step in the extraction proccdure. Evaporate the 7 5 to 85% alcoholic extract to dryness, cool, tease t h e residue thoroughly with 75 ml. of absolute ethyl alcohol, and filter. Concentrate the filtrate on a water bath to about 20 ml., cool, add an excess of acetone, and proceed as described before. DISCUSSION A N D RESULTS
Of the two tests described, the second one is more sensitive and was therefore applied to all the determinations. The course of the first reaction is explained by Schechter (11, 12), only sodium methylate has been replaced by potassium hydroxide. The second reaction, the chemistry of v, hich has not been studied, is a n application of the mell-known Janovsky reaction. although sodium methylate can also be used in place of potassium hydroxide for these spot tests, n i t h equal sensitivity, the latter was preferred because of the folloning advantages. The successive color changes are distinct and relatively slow, for close observation of the transition of colors, while in the case of sodium methylate the successive color changes are indistinct and too quick to follow; also, the reagent can be prepared easily. Applications. T h e tests are not specific for a n y single component, and are applicable to the nitrated products of P,P’-DDT and its degradation products, such as P,P’-DDE and P,P’-DDA; they are also applicable to D D D , D F D T , and methoxychlor. Detection Limit and Recovery. T h e loner limit of detection of D D T is
1 pg. (Table I). If the nitrated product is free from interfering radicals, quantities less than 1 fig. can be detected and the results reproduced. Recovery by this method is almost complete as compared to the color intensity of spots developed with standards. Quantities up to 3 or 4 pg. can be determined semiquantitatively by comparing the color intensity of the test spot with those simultaneously developed with standards. Interferences. Although 1 gg. of the nitrated product of D D T could be detected by t h e tests described, a t times t h e sensitivity is decreased slightly in t h e case of putrified viscera. T h e nitrated product of its extracts are contaminated with intensely colored degradation products of tissue despite t h e removal of proteins and colored bodies which are likely to mask the color tests. They are removed by precipitating them from the alcoholwater extract by the addition of excess acetone, and further by oxidation with nitric acid during nitration. Also, in the cases of liver and kidney tissue with fatty deposits and viscera of exhumed bodies, the final nitrated product is colored yellow or light brown and renders the test less sensitive. It is difficult sometimes to distinguish the sequence of colors, as some of the colors are nearly masked. I n such cases the modified procedure is recommended whereby the interference can be greatly minimized. Normal constituents of tissue and urinary excretions do not interfere with the color tests. In our laboratory, cases of D D T poison-
ing have been analyzed by the above method with satisfactory results. This method with suitable niodifications can be used to advantage in other fields such as residue analysis, where interfering substances are less when compared to biological materials. ACKNOWLEDGMENT
The authors thank Satesa T’edachalani of this laboratory for his assistance. LITERATURE CITED
(1) Alessandrini, M. E., Arm. Chim.
applicata 38,53-4 (1948). (2) Amsden, R. C., Ralbridge, D. J., J. Agr. Food Chem. 2, 1323 (1954). (3) ~. Claborn. H. V.. J . Assoc. Ofic. - 1 ~ . Chemists 29, 330-7 (1946). f41 Downing. G.. Norton. L. €3.. -&SAL. CHEM.23T1870-1 (195f). ( 5 ) Iling, E. T., Stephenson, W. H., Analvst 71, 310 (1946). (6) Luis, Policarpo, Rev. asoc hioqitim. arg. 17, 334-8 (1952). ( 7 ) Martin, J. T., Batt, R. F., -inn. Rept. Agr. Hort. Research Sta., Long Ashton, Brzstol 1953, 121 (1954). ‘ (8) Pontoriero, P. L., Ginsburg, J. M., J . Econ. Entomol. 46, 903-5 (1953). (9) Prickett, C. S., Kunze, F. M., Laug, E. P., J . -4ssoc. Ofic. Bgr. Chemasls 33, 880-6 (1950). (10) Quintana Y Mari. A,. Cid Caoella. ‘ AnaMaria, Cz~uderno’81,229-51 ((946): (11) Schechter, M. S., Haller. H. L., J . Am. Chem. SOC.66, 2129 (1944). (12) Schechter, M. S., Solowaq, S. B., Hayes, R. A., Haller, H. L., 1 s ~ ENG. . CHEM.,ANAL.ED. 17, 704-9 (1945). (13) Stiff, H. S., Jr., Castillo, J. C., Science 101, 440-3 (1945). \ - ,
~
for review May 24, 1960. AcRECEIVED cepted December 16, 1960.
An Absolute Method of Turbidimetric Ana lysis E. J. MEEHAN and W. H. BEATTIE’ School o f Chemistry, University o f Minnesota, Minneapolis 7 4, Minn.
b An absolute method of turbidimetric analysis is described which eliminates empirical comparison of known and unknown suspension, and is applicable to heterodisperse suspensions. It is based upon measurement of turbidity a t an experimentally determined wave length at which the turbidity is proportional to the reciprocal of the wave length. The observed turbidity at this wave length is related to weight concentration through the M i e scattering coefficient. An example is given of application to a silver bromide sol in which the precision of determination is a few per cent.
1 Present address, Shell Chemical Co., Box 211, Torrance, Calif.
632
ANALYTICAL CHEMISTRY
based upon turbidity traditionally depends on the comparison of intensity of light transmitted by a suspension of the unknown Kith that transmitted by a suspension of known concentration, both suspensions being prepared under identical conditions. I n nephelometric analysis a similar comparison is made of intensities scattered more or less at right angles to the incident beam. With modern instruments both kinds of measurements can be made precisely. The principle difficulty is that the intensity scattered by a given substance depends on the weight concentration and on the particle size and shape. The present discussion is restricted t o nonabsorbing spherical particles. For such particles with radii much smaller than the wave length, the scattering NALTSIG
per unit n eight concentration (specificscattering) increases with the cube of the radius. For particles much larger than the wave length the specific scattering is inversely proportional to the radius. For intermediate sizes t h r behavior is complex (10). Therefore, turbidimetric and nephelometric comparisons are meaninples. unless the particle size is the same in both suspensions. For heterodispersed suspensions the size distribution must be the same. As is well known, i t is difficult to form a suspension-e.g., barium sulfate, silver chloride-in a manner which yields reproducible particle sizes. illso, both the initial particle size and rate of change of size with time are subject to a variety of chemical and physical influences. It is evident why empirical turbidimetric and nephelo-
metric methods yield reliable results only under special and strictly controlled conditions. T h r present paper describes a simple method n hich, under certain conditions of pnrtirle qize and size distribution. cliniinates empirical comparison and yields an ab-olute measurement of concentration. It rl-as developed during a n investigation of the light scattering proprrties of silver bromide sols (6) and its practical use is illustrated by applicaation to such a sol.
3 34 I 1
I
I
8
9
I
I
1:
I
PRINCIPLE
The principle of turbidimetric measurements n ith particular reference to si1vc.r bromide sols has been described previoidy (6). The turbidity T (diof a suspension of nonmcnsion :tbsorl)ing particles is defined by Equation I ,
in w111(*11I , and I are intrnsities transmitted through thickness, I , of suspension. For a monodisperse suspension of sphews of radius T, r is given by Equation 2: r = n d K
Figure 1. K / a vs. o( for refractive index ratio m = 1.20, 1.44, and 2.0 K = scattering coefficient,
01
= 2 a r/X
(2)
f(r). I n some cases suitable scattering
in nhich n is the number of particles per milliliter and K is the scattering c~oeficient (10). For a suspension of ('oncentration, c, in grams per milliliter, the specific turbidity is defined as r/c. When the Suspension is heterodisperse (usual case) the following relations hold. The specific turbidity of a suspension containing nL particles per milliliter of radius r, and scattering coefficient K , is given by Equation 3:
measurements may yield f(r). and therefore c (6). The present paper is restricted to a special case that is applicable when the size distribution f(r) is not too broad. Any narrow portion of the complex K-a curve can be represented by the equation K = ka". When x = 1, Equation 4 assumes the special form:
For practical purposes sums are replaced by integrals. The particle size dietribution is defined by dn = f ( r ) d r , dierc, dn is the number of particles per milliliter of radius between r and r dr. Then the concentration c is given -
+
by
f(r)r3dr, and the specific
till M i t y is given by Equation 4: Pcc
Therefore sprcific turbidity can be calculated for any wave length for any distribution f(r), given the values of K . For analytical purposes we have the reverse problem to determine c from experimental transmission or scattering measurements. Obviously i t is impossible to obtain c from a single measurement, arid i t is necessary to obtain
Therefore, if a range of a exists over which K is approximately proportional to a-Le., K / a is constant-and the size distribution is narrow enough that practically all the particles are I\ ithin this range of a, then the specific turbidity in this range is independent of the particle size distribution, and r is directly proportional to e. The validity of the first condition is readily ascertained using available tabulations of K Figure 1 is a plot of K / a us. a for m = 1.20, 1.44, and 2.0. Values for m = 1.75 are shown separately in Figure 2. The value m = 1.73 corresponds to silver bromide in water at n-ave length (vacuum) of 450 mp. The K values were taken from those of LIeehan and Beattie ( b ) , supplemented by additional values obtained on a Univac 1103 computer. For m = 2.0 values of K were taken from Lowan ( 4 ) and Penndorf (8). For this m, the values of K / a fluctuate so widely with CY that no meaningful average value can be assigned t o K,/a. Therefore the principle outlined in this paper does not apply t o suspended material of
m = 2.0 (or larger). This is not a severe restriction since most suspensions of analytical importance have m values smaller than this. For example, suspensions of barium sulfate, silver chloiide, and silver iodide in water have m values (A,,, = 589 nip) of about 1.23, 1.5. and 1.7, respectively. For 7n + 1, a simple analytical expression for K exists (10, chap. 11) in terms of the parameter p = 2a ( m -1). From this expression, K / p has its maximum value of 0.899 a t p = 2.93. From the K values of Penndorf ( 9 ) , over the twofold range P = 2.0 to 4.0, K / p varies smoothly and relatively little, the average value of K / p being 0.856 =t0.026 (3.0%). The exact expression for m + 1 is useful in estimating the general feature of the IC- a curve for m values different from unity (10). From the numbers given above and the definition of p , for m = 1.20 the maximum K / a should be about 0.36 at a = 7.3. From the extensive tabulations for m = 1.20 ( 7 ) , the maximum K / a actually is 0.414 a t a = 7.3 to 7.4. Over the nrarly twofold range of a from 5.4 to 9.8, K / a varies smoothly and the average value of K / a is 0.398 1 0 . 0 1 1 (2.8%). For m = 1.44, the results for m -+ 1 similarly lead to the approximate predicted values: ( K / C Y ) about ~ ? ~ 0.8 a t a about 3.3. For this m, K no longer varies smoothly with a, and also the number of K values available is less than for smaller m. From the tabulation of Lonan ( 4 ) the range of approximate constancy of K / a is about a 2 7 to 4.5; the maximum K/cy is about 1.0, the average K / a is 0.95 with an average deviation of a fen- per cent. K i t h increasing m the irregularities in the K a us. a curve become still more pronounced, as indicated above for m = 2.0. Honever, for m = 1.75, K / a is approximately constant over the range of a from 1.6 to 2.3 (Figure 2). The average K / a over this range is 1.66 i 0.06. From all the above it is seen that K / a is constant aithin a few per cent over a 1.5- t o 2.0-fold range of a . or, a t a given X, over a corresponding range of r. Therefore Equation 5 may be applied to any heterodispersed suspension of known m for which the distribution of size is no wider than corresponds to the allowed cy range, and a n a r e length may be used corresponding to the desired a . I n conventional visible-ultraviolet spectrophotometers, about a fivefold range of A is available (200 to 1000 mM). It is assumed for the present paper that the distribution of a is known to be sufficiently narrow. [Determination of f(r) is possible from scattering (6), and refinement of the calculation by use of f(~)will be discussed in a later paper.] The choice of proper wave length is illustrated here for the VOL. 33, NO. 4, APRIL 1961
633
I. 2
1' /
I1
1
I
O
1'4
116
Ij8
210
212
U
Figure 2.
Table I. Refractive Index Ratio and K/a Values for Silver Bromide Sols in Water
350 400 436 450 546
m
1.89 1.79* 1.76 1.75 1.70
iRange of a 1.5-2.1 1.5-2.1 1.6-2.2 1.6-2.2 1.6-2.2
Average
K/aa 1 . 9 5 =k 0.10 1 . 7 8 =k 0.07 1 . 6 9 =k 0 . 0 6 1.66 + 0.06 1 . 5 2 =k 0 . 0 5
Over the range of in column 3. Based on extrapolated value of refractive index of silver bromide. a
634
218
'
K / a v s . afor rn = 1.75
case of silver bromide in water. In this case the measurement must be made at a = 1.9. This means that for spheres of radius 0.05, 0.1, or 0.2 microns, respectively, the corresponding vacuo wave length must be 220, 440, or 880 mp, respectively. The upper limit is set by the absorption of water. The further restriction for the present purpose is that the particles must be nonabsorbing. [Extension to absorbing particles is being studied.] Since silver bromide absorbs appreciably in the ultraviolet ( 2 ) , the present method is restricted to suspensions with radii in the approximate range 0.07 to about 0.2 micron. Given a suspension of unknown T , the proper X is selected by measuring the transmission in a spectrophotometer with a suitably restricted angle of view. The observed turbidity 7 (Equation 1) is plotted against 1/X. For a homogeneous suspension this is exactly equivalent (Equation 2) to a plot of K us. CY. For a heterodisperse suspension it is only necessary to note A at which T is proportional to 1/X, and apply Equation 5 a t this X. [-41ternatively, the product T X may be plotted against 1/X. The proper X corresponds to the maximum TX.]
Lao
216
Figure 3. Observed turbidity of a silver bromide sol plotted against reciprocal wave length Abscissa scole is 1 /X, where wave length is in medium, in microns W a v e lengths shown on graph a r e vacuo wave lengths
APPLICATION TO SILVER BROMIDE SOL
Data on the refractive index of silver bromide are available at 436 mp and longer wave lengths ( 3 ) . These data were extrapolated b y conventional means to estimate the refractive index at 400 and 350 mp. Values of m in the range 350 to 546 mp are given in Table I. Average values of K / a and the corresponding ranges of CY also are given in Table I for each m value. The values of average K / a were obtained from exact Mie theory calculations for m = 1.65, 1.75) and 1.85 at intervals of 0.1 in CY. These values were calculated on a Univac 1103 computer. The average K / a for intermediate m were obtained by interpolation. A silver bromide sol, the concentration of which was known t o be 1.69 X IO-5 gram per ml.. has t h e T -1/X dependence shown in Figure 3. These measurements were made in a Beckman DU spectrophotometer using a 10-em. cell and apertures of 4 mm. in diameter to limit the angle of view. T was approximately proportional to l / h a t ,,A, = 350 inw, or X = 260 mp in the medium. While silver bromide absorbs a t this wave length, for the particle sizes involved the absorption is negligible relative to the scattering. The transmission at this wave length was 40.57, (10-em. cell), corresponding to T = 0.0904 em.-' The average value of K/CYappropriate to this wave length and m (Table I) is 1.95. Taking D = 6.47 grams per ml., Equation 5 assumes the form:
O(
ANALYTICAL CHEMISTRY
3?r X 1.95 T - = 2 X 6.47 X 2 . 6 X c
= 5.46
x
104
Hence c=--= 0'0904 5.46 x 104
1.66 X 10-bgramperml.
The agreement with the correct value is even better than might be expected, considering the uncertainty in the extrapolated refractive index a t this wave length, and the use of the average value of K / c Y . DISCUSSION
I n the application of the principle described above no knowledge is needed of the manner of preparation of the suspension, since no comparison with a known suspension is involved. -41~0,so far as the stability of the sol is concerned, it is required only that the scattering properties be constant during the time to obtain the spectrum. K h e n a recording spectrophotometer is used, this amounts only to a few minutes. The restrictions summarized briefly, are as follows. The method can be applied to heterodisperse suspensions of spherical, nonabsorbing particles of known refractive index, for ivhich accurate Mie theory scattering coefficients are available. The size must he in a range such that a waye length can be used corresponding to a specified CY (CY = 1.9 for aqueous sols of silver bromide, about 7 for aqueous sols of barium sulfate, etc.). The nature of the size distribution is of no consequence, provided only that it is no wider than that which corresponds to the allowable range of a (about 1.6 to 2.3 for silver bromide, etc.). Future papers will describe improvements and refinements, including extension to wider distributions and use of
wave lengths at which absorption may occur. REFERENCES
(1) Beattie, W. H., Ph.D. thesis, Uni-
versity of Minnesota, October 1958.
( 2 ) “International Critical Tables,” Vol. V, p. 270, McGraw-Hill, New York,
1926. (3) Landolt, Hans, Bornstein, Richard, “Physika1isch.-chemischeTabellen,” 5th
ed., 3rd suppl., p. 1477, J. Springer, Berlin, 1935. (4) Lowan, A., “Tables of Scattering Functions for Spherical Particles,” National Bureau of Standards A.M.%4, Washington, D. C., 1948. (5) Meehan, E. J., Beattie, W. H., J . Opt. SOC.Am. 49, 735 (1959). (6) Meehan, E. J., Beattie, W. H., J . Phys. Chem. 64, 1006 (1960). (7) Pangonis, W. J., Heller, W., Jacobson, A., “Tables of Light Scattering Func-
tions for Spherical Particles,” Wayne State Universitv Press, Detroit. Michigan, 1957. (8) Penndorf, R. B., J . Opt. SOC.Am. 46, 1001 (1956). (9) Penndorf, R. B., Ibid., 47, 603 (1957). (10) van de Hulst, H. C., “Light Scattering by Small Particles,” Wiley, New York, 1957. RECEIVEDfor review March 28, 1960. Accepted November 17, 1960.
Detection of Olefins by Epoxidation and Hydroxamation and Characterization by Rearrangement of Epoxides to Carbonyl Compounds JACOB G. SHAREFKIN and HARRY E. SHWERZ Department of Chemisfry, Brooklyn College, Brooklyn, N.
Y
b The two most general tests for the
QUALITATIVE DETECTION OF ALKENES
olefin bond, decolorization of bromine and permanganate, give positive signs of reaction with many nonolefinic reducing reagents. Such false positive tests are avoided b y devising tests in which the sign of a positive reaction depends on a chemical change in the substrate rather than the reagent. Commercial 40% peracetic acid transforms olefins to their glycol monoacetates which are treated with hydroxylamine and the hydroxamic acids are detected as the wine red ferric salt. Olefins are also characterized by rearranging the epoxides with boron trifluoride to carbonyl compounds that are then converted to solid dinitrophenylhydrazone. Reactions may b e carried out on a semimicro scale and permit identification of olefins in the presence of acetylenes and other reductants.
Sharefkin and Sulzberg (14) have pointed out that an ideal functional group test should be specific in avoiding both false positive and false negative tests and should be capable of both detecting and characterizing the function in the presence of other functional groups that are either in the same molecule or in different compounds in a mixture. Their test for olefins is based on a Friedel-Crafts acetylation of the carbon to carbon double bond and detection and characterization of the unsaturated ketone with 2,4-dinitrophenylhydrazine. I n accordance with the reactivity selectivity principle, reaction with the more nucleophilic olefinic substrates occurs with all Friedel-Crafts catalysts, but ketone formation with the least nucleophilic substrates requires the most electrophilic catalysts. The classification of olefins into a reactivity spectrum based on the difference in electrophilicity of the catalyst that effect reaction underscores the need for developing a reagent selectivity principle. Detection of organic functional groups that vary widely in their reactivity requires a group of complementary reagents of different reactivity that may be used for the various parts of the reactivity spectrum. i i n ideal reagent for a functional group should also he effective a t the lower limits of the micro or semimicro level for each band of the reactivity spectrum and should also meet the demand of the reagent selectivity principle in permitting the preparation of a suitable derivative in good yield and of sufficient purity for characterization of the substrate. Commercial 40% peroxyacetic acid (4) is specific in transforming olefin substrates to their epoxides under mild
D
the considerable literature, there is no general and reliable reaction for the detection and characterization of olefins (alkenes). The tn-o most general tests for olefin detection, decolori7ation of a 2% aqueous permanganate solution (8, 1 6 ) and a 5% solution of broniine in carbon tetrachloride (9, I S ) , are not specific for the olefin bond and do not occur with many alkenes. Both reagents are oxidants that are reduced by aldehydes, polyhydric phenols, and other reductants as n-ell as by alkynes. The bromine reagent may also be decolorized by substitution with slow evolution of HBr. The nonspecific character of these tests for the olefin functional group is similar to that of tests in which the sign of reaction is based on a chemical change in the reagent rather than in the substrat?. ESPITC
or alkaline conditions and to glycol monoacetates on heating or in acid medium. These monoacetate esters are detected by the Davidson hydroxamic acid test ( 7 ) , which is based on hydroxamation in alkaline methanol solution.
+ CHSCO~H
+
\C-C/
+ HzO
I\\
/I
0
0 .
H
~OCHB
I
1
‘c-c/
/A A‘ I
k
I
+ NH~O-+
‘c-C’ /I
bCHa
I\
+
0 0
I
3 CH3CONHO-
+ Fe+3
I
H H CHBCONHO+
( CH3CONH0)3Fe
The reaction is also used to discriminate alkynes n-hich are epoxidized more slowly, and are cleaved to carboxylic acids under these conditions. Peroxyacetic acid is the acid anhydride of acetic acid and hydrogen peroxide and reacts with hydroxylamine under these conditions to form a hydroxamic acid and give a positive test. The excess of peroxyacetic acid is destroyed before hydroxamation by addition of a trialkylamine which forms an amine oxide. RIN
+ CHaCOpH SPECTRUM
+
OF
RaK40
+ CHsCO9H
ALKENE REACTIVITY
Olefins are nucleophiles and the reactivity of their pi electrons depends on the substituents on the unsaturated carbon atoms. Swern (19) has shown VOL. 33, NO. 4, APRIL 1961
635