“Methods in Enzymology,” Academic Press, New York, 1955. (3) Feinstein, H. I., U. S. .itomir Energy Comm., TEI-555(1955). (4) Kato, Takeshi, Nomiao, Y., Shinra, K., J . Chein. SOC.Japan, Purr C h m . Sect. 76, 373 (1955). (5) Kushner, D. J., Bioehim. e1 Hiophys. Aclo 20, 551 (1956). (6) Monnier, D.,Fasel, M., .lfitt. Lehmsm. Hyg. 47, 141 (19,561.
mination df Traces of Metals,” 3rd ed., Interscience, New York, 1!158.
RECEIVEDfor review October 31, 1958. Accepted March 16, 1959. Work s u p ported by the Howard Hughes Medical Institute.
Isomer Ratio Analyzer for Toluenediisocyanate Based on Dielectric Constant Measurements SAMUEL STEINGISER, W. C. DARR, and E. E. HARDY Research Department, Mabay Chemical Ca., New Martinsville, W. Va. .In the production of toluenediisacyanate, a rapid reasonably precise method far the determination of isamer content was needed. Freezing points and infrared spectroscopy had certain disadvantages. The dielectric isamer ratio analyzer was developed to fill the need for a continuous record of isomer ratios far the toluene-2,4and -2,6-diisocyanate mixtures. This instrument can be used in the laboratory ar in the line in production. Its accuracy i s to about *0.20/, isomer unit, better than methods previously used.
~o~UENEnIIsocYANATE is one of the basic ingredients in the manufacture of urethane polymers, principally urethane foams. A very important aspect in the creation of these foams is the interrelationship of the rates of reaction of the toluenediisocyanate isomers with hydroxyl resins in forming the polymer. Toluenediisocyanate exists in several isomeric forms, having different reaction rates in terms of the desired reactions. As manufactured in this country, it contains mainly two isomers, the 2,4and 2,6- structures. A trace of the 2,5- isomer may be present, which will affect all the methods of analysis used. The main consideration has been given to the reaction rates of the 2,4and 2 , 6 isomers and their mixture in thetoluenediisocyanates. Normalproducts sold in this country consist of nominal 100% 2,4- isomer (actually nreater than 97.5%); nominal SOYo 2,4- and 20% 2,6- isomer mixture (actually 80 + 2%); and nominal 65% 2,4- and 35% 2,6-isomer mixture (actually 66 =t2%). For trouble-free production of urethane polymers, especially the very sensitive foams, careful control of isomer content is important,
The first step, then, is measurement of the isomer content of the toluenediisocyanate. Previous methods of measurement consisted of determining the freezing point of the mixture and comparing it with a calibration chart based on the values of the pure components ( I ) , or using the infrared absorption peaks a t 12.35 and 12.80 microns for each of the pure components and the ratio of the relative heights of each peak (3). I n each method, a sample of diisocyanate must be removed and analyzed for either the freezing point or infrared absorption with its consequent
Toble 1. Dielectric Constant as a Function of Frequency for Pure Toluene-2.4diisocyanate and Toluene-2,6-diisocyanate (Boonton Q-Meter, temp., 23.1”C.)
FIP quency, Kc. 90 500
Dielectric Constant,p 100% 2,4-
100%
8.15 8.48
5.li
Figure 1. FoxbaroCapacitance Dynalog analyzer
VOL. 31. NO. 7, JULY 1
2,h
5.36
time delay. However, in the dielectric method, a direct reading of isomer ratio to a greater precision, can be available instantaneously if the probe is mounted in the pipeline of the production equipment. The dielectric isomer ratio method is based on the dielectric constants of the two isomers in question-2,4- and 2,6-. The variation in capacity of a cell with variation in isomer ratio records directly this value on a chart calibrated in per cent 2,4- isomer units. This method is subject to some of the errors inherent in the freezing point procedure, in that it must be applied to a pure system consisting of two components only-the 2,4- and 2,6- isomers. I n infrared, additional components will be a problem only if they have bands overlapping those
assigned in the analyses. Interferences from other materials will cause proportionate errors depending upon the degree of contamination. However, these can be allowed for, by calibration, up to some limiting value. The accuracy of the dielectric isomer ratio method is better than that of the freezing point method and the infrared method for mixtures in the range of 5 to 95% of the 2,4- isomer, and equivalent to the infrared technique in the 95 to 100% 2,4- isomer range. The dielectric method is less costly to operate in terms of manpower and time, and can give a continuous record if desired. The resultsare obtained instantaneously, a most important factor in production. DISCUSSION
The basis for this method is t,he
Figure 2. Capacity measuring cell Cell Factor A 5.2
0.219
14.3
0.094 0.063
21.5
Table It.
2 , 4 T D I = 8.27 2,6-TDI = 5.21
TDI Isomer Ratio Measurements
(Calibration data on known mixtures of pure 2,4- and 2,6-TDI) Isomer Ratio, Direct Isomer Ratio, Wt. yo Chart Reading, % Isomer Ratio, Infrared, yo 2J4 2,6 2J4 2,6 2142160.0 0.0 100.0 100.0 0.0 100.0 88.2 11.8 87.5 12.5 87.8 12.2 7 6 . 6 2 3.4 76.6 23.4 76.6 23.4 66.8 33.2 33.8 66.2 33.8 66.3 60.8 39.2 41.3 41.0 58.7 59.0 0 . 0 100.0 0.0 100.0 0.0 100.0 73.0 27.0 72.5 27.5 72.4 27.6 37.3 37.9 62.7 37.6 62.1 62.4 0.0 100.0 0.0 100.0 86.7 13.3 86.9 13.1 75 7 24 3 75.4 24.6 34 2 34 3 65 8 65 7 99.6 0.4 ... 100.0 0.0 84.0 16.0 ... 15.8 84.2 76.3 23.7 76.4 23.6 ... 63.4 36.6 ... 63.4 36.6 100.0 0.0 100.0 0.5 0.0 99.5 9.4 8.7 90.6 91.4 8.6 91.3 83.5 16.5 84.0 16.0 83.9 16.1 76.0 24.0 24.1 24.3 75.9 75.7 69.3 30.7 69.3 30.7 69.4 30.6 35.2 36.2 64.8 63.6 36.4 63.8 94.0 6.0 95.0 5.1 5.0 94.9 87.8 12.2 88.5 11.5 88.5 11.5 53.4 46.6 50.5 49.1 49.5 50.9 100.0 0.0 99.8 0.2 ... ,.. 2.0 98.0 2.0 ... ... 98.0 4.0 96.0 4.0 ... ... 96.0 6.0 94.3 5.7 ... 94.0 t . .
1262 *
ANALYTICAL CHEMISTRY
fact that the dielectric constant of ii. mixture is proportional to the values of the pure components, depending on the percentages of these components in the mixture. Because a single dielectric constant is measured for a mixture, it must be a binary mixture only and have no third component which would change the value of the constant beyond the precision of the measurement. Measurements of the pure 2,4- and pure 2,6- isomers a t 23.1" C. show them to have dielectric constants as a function of frequency as given in Table I. It is beneficial to use as high a frequency as possible, so as to minimize the effect of conductance in the liquid being measured. Other factors, such as design of equipment, limit the upper level of frequency one can use. As a result of these considerations, the frequency of 1.6 megacycles was chosen, which was commercially available in the Dynalog capacitance bridge (Foxboro Co., Foxboro, Mass., Model 9850 Dynalog) (2). However, this choice was made after careful experimentation with a General Radio capacitance bridge (too low a frequency), Marconi dielectric test set (too expensive and complicated to operate) , and a Boonton &-Meter (nonrecording, too insensitive). Measurements of the pure 2,4- and 2,6- a t 23.1' C. and a t 1.6 megacycles show them to have the following dielectric constants:
This is a sufficient separation in value so that with the proper cell and a sensitive detector, a correlation with isomer content can be established. The dispersion provided for a t least a 0.5% isomer unit separation, if not better. As a result of this work, it was concluded that the Foxboro Capacitance Dynalog (Figure 1) with a capacity span of 20ppf. (both span and zero adjustable) operating a t a frequency of 1.6 megacycles would be satisfactory in this application. This is used in conjunction with a stainless steel dipping cell (factor 14.3) (Figure 2), and having temperature compensation over the range of 20 to 30' C. Such an instrument was put into service in this laboratory, adjusted for zero and span, and set for room temperature compensation. The unit was calibrated by preparing known mixtures by weight of isomers using pure 2,4and 2,6- compounds obtained by recrystallization, until no 2,g-infrared band showed in the 2,4- and no 2,4infrared band showed in the 2,6- on six different occasions over a long period of time in two different laboratories using different operators. The data are given in Table I1 and shown graphically in Figure 3. Proper ad-
100
IO0
ISOMER RATIO DYNALOG-2,4% INFRA RED-2,4%
ISOMER’ RATIO
-
WEIGHED-2.4% (CALIBRATION) UFFERENCE INFRA RE0-2,4% OFFERENCE ,FREEZING POINT-2,4% axax
0.
@--d
90
8
2 B 80 H Y P
3 70 W K
3
v,
60
50
4
1
I
I
1
60
70
80
90
)
ISOMER RATIO
Figure 3.
TDI
isomer ratio bration curves
50
measurements-cali-
Figure 4.
TDI isomer ratio measurements
Plant lot determinations and comparison of methods
M o b a y isomer ratio analyzer
justments of both zero (at the 100% 2,4position) and of the span, allowed the instrument to read directly on the chart in isomer units to within f 0.2 unit along the major portion of the curve. A calibration correction curve was also constructed and could be used if one wished to go further than the above figure in precision a t the upper end of the curve (about 92%). The weighed isomer concentrations \%iththe Dynalog values were closer t o R theoretical relationship than the infrared values. Whether the infrared M RS off because of improper calibration has not been determined, but experience has been that the infrared tends t o “drift” more easily off calibration. Using the Dynalog to determine “unknon n” concentrations, a series of 57 samples has been run with the Dynalog US. infrared us. freezing point methods. I n Figure 4 the Dynalog calibration line is drawn in for comparison. Very good concordancy is observed at the 80% 2,4region of‘ the calibration line. The infrared data fall somewhat off the line a t the high and low regions, but follow the infrared calibration curve shown in Figure 3 closely, indicating that a consistent calibration error may be causing the deviation. The freezing points are also off in the high and low regions, but in the opposite direction to infrared. In addition, a t the 65% 2,4- region the scatter of freezing points is evident as the method became more tedious and subject to greater error. The primary calibration curves of the freezing method could be contributing a large share of the noted deviation from the theoretical line. The next important question was the effect of contaminants, which one might expect in TDI, on the use of the Dynalog as a precision indicator of isomer ratios.
0
ISOMER RATIO
Two general groups of contaminants were studied. The first included various solvents such as n-hexane, benzene, acetone. Solvatone, and orthene (odichlorobenzene). The second included acidic materials such as hydrochloric acid, organic acid chlorides, and also reaction products such as the dimers.
Table 111.
Xominal Isomer Content,
and urea from water addition. The data for the effects of solvents are compiled in Table 111. Orthene has the least effect, showing 0.4 to 0.6% deviation in isomer unit value from the pure sample for all isomer ranges from 60 to 100% The other solvents listed had larger effects in different directions ranging
Effect of Solvents
Apparent Isomer Ratio Dmlog, %
Contaminant
or, 2 > 4
Wt. %
95
n-Hexane
so
Benzene Acetone Solvatone
0.4 0.5 0.0 0.1 0.2 0.4 0.5 0.0 0.1 0.5 0.0 0.1 0.5 0.0 0.5
Orthene
0.0 0.28 0.38 0.58 0.0 0.5
95
Orthene
80
Orthene
0.0 0.1 0.3 0.5
65
0rthene
0.0 0.5
2466.4 66.1 65.9 65.4 65.1 64.9 95.2 95.0 94.7 94.2 93.7 78.8 79.0 77.9 78.8 80.0 82.2 78.8 82.0 78.9 79.0 79.0 79.1 95.8 96.4 78.8 78.9 79.1 79.2 65.8 66.4
2,633.6 33.9 34.1 34.6 34.9 35.1 4.8 5.0 5.3 5.8 6.3 21.2 21.0 22.1 21.2 20.0 17.8 21.2 18.0 21.1 21.0 21.0 20.9 4.2 3.6 21.2 21.1 20.9 20.8 34.2 33.6
VOL. 31, NO. 7, JULY 1959
1263
Effect of Acid Chlorides, Dimer, and Urea
Table IV.
Nominal Isomer Content,
Apparent Isomer Apparent Isomer Ratio Dynalog, Ratio Infrared, %I %
Contaminant
%, 2,480
Wt. 7%
2,4-
2,6-
Hydrolyzable chloride from benzoyl chloride
0.00 0.12 0.25 0.37
78.8 81.5 83.6 85.4
21.2 18.5 16.4 14.6
Hydrolyzable chloride from HCl
0.00 0.01 0.09 0.15 0.30 0.43
78.9 79.0 81.0 82.5 86.3 89.8
21.1 21.0 19.0 17.5 13.7 10.2 5.2 5.0 4.7 4.5 3.8 2.8 2.0
95
Hydrolyzable chloride from HCl
0.00 0.01 0.02 0.03 0.05 0.075 0.10
94.8 95.0 95.3 95.5 96.2 97.2 98.0
80
Hydrolyzable chloride from HCl
0.00 0.002 0.01 0.02 0.03
79.1 79.3 79.5 79.7
Hydrolyzable chloride from HCI
0.00 0.01 0.02 0.03
65.5 66.0 66.5
Hydrolyzable chloride from HCl
0.00 0,004 0.015 0.03
53.9 54.4 54.9
46.1 45.6 45.1