZ~-C-
(4 (b) 0
0
I1
I1
-C-OCHzCHzOCHzCHz0-C-
(4(b) 0
0
I/
/I
-NCOCHaCHsOCN-
0
0
I1
I1
>NCOCHzCHeCH2CHzOCN< (a) (6)
51 8
(8) N. F. Chamberlain, Esso Research and Engineering Co., Baytown, Texas, 77520. Private communication, 1967.
Chemical Shift Data for Different Polyether-Based and Polyester-Based Polyurethanes Varian Code Chemical shift (6) Structure 1-CbO 1.15 --O(CHCHzO),,3.64 2-Ca00, I 3.6 to 3.8 CHa 2-BbBo 1.62 -O(CHrCHzCHzCHz-O)n2-BbOb 3.6 (a) (b) 0
Polyurethanes
The 1,2-propane diol (PD) moiety-OCH2CH(CHa)Ois revealed by a methyl doublet at high field and a characteristic pattern for the CH2 and C H protons in the 3.5 to 4.2 ppm region. In polyurethanes based on PPG, the methyl doublet is at 1.15 ppm and the pattern for the other three protons is at 3.6 to 3.8 ppm (Figure 2) except for the C H group in PPG which shows a broad line at 4.8 to 5.2 ppm (8). These latter resonances are attributed to units situated between ether links. Weaker resonances on the downfield side of the two major patterns are attributed to propylene units attached to the oxygens of urethane groups. In polyesters and the corresponding polyurethanes, the PD moiety produces the methyl doublet at 1.20 ppm and additional multiplets between 4.1 and 5.5 ppm (Figure 4-D). The broad line at about 5.0 ppm can be used to measure the number of polyester linkages in the polyurethanes.
ANALYTICAL CHEMISTRY
2-BbBk 2-BbKb
1.62 2.30
2-BbQb
4.19
1-Cbq 2-CaqQb 3-ABqQb
1.20 4.23 4.8 to 5.2
2-BbBq 2-BbQb
1.69 4.0 to 4.3
2-BqOb 2-BoQb
3.62 4.13
2-BqQn
4.30
2-VhhVhh 14-HB 14"
3.88 7.07 7.25
1-Vhn
2.13
2-BbBqn 2-BbQn
1.68 4.10
A
B
e 8.0
7.0
6.0
5.0
l-
B 4.0
3.0
2.0
1.0
0 PPM
Figure 1. 60 Mchec:. PMR spectra and integrals (ASCIs solutions, 100" C) A. Polyurethane from pldy(tetramethy1ene ether)glycol and methylene-bis(4-pheny1)isocyanate B. Polyurethane from poly(tetramethy1eneether)glycol and toluenediisocyanate.
In polyurethanes bas;ed on 1,Zethanediol (ED), a singlet is observed at 4.30 ppm. In polyurethanes based on polyesters of E D , two sharp resonances are observed at 4.19 and 4.30 ppm (Figure 4). The former is attributed to CH2 groups attached to ester oxygens, while the latter is attributed to CH2 groups attached to urethane oxygens. The methylene bis(4-phenyl isocyanate) (MDI) moiety
produces a characteristic AA'BB' pattern centered at 7.16 ppm and a CH2resonance at 3.88 ppm (see Figures I-A, 2-A, 4-A, and 4-B). The toluene diisocyanate (TDI) moiety produces a characteristic aromatic resonance pattern in the 7.0 to 8.0 pprn region and a methyl resclnance at 2.13 ppm (Figures 1-B, 2-B, 4-C, and 4-D). Even though toluene diisocyanate is known to exist in two different isomer forms (:2,4and 2,6) the patterns shown in the region of 7.0 to 8.0 pgm for TDI based polyurethanes are similar enough to suggest that the isomeric ratio is essentially constant in these examples. Other examples have also shown this pattern to be highly characteristic of TDI moiety. The diethylene glycol moiety (OCH ?CH?OCHzCHz&) produces an XCHzCHIY pattern (X = -00
and Y =
/I
8.0
7.0
6.0
5.0
4.0
3.0
2.0
1.0
0 PPM
Figure 2. 60 Mc/sec. PMR spectra and integrals (ASCIS solutions, 100" C) A. Polyurethane from poly(propy1ene ether)glycol and methylenebis-(4phenyl)isocyanate B. Polyurethane from poly(propy1ene etherklycol and toluenediisocyanate
The methylene groups for this moiety have not been unambiguously located, but are presumed to appear in the region between 3.6 and 4.3 pprn. In Table I the chemical shift data for different polyetherbased and polyester-based polyurethanes are summarized. Also included are the structural formulas and the Varian Code (9). Quantitative Applications. The integral intensities of the resolved groups of resonances provide a means of calculating the compositions of polyurethanes. Small uncertainties arise when the NH resonances have not been unambiguously located. The analysis of two-component systems is sometimes useful in providing further verification of the identification made by chemical shift correlations. It also can be used to estimate the molecular weight of the polyether or polyester diol employed. For the PTMEG-MDI polyurethane of Figure 1-A, three integral amplitudes can be measured: I, 1.0 to 2.5 ppm; 11, 2.5 to 5.5 pprn; and, 111, 5.5 to 8.0 ppm. The relationships are: kI = 4[C4] (1) kII = 4[Cd]
+ 2 [MDI]
(2)
kIII = 8[MDI]
-0C-), with multiplets centered at 3.62 and 4.13 ppm (Figure 3-A). The resonance at 6.82 ppm in Figure 4-A is attributed to the aromatic protons of a fourth component, bis-(hydroxyethyl hydroquinone) (HEHQ)-i.e., HOCH~CH~O~OCH~CH~OH
(3) Where k is a proportionality constant including the solution concentration and instrument variables, [Cd is the mole fraction of -0(CH2)40- units, and [MDI] is the mole fraction of the aryl isocyanate. Rigorously, a correction should be made for the terminal groups of the polymer-i.e., aryl (9) N. S. Bhacca et al., Varian Spectra Catalog (1962 and 1963). '401. 39, NO. 4, APRIL 1967
519
8.0
,
70 I
60
6.0
4.0
3.0
20
1.0
0 PPM
I
B
1 8.0
7.0
6.0
5.0
4.0
3.0
20
1.0
0 PPM
Figure 3. 60 Mclsec. PMR spectra and integrals (ASCI* 801UtiOIls,
loo" C )
A. Polyurethane from polyester (adipic acid-diethylene glycol) diol and toluenediisocyanate B. Polyurethane from polyester (adipic acid-1,4-butanediol) diol and toluenediisocyanate
NCO or FTMEG-based on an independent end group analysis. However, for a high molecular weight polyurethane, [MDI] =[PTMEG], and the degree of polymerization of PTMEG, D P ==[Cd/[MDI]. For the PTMEG-TDI polyurethane of Figure 1-B, four areas can be measured: I, 1.0 to 2.0 ppm; 11,2.0 to 2.5 ppm; 111, 2.5 to 5.5 ppm; and IV, 5.5 to 8.5 ppm. The relationships are: kI = 4[C4]
(4)
kII = 3 p D I ]
(5)
kIII = 4[C4]
(6)
kIV = 3[TDI]
(7)
The analysis is overdetermined by four equations, but they are useful for consistency tests and for the detection of possible additional components. Since the measurement of the methyl resonance (area 11)tends to be inaccurate due to overlap with area I, the mole fractions are most reliably estimated by solving the pair of Equations (7) and (4) (5) (6). The analyses of polyurethanes of Figures 2 and 3 are similar. They can be deduced from the chemical shift data in the discussion or Table I. The analysis of the polyurethane of Figure 4-A is based on the areas I, 1.0 to 3.0 ppm; 11, 3.0 to 5.0 ppm; and 111, 6.0 to 8.0 ppm, and the relationships
+ +
kI = 8[AA] kII = 4[ED]
+ 2[MDI] + 8[HEHQ]
kIII = lO[MDI]
+ 4[HEHQ]
(8) (9) (10)
In this case, the NH resonances appeared at 6.82 ppm, overlapping with the aromatic resonance of HEHQ. The results of analysis on the polymer of Figure 4-A were [AA] = 0.42, [ED]= 0.45, [MDIJ = 0.10, and [HEHQl = 0.04. From the mole fractions of adipic acid and ethanediol, one can readily calculate, for the polyester diol HO(CH& [00C(CH~)4C00520
0
ANALYTICAL CHEMISTRY
1
I
I
I
I
I
8.0
7.0
6.0
5.0
4.0
3.0
I
2.0
I
LO
I
o
PPM
Figure 4. 60 Mc/sec. PMR spectra and integrals (ASCIS solutiom, loo" C ) A. Polyurethane from polyester (adipic acid-l,Z+thanediol) diol, methylene-bis(4-phenyl isocyanate), and bb(hydroxyethy1 hydroquinone) B. Polyurethane from polyester (adipic acid-1,Zethanediol) diol and methylene-bis(4-phenyl isocyanate) C. Polyurethane from polyester (adipic acid-1Jethanediol) diol and toluenediisocyanate D. Polyurethane from polyester (adipic acid-1,Z-ethanediol and 1,Z-propanediol) diol and toluenediisocyanate
(CH2)2]nOH, that 2 = 28. However, calculation of the molecular weight of the polyurethane would require identification of the end groups, or an end group analysis. The analysis of the polyurethane of Figure 4-D was made by measuring the integrals I, 1.0 to 1.4 ppm; 11, 1.4 to 2.0 ppm; 111, 2.0 to 3.0 ppm, and IV, 3.0 to 5.5 ppm; and the relationships kI = 3[PD]
(11)
kII = 4[AA]
(12)
+ 3[TDI] kIV = 4[ED] + 3[PD]
kII1 = 4[AA]
(13)
(14)
Table II. Mole Fractions of Components in Polyester-Based Polyurethane Component NMR Hydrolysis AA 0.37 0.35 ED 0.44 0.46 MDI 0.14 0.15 HEHQ 0.05 0.04
The aromatic region can also be employed in the analysis, if the N H resonances hilve been located. The composition of the polymer of Figure 4-D calculated from the four equations above, was [AA] = 13.43, [ED] = 0.39, [PD] = 0.10, and [TDI] = 0.08. The NMR analyses on polyurethanes prepared in our lab-
oratories agreed with the known compositions, within the experimental accuracy. We have obtained a verification of the reliability of the NMR method on one “unknown” commercial polyurethane by an independent analysis, based on hydrolysis of the polymer and quantitative isolation of the hydrolysis products. The composition of that polyurethane, as found by the two methods, is given in Table 11. ACKNOWLEDGMENT
The authors are indebted to Drs. M. Brown, J. W. Crary, and R. J. Athey for supplying the polyurethanes, and for helpful discussions. J. W. Crary and M. DeBrunner performed independent hydrolysis analysis on some of the polyurethanes including the sample reported in Table 11. V. A. Brown did most of the experimental work.
RECEIVED for review November 23,1966. Accepted February 6,1967.
Gas Chromlatographic Analysis of Dilute Aqueous Systems W. G. Jennings and H. E. Nursten Department of Food Science and Technology, University of California, Davis, Calif.
WHILEGAS CHROMATOGRAPHY has contributed significantly to the analysis of volatiles in many systems, this degree of success has not been achieved with dilute aqueous systems, particularly those such as milk, which contain solventextractable nonvolatiles. Many attempts have been made to surmount these difficulties (Id), but they have not been too successful. Methods involving the injection of head-space vapors ( I , 3, 6, 7) are limited by the fact that in order to obtain detectable quantities of dilute vapor components, one must use such a massive quantity of vapor that the injection requires a considerable period of time, and the separated components are in consequence so diluted by carrier gas that they may escape detection. The use of low temperature precolumns has been suggested (3, 5, 7, 8) to permit the concentration of volatiles from head-space vapors prior to their chromatographic sepacltion, but these suffer from obvious disadvantages. While a degree of concentration can be achieved, sufficient diffusion occurs in the pre-column to limit the quantity of gas distillate 01- head-space vapor that can be used, and in dilute aqueous systems the major volatile in the pre(1) R. Bassette, S. Ozeris, and C. H. Whitnah, J. Food Sci., 28, 84 (1963). (2) I. Hornstein and P. F. Crowe, ANAL. CHEM., 34, 1354 (1962). (3) . , W. G. Jenninas. S. Vilihalmsson, and W. L. Dunkley, J. Food Sci., 27, 306 (1962). (4) L. M. Libbey, D. D. :Bills, and E. A. Day, Zbid.,28, 329 (1963). (5) M. E. Morgan and E. A. Day, J. Dairy Sci., 38,1382 (1965). (6) R. Teranishi, R. G. ELuttery, and R. E. Lundin, ANAL. CHEM., 34, 1033 (1962). (7) J. M. Mendelsohn, Pd. A. Steinberg, and C . Merritt, Jr., J. Food Sci., 31, 389 (1966). (8) D. E. Heinz, M. R. Sevenants, and W. G. Jennings, Zbid., 31, 63 (1966).
column is ineviti-.; water, which saturates the pre-column and again limits the quantity of vapor that can be used. Although the flame ionization detector is relatively insensitive to water vapor, massive slugs of water cause major deflections and effectively blank out at least the first portion of any chromatogram. Scott and Phillips (9) proposed the use of gas-solid chromatography to achieve a concentration of volatiles from dilute systems, and their subsequent elution to a gas-liquid chromatography column for analysis. Several investigators have used activated charcoal for adsorbing volatiles for subsequent analysis--e.g., Walls (IO); the work of Dhont and Weurman (11)on model systems indicated that recovery efficiencies were quite high. A major advantage of charcoal for this purpose lies in its relatively low affinity for water, and Heinz et al. (8) employed it to adsorb volatiles from several dilute aqueous systems for subsequent elution and analysis. It appeared logical that a significant advantage might be gained by substituting charcoal in the gas-solid chromatography step of the method proposed by Scott and Phillips (9). This work was directed toward the development of a method that would permit the concentration of volatiles from very large vapor samples and their elution for chromatographic analysis. METHODS AND PROCEDURES Gas Chromatography, Gas chromatographic separations utilized a modified Aerograph 600B with flame ionization detection. The output impedance was increased to 20 kohms (9) C. G. Scott and C . S. G. Phillips, “Gas Chromatography,” A. Goldup, Ed., Brighton, 1964, p. 273. (10) L. P. Walls, J. Pomol. and Hort. Sci., 20, 59 (1942/3). (11) J. H. Dhont and C. Weurman, Analyst, 85,419 (1960). VOL. 39, NO. 4, APRIL 1967
521
