V O L U M E 2 5 , NO. 3, M A R C H 1 9 5 3
511
Table I. Spectrophotometric Determination of Molecular Extinction Coefficient a n d Wave Length of Maximum Absorption of 2,9-Dimethyl-4,7-diphenyl-l,lO-phenanthroline Cuprous Complex Cation Cu N a v e Length E x D ~ . Taken. Max. Absoru..
.
sb.
RIg.
1
0.06260 0.05392 0.05652 0.06332 0.01626 0,03042 0.04623 0.06069 0,08132
2
3
4
a 0
-k 9
479 479 479 479 479 479 479 479 479
Molecular Extinction Coefficient Found 13,980 13.830 14,060 14,240 14,650 14,720 13,880 14,190 13,910 Av. 14,200
Optical Densitv Found 1.373 1.175 1.255 1.415 0.376 0.705 1.010 1.355 1.780
Optical Density a f t e r 111 Hours' Aging S o t determined N o t determined Not determined Not determined 0.344 0.665 0.966 1.295 1.708
Table 11. Determination of Copper i n Iron
The results are shown in Table 11.
(0.250 gram containing 0.04 mg. of Cu)
0.0188 0.0260 0.0260 0.0292 0.0328 0.0380 0.0456 0.0516 0.0016b
0.047 0.065 0.065 0.073 0.082 0.095 0.114 0.129 0.004
0.047 0.062 0.071 0.072 0.081 0.094 0.113 0.124 0.004
0.0188 0.0248 0.0284 0.0288 0.0324 0.0376 0.0452 0.0496 0.0016
Interference by Other Cations. The presence of commonly occurring ions such as chloride, nitrate, perchlorate, and phosphate as well as sulfate and citrate does not interfere. The influence of commonly occurring metal ions was not studied, as in general the phenanthroline-type complexing agents have been shown to be practically free from such interferences. Because the new type of cuproine reaction described involves extraction of the color complex, the presence of colored metal cations would not be expected to interfere. The use of the new reagent for special applications involving copper determinations in biological, medicinal, and food products requires special investigations.
AO. 000 -0.003 k0.006 -0.001 -0.001 -0.001
-0.001 -0.005
+o.ooo Av.
acid and evaporated to a volume of a fen milliliters. Citric acid (1.5 grams in 10 nil. of water) nas added. The acid was then neutralized with ammonia to a pH of approximately 5, employing the green color of ferric citrate as a suitable pH adjustment indicator. JVeighed portions of standard copper solution were then added and the samples mere transferred t o 60-ml. separatory funnels and diluted to 25 ml. One milliliter of 0.01 M complexing reagent solution in n-hexyl alcohol was then added and 5 ml. of hydroxylamine solution (0.05 gram per ml. in water) completed the reaction mixture. The solutions were shaken for 2 minutes, and allowed to separate into two phases for 5 minutes. The lower aqueous layer was drawn o f f and the alcohol layer washed with 5 ml. of ammonium acetate solution plus 1 ml. of hydroxylamine solution. The aqueous lower layer was drawn off and the alcohol layer transferred quantitatively t o a 25-ml. graduated flask and diluted to volume with n-hexyl alcohol. The optical density of these solutions was determined a t 479 mp and the copper determined by reference to the calibration data.
0,002
LITERATURE CITED
cations are not entirely free from the effects of air oxidation (0.05% per hour under ordinary laboratory conditions).
(1) Breckenridge, Lewis, and Quick, Can. J . Research, B17, 258 11939). (2) Case, Francis, Temple University, Philadelphia, unpublished
data.
DETERMINATION OF COPPER IN IRON
(3) Hoste, A n d . Chim. Acta., 4 , 2 3 (1950).
Procedure. Synthetic samples of copper in iron were prepared by adding weighed amounts of the standard copper solution to 0.250-gram samples of iron. The copper content of the iron had previously been determined from spectrographic examination (6). The samples of iron were dissolved in excess hydrochloric
**'
and 779 (1952). 21v 1313 (1949). ( 5 ) Smith and Brandt*ANAL. 24p3i1 (1952). (6) Smith and h4cCurdy* (7) "cCurdyt and Diehl, Anazysti 77* 418 (1952). (*)
RECEIVED for review August 15, 1 9 ~ 2 . Accepted October 20, 1952.
Estimation of Fructose by the Cyanohydrin Reaction R. D. COOMBS 111, Massachusetts Institute of Technology, Cambridge, Mass.; A. R. REID AND C. B. PURVES, McGill University and Pulp and Paper Research Institute of Canada, Montreal, Canada.
be satisfactorily estimated by measuring the ammonia expelled in the cyanohydrin reaction, T IS known that glucose and other aldose sugars can
KCN RiRZCO
--+
2H20
RiR&(OH)CN
--+
RiRzC(0H)COOH
+ NHB
and that neither the concentration nor alkalinity of the aqueous potassium cyanide is critical provided sufficient time is allowed. The present article shows that fructose, and presumably other keto sugars, give erroneously high results with 0.3 S potassium cyanide unless the pH is a t least 11. The condensation proceeds normally, although slowly, within the range p H 8 to pH 11 when the concentration of the cyanide is 0.03 ;V or less. Frampton, Foley, Smith and Malone ( 1 ) recently described an estimation for aldose sugars that was based upon the cyanchydrin reaction. This estimation (hereafter called Method A) involved condensing the sample a t pH 8.5 and 39" C. for 3 hours with an excess of approximately 0.3 .V potassium cyanide. Unreacted hydrogen cyanide was then expelled from the acidified liquor, and ammonia was quantitatively recovered by steam distillation after the addition of an excess of 20% sodium hydroxide. A blank containing no sugar revealed the amount of ammonia contributed by the original cyanide solution. Al-
though Method h gave very good results with aldoses, those for the keto sugars, fructose and sorbose, TT-ere 110% to 196% of theory depending on the time allowed for the condensation. Since a closely similar method (hereafter called Method B) was found in this laboratory to he satisfactory for fructose as well as for aldoses, it became of importance to investigate both methods in greater detail. METHOD B
A solution of 0.101 gram of pure fructose (0.561 millimole) in 100 ml. of freshly prepared 0.00982 S potassium cyanide (75% excess), buffered to p H 8.0 with acetic acid, was kept for 48 hours a t 45' C. The pH of the solution was then adjusted to 11.5 with S sodium hydroxide. This adjustment was necessary to complete the expulsion of ammonia in the subsequent distillation. When more concentrated cyanide solution was used, the volume was increased a t this stage to 100 ml. with distilled water. The liquid was distilled through a Liebig condenser into a graduate cylinder containing 100 ml. of 0.01 S sulfuric acid, a t a rate that was strictly controlled, preferably by an electrical heating mantle operating a t constant voltage. After about 25 minutes, when 50 ml. had been collected, the contents of the receiver was titrated with standard alkali to a methyl red or methyl purple end point in order to determine ammonia. If w, z,y, z millimoles represented the original cyanide, the original fructose, the ammonia
512
ANALYTICAL CHEMISTRY
recovered in the estimation, and the aniinonia recovered in a blank containing no fructose, respectively, then z = 21
- z(w
- z)/w,or z = w(y
- z)/(w
-
Table I.
Z)
These equations expressed the fact that only the unused poition of the cyanide contributed to the blank in the actual estiniation. I n the experiment quoted, w, y, and z were 0.982, 0.591 and 0.056 millimole, respectively, leading to a value of 0.568 millimole, or 101% of theory for the fructose, z. Duplicate estimations gave recoveries of 1007, and lOl$&. The fraction of cyanide directly hydrolyzed to ammonia in the estimation was not greatly affected by moderate changes in the concentration or by changes in pH between 7.5 and 11.5, but was markedly influenced by change in the rate of the distillation. When the temperature or the time of the c anohydrin condensation was increased, the data altered in sue{ a way as to leave the result almost unchanged, provided, of course, that the conditions were adequate for complete reaction. The condensation of 10 mg. of glucose (0.056 millimole) with 0.098 millimole of sodium cyanide ( i 5 q excess) in 25 ml. of water a t pH 8 and 45“ C., for example, nas only 32% complete after 24 hours, and 60% complete after 70 hours The method was therefore not adapted to the microscale. JVhen 25 ml. of 0.3 N potassium cyanide was distilled a t p H 11.5 after having been kept a t p H 8 for i o hours, the blank was 0.082 millimole of ammonia, but was only 0.069 millimole when 0.3 gram of purified cotton cellulose was present; after about 43 hours at pH 10, the figures were 0.09 and 0.056 millimole, respectively. These and similar results showed that the presence of cellulose tended to diminish the hydrolysis of the cyanide. The unsuitability of the method on the microscale and uncertainty about the blank probably explained why attempts to estimate the number-average degree of polymerization of hydrocelluloses gave very high results. Method A perhaps suffered from the same defect, because the viscosity averages quoted in Figure 4 of reference ( 1 ) were less than, instead of being greater than, thc cyanohydrin number averages. Method B differed from Method A in that the esccss cyanide was not removed prior to the distillation, but this difference was immaterial when 0.3 ;L’ potassium cyanide was used a t pH 8, 10, or 11.2 with fructose (Table I). Both methods gave an excessive recovery of ammonia at pH 8, and, to a lesser extent, a t p H 10, but Method A was correct a t pH 11.2. When more dilute, 0.03 M , potassium cyanide was used, Method B gave recoveries within 5 3 % of theory, both a t pH 8 and pH 11, and presumably Method A would behave in a similar way. Yundt (3’))who employed 0.11 N potassium cyanide a t pH 8.3 for the estimation of fructose by noting the amount of cyanide consumed
Estimation of Fructose by Condensation with Potassium Cyanide
(Condensation Fructose, JIetliod Millimole Hours With 0.30 M KCNb A 1.019 70 0.601 70 B 0.979 70 0.688
70
B
1.065 0.679
06 96
A
0.926 0.853
42 42
B
0.934 0.712
44 44
A
1.023 0.677
75
With 0.03 ,M K C N
B
0.561 0.474
7.5
a t 29OC.) pH
8 8 8 8 8 8 10 10 10 10 11.2 11.2
h-Hs Recorercd Millimolesa % theory 1.449 0,823
142 137
1.301 0.903
133 131
1.431 0,905
134 133
1.011 0.915
107 109
0.998 0.751
107 106
1.018 0.694
99 103
8 0.573 102 8 0.488 103 B 0.592 96 10 0.571 100 0.431 96 10 0.429 97 Alethod A, corrected by simplc subtraction of blank; Method B, blank prouortioned t o unused cyanide b y formula in text. b The times a t this concentration, excessive for fructose, happened to be similar t o those used in parallel estimations with oxycelluloses. Other experiments employed 48 hours at 45‘ C. 96 96
rathcr than the ammonia produced, also obtained results high by 25%. Militzer’s ( 2 ) similar estimation was a t pH 9, and fructose rcacted rapidly and completely. The above ohservations suggested that Yundt’s conditions were unsuited to fructose, although well chosen for the estimation of aldoses. KO explanation of the curious behavior of the keto sugar appears to be available a t the present time. LITERATURE CITED (1) F i a m y t o n , V.
I,.,Foley, L. P., S m i t h , L. L., and Malotic, J. G.,
AXIL. CHEM.,23, 1244 (1951). ( 2 ) Rlilitaer, W., Arch. B i o c h e ? ? ~9, , 91 (1946); 21, 143 (1949). (3) Yundt, 9. P., T a p p i , 34, 95 (1951). RECEIVED for review September 29, 1952. Accepted November 12, 1952. Method B was abstracted from a B.S. thesis submitted b y R. D. Coombs I11 in M a y 1941. t o the Massachusetts Institute of Technology-. T h e comparison rvith Method A was cairied out in Montreal.
Correlations of Infrared Spectra with Structure
‘
JOSEPII BOMSTEIN, Sinclair Research Laboratories, Inc., Harcey, I l l .
and extended corre1:ttioiis for infrared spectra. are presented, based on spectrograms of the .4merican Petroleum Institute; the Xational Advisory Committee for Aeronautics: “Infrared Spectroscopy,” by Barnes, Gore. Liddel, and Williams; and unpublished work of this laboratory. Among relationships developed are spectra-structure patterns for methylene chains, para-disubstituted aromatics, and disubstituted :ironlatics generally. Many v orkers in recwit years have denionstr:ttcd the usefulness of the infrared instrument as an aid in identification. The practical application of the method depends on the fact that many chemical structures have absorption maxima a t approximately the same wave length, regardless of the molecule with which they are associated. In order to extend the range of applicability, empirical studies wcre made of existing data in fields of interest to this laboratory, and the data given herein are the first results of such studies. Since the data were collected from a variet) of sources, instrumentation is of several types. Rather than enumerate all types employed, it suffices to bear in mind that data from several sources must be used judiciously since different instrument@and operating conditions can produce n idely varying results. For this reason, EW
all nunierical values given in this paper should be considcrcd as rough approximations, not as correct absolute values. Data from this laboratory \%ereobtained either with a Model 12-B or Model 12-C Perkin-Elmer recording infrared spectrometer, equipped xvith rock salt prism and cells. SPECTROGR4JI SOURCES
Data for the compounds listed in Table I were obtained from spectra published by the S.rl.C.A. (8). In Table I1 the data for the sulfonates, phenols, p-di-tert-butylbenzene, p-tertbutyltoluene, and p-xylene n-ere obtained in the Sinclair Research Laboratories, and all others N ere taken from the catalog issued by the American Petroleum Institute ( 1 ) . I n Table I11 data for xylene and cresol were obtained in the Sinclair Research Laboratories; data for other hydrocarbons were taken from ( 1 ) ; all other data were obtained from ( 2 ) . I n each case mentioned in Table 111, data for the three isomers of one molecule were dran n from a single source. ABSORPTION POSITIONS FOR METHYLENE CHAI\S
Recent work of McPIIurry and Thornton (6) has shown thnt a definite correlation exists betn een position of absorption and the