Table 11.
Sumher 1
2
Determination of Copper and Zinc in Brass
Present, % Sample Copper Zinc SInthetic" 60 05 36 54 Studentstandardb 59 49 38 65
3
Found, % Copper Zinc 60 18 36 54 59 62 37 90 59 24 38 78 59 58 38 52
Difference, c& Copper Zinc $0 13 0 00 $0 13 -0 7 5 -0 25 $0 13 + O 09 -0 13
Silicon bronze, -0 03 90 83 2 05 90 86 2 07 SBS 158 4 dluminum biass, -0 56 63 20 21 83 63 76 21 89 XBS 16-1 a Similar t o 2. b Thorn-Smith student sample contained A1 l . 1 5 ~ cPb , O.lO~c, and traces of Sn and Xi. organic, solvclnts, tht. folloning results werr obtaincd : Solvent Methanol Ethyl alcohol Isopropyl alcohol 1,4-Dioxane S,S-Dimethvlforniamide
Copper Found, Mmole 0 2504 0 2502 0 2498 0 2500 0 2499
These solvents were equally effective in improving the rnd point for the direct titration of copper. The direct titration is simpler. Hon-ever, the back-titration technique is cssrntial for titrating cobalt, lead, or bismuth with the dye as indicator. It is not necessary to add the organic
-0 03 -0 06
solvrnt for titrating zinc directly in the absence of copper or if the copper is complexed by thiosulfate. The masking of copper has bern similarly applied to the titration of the mixtures of cadmium and copper, and nickel and copper. The difficulty of obtaining a sharp end point for the direct titration of nickel with E D T A with the dye as indicator was overcome b y heating the solution. The optimum pH range and trniperature for the titration of nickel were investigated (Figure 1). Below 50" C.. higher results a t p H 3 or4indicate that much rxcrss E D T A was needed to wqurster nickel from thr stable nickeldye complex a t l o r e r temperatures. Above 60" C., low results a t pH 3
indicate that the cnd point \vas rraclicd too soon, because the stability of nickeldye complex n as decreased by both lo^ acidity and high temperature. To denionstrate the proposed procedure, an attempt w s made to determine copper and zinc in brass. Satisfactory results n e r r obtained (Tahle 11). Silver and uranium formed strongw complexes n-ith the dye than v i t h E D T A in the alkaline medium. They did not interfere in slightly acid nicdliim. The use of t h r dye a' a ,ensitiic ant1 selective method for uranium IT ill 1)c covered in a scqiarate report. LITERATURE CITED
(1) Cheng, I-, R. H., A s . 4 1 . . CHEX 27, 782 (1955). (2) Cheng, K. I,., Killiama, T. R., Jr., Cheniist-dnal?jst 44, 96 (1965).
(3) Flaschka, H., .4bdine, H., Ibid., 45, 2 (1956). 14) Ibid..D. 58. (5) FlascGka, H., Abdine, H., J I i k r o c h i u ~ . Acta 1956, 770. (6) Schn-arzenbach, G., Freitag, 1: , Helv. Chinz. Acta 34, 1503 (1951). RECEIVED for revien March 12, 1957. Accepted September 26, 1967. Presented in part before the Pittsburgh Conference on -4nalytical Chemiqtri- anti Applied Spectroscopv, March 1957. Preliminary ivork carried out a t Chemical Tahoratory, Westinghouse Electric Corp., East Pittsbnrgh, Pa.
Determination of Alkyl pyridines by Infrared Spectroscopy Rapid Methods of Analysis ROBERT L. BOHON' and RAYMOND ISAAC The Anderson Physical laboratory,
609 South Sixth St., Champaign, 111.
HENRY HOFTIEZER2 and ROBERT J. ZELLNER The Ansul Chemical Co., Marinetfe, Wis.
b Analyses are described for a number of synthetically produced alkylpyridines, utilizing both the sodium chloride and potassium bromide regions o f the infrared spectrum. These rapid, flexible techniques are useful for analyzing a limited number of samples of widely varying composition. High-concentration components are successfully determined without the customary solvent-dilution methods in several cases b y using thin cells, appropriate standards, and the base-line method o f measuring 10. Differential spectra are used for impurity detection and identification, as well as for trace analyses in highly purified pyridines.
I
spectroscopy lend3 itself to the analysis of many complex mixtures, and the pyridines are no exception. Wet chemical methods are time-consuming and nonspecific in most cases (1, 17). The infrared method is rapid, specific, and accurate, and enables the analyst to check for the presence of unsuspected impurities. This last property is invaluable when TT orking with samples from the laboratory or pilot plant where evperimental conditions, and hence product composition, vary greatly and unpredictably. Processes for producing alkylpyridines (2, 3, 19, 20) involve reaction a t elevated temperatures of saturated XFRARED
aldehydes or ketones with ammonia-in the presence of a suitable catalyst. Variation of reactants and experimental conditions permits product'ion of mixtures rich in specific nlkylpyridines, such as t8he picoline^. lutidines. or collidines, and fractional distillation is generally sufficient for isolation of t'lie desired components in a state of high purity. This obviat'es costlj- crystallization (14) or clirornatographic (18, 21) separations. Previous infrared determinations for
PE
Present address, Llinnesota Mining
Mfg. Go., St. Pan1 6, SIinn.
* Present address, IIarathon C o r p , Rot,hschild, \Vis. VOL. 30, NO. 2, FEBRUARY 1958
245
pyridines have referred to fractions obtained from coal tar (8-11, 86) or oil shales ( 7 ) . They involved the usual procedures of dilution with solvents before analysis, use of successive approximations ( I S ) t o compensate for background absorption, etc. I n the present case, many of these time-consuming, and sometimes error-promotingJ steps have been eliminated. The determinations presented are rapid, reproducible, flexible] and sufficiently accurate for most purposes. The techniques should be applicable to many systems other than the pyridines. The potassium bromide region (15 t o 25 microns) was extremely useful. Differential methods in the sodium chloride region were valuable in identification and determination of certain trace impurities. REFERENCE COMPOUNDS
The reference materials and their methods of purification are listed in Table I. Spectra on these substances are included in Figure 1. The acid insolubles were a mixture of compounds produced in the catalytic reactor which were insoluble in aqueous hydrochloric acid. They were largely aromatic;hydrocarbon by-products and varied in composition with reactor conditions and feed. ANALYTICAL METHODS
Most of the spectra and analyses in the 2.5- to 15-micron range were obtained on a Perkin-Elmer Alodel 21 Spectrophotometer. The 15- to 25micron range mas covered with a PerkinElmer single-beam, double-pass spectrometer. Cells were of the permanent type with the exception of a variable space-liquid cell (Research & Industrial Instrument Co.), which was used in the
reference beam of the Model 21 for most differential recordings. Figure 1 is a line chart of the principal absorption bands for a number of alkylpyridines and benzenes. Each set is arranged in order of increasing boiling point. Relative band intensities and data sources are indicated. Where more than one reference is given, the data were compared and the most reliable used. All spectra reported in the potassium bromide region for the pyridines were determined during the present study. Boiling points are from literature. The first table in reference (7) contains a few characteristic and useful absorption peaks for some pyridines not included in the line chart. As fractional distillation is the customary final purification in the industrial processing of pyridines, only those compounds with boiling points in the same vicinity as the major ingredient constitute serious impurities (neglecting azeotropes). Reference to the line chart shows what spectral interferences might be expected. Infrared curves were unavailable for those alkylpyridines not included in the chart. I n setting up a n analytical procedure] the spectrum was scanned on a few typical samples. and the impurity bands were noted. The band (or bands) that seemed most promising under the circumstances was chosen for analysis. When more than one component was to be determined] this consideration often determined the selection. Such factors as impurity interference and Beer’s law adherence also affected the choice of band (examples given below). The use of solvents was avoided wherever possible to reduce analysis time, volumetric errors, and excessive handling of the samples. This last
Table 1.
Purification Method or Source Compound 2-Picoline (2-methylpyridine) Vogel (27) 4-Picoline (4-methylpyridine) Vogel(d7) Vogel (27) 8-Collidine (3-ethyl-4 methylpyridine) 5-Ethyl-2-picoline (5-ethyl- Vogel (27) 2-methylp yridine )
Reference Compounds
B.P., C. (Obsd.) 128.5 144.0 196.0
nF ZnClp 1.4995 165-8 1.5059 166.5-7.5” 1.5079 146-7
176-176.5
1.4968
3-Picoline (3-methylpyridine) Reithof and associates ( 2 3 ) 143.5-144.0 170.0 3,5-Lutidine (3,5-dimethyl- Reithof and associates (23) pyridine) Brown and Murphey, 163-4 3-Ethylpyridine Method B (6) Brown and Murphey, 165-6 4-E thylpyridine h4ethod B (6) NaOH. Steam 142.0-.5 Picrate 2,6-Lutidine (2,6-dimethyldistillation. fractional pyridine) distillation 167.0 s-Collidine (2,4,6-trimethyl- Engel (12) pyridine) 163-6 Eastman, White Label Mesitylene 115.2 + 1 . 0 Mallinckrodt, reagent Pyridine urified grade ... Pyrrole ... Bis(2-methoxyethyl) ether ?%f&emical Co., E-141 ... 4-Picoline preparation Acid-insolubles e A second crystalline form melts at 120-130’ C. 4-Picoline regenerated
+
246
ANALYTICAL CHEMISTRY
item can be very important with the hygroscopic pyridines. \\here a very strong absorption band seemed beste.g., 3,5-lutidine-a solvent was used to avoid handling liquids in cells less than 0.02 mm. in length. If a nonpolar solvent such as carbon tetrachloride were used, traces of water in the unknowns often would form emulsions which scattered visible light badly. Unknowns and standards were dried with sodium hydroxide, silica gel, or metallic sodium just before use, but occasionally the water content of an unknown would exceed the drying agent’s capacity. High background absorption was a sign of insufficient drying. The series of glycol ethers manufactured by Ansul Chemical Co. were ideally suited for analyses from 860 to 650 cm.-* They are transparent in this region and miscible with both polar and nonpolar compounds. The absorption maxima of some of the pyridines were found to shift a few wave numbers when dissolved in the ethers. Vials with aluminum-liner screw caps were satisfactory for samples and standards if filled more than half full of liquid and fresh desiccant added. It was usually impossible to use standards more than once after opening, because of the rapid water uptake from air. Fresh standards were prepared the day before use and desiccated overnight. The “base-line” technique of measuring lo/l (16) was found to be essential for correction of both the background effect of absorbed atmospheric water and the unpredictable presence of impurities. The base-line method compensated well for most of these interferences whereas cell-inJ cell-out methods did not. An exception was found when an unsuspected impurity
...
1.5059 1,5055 212.0-5 1.5026
...
Salt, M.P., ’ C. Picrate Hg2C12 HnPtClz AuCli
...
... ...
...
124.5
...
238:O
... ...
... ...
... ... ... ... ...
195(d)
...
131.0-.5 130-1
... ...
..
... ... ...
...
1.5020
...
164-5
147.0-.5
...
160.0-.5
... ...
...
1.4980
...
...
1.4980
...
...
...
...
... ... ... ...
...
... ...
...
...
...
...
... ...
... ...
...
...
...
... ...
...
...
... ... ...
... ... from the two forms had identical spectra.
... ... ... ...
(mesitylene) had a prominent absorption superimposed exactly upon the analytical band of the major ingredient (s-collidine) , The base-line method gave insufficient correction, and a n alternate band was used. Figure 2 illustrates the authors' method of drawing base lines, which they have arbitrarily called "humpto-hump." Khile this method requires mnie intuition, it is more generally applicable than the inflexible rule of choosing two set TT aye lengths between
which the base line is always drawn, regardless of circumstances. I n Figure 2a, no band exists a t 710 cm.-l, and hence the hump-to-hump base line is drawn from about 710 t o 740 cm.-l I n Figure 2b, a fairly strong 710 em.-' band is present, and the hump-tohump base line extends from about 715 to 735 cm.-l Figure 2c, illustrates a more typical sample in which both the 688- and 710-cm.-' impurity bands were present. Excellent Beer's law plots resulted
using this technique, in spite of deliberate addition of impurities absorbing near the analytical band. The interference can become so severe that other methods of correction must be employed. Constant slit widths were used for all quantitative work, except in the few cases where extremely long scans were made on the Model 21. Automatic programing of slits introduces some errors in reproducibility which are objectionable for precise quantitative
0.02 mm.
0.07mm KBr
NaCl
2,3,6-T r i m e t hy I py r id I n e
5 - E t hyl -2-Met hy I p y r i d i n e 3 - E t h y I - 4 - M e t h y l p y r i d i ne
- - - - - - - - - - - - - -- -
172.8
c
I74
o.b
190
I
I
I
a
1 I I I
p-Xylene
138.4 o , d
m-Xy lene
139.1 o , d
I
o-Xylene
144.4 o , d
I
Mesitylene
164.7 a , d
I
I I
II
I
I
I
I
I
I l l ,I
I
I(
4-Picoline
Process
105-35 a
I 1
0 0 0
I
I
- --- - - - - - - - - - - -- - -
1
--
- - - - - - -I
0 0 0 0 0 0 0 0 0 0 0 ~ 0 ~ c- c - ) N ~
~
NaCl 0 . 0 2 m m . Figure 1.
- --
Ill
1
e- a. _ _ - _- Doto _ unavailable
I
- - - - -- -- - -- - - - - - - - - - --
I
I
I
II
I
Acid lnsolubles f r o m t h e
- - - - - - - - -- - -
I
I
0 0 0
0
O
o
0
0 7 .
"","," C
r
-
KBr
Principal infrared absorption bands
Height of line indicates relative intensity of absorption Acid-insolubles intensities varied with b.p. a. Anderson Physical laboratory b. ( 2 5 )
e.
(6) API-NBS Project 44 (4)
f.
(22)
E.
d.
VOL. 30, NO. 2, FEBRUARY 1958
247
analyses, but nhich can be tolerated during combined qualitative and quantitative runs on unknowns-e.g., 4picoline. Analysis conditions are summarized in Table I1 and discussed below. The last three analyses here were set u p before the double-beam spectrophotometer was available. Major components exceeded TOPc in all samples.
5
L
..
.
..
.
.. .
.. .
.. .
..
.. .
...
...
.
-c
M
c
2
E
248
ANALYTICAL CHEMISTRY
Figure 2. Hump-to-hump base lines for determination of s-collidine and mesitylene in s-collidine 724 ern.-' 710 crn.-l 688 ern.-'
s-Collidine Unknown impurity (see text) Mesitylene
Analysis conditions must be flexible to meet the variability of samples from laboratory or pilot-plant operations. Therefore, the conditions given in Table I1 should serve only as a guide. 3-Picoline (3-Methylpyridine). As a precursor of niacin, 3-picoline must be supplied essentially free of its homolog, 4-picoline. Because of the almost identical boiling points of these two compounds, separation on a commercial scale is costly. I n contrast t o 3-picoline obtained from coal tar, synthetically produced 3-picoline contains practically no 4-picoline. Therefore, an extremely sensitive analytical method for detecting small amounts of the 4- isomer in 3-picoline must be established. The percentage of both 3- and 4picoline was determined directly, the difference from 100% representing "other" impurities. The sodium chloride region of the spectrum contains no powerful 4-picoline bands n-hich are
S3,10.04%3-PicoIine
1
I
1
Figure 3. Hump-to-hump base lines for determination of 4-picoline in 3-picoline
I sufficiently removed from strong 3picoline bands t o be useful for trace analysis. The 995- and 1212-ern.-* bands can be used with some success if the 4-picoline concentration exceeds about 2%, but this is not the case in most of the present samples. I n the potassium bromide region, the strong 515-crn.-' band of 4-picoline gave a good Beer's law plot. Synthetic standards were prepared from pure 3- and 4-picoline, and a quantitative procedure was set u p as outlined in Table 1Ij-i. 3-Picoline n-as determined simultaneously by esteiiding the scanning region to include bhe relatively weak 539-cm.-l band of this major constituent. Two high absorptivity bands of 3-picoline in the sodium chloride region (705 and 3 6 cni.-l) can be used advantageously for more precise determination of thin component. [A mean deviation of uliout &0.5% by weight of 3-picoline as obtained using the 705-cm.-1 bari~l. bis(2-methoxyethy1)ether (diethylene glycol dimethyl ether) as the solvent. and a cell 0.15 mm. in length.] Figure 3 illustrates t,he hump-tohump method of drawing base lines for this determination. Independent calculations by different individuals using their orrn versions of the hump-tohump base line produced negligible differences in the aiialytical results. Table 111 presents some analyses of laboratory-scale samples on which titration results (reported as 3-picoline) were arnilable. High water content interfered with some of these analyses, and higher precision was obtained on thoroughly dried samples. s-Collidine (2,4,6-Trimethylpyridine). From chemical a n d spectral analyses, it was established t h a t synthetically produced s-collidine contained mesitylene as its principal impurity. T h e strongest s-collidine band occurs a t 540 cm. --I; hiit mesitylene also
I
1200
1\00 cm:,
IO00
9 00
Figure 4. Long-scan, differential determination of 3-picoline in 4-picoline Pure 4-picoline in reference beam, cells ca. 0.07 mm. S3 was a standard (10.0470 3-picoline), No. 3 5 was derived from coal tar, and No. 4 4 was prepared synthetically. Four marked bands due to 3-picoline
has a powerful absorption at this same location. Becau-e the mesitylene content was of interest too, a dual analysis was chosen using the 724-cni.-' band of s-collidine and the much stronger 688-cm.-' band of mesitylene. By extending the scanning range from 750 to about 650 em.-', i t was possible to include the region around 710 cm.-l M here 3,5-lutidineJ 3-ethj-lpyridine, and 2,3,6-trimethylpyridine have strong absorptions (Figure 1). This procedure permitted a twocomponent quantitative analysis for s-collidine and mesitylene, as well as a semiquantitative indication of the possible impurities absorbing around 710 em.-' (Figure 2). Table II,B+gives the specific experimental conditions used. A differential technique (24) was employed with a Table 111.
Analysis
purified sample of s-collidine in the reference beam of the Model 21. A deliberate uncompensation with a slightly longer path length in the sample beam assured a downward pen deflection for all samples. The samples analyzed by this technique varied from about 82 t o 100% s-collidine. Mesitylene ranged from 0 to 10%. Cell lengths were kept low to cover the high percentage range of mesitylene, but much more accurate analyses are possible for s-collidine by using thicker cells and adequate compensation in the reference beam. Table I V illustrates an analysis on six samples. I n this particular run no compensating material was used in the reference beam. The synthetic s-collidine Kas free of 3,j-lutidine (no 711- or 860-cm.-'
of 4-Picoline in 3-Picoline
(Experimental Conditiona: Table I I , A . R u n made on different days. Values approximate)
sample
Wt. % 3-Picoline Wt, V0 4-Picoline Run 2 h v . Run 1 Run 2 Run 3 94.5 94.6 94.5 22P 0.5 0.5 ... 94.7 98.4 96.5 c u t 221 0.9 1.1 0.8 98.2 c u t 225s ... ... -0.2 98.2 0.1 Cut 226s 9 5 . 6 9 3 . 6 91 6 0.3 0.4 0.8 87.97 2.7 96.6 9 2 . 2 227s 1.1 0.9 98.3 Cut 228 0.3 97.9 9 8 . 1 0.3 0.6 99.9 99.9 Cut 229 ... 0.8 ... 1.3 Cut 23OU 97.0 . . . 97.0 1.6 ... 2.2 Mean deviation. 4-Picoline i 0.2y0 3-Piroline ~ k 1 . 5 7 ~ Seriously affected by water, particularly No. 227.
KO.
Run 1
.Iv. 0 5 0.9 0.0 0.5 1.6 0.4 1.o 1.9
Sum
Total Base (as 3Picoline)
by Av. Titration 95.0 90.0 97 4 93.0 98.2 97.5 95.1 99.7 93.8? 96.6 98.5 97.6 100.9 98.6 98.9 92.17
VOL. 30, NO. 2, FEBRUARY 1958
0
249
Table IV.
s-Collidine and Mesitylene Analysis
(Experimental Conditions
Sainplc
Kt.
Run 1
so.
246-1 246-2 246-3 246-4 246-5 246-6 Mean deviation.
Table V.
95 0 98 6 95 4 95 4 94 0 95 0
Table II,B, except reference beam was blank)
s-Collidine Run 2 Av. 95 2 95 1 98 0 94 8 96 1 91 5 95 0
98 3 95 1 95 8 94 3 95 0
0 io 0 20
0 70 0 10
0 io 0 15
su111
.IV. 95 5 98 4 95 2 96 1 95 0 95 1
s-Collidine 10.23% Mesitylene +O 0 2 5
Determination of 3,5-Lutidine
(Exlieriinental Conditions
Table I1,C)
Kt. cG 3,S-Lutidine SO. Run 1 Run 2 Av. i 89 1 89 1 89 1 8 96 0 96 8 96 4 9 96 1 96 1 96 1 10 93 7 93 i 93 7 Mean deviation. 1 0 1 %
Sample
bands), whereas this impurity is rather prominent in the commercially available samples used by Brown, Johnson, and Podall (b). 3,5-Lutidine (3,SDimethylpyridine). Analysis of this compound illustrates a case where i t was essential to dilute samples and standards with a solvent because of the intensity of the band chosen for analysis and the lack of satisfactory cells less than 0.02 mm. in length. Attempts t o use weaker bands without dilutione.g., 855 cm.-'--were only partially successful, because of severe impurity interferences which base-line methods could not fully correct. Samples containing over 90% 3,5lutidine were analyzed using the 715cm-1 band; the experimental conditions are described in Table I1,C. A low concentration standard was used in the reference beam for this differential procedure (24, and bis(2-methoxyethyl) ether served as the solvent. Table V lists a few typical analyses. Differential scans with pure 3 5 lutidine in the reference beam suggested that the principal impurities were trin-propylamine, 2,5-lutidine, and 2,3lutidine. 3-Ethylpyridine. which might be expected as an impurity, was absent in these fractionated samples. +Picoline (4-Methylpyridine). I n contrast to t h e detection of tiace amounts of 4-picoline in 3-picoline, there are several strong 3-picoline bands in the sodium chloride region which can be used for its trace determination in 4-picoline. T h e choice of rvhich band t o use will depend upon interferences encountered from other impurities. The "long-scan" technique described below allon s one a 250
Kt. 7cMesitylene Run 1 Run 2 Av. 0 40 0 40 0 40 0 10 0 10 0 10 0 10 0 10 0 10 0 30 0 25 0 20
ANALYTICAL CHEMISTRY
considerable degree of freedom in this respect-but with a corresponding loss in accuracy. Unkno\\ ns and several synthetic standards were scanned over most of the fingerprint region (1300 to 650 cm.-l) on the Model 21 with pure 4-picoline in the reference beam using autoniaticslit programming (Table 11,D). Because a wide variation in sample composition was encountered, slit errors were insignificant. There are a t least five 3-picoline bands within this region: 705, 1030, 1105, 1125, and 1192 cm.-l Generally one or more can be found which is free from obvious impurity interference and is, therefore, usable for quantitative measurements. Typical differential scans are shown in Figure 4. The four sharp 3-picoline bands in the standard S3 were readily apparent in a sample ( S o . 35) of 4picoline derived from coal tar and specified by the manufacturer as "pure." The redistilled, synthetic sample KO. 44 showed no sign of these bands under the same instrumental conditions. Sample 35 assayed 5.0% 3-picoline. The principal impurity band in the synthetic 4-picoline samples occurred a t 1000 cm.-' and is probably due to 2-ethylpyridine, which boils onlp 3" C. from 4-picoline. There was also evidence of some 2-picoline in some of these samples. The 1000-cm.-l inipurity band was readily isolated from the nearby 4-picoline absorptions by differential methods as shown in Figure 4 for a typical coal-tar sample. The 1 1 9 2 - ~ r n . -band ~ permitted detection of about 0.7% 3-picoline in 4picoline using the long-scan method and an 0.07-mni. cell. The detection limit can be improved a t least tenfold by using high instrumental sensitivity, 0.15-mm. cells, and pure 4-picoline in the reference beam. The purity of the reference 4-picoline becomes the limiting factor for absolute analysis IThen using the differential technique. Analysis of the major ingredient, 4picoline, can be carried out conveniently a t 800 or 515 em.-' A differential technique, such as described by Hammer and Roe (Ib), is recommended. 2-Picoline (2-Methylpyridine), 5Ethyl-2-picoline (5-Ethyl-2-methyl-
pyridine), and p-Collidine (3-Ethyl4-methylpyridine). T h e analytical methods used during the early stages of this project (reported below) utilized only a Perkin-Elmer single-beam spectrophotometer and the conventional solvent-dilution methods. Synthetic 2-picoline was determined using the intense 758-cm.-* band (Table 11,E). Bis(2-nietho~yethy1)ether served as the solyent. Mean deviation on samples containing 95 to 100% 2-picoline was about +0.5%. hletallic sodium can be used to remove the last traces of water from reference samples of 2-picoline if contact time is limited to a few hours a t room temperature. The 1 1 3 5 - ~ n i . - ~band of 5-ethy1-2picoline (aldehydine) was used for its analysis (Table 11,F). Both the 740and 8 2 8 - ~ m . - ~bands were unsatisfactory because of the severe interference from 6-collidine, \vhich was present in appreciable amounts before fractionation. The 1135 em.-' band gave a good Beer's law curve using chloroform as the solvent and a 1.0-mm. cell. The mean deviation was +0.3% on a series of samples ranging from 73 to 96% 5ethyl-2-picoline. Synthetic p-collidine was determined satisfactorily using the 1195-cni.-' band and carbon tetrachloride as the solvent (Table 11,G). Samples must be dry to prevent emulsions. Cells from 0.15 to 0.25 nim. gave good Beer's lam curves. AGING AND COLOR FORMATION
Many alkylpyridines develop a yellow color upon standing for several months in clear glass bottles, but this could be readily removed from aged samples of 4-picoline by redistilling. Direct comparisons were made of the infrared spectra before and after distillation. I n every case the redistilled samples had less background absorption than the aged material, particularly around 700 to 900 cm.-l, but no discrete bands below lGO0 cm.-' disappeared or weakened noticeably. Above 1600 ern.-' two bands decreased markedly, 1650 and 3400 em.-', and this was attributed to the removal of water. The high background absorption below 1600 em.-' of the yellowed samples also was assigned to water. (Merely adding water to an alkylpyridine does not cause yellowing per se.) Apparently the color-body concentrations vere too lo^ to influence the infrared spectrum under these typical instrumental conditions. A colorless sample of purified 4-picoline was irradiated in air kl-ith a watercooled mercury arc of high intensity (General Electric A-H6 with borosilicate glass envelope) for 18 minutes. This produced no visible coloration nor did it alter the infrared spectrum in any way.
It was concluded that color-botly formation as a result of aging would have no effect upon infrared analyses, although atmospheric water-pickup can be serious. DISCUSSION
The technique.. described for alkylpj ridines provide useful procedures for analysis of a limited number of samples of widely varying, and unknown, impurity content. Differential niethods are particularly useful for this sort of determination as I\ ell as for trace :tnalyses. Unexpected constituents, whicli absorb exactly a t the chosen analytical band, constitute the greatest hazard in any spectroscopic analysis method. The long-scan technique described for determining 3-picoline in 4-picoline (Table II,D) and charts like Figure 1 can aid the analyst greatly in detecting such situations. The advantages of base line techniques for measuring are many, particularly in the potassium bromide region where water depresses the background transmission severely. The hump-to-hump method of drawing base lines was found superior to choosing fixed n-ave lengths, betn-een which the base line was inexorably draTvn. When it failed, one could suspect such serious interference that any base-line method of correction would be insufficient. Vsing undiluted samples has not affected Beer’s law in the case of the pyridines, providing sufficiently thin cells are used and the concentration range is not too great. The elimination of solvent dilution has the threefold advantage of reducing water absorption through sample handling, reducing sample-preparation time, and eliniinating dilution errors. n7it11 strong absorption bands it might require the use
of very thin cells (
