V O L U M E 28, N O . 6, J U N E 1 9 5 6 4). With increasing concentrations of aluminum a significant increase in absorbance is observed, except for the 0-*( series. This behavior means that aluminum forms stable complexes with fluoride but not n i t h -1lizarin Red S to any extent a t the pH utilized. DISCUSSION
The thorium-Alizarin Red S reagent was originally introduced in this laboratory to determine comparatively large amounts of fluoride, around 500 y. The color reagent was similar to the one described here, except that twice the amounts of the thorium and the hlizarin Red S stock solutions were used. This reagent had to be made up fresh for each determination. As expected (8)>the precision was very good at the higher concentrations. The coefficient of variation was only O.2YGfor 10 samples of 400 y of fluoride, the readings being taken against a 500--i reference standard. The procedure described here has proved to be reliable even for amounts of fluoride of around 20 y. The stability of the reagent for a t least a month makes it convenient for laboratories engaged in routine determinations of fluoride, as it can be made up in large quantities. The main disadvantage of methods based on thorium-iilizarin Red S reagents is the great pH sensitivity. Furthermorc, the reagent is not as sensitive to small amounts of
953 fluoride, as is the thoron reagent, for example, proposed by Horton, Thomason, and Miller (6). ACKNOWLEDGMENT
The author gratefully acknowledges the valuable aid and assistance of Celia Gorl6n and Kirsten Matheson during the course of this investigation. Appreciation is expressed to A/S L d a l og Sunndal Verk for support of this study. LITERATURE CITED
Armstrong, W. D., IND.ENG.CHEM.,ANAL.ED. 8 , 384 (1936). Barr, G., Thorogood, A . L., Analyst 59, 378 (1934). Bumsted, H. E., Wells, J. C., ANAL.CHEM.24, 1595 (1952). Danielsen, M. E.,. Univ. Bergen &bok, h‘aturvitemkap. Rekke KO.9, 1-10 (1953). Horton, A. D., Thomason. F. D.. lliller, F. J., ANAL.CHEN. 24, 548 (1952). Hoskins, W. hl., Ferris, C. A., IND.ENG.CHEM..ANAL.ED. 8, 6 (1936). Icken, J. M., Blank, B. M., ANAL. CHEM.25, 1741 (1953). Lothe, J. J., Ibid., 27, 1546 (1955). Nommik, H., Acta Polytech., C h a . M e t . Ser. 3. No. 7, 7-121 (1953). Sanchis, J. hl., IND.ENG.CHEM.,ASAL. ED.6, 134 (1934). Talvitie, N. A,, Ibid., 15, 620 (1943). Thompson, T. G., Taylor, H. J., Ibid., 5. 87 (1933).
RECEIVED for review December
6, 1955.
Accepted March 16, 1956.
Carbon-Hydrogen Stretching Bands in High-Resolution Infrared Spectra of Heterocyclic Compounds W. H. TALLENT and IRIS J. SIEWERS National Heart Institute, National Institutes o f Health, Bethesda 14,
Wave lengths for carbon-hydrogen stretching bands found in high-resolution infrared spectra of a number of heterocyclic compounds are given. Comparison of these spectra with those for analogous hydrocarbons suggests that assignments for carbon-hydrogen stretching bands in hydrocarbons can be extended to heterocyclic compounds, with certain limitations. The detection and characterization of nonaromatic double bonds in compounds containing both aromatic rings and nonaromatic double bonds, by comparing spectra of these compounds with those of their dihydro derivatives, are illustrated.
D
URING the past fen- years infrared spectrophotometers with high resolving pon er in the 3- to 4micron region have become commercially available. The possibility of using C-H stretching bands found in this region for determining the nature and location of C=CH bonds, and for detecting the presence of three-membered carbocyclic rings, was recognized sometime ago. .4 number of papers and reviews contain structural assignments for bands in this region (1, 8, Q , Z O , d l ) ; of these the data pertaining to steroidal systems, as recently summarized by Jones and Herling (11)and by Cole ( 5 ) ,are perhaps the best knon-n. Much of the work of this laboratory is concerned with nitrogen-containing compounds, and in the course of a separate study on the structure of pinidine [a nen- alkaloid isolated from leaves of Pinus sabzniana Dougl. (WS)]the problem arose of determining the nature and position of an unsaturated bond present in this compound. T o use the available infrared data effectively, it was necessary to compare the C-H spectra of a number of analogous
Md.
carbocyclic and heterocyclic compounds. The information obtained, which is potentially useful in structural studies on alkaloids and related compounds, is summarized in this paper. The overlap in =CH absorption originating in an aromatic system with nonaromatic =CH absorption is also considered, and a method for detecting nonaromatic =CH absorption bands in the presence of interfering aromatic =CH bands is illustrated. SPECTROSCOPY
A Beckman IR-3 spectrophotometer (which achieves high resolution by unique optical and electronic systems without recourse to lithium fluoride or calcium fluoride prisms) was used in these studies. The spectra were obtained using sodium chloride optics and liquid samples or carbon tetrachloride solutions and cell thicknesses of 0.03 and 0.10 mm., respectively (with the exceptions given below). The spectra of pinidine and dihydropinidine are given to illustrate the type of results obtained. The vinyl hydrogen stretching absorption band in the spectrum of pinidine (3.31 microns) is relatively low in intensity as compared with that found for most of the other olefins. The s ectrum of 1-ethyl-2-ethylidenecyclohexanealso showed low vinyr hydrogen absorption. Integration %-asnot considered a feasible method for obtaining data for comparison, as in Table 111, because the absorption bands in question were very sharp and very close to adjacent and much more intense bands. Molar absorptivities were calculated in the usual manner, using transmittances recorded by the spectrophotometer. They are reported here not as physical constants for the compounds in question but merely as measures of absorbances a t given wave lengths in the spectra obtained by the method used in this work. The Beckman IR-3 is a single-beam instrument with which the pattern of variation of slit width with wave length, established in a “blank” run giving a base line (zero transmittance) for a given cell (in this case 0.40 mm. thick) and solvent (carbon tetrachloride), is repeated exactly when the spectrum of a com-
ANALYTICAL CHEMISTRY
954 pound is determined using the same solvent and cell. The E values of the bands for a given compound and its dihydro derivative were therefore determined a t the same slit width as well a8 the same concentration level. The methods illustrated in this paper are a t present limited to use with liquids or compounds that are soluble in carbon tetrachloride. Preliminary work in this laboratory indicates that these methods can be made more general by use of mulls in hexachlorobutadiene.
Thus, piperidine and cyclohexane gave very similar spectra, and the same effect held for the noncyclic analogs, di-n-propyl ether and hexane. Norpholine, however, with a relatively large ratio of noncarbon to carbon atoms, showed much less correspondence with the C-H bands of cyclohexane than did piperidine.
IO0
I
I
I
I
MODEL C O M P O L S D S
1-Ethyl-2-ethylidenecyclohexane (15), l-butyl-1,2,5,6-tetra1,2-diethyl-1,2,5,6-tetrahydropyridine, 1,2hydropyridine, diethylpiperidine ( 19 ) , pinidine, and dihydropinidine ( 2 3 ) were obtained as described. Dehydration of 1-phenylcyclopentanol (25) to obtain 1-phenylcyclopentene was effected by using 85% phosphoric acid and conditions described (18)for the dehydration of 1,2-diphenylcyclohexanol. The sample of 3-phenylcyclopentene was prepared by the method of von Braun and Kuhn ( S ) , except that 3-bromocyclopentene rather than 3-chlorocyclopentene was used as the starting material. Dihydroeugenol, 1,2-diethylcyclohexane, and phenylcyclopentane were prepared by hydrogenating eugenol, l-ethy1-2-ethylidenecyclohexane,and 3-phenylcyclopentene1 respectively, a t room temperature and atmospheric pressure in ethyl alcohol solution Kitha 10% palladiumcarbon catalyst. All other compounds used, except codeine and dihydrocodeine, were obtained from commercial sources. The boiling points and indices of refraction of all compounds prepared for these studies were in good agreement viith the values reported in the literature.
80
Y Z
2-
60
I
4 2 I-
5w
40
U P
n w
20 R E S U L T S AND DISCUSSIOY
Table I s h o w the observed bands for a number of nitrogenand oxygen-containing compounds and their hydrocarbon ann alogs. The degree of resolution in these spectra was relatively 30 32 34 36 38 4.0 good, although bands in the 3.53- to 3.58-micron region were WAVE LENGTH, MICRONS occasionally obscured by strong absoiption a t 3.47 to 3.51 Figure 1. Pinidine spectrum in mici one. For example, a 3.56-micron band for methylcyclocarbon tetrachloride hexane became evident as a shoulder on a 3.48-micron band only xhen a high concentration of the compound was employed; the same effect was found in the case of myosmine ( 7 ) . The B comparison of the spectra for five- and six-membered satuabsence of a 3.53- to 3.58-micron band in a spectrum where it rated cyclic systems indicated that i t might be possible to draw might be expected to occur by comparison with spectra of anpreliminary conclusions about ring size from C-H spectra. alogous compounds, as is the case for 1,2-diethylcyclohexane, is Six-membered ring compounds showed strong absorption near therefore not serious and is normally due to too great absorbance 3.41 microns, while for compounds with a five-membered ring a for the neighboring band. A similar effect presumably accounts band of comparable intensity was present near 3.38 microns and for the absence of the 3.38-micron band in the spectra of methyla much weaker band or no band a t all was found a t 3.41 microns. cyclohexane and 2-methylpiperidine. Simple shortcchain noncyclic systems containing both CH, and According to the work of Fox and Martin (8) the spectra of CHZgroups showed bands, usually of approximately the same hydrocarbons containing methyl groups have characteristic bands intensity, near both 3.38 and 3.41 microns. For cyclic comat 3.38 microns (asymmetrical stretching) and 3.48 microns pounds, the major band was apparently due to a cumulative effect (symmetrical stretching). The corresponding bands for a of several structurally similar CH, groups, and the replacement methylene group are at 3.41 and 3.51 microns, respectively. The C-H stretching- band for a hydrogen at'om attached to a tertiary carbon atom is usually Carbon-Hydrogen Stretching Bands in IR-3 Spectra of Hydrocarbons and Table I. found at 3.46 microns and is Analogous Oxygen- and Nitrogen-Containing Compounds weak. Bands in the 3.53- to Ware Length of Ahaorption Bands, Microns Compound 3.58-micron region have not Di-n-propyl ether 3.36 3.10 3.49 3.56 been assigned ( 1 ) . Since a Hexane 3.37 3.41 3 48 3 32 3.37 3 44 3 47 3.58 hetero atom would be expected $~;~$$~~ 3.41 3.50 3.57 3.41 3 51 3.57 to influence the C-H absorpCyclohexane 2-Methylpiperidine 3.41 3 50 3.57 tion bands arising from an adhfethylcyclohexane 3.41 3.18 3.56 jacent methylene group, the 1,2-Diethylpiperidine 3 36 3.39 3.4s 3.48 3.57 1.2-Diethylcyclohexane
relatively close correspondence in major bands shown for analogs in Table I is encouraging and suggests that hydrocarbon assignments may be extended to simple heterocyclic compounds containing one nitrogen or oxygen atom in the ring.
: ; : 2
3-Meth~ipyridine Toluene Qui n o 1in e ~ ~ ~ $ ! ! ~ . ~ e
3.24 3.23 3.23 3.24 3.23
g;;$$Fofuran Cyclopentane %-ornicotine7 Plwnylcgclopentane -
3.23 3.24
3.25 3.25
3.27 3.27 3.26 3.27 3.27 3.27
3.26 3.26
3.31 3.30 3.29 3.30 3.29 3.31
3.32 3.35 3.32 3.34 3.35
3.30 3.30 ~~~
3.37
3.41
3.4i
3 37 3 38 3.39
3.41 3.41 3.41 3.40
3 48 3 17 3.49
3.37 3.37 3.36 3.37 3.37 3 38
3.41
3.47 3.47 3.47 3.49 3.48
3.33
~~~~
3 41
3.54 3.54
955
V O L U M E 28, NO. 6, J U N E 1 9 5 6 of one CHI group with a hetero atom did not displace this band. These data should be used with care in extension to unknown structures, but useful leads may be obtained in some cases. For example, from the spectra of pinidine (C9Hl?S)(Figure 1) and dihydropinidine (COHlgS) (Figure Z ) , these compounds were judged to contain a six-membered system with one or more alkyl siihstituents, and this was subsequently found to be correct.
Table 11. Vinyl Hydrogen Stretching Bands
Type
-1
Structural Unit
RL"=CH*
Examples D-Plnene
Peak Wave Lencths, 3Iicrons Reported rangea Found 3 23-3 25 3 2.5 ( 8 , f 2 ,1 9 , 21) 3 27
were calculated for each =CH- band. With one exception there mas an easily recognizable fall in the absorptivity of one of the =CH- bands for each olefin when i t was hydrogenated. This band (in italics in Table 111)was clearly related to the site of reduction, and from this fact the type of isolated =CH- bond could be deduced. Thus, for codeine a =CH- bond was evidently present in a six-membered ring or in a linear system (the bond is in fact in a six-membered alicyclic system); for 3-phenylcyclopentene unsaturation was located in a five-membered ring (correct); for eugenol and isoeugenol changes nere found in bands due to a terminal methylene group and a double bond 111 a linear system or a six-membered ring, respectively (correct for both isomers). Since this procedure is based on observed c1i:tngrs in intensity of absorption bands, it is clearly most satisfactory n-heie the change is large and easily recognizable. Thcre n-ill be instances, such as the case of 1-phenylcyclopentene, wheie the effect r i l l be small and difficult to interpret. The usefulness of this method in structural determinations can be determined only by future applications.
I
D
a
RCH=CHR
Cyclohexene a-Pinene I-Butyl-1,2,5,6tetrahydropyridine I ,2-Diethyl-1,2,5,6tetrahydropyridine Pinidine
3 29-3.32
Oleic acid 1-Ethyl-2-ethylidenecyclohexane
3 29-3 32 (8, 8 1 )
(8. 1 0 )
1
I
3 31 3 29
3 30 3 30 3 31 3 29 3 31b
References given in patentheses.
b Shoulder.
-
-___-
Table I1 contains reported and observed data for =CH stretching bands for several representative olefinic structures. The types of =CH- bonds are (A) terminal methylene, (B) in a fivenienibered system, (C) in a six-membered system, and (D) in a linear structure. From these data it is evident that it is not possible to distinguish a =CH- group in a six-membered system from A linear =CHgroup, but a terminal methylene group and a =CH- in a five-membered system may be distinguished from each other and from the six-membered or linear case. Three heterocyclic compounds were taken for comparison with their carbocyclic analogs. A =CH- band was found for each one a t 3.30 to 3 31 microns. I n two cases the structure was known to be that of :in alkyl-substituted tetrahydropyridine, and in these cases the observed band corresponded to what would be expected for a sixwembered carbocyclic analog. The third case was t h a t of a c oinpound (pinidine) whose structure was unknown a t the time. The position of the =CH- band indicated that the group \vas p~esenteither as part of a six-membered system or a linear system, and it was later found that the group was linear (in -CH= C'H-CH3 attached to the 2-position of piperidine). I n the authors' experience there are several limitations to the us? of these properties in structural studies. For example, vinyl amines are best studied by the excellent procedure of Leonard m d Gash (16), rather than by =CH- absorption. Perhaps the chief limitation, however, lies in the fact that isolated =CHbonds and aromatic =CH- bonds (in aromatic systems or in heterocyclic systems with aromatic character), when present in the same molecule, will often result in spectra in which there is a considerable overlap of =CH- bands. I n the absence of further information these spectra are largely valueless for the detection and characterization of the nonaromatic double bond. A solution to this problem was sought by comparing spectra for parent and dihydro compounds. The spectra were taken a t the eume concentration level in the same cell, and the absorptivities
I
O
30
32
34
36
L
38
40
WAVE LENGTH, MICRONS
Figure 2. Dihydropinidine spectrum i n carbon tetrachloride
The effect of small ring size on C-H absorption bands in heterocyclic systems rras not investigated in detail. The spectrum of cyclopropane s h o w bands a t 3.22 and 3.31 microns ( 2 2 ) , and the latter band has been reported to depend on the presence of a methylene group in the cyclopropane ring (4). The presence of one or the other of these bands has been used to detect or confirm three-membered rings in triterpenoids (41, 3,5-cyclosteroids ( I d ) , and norpolycyclenes (6, 17). The spectrum of propane epichlorohydrin contained bands a t 3.27 and 3.33 microns which were compared with bands a t 3.23, 3.29 (shoulder), and 3.33 microns for methyl cyclopropyl ketone (compare 2 4 ) . The correspondence of the oxide three-membered methylene band with that of the cyclopropane system is therefore rather close. The spectra of 01- and @-pinene %*ere taken for comparison; these two compounds contain a fourmembered carbocyclic system and presumably shoiild possess a C-H band lying betn-een the three-membered methylene region (3.27 to 3.31 microns) and thp five-membered region ( 3 36
ANALYTICAL CHEMISTRY
956 to 3.38 microns). Bands were observed a t 3.34 and 3.35 microns, respect.ively, for aand ,+pinene.
ACKNOWLEDGMENT
Table 111. Detection and Characterization of Double Bonds in Aromatic Compoundsa -4bsorption Bands in Vinyl Hydrogen Stretching Region Cb x c € xc t xc € xC T Codeine 0.114 3.27d 3.30 67.0 3.33 45.3 Dihydrocodeine 0.114 3.27 17.8 3.30 27.4 8.33 44.0 1-Phenylcyclopentene 0,337 3.25 29.6 3.27 87.7 3.30 34.7 3-Phenylcyclopentene 0.336 3.246 3.27 65., 3.30 47.5 Phenylcyclopentane 0.326 3.24 21.2 3.27 27.4 3.30 46.2 Eugenol 0.286 3.24 26.3 3.26 26.3 3.30 20.6 3.32 40 8 Isoeugenol 0.289 3.26 21.8 3.31 56.8 3.32d Dihydroeugenol 0.314 3.24* 6.9 3.27 20.0 3.30 84.7 3.32 36.2 a D a t a which would indicate presence and nature of double bond in structure if compound were unknown are Compound
The authors &h to thank Claudia House for assistance in the preparation of some of the m o d e l c o m p o u n d s . b Concentration in millimoles per milliliter of carbon tetrachloride. Wave length in microns a t band peak. The tetrahydropyridines and d Band was obscured b y strong adjacent band. 1 , 2 d i e t h y l p i p e r i d i n e were * KO band a t this wave length. Molar absorptivity given for comparison purposes. generously supplied by Raymond Paul, I-ethyl-2ethylidenecyclohexane by J. V. Karabinos and codeine and dihydrocodeine by 9. P. Findlay. (12) Jones. R.N.,Humphries, P., Herling, F., Dobriner, K.. J . Am. C
~~
~~~
Chem. Sac. 74, 2820 (1952). (13) Jones, R. X., Humphries, P., Packard, E., Dobriner, K., Ibid., 72,86 (1950). LITERATURE CITED (14) Josien, AI. L., Compt. rend. 231, 131 (1950). ~ ~ lL, Jl, , .~ , ~~ ~~spectra f ,~ ~ of complex ~ ~ d ~ l o l e c u l e s , ~ ~ (15) Karabinos, J. v., Ballun, A . T., J . Am. Chem. Soc. 76, 1380 (1954). Wiley, New York. 1954. (16) iY. Gash, l'. mT., Ibid.* 767 2781 (Iga). Bladon, P., Fabian, J. SI.. Henbest, H. B., Koch, H. P., Wood, (17) Lippincott, E.R., Ibid., 73, 2001 (1951). G.W., J. Chem. Soc. 1951, 2402. (18) Nueller, G. P., Fleckenstein, J. G., Tallent. W. H., Ibid., 73, Braun, J. von, Kuhn, AI,, Ber. 60, 2551 (1927). 2651 (1951). Cole, A. R. H.. J . Chem. Soc. 1954, 3807,3810. cole, A, R,H,, R ~pure ~ and , ~ ~Chem,~ ( ~l ~ ~, t4,~111~ l i (19) ~ )Paul, R., Tchelitcheff, s., Bull. SOC. h i m . France 1954, 982. (20) Pinchas, S..ANAL.CHEM.27, 2 (1955). (1954). (21) Sheppard, N., Simpson, D. S . .Quart R e t s . (London) 6, 1 (1952). Cristol, S. J., Snell, R. L., J . Am. Chem. SOC. 76,5000 (1954). (22) Smith, L. G., Phys. Rev. 59, 924 (1941). Eddy, C. R..Eianer, -1.. .XN.AL. CHEM.26, 1428 (1954). '. L., Homing, E. C., J . A m . Chem F ~ J. ~J,,,~ ~-4,E , , ~pro,-, R~ ~ sot, ~ i ( , ~ ~ ~ ~, 208 d ~ (23) ~ )Tallent, W. H., Stromberg, 1 SOC.77, 6361 (1955). (1940). (24) Wiberley, S.E., Bunce, S.C., h-u..CHEX 24,623 (1952). ~ .~ ,p A , , J ,~c'hellL, p~/ l y s ,18,~861 (1950): i 19,~942 (1951). , (25) 'elinsky. N . D., Ber. 2775 (lgZ5). Johnson, D. R . . Idler. D. K.,Melorhe, V. W., Bauman, C. A , , J . A m . Chem. SOC.75, 52 (1953). R E C E I V Efor D review January 28, 195G. dccepted 1Iarch 23, 1956. Jones, R . X..Herling. F., J . O r g . Chem. 19, 1252 (1954). J.l
581
Differential Spectrophotometric Determination of Beryllium 1. C. WHITE, A. S. MEYER, Analytical Chemistry Division,
JR., and
D.
L. MANNING
Oak Ridge National Laboratory, O a k Ridge, Tenn.
Differential spectrophotometry was applied to the determination of beryllium as the p-nitrobenzeneazoorcinol lake in basic solution. The absorbance of the complex is measured against a reference standard which contains 1.0 mg. of beryllium per 100 ml. of solution. The coefficient of variation of the method is less than 1% on duplicate determinations. The method is essentially free from interferences and is applicable to the determination of beryllium in the presence of moderate amounts of uranium and aluminum. Relatively large amounts of fluoride can be tolerated without interference.
B
ERYLLIUM and beryllium compounds have many unique
chemical and physical properties ( 5 , 1 5 )which have won for this element increasing applications and widespread interest. As the use and availability of beryllium have increased, many new problems have arisen in the analysis of this element, which have resulted in the development of new and modified methods of determination. The gravimetric methods ( I S ) most widely used for the determination of beryllium include the precipitation of beryllium as the hydroxide and ignition to beryllium oxide for weighing, and the precipitation of beryllium as the ammonium phosphate salt with
subsequent ignition to berylliuiii pyrophosphate (Bed'&,) for weighing. When beryllium is determined as the oxide, the solution from which beryllium hydroxide is to be precipitated must be free from all other cations that form insoluble hydroxides with ammonium hydroxide. The precipitation of beryllium ammonium phosphate with subsequent ignition to beryllium pyrophosphate has the advantage of almost four times more mass per equivalent beryllium content as in the beryllium oxide method. If the conditions of precipitation are not meticulously controlled, however, slight departures from the theoretical composition of beryllium pyrophosphate may occur, resulting in a loss of accuracy. A4volumetric method (8, I S ) for beryllium has been reported in xhich beryllium is first precipitated as the hydroxide at a pH of 8.5. If an excess of sodium fluoride is then added, beryllium is converted to the very stable, weakly ionized fluoride complex, and an amount of hydroxyl ions equivalent to beryllium is found as sodium hydroxide. This alkalinity is determined by titration n i t h standard acid and calculated as a measure of the concentration of beryllium present. The method, although empirical, IS rapid and precise under ideal conditions. Aluminum reacts in a similar manner to beryllium and will be included with the beryllium. Although it is possible to correct for the aluminum present, the precision of the method is seriously affected. Zirconium, hafnium, rare earths, uranium, and thorium interfere and must be separated.
