a t this time whether this was due to traces of other fractions trailing over; fraction 1from the initial separation also showed positive results. The main product of the pulegone-orcinol condensation previously mentioned produced a violet color which turned to blue-violet after standing 15 to 20 minutes; after dehydrogenation of the main product with sulfur (which should result in the methyl homolog of cannabinol) a blue color was produced. The acetylated cannabinol did not react with the reagent. The color sequences observed when an extract from the plant material is tested with the reagent are obviously results of the interaction of the various components. Similar colors can be obtained with other compounds. Several resorcinol derivatives produce a violet color-orcinol, for instance, goes through the sequence of orange to pink to violet-and a blue color can be obtained with thymol, which goes through the sequence of pink to violet to blue. The mechanism of the reactions is probably a condensation of the type experienced between aldehydes and phenols; the increased reactivity of the resorcinol-type structure produces a rapid condensation under the acidic conditions of the test. Although they are useful such color tests cannot be considered as rigorous proof for the presence of a complex botanical material. The use of gas-liquid chromatography provides an identifica-
34
20
6
5
b
io
I5
-----t
rnin
I
3
4
Figure 1 .
2
I
Fractions detected in Cannabis sativa resins
Temperature 225' C.; flow rate 40 cc./mln. Numbered lines indicate the six initial fractions
tion based on the retention patterns which is superior to that of color tests and allows separation of a complex mixture, as in the case of Cannabis resin, and an identification to be made by establishing the presence of some particular substance. ACKNOWLEDGMENT
The authors express appreciation to the State of California Department of Justice, Bureau of Karcotic Enforcement, for material used. LITERATURE CITED
(1) Adams, Roger, Cain, C. K., Baker, B. R.. J. Am. (:hem. SOC. 62, 2201 (1940): (2) Adams, Roger, Hunt, Madison, Clark, J. H., Ibid., 62, 196 (1940).
(3) Adam, Roger, Smith, C. M., Loewe S. Ibid., 63, 1973 (1941). (4) hou uet, R. J., U. N. Bull. on Nar4 , 1 4 (1950). cotics (5) Duquenois, P., Moustapha, H. N., J. Egypt. Med. Assoc. 21,224 (1938). (6) Matchett, J. R., Levme, J., Benjamin, L., Robinson, B. B., Pope, 0. A., J. Am. Pharm. ASSOC., Sci. Ed. 29, 399 (1940).
a:
CHARLES R. KINGSTON
PAULL. KIRK School of Criminology University of California Berkeley, Calif. RECEIVEDfor review July 31, 1961. Accepted September 21, 1961. Taken in part from the Master's thesis of Charles R. Kingston, University of California, 1961. Work supported by grants from the National Institutes of Health, U. S. Public Health Service (RG-4372 and RG5802), and from the Research Committee, University of California.
Experimenta I Eva I ua ti o n of Liquid-J unctio n PotentiaI SIR: The increasing use of nonaqueous solvents for investigating the nature of ion-solvent interactions and mechanism of organic electrode reactions has prompted us to submit the following communication. Although a large number of electrochemical investigations have been conducted in various solvents, most of the studies are of little value in furnishing information fundamental to understanding better the nature of ion-solvent interactions and organic electrode reactions, because of our inability to ascertain the magnitude of the liquidjunction potentials included in the potentials obtained. The complicated nature of junctions betu-een reference electrode systems and solutions being examined excludes any, but probably highly complicated, theoretical means of evaluating junction potentials. A number of attempts have been made to compare the electromotive force series in various solvents (5). Pleskov (5) has suggested the use of the rubidium couple as the standard
reference electrode, because of the low and nearly identical solvation energy of the large, slightly polarizable, noncomplex-forming rubidium ion in all solvents. Because of the limited number of solvents in which the potential of the rubidium couple can be evaluated and not constant nature of the solvation energy of rubidium ion in all solvents, use of the rubidium couple for the evaluation of liquid-junction potential, however, has ,attracted little support. The reversible nature of the 4,7dimethyl-1, 10-phenanthroline ferric4,7-dimethyl-I, 10-phenanthroline ferrous [DMFe(III), DMFe(II)] couple, its more accessible potential, and the nearly identical free energies of solvation of the two species in various solvents-because of their large size, slight polarizability, and noncomplexforming properties-make the DMFe(111), DMFe(I1) couple an eminently more superior couple for the experimental evaluation of liquid-junction potentials than the rubidium couple. Table I summarizes the half-wave
potentials (us. S.C.E.) observed for the oxidation of DMFe(I1) at a rotating platinum electrode in a number of solvents. The uncertainty of the potentials reported is less than =kt5 mv. The electrode, a 1-cm. length of 22-gage platinum wire extending from the tip of a soft glass tubing, was rotated a t ca. 600 r.p.m. DMFe(II)(ClOd)z was prepared by mixing 4,7-dimethyl-1, 10-phenanthroline and ferrous perchlorate-hexahydrate in a 3:1 mole ratio in acetone. The product was air-dried, and analysis showed it contained 6.36% of Fe(I1). The theoretical value is 6.36QjO0. The close to 0.06 value of the slopes indicates that the couple is reversible in practically all the solvents. In solvents in which DMFe(I1) perchlorate is not sufficiently soluble to give a measurable wave, the halfwave potentials were evaluated by extrapolating the El 2 values in several solvent-acetonitrile mixtures to zero acetonitrile concentration. The validity of this extrapolation technique has been checked with a number of systems in VOL. 33, NO. 12, NOVEMBER 1961
0
1795
Table 1. Half-Wave Potentials for Oxidation of 4,7-Dimethyl-l,l O-Phenanthroline Ferrous, in Nonaqueous Media
DielecE,12 tric (Volt us. ConSolvent“ S.C.E.)c Slopec stant Kat er 0 . 830d 0 . 0 6 ~ 78 Acetonitrile 0.860 0.063 38 Nitromethaneb 0.880 0.066 36 Methanol 0.890 0.120 33 Allyl alcohol 0,890 0.077 22 Ethanol 0.895d 0.068 24 A.
c 4..c anhv. . . .
dride 0.900 0.058 21 0.065 26 Acetyl acetone 0.905 1-Propanol 0.915d 0.068 20 1-Butanol 0.920d 0.06a 17 2-Propanol 0 . 925d 0.06a 18 0.060 21 Acetone 0.940 Pyridine 0.985 0.099 12 0.1M LiC104supporting electrolyte. * 0.1M Et,NC104 supporting electrolyte. c Average of two or more values. d Obtained by extrapolation. Average slope of waves for solutions of different solvent composition. 0
which the half-wave potential of the couple was knoxn. Essentially identical ultraviolet and visible spectra for Fe(o-phen)S+* in water, methanol ( I ) , and acetic acid (2, 6) suggest very little, if any, solvent effect on the reduction potential of DMFe(III), DMFe(I1). The solvent molecules making up the secondary solvation sphere do not appear to modify the properties of the phenanthroline complexes. Because the liquidjunction being considered here is an aqueous saturated potassium chloride solution in contact with a nonaqueous 0.1M lithium perchlorate solution, one would expect the junction potential to be far more sensitive to variation in solvent than the potential of DMFe(111), DMFe(I1). Even if, on the basis
of these arguments, the entire difference in half-wave potential from one solvent to another is attributed solely to change in junction potentials, the magnitude of liquid-junction potentials does not appear to be as large &s is erroneously believed to be by many. In fact for nonaqueous solvents of dielectric constant ca. 80 to 10, the difference in junction potential for the type of junction involved here is a t the most ca. 0.15 volt and on the average only 0.05 volt. Identical halfwave potentials for DMFe(I1) in nitromethane 2 X 10-3dl and 1.23M in water also demonstrate the rather insensitive nature of liquid-junction potentials to the solvent system. The &(I), Cu(Hg) couple in nitromethane is affected very slightly by water (6); in 10-aM water solution it is +0.220 volt, and in 1M water solution $0.205 volt us. S.C.E. The solvation energy of Cu(1) ion in water and in nitromethane, therefore, a p pears to be similar, which means that the 0.077-volt difference in potential of the Cu(I), Cu(Hg) couple in water (Eo= $0.143 volt us. S.C.E.) (4) and nitromethane (El,z = +0.220 volt us. S.C.E.) must be essentially difference in liquid-junction potential. Taking into account the fact that the potentials E o ~ u ( ~ ) .and ~ u ( ~ g C)~ ( I ) , C , , ( R ~ )above are not strictly comparable and that the solvation energy of Cu(1) ion is not exactly constant, we find surprisingly good agreement between the difference of 0.077 volt and the 0.050 volt-difference in the half-wave potentials of DMFe(I1) in water and nitromethane, Le., difference in junction potential. The half-wave potentials for the reduction of rubidium ion in water and in acetonitrile differ by 0.05 volt (3). The 0.030-volt difference in the halfwave potentials of DMFe(I1) in these two solvents, Le., difference in junction-
potential, suggests difference in solvation energy of rubidium ion equivalent to 0.020 volt, which is not an unreasonable value. Finally, in a similar study with the ferricinium ion-ferrocene couple, close to identical AEl,z values in most cases were observed between solvents as were observed for the DMFe(III), DMFe(I1) couple, a fact which strongly suggests that the difference in potential observed for the half-wave potential of DMFe(I1) from one solvent to another is due to difference in liquid-junction potentials. Admittedly more work has to be done, and is being carried out, to check on the reliability of this experimental method of evaluating junction potentials, but, certainly, from the preliminary information available, use of the half-wave potential of DMFe(I1) in different solvents for the evaluation of liquid-junction potentials merits strong consideration. LITERATURE CITED
(1) Bjerrum,
J., Adamson, A. W., Bostrup, O., Acta Chem. Scand. 10, 329 (1956). (2) Brandt, F.W., Howsmon, W. B., Jr., J. Am. Chem. SOC.76,6319 (1954). (3) Kolthoff, I. M., Coetzee, J. F., Ibid., 79, 870 (1957). (4) Kolthoff, I. M., Lingane, J. J., “Polarography,” p. 227,2nd Ed., Interscience,New York, 1952. ( 5 ) Strehlow, H., 2. Elektrochem. 56, 827 (1952). (6) Weatherby, G., unpublished data, this laboratory, 1961. IVORY V. EELSON REYNOLD T. IWAMOTO Department of Chemistry University of Kansas Lawrence, Kan. RECEIVED for review August 7, 1961. Accepted August 21, 1961. Work supported in part by the Directorate of Chemical Sciences, Air Force Office of Scientific Research.
Determination of Calcium and Magnesium in Plant Material with EDTA SIR: Most plant material digests contain enough heavy metals and orthophosphate to prevent the successful titration of calcium and magnesium with EDTA. The interfering ions cause unstable end points and discoloration of the indicators. Methods for their removal, prior to titration, such as by ion exchange (6),precipitation (1, 3, 7), or solvent extraction of heavy metal complexes (d, 4,6) are tedious and t i i e consuming. 1796
ANALYTICAL CHEMISTRY
Orthophosphate can be removed from plant digests as the insoluble zirconium salt a t p H 5.5 to 6.5 (6, 8). However, the ability of the gelatinous hydrolysis product of quadrivalent zirconium salts formed a t this p H to remove heavy metals, apparently by adsorption, seems to have been overlooked. The interfering ions are removed to such a degree that on titration for calcium and calcium plus magnesium, using murexide (or Calver 11) and Eriochrome Black T,
respectively, there is no indicator discoloration or unstable or indistinct end points. This is true without the addition of inhibitors such as sodium diethyldithiocarbamate or potassium cyanide. The following method, which uses zirconyl oxychloride to remove orthophosphate and heavy metals simultaneously, is convenient, accurate, and especially suited for routine analysis of large numbers of samples. The
