CONCLUSION
judging the adrquacy of gas chromatography detectors, both flow time should be sellEitivity and under the operating conditions to be used, ~l~~~ sensitivity is acceptable if output is not significantly h some ~ affected by flow ratc, ~ detectors are not sensitive to flow rate, other they may be sensitive to operating variables as pressure. The pressure sensitivity of such detectors may be analyzed like flow sensitivity. The response time of any detector influences peak height, which is especially important in trace analysis; it affects band broadening and symmetry. *and asymmetry is not always caused by column factors; diffusion-type detectors can produce
some effects. Adequate representation of band shape is obtained where the 7 / 0 ratio is 0.2 O r less. The fastest response is needed for the sharp bands appearing early in a chromatogram. If the u of such bands is 5 seconds, then the response time be about second. ~ ~ A recorder h pen speed of 1 second limits over-all speed of detection even with a very fast detector. Faster response n-ill not improve detection Significantly unless the speed of the associated equipment is also improved. LITERATURE CITED
(1) Daynes, H. A , , “Gas Analysis by Measurement of ThPrmal Conductivity,” p. 93, Camhridge Press, London,
1033.
(2) Desty, D. H., .Tatwe 180, 22 (1957).
(3) Deety, D. H., “Vapour Phase Chrorna~ ~ @ $ ‘ & ~‘ji,, p Academic . Press, Sew (4) Dimbat, M,, Porter, p. E., Stress, F, H., ANAL.CHEW 28,290 (1956). (5) Johnson, H- W., Jr., Str08S, F. H.. Pittsburgh Conference
On
Analytiral
Chemistry and Applied Spectroscopy, March 1958. (6) Kepp!er, J. G., Dijkstra, G., Scholj, J. -4.7 In T a p o u r Phase Chromatograph~,’’ D. H. DestyJ ed.2 p. 22’s Academic Press, Yew York, 1957. (7) Keulemans, -4.I. M.,Kwantes, A , , Ibid., p: 15. (8) Martin, A. J, P., James, A. T., Biochem. J. 63, 138 (1956). (9) National Bureau of Standards (Supt. Documents, Washington 25, D. C J , Tables of Normal Probability Functions, Applied Mathematics Series 23 (1953). R~~~~~~~ for M~~ 5, 1g58, Accepted October 6, 1958. Division of Petroleum Chemistry, 134th hfeeting, ACS, Chicago, Ill., September 1958.
Correlation of Thermal Conductivity Cell Response with Molecular Weight and Structure Quantitative Gas Chroma tog ra phic Ana lysis A. E. MESSNER,
D. M. ROSIE,‘
and P. A. ARGABRIGHT
Esso Research und Fngineering Co., linden, N. J.
b A correlation exists between the relative thermal detector response and molecular weight within a structurally similar homologous series. This enables one to predict response values for given compounds where related information is available. The relative response of a thermal detector is independent of temperature, concentration, and carrier gas flow rate over the range investigated. Because it is also independent of the thermal detector used, data here are applicable to all gas chromatographs using thermal detection and helium as a carrier gas.
R
Rosie and Grob (1) have shown that the molar response of a thermal detector to hydrocarbon vapors is unique for each compound T o obtain more acceptable quantitative analytical results, it is necessary to correct observed areas rather than assume that the area is directly pioportional to the weight per cent. Results obtained with multicomponent blends showed that the avECENTLY
1 Present address Department of Chemistry, University o! Rhode Island, Kings-
ton, R. I.
230
ANALYTICAL CHEMISTRY
erage error may be reduced by more than 50% by applying relative response values. A careful examination of the data previously reported has shovn that a h e a r correlation exists between thermal detector response and molecular weight within a structurally similar homologous series. As an extension of this work, relative response data for a number of compounds differing in type as well as structure have been determined (Table I). The average relative response values of all compounds investigated are reported in Table I. The average coefficient of variation for these compounds was 2.8. Differences among observed responses less than this amount were assumed to be due to normally encountered experimental errors. I n the course of this investigation i t was deemed necessary to determine the dependence of relative response on concentration, individual sensing unit, and carrier gas flow rate. APPARATUS A N D MATERIALS
The instruments used were for the most part Perkin-Elmer 154 Fractometers, equipped with thermistor-type detectors. I n one instance, however,
experiments were performed on a Wilkens Aerograph which employs a hot wire thermal detector. I n all cases, the carrier gas a a s helium. Liquid samples of high purity were obtained from the National Bureau of Standards, Washington, D. C., or from the American Petroleum Institute, Carnegie Institute of Technology, Pittsburgh, Pa. When not available in high purity from either of the above sources, the compounds were rectified by distillation. Gas samples of research grade were obtained from the Phillips Petroleum Co., Bartlesville, Okla., or the Matheson Co., East Rutherford,
N. J.
EXPERIMENTAL PROCEDURE
Benzene was chosen as the internal standard and arbitrarily assigned a signal response of “100.0” units per mole. Each compound to be ‘investigated was mixed with a known amount of benzene and the resulting blends were run from two to five times, A minimum of two blends was prepared for each compound. The areas were determined by cutting out the peaks and weighing them on a n analytical balance. Reproducible analyses required that the operating conditions remain constant for the duration of the individual run (5 to 30 minutes).
The volume or weight of the sample need not be known, because the ratio of the area under the benzene peak t o that of the compound being investigated was found t o be invariant with sample size. This is illustrated in Table 11, where the sample charge mas held nominally constant (ca. + 10%) while the amount of each component was varied over a tenfold range. Total sample size was kept below 0.03 ml. for liquids and 5 ml. for gases.
Table
Type n-Parafins
RESULTS AND DISCUSSION
TO evaluate fully the applicability
Branched painfin3
of the results presented in both this and the previous investigation, i t was decided to investigate the effect of varying the following parameters on relative response: 1. Detector operating temperature 2. Concentration (linear response) 3. Carrier gas flow rats 4. Individual sensing unit
It was indicated b y Rosie and Grob (1) that relative response values were
Naphthenes
independent of temperature over the range of 85" to 125OC. At that time it was suggested that such a temperature independence would hold true over a much wider range. The present investigation supports this view, because the correlations presented below were made with data obtained a t arbitrarv Olefins temperatures covering a range from 30" to 160" C. No significant deviations from linearity were observed in relative response, when the concentration was changed tenfold (Table 11). Recent experiAromatics ments indicate that relative response is invariant over a 50,000-fold concentration range covering charges varying from 5 gaseous ml. to 0.1 gaseous ~ l . This is illustrated b y a coefficient of variation of 2.3 derived from the relative responses obtained with nine charge sizes covering this range, Relative response data are not only independent of the individual thermal Ketones sensing element used, but also of the type of thermal sensing element. Thermistors and hot wires give interchangeable data (Table 111). Table IV illustrates that relative Alcohols response is, ' i ithin ~ assumed limits, independent of flow rate over a range of from 33 to 120 mi. per minute. Further proof is exhibited b y the correlations discussed below, because these are based upon data obtained a t arbitrary flow rates. I n so far as thermal detectors have been shonn to be independent of variation in the parameters set forth previously i t is concluded that the data Acetate? presented in this report are applicable to all gas chromatographs using thermal detection and helium as the carrier gas. Therefore, relative response values may be applied without concern of the
I.
Relative Thermal Response Data
Relative Response per Mole" 36 51 65 85 105 123 143 160 177 199
Compound Methane Ethane Propane Butane Pentaneh Hexaneb Heptaneb Octaneb Nonaneh Decane* Isobutane Xeopentane 2,2-Dimethylbutam+ 2,3-Dimethylbutane5 2-Methylpentaneb 3-Methy lpentaneb 2,2-Dimethylpentane* 2,CDimethylpentaneb 2,2,3-Trimethylbutaneh 2,3-Dimethylpentaneb 2-Methvlhexane* 3-Methylhexaneb 3-Ethvlpentaneb 2,2,4-Trimethylpentaneb Cvclopentaneb hIethvlcyclopentane~
82
99 116 116 120 119
133
1,l-Dimethylcyclopentaneb
Ethylcvclopentaneb
1,2,4-Trimethylcyclopentaneb(cis, tranq, cis) 1,2,4-Trimethylcyclopentaneb(cis, cis, trans j
Cvclohexaneb Methylc yclohexaneb
1,l-Dimethylcyclohexane6 1,4-Dimethglcyclohexaneb(cis)
Ethylcyclohexaneb Ethylene Propylene Isobutylene 1-Butene trans-2-Butene cis-2-Butene 1,3-Butadiene Benzeneb Tolueneb Ethylbenzeneb o-Xyleneb m-Xvleneb p-Xyleneb Isopropylbenzencb n-Propvlbenzene5 p-Ethvltolueneb 1,2,4-Trimethylbenzene" 1,3,5-Trimethylbenzene' sec-Buty Ibenzeneb Acetone Methvl ethvl ketone Diethvl ketone 3,3-Dimethvl-2-butanone hiethvl n-amvl ketone Methi 1n-hexyl ketone Water Methanol Ethanol 1-Propanol 2-Propanol I-Butanol 2-Butanol 2-Methy1-2-propanol 3-Methvl-1-butanol 3-Pentanol 2-llIethyl-2-butsnol I-Hexanol 1-Heptanol Ethyl acetate Isopropvl acetate n-Butvl acetate n-Amvl acetate Isoamvl acetate n-Heptyl acetate (Continued on page $38)
VOL. 31, NO. 2,
129 129 135 136 133 131 147 97 115 124 126 136 143 114 120 141 146 145 48 G3 82 81
85 87
80 100 116
129 130 131 131 142 145 150
150 149 158 86 98
110 118
133 147 21 55 72 83 85 95
97 96 107 109
106
118
128
111 121 138
146
145 1'io
FEBRUARY 1959
231
Table I.
Relative Thermal Response Data (Continued)
Relative Response Type
Compound Diethyl ether Diisopropyl ether Di-n-propyl ether Ethyl n-butyl ether Di-n-butyl ether Di-n-amyl ether Inorganic gases Argon S i t rogen Oxygen Carbon dioxide Carbon monoxide -411 values measured relative to benzene (benzene = 100.0). Values reported hy Rosie and Grob ( 1 ) .
per Mole' 110
Ethers
1 43 1 25
Relative Molar Response 131 129 130
Table V.
Relative Thermistor Response 54
48 $2
Response
=
A
+ B (molecular weight)
nherc A represents the intercept and B is the slope of the line. This correlation \vas found to hold n-ith the classes of compounds listed in Table V > n here the data have been further suiiiniarized in slope-intercept form. It is probable that the type of correlation demonstrated above ~ i l l include other classes of compounds such as cycloparaffins when sufficient data become available. The l o w s t
86 8\i
163
Pro-
64 66 65 64
-I D
120
B , Slone
c1-c3 CI-ClO
Relative llolar Response
60
Range of Experimental Points
c,-c: SIethX 1 paraffins c j-C; Dimethx 1 paraffins C-C4 a-Olefins Cy-C* Trimethvl pztraffins C;-Cg ?\lethi1benzenPe Xono-n-alkx 1 benzence ci-c 9 llono-see-alkyl benzenes Cs-cia n-Ketones c3-C8 C2-c, Primarv alcohols Ca-Cj Tertiarv alcohols Secondary alcoholq c,-c!, C2-G n-Acetates n-E t hers CI-ClO Determined by method of least squares. Fraction of variation in relative b Correlation. veight.
55
85 85 160
Constants" for the General Equation Relative Response = A (Molecular Weight)
Sprire n-Pai affins
Relative Hot-Wire Response
Relative Response of pane vs. Flow Rate
F l o Rate, ~ 3ll./Min. 33
44
42 10
131
Table 111. Relative Response of Thermistor and Hot-wire Detectors
Table IV.
183
131
0 600 0 595 0 140
Compound Methanol Acetone Isopropyl alcohol n-Butyl ether
160
four parameters discussed abox e, so long as the operating conditionq remain constant for a given run. An examination of the relatire response data collected on various h>drocarbons. reported earlier, shon ed that the molar response increases remlarly
Table 11. Relative Response of Diisopropyl Ether over a Tenfold Concentration Range
Mole Ratio Ether /Benzene
130 131 130
with iiioleculnr weight, as shown in Figure 1. rllthough branching seemed at. first to cause appreciable inconsistencies, subdividing the hydrocarbons into groups varying in the degree of branching allel-iat'ed this difficultmy. This is shown for both paraffins and aromatics in Figure 2. -4straight line is obtained d i e n the response of a st'ructurally similar homologous series is plotted against molecular weight. It ran be expressed by the simple equation:
1.25 1.20
d.IntereeDt 20.6 6.7
1.20
1.16 1 16
1.06 1 04 0 861 0.808 0.808 0.857 0.841 0 886
+6 I P 0.06'2 0.992 0.999 0.998 0.998 0.999 0 999 0.999 0.999 0.999 0 nso 0 999 0.997 0.997 0,904
10.8
13.0 13.0 13 9
9 7 17 9 18.1 35,9 34.9 34.9
33.6 37.1
43.3
response due to variation in molecular
130 0
g N-No7' - 120
120 -
c
I
9
s
5
100
/*
-
/*-
80N-PENTANE.
3
/*
-
W
5 110-
N-°CTANE N-uEPTANE
N W z
m 100 -
E XAN E
v) W
$ 90v)
a W 40
I
MOLAR
I
I
"ESPONSE (BENZENE=10001
Figure 1 . Relative molar thermal conductivity cell response of n-paraffins vs. molecular weight
232
/
*PROPANE
60
ANALYTICAL CHEMISTRY
Figure 2. aromatics
50
,
60
I
1 70 80 MOLECULAR WEIGHT
90
100
Branching effects response of paraffins and
members-e.g., methane, ethane, and methanol-deviate appreciably from the correlation line. On the basis of these correlations, it is non possible to predict, b y a simple calculation, the relative response of coiiipouiids heretofore unknon 11. The striictural features-Le., degree of
branching-must be considered when such a n extrapolation or interpolation is used. ACKNOWLEDGMENT
The authors \\auld like to thank .D.' f Winters for hi5 assistance in the experimental work.
LITERATURE CITED
(1) Rosie, D. 11,
R, L,,
C H E ~29,1263 I. (1957). RECEIVEDfor review Xarch 11, 1958. Accepted September 8, 19%. Conference on Analytical pittsburgh, Chemistry and Spectroscopy, pa., Applied l \ ~ ~ ~ ~ h 1958.
Analysis of Mixtures of Sulfuric Acid with Other Acids by Nonaqueous Titration MlHlR NATH DAS1 and DEBABRATA MUKHERJEE2 Indian Associafion for the Culfivation o f Science, Jadavpur, Calcutta 32, India
b Sulfuric acid can be potentiometrically titrated as both a mono- and dibasic acid in glycolic media with sodium hydroxide or an organic base like piperidine as the titrant. Conductometric titration of sulfuric acid with piperidine in ethylene glycol or a glycol-acetone (2 to 1) mixture also gives two breaks in the titration curve. The potentiometric titration has been utilized for quantitative estimation of several acid mixtures containing sulfuric acid as a common constituent. These include binary mixtures of this acid with nitric, hydrochloric, perchloric, p-toluenesulfonic, phosphoric, salicylic, and acetic acids. Some ternary mixtures have also been titrated. The method has been used for studying the reaction kinetics of sulfonation of cresols, and should be applicable to analysis of acid mixtures encountered in other organic reactions, such as sulfation and nitration.
MUG RE^ n-ith sulfuric acid as a cominon constituent are frequently encountered in the course of many organic reactions. The rwction mixtures obtained during sulfonation. sulfation, nitration and estc.rification are typical and common examples of such mixtures. These mixture< cannot be analyzed b y direct acid-base titration in watc>r. Sulfuric acid. when titrated potrntiometrically in water, gives only one inflection corresponding to the total aridity. Mixtures of sulfuric acid with other mineral acids or common organic acids behave similarly> and hence can not be analyzed by differentiating titration in n ater.
ARIOL-Y ACID
Present address. Jsdavpur University,
Calciitta 32, India.
Present address, Central Leather Resrnrch Institute, 3Indrae 20, India.
Sumerous cases, however, demonstrate t h a t it is often possible to achieve differentiating titration of different types of acid mixtures in nonaqueous or even semiaqueous media. Izmailor and coworkers (fa,I S ) determined binary mixtures of hydrochloric acid with monochloro-, dichloro-, and trichloroacetic acids, respectively, by differentiating titration in ketonic solvents. Shkodin and Izmailov (bo) titrated mixtures of pprchloric-hydrochloric, perchloric-sulfuric, and p-toluene-sulfonicnitric acid mixtures in glacial acetic medium, using pyridine or dimethylaniline as the base. Evans and Davenport 16, 7 ) used butyl alcohol and water-butyl alcohol mixtures to differentiate picric-benzoic and hydrochloric-stearic acids. Acetonitrile was used b y Lavine and Toennies (16) as the medium for the titration of a mixture of perchloric and acetic acids. Critchfield and Johnson (2) used the same solvent for sulfurichydrochloric and sulfuric-nitric acid mixtures, with morpholine as the titrant. Higuchi and Rehm (11) analyzed several acid mixtures b y conductometric titration in glacial acetic acid, and were able to differentiate between the two hydrogen atoms in sulfuric acid. Moss, Elliott, and Hall (17) used ethylenediamine as the titration medium and could differentiate mineral or carboxylic acids from phenols. Dimethyl formamide and pyridine have since been used (5,5, 9, 15) for the differentiation of all these types of acids. Ketonic solvents ha\ e been found to be the most effective differentiating media for the titration of acid mixtures. Using methyl isobutyl ketone, Bruss and Wyld ( 1 ) obtained as many as five inflections when titrating a mixture of perchloric. salicylic, and acetic acids and phenol. Fritz and Yama n u m (8) used acetone for several
acid mixtures. Harlow and Wyld ( I O ) studied the influence of sokents on the resolution of acids by potentiometric titration 11ith qiiatrriiary animonium titrants. Palit (18) reported that. nhen a solvent niivture like ethylene glycolisopropyl alcohol is used as the titration medium, t n o inflections can be obtained -:1 titrating sulfuric acid nit11 sodium hydroxide. This is perhaps the first case reported of the titr a t ion ' of sulfuric. acid a. both a mono- and dibasic acid. Palit also observed that. in the same solvent medium, t h r w infleetions are obtained for a mixture of sulfuric and acetic acids. Kalidas and Das ( 1 4 ) have used ethylene g l p o l isoprop! 1 alcohol as the solvent for several acid mixtures. As glycolic solvents h a r e been found to possess certain advantagrs over many other organic solvents as titration media ( I $ ) , the present investigations were undertaken to find the possibilities of utilizing this method for quantitative analyses of acid mixtures containing sulfuric acid as a common constituent. REAGENTS
The standard alkali solution v\as prepared by rapidly washing beads of sodium hydroxide with distilled water three or four times to remove the superficial coating of carbonate, and then dissolving in a rei-\- small amount of n ater in a stoppered bottle. The solution n-as first diluted with ethylene glycol and then with twice the 1-olunie of isopropyl alcohol. The strength of the solution n-as roughly determined and, if necessary, it \%-asfurther diluted with ethylene glycol-isopropyl alcohol (1t o 2) to make it nearly decinormal. The higher proportion of the alcohol in the mixed solvent helps t o reduce the viscosity of the solution and thus minimize the drainage error. The alkali solution was standardized against potassium acid phthalate disqolred in water, using VOL. 31, NO. 2, FEBRUARY 1959
233
