(18) Griebel, C., Apoth. Zt 3, 172 (1951). (19) .Grill, F., J . Am. $harm. ASSOC., Scz. Ed. 21, 1012 (1932). (20) Haley, T. J., Ibid., 36, 301 (1947). (21) Ibid., 37, 378 (1948). (22) Hanson, A., S v a s k Kern. Tidskr. 58, 10 (1946). (23) Hey, D. H., J . Chem. SOC.1930,18. (24) Hun, M. C., Anales fac. farm. y
(34) Nagasawa, K., Ohkuma, S., Bull. Rail. Hyg. Lab. (Tokyo) 73, 113 (1955). (35) Nagasawa, K., Ohkuma, S., J . Pharm. SOC.Japan 74, 773 (1954). (36) Ogata, A,, Ibid., 451, 751 (1919). (37) Pharmacopeia of the United States XV, Mack Printing Co., Easton, Pa., 1955, pp. 55, 57. (38) Portnoy, E. O., Rev. guim. farm. (Santiago, Chile) 35, 21 (1945). bioquim. Univ. nacl. mayor Sun Marcos, Lima, Peru 3, 417 (1952). (39) Rathenasinkam, E., Analyst 7 6 , 115 (1951). (25) Ikuma, S., J . Pharm. SOC.Japan 72, 951 (1952). (40j Ibid., 77, 135 (1952). (41) Rotondaro, F. A., J . Bssoc. Oqic. Agr. (26) Itai, T., Igetai, H., Bull. Natl. Chemists 40,824 ( 1957). H y g . Lab. (Tokyo) 73, 127 (1955). (27) Keenan, G. L., J . Am. Pharm. ASSOC., (42) Sanchez, J., Semana m&d. (Buenos Sci. Ed. 37,519 (1948). Aires) 1932, 11, 1183. (28) Keenan, G. L., J . Assoc. Oqic. Agr. (43) Sandri, G. C., Mikrochim. Acta 1956 Chemists 25. 830 f 1942). ( 1-3), 244. (29) Kjeldgakrd,. E:, Fakm. Tidende 63, (44) Schmidt-Hebbel, H., Schmidt, E., 545, 561, 577, 593, 609 (1953). Bol. SOC. chilena qutm. 3, No. 1 , 5 (1951). (45) Shriner, R. L., Fuson, R. C., "Syste(30) I m a s , G. H. TV., Can. J . Research 288,37 (1950). matic IAentification of Organic ComDounds, 3rd ed.. Wilev, New York, (31) Martin, F., J . pharm. Belg. 6 , 283 i948. ' (1951). (32) Moraes, E., Palma, E., Anais fac. (46) Skaliks, H. C., Arzneimittel-Forsch. farm. e odontal univ. Sbo Paulo 12, 7, 386 (1957). 149 (1954). (47) Smodinos, E., Vuillaume, R., Bull. (33) Nagai, N., Pharm. Ztg. 32, 700 SOC. chim. biol. 32, 409 (1950). (48) Tsukamoto, H., Tsutsumi, IC., Ara(1887). -
I
maki, S., Sci. and Crime Detection 7, 16 (1954). (49) Uffelie, 0. F., Chem. Weekblad 41, 101 (1945). (50) Van Espen, J., Mededel. Vlaam. Chem. Ver. 15, 66 (1953). (51) Vidic, E., Klin. Wochschr. 30, 223 (1952). (52) Vignoli, L., Delphant, J., Sice, J., Bull. S O C . chim. biol. 28, 768 (1946). (53) Vincent, M. C., Krupski, E., Fischer, L., J . Ana. Pharm. ASSOC.,Sci. Ed. 46. 85 (1957). (54j'vitolo, A., Boll. chim. farm. 88, 103 (1949). (55) Vogel, G., Deut. Z. ges. gerichtl. Med. 41, 420 (1952). 156) Reiser. M.. Zacherl., M.., Mikrochim. Acta 1957 (3-4), 577. (57) 'Il'ickstrom, A., Salvesen, B., J . Pharm. Pharmacol. 4, 631 (1952). (58) Wimer, D. C., ANAL. CHEM.30, 77 (1958). (59) Yamagishi, M., J . Pharm. SOC.Japan 74, 1233 (1954). ~
RECEIVED for review December 5, 1958. Accepted April 16, 1959.
Stable Apparatus for High-Frequency Analysis KO ISHII,' SHOlCHl HAYASHI, and SHIZUO FUJIWARA University o f Electro-Communicafions, ChZifu, Tokyo, Japan
,A
general study of high-frequency analysis was undertaken and an apparatus was constructed. The responses of the instrument measured as the current and the shift of the tuning frequency were interpreted mathemaiically as the functions of the dielectric and the resistive characters of the sample contained in the condenser of the tuning circuit. Because the method of analysis is general and applicable to any kind of sample, the photochemical reactions in the leaves of waterpot plants were investigated. The reactions in soybean and buckwheat leaves illuminated by an electric lamp with and without a blue filter are accompanied by a change in resistance. Reactions of leaves illuminated b y the lamp covered b y a red filter are intermediate between those of capacitive and resistive natures.
I
high-frequency analysis, changes in the sample are detected by the detector circuit as measurable responses of current and frequency shift. I n the experiments reported here, the following two points must be considered: The loading effect of the sample on the N
Present address, Nip on Telegraph and Telephone Public &rp., Aoicha, Minatoku, TBky6, Japan.
1586
0
ANALYTICAL CHEMISTRY
oscillating circuit should be kept as small as possible, and the responses should be linearly proportional to the magnitude of the change in the sample. The first requirement is due to the fact that the sample forms a part of the oscillator and competes with the sensitivity of the apparatus. The use of the infinite impedance detector system and the mechanism of high-frequency titration have been discussed before (2, 3). This article presents some results .of a general consideration of high-frequency analysis and shows its application.
V2,is therefore used as the detector to eliminate this difficulty; the signal voltage, eout, the output voltage from the cathode of the left half of V2,is put in parallel t o the reference voltage, er, which is the output of the cathode of the other half of Vz, and the difference between e, and eout is applied to the meter. The function of each component of the apparatus may be explained by Figure 2, where et is the input voltage to the grid of VI and is assumed constant so long as the oscillating voltage is kept constant. Then, the input voltage to the grid of V 2 ,e,,, may be written as
APPARATUS
The block diagram of the apparatus is shown in Figure 1. The oscillator, Vo,is the Franklin type which has high stability of the oscillating frequency against the fluctuations of the constants of the oscillating tubes. It is suitable for measurements of long duration. A pentode tube, VI, is used as the buffer amplifier. The sample is contained in cell, C, of the tuning circuit, C , and LpIconnected to the plate of VI. The detector circuit and the measuring system consist of V zand the meter, M . The detector circuit is the infinite impedance type which works with good linearity when the input voltage t o V?is appropriate (higher than 6 volts). The disadvantage of this detector is loss of sensitivity with use of the cathodefollower scheme. A parallel triode tube,
where p, gm, and 2 are the plate resistance, mutual conductance of VI, and the tctal impedance of the tuning circuit, respectively. Because a pentode is used for VI, p >,Z and ei, = gm e, 2. As mentioned above, ei, is linearly proportional to e,, and hence, eout = k 2, where k = c g m e, and c is a constant. The current to be measured at the final stage of the measuring system, i, is given by i
= (eout -
er)/(T
+ ro)
=AZ+B A = gn e,[cl(r
+ ro)l
V*
VI tB 200 v
Ria
I
I
4 R5
I ntm
c,
Figure 1.
Block diagram of apparatus
c7, cs, ClO,
1 0 0 kfl
c2,
10 kR 1 kQ 1 MQ 50 kfl 200 ppf.
CI.
Cll, CY..
0 . 0 1 pf. 150 Npf. 100 amperes 4 mh.
6SJ7 6SJ7 6SN7 Cell
1 5 0 ppf.
50 ppf. 5 MPf.
+
B = - edr To), where r and ro are the equivalent internal resistances of the whole detector and measuring system, respectively. Impedance, 2, is equal a t resonance to the real part of the admittance of the tuning circuit and i=ARt+B
(1)
d i = A dRt
(2)
if C , (Figures 1 and 2) is adjusted to keep the circuit resonant. I n Equations 1 and 2, Rt is the impedmce of 2 a t l / R ) ,where resonance, Rt = l/(l/Ro R, and R are the resistances of the tuning coil and the cell, respectively. The derivatives dRI and di with respect to time are written as
+
dRt = [ ( R , / ( R
Figure 2. Schematic diagram showing function of apparatus
If the change in the sample is of the resistive nature,
+ R,)) I'dR + R,)] dR
dRz >>de, ( = 0) d i = A'(dg/dR,) dR,
d i = A (R,/(R =
A'dR
dCu = - ( d h / a R z ) dR,
(3)
The details of the responses of the apparatus when the sample is loaded in the capacitor, C (Figures 1 and 2), must be considered. The loading effect may be illustrated by the equivalent circuit (1) (Figure 3J) where C, C,, and R, refer to the capacitance between the condenser and the sample, the equivalent capacitance, and the resistance of the sample, respectively. Figure 3,A, may be rewritten as Figure 3,B, where C and R are the functions of R, and C,, and are written ais g(C,, R2)and h(C,, RI), respectively. Strictly speaking, the equivalent resistance of the condenser, R,, should be taken into account as well as C,, but in ordinary cases, it is neglected. According to the equivalent circuit of Figure 3,B, the following relation (4,5) exists
where d i and dC, are the readings of the current and of the shift in the resonant frequency. Hence.
B
A Figure
3.
Tuning system
A.
Equivalent circuit
E.
Rewritten form
(10)
di/dC, = -A' (ag/aR,)/(ah/aR,)
(11)
Similarly, if the change in the sample is of a capacitive or dielectric nature, d i / d C , = -A' (ag/aC,)/( ah/aC,)
(4) C = h ( R z ,C,) = Cc
1
+ ~ ~ R z ~ C c C z+( c zCc) + w z ( C , + CCl2R z 2
(12)
The derivatives of g and h with respect to C, and R, are given by Equations 13 and 16,
(5)
Equations to express the changes in the sample (dR,, dC,) are obtained as follolvs : From Equations 3 and 5, VOL. 31, NO. 9, SEPTEMBER 1 9 5 9
1587
Table I and Figure 5 show the results of the experiments. It is clearly shown that the di/dC, value for type 1 refers to the resistive change in the sample, whereas that for 2, to the one intermediate between ionic and dielectric natures.
It is assumed that R, and C, in Equations 13 and 16 can be replaced by con-
Fz,
stant values of 2,and respectively, if the changes in the sample are small. Then, with the aid of the procedures described below, the magnitudes of the right-hand sides of Equations 11 and 12 can be calculated. Results of the calculation of di/dC, can be compared to the value experimentally obtained and the nature of the change in the sample is interpreted.
4
RESULTS
Table 1.
Figure 4. Structure of condenser which holds sample Two copper plates are connected b y two polystyrene bars 2 mm. thick. Leaf i s held between copper plates. Light of the lamp reaches sample through 135 holes made in one plate. Numbers shown are given in millimeters
--I\
I
Determination of A' and R,. the circuit without the sample.
Tune Read
B ~ ~ n d ~ ~ ~ e ' r e ~ s s t ~ n ~ e s , c ~ circuit in parallel to the circuit and retune it.
Read the current ij. Then,
80
+=H=--
As an example of the application of the analytical method shown above, photochemical reactions of living plank were inqestigated. Waterpot Glycinr: rnax., Fagopyrum vulgare, and F . esculentum were kept inside of a large dark box, and a leaf of each was mounted in the tuning condenser with a structure as shown in Figure 4. The condenser mounted with the sample was illuminated by a 300-watt electric lamp. The output current and the tuning condenser were measured during the on and off periods of the light To prevent the heat of the lamp from reaching the sample, a water layer 5 cm. thick was placed between the condenser and the lamp. Experiments were carried out with the electric lamp uncovered and covered by violet or red filters so that the spectrum of the light was changed. Sizable effects of the light were observed with good reproducibility. The results of the experiments were classified into two groups of illuminations: (1) violet and normal electric light and ( 2 ) red light. In all cases. the output current, i, and the tuning capacitance, C,, were measured as the functions of time. I n the case of 1. i reached an equilibrium value as the light was turned on and returned to the former value as the light was turned off. Two types of current behavior, ascending and descending, were observed in 1. I n the present experiments, the ascending behavior was observed for a strong young plant and the descending, for an old and infirm plant.
PROCEDURES
,
,
5
Solve Equstion 17 for different Rj's and obtain A' and R, by using the least mean square method. Determination of C,. As seen in Figure 3 , A , C , is obtained by putting a dummy sample which has a shape similar to the sample itself and satisfies the condition R, = 0. I n the experiments described here, two dummy samples were prepared by the copper plates, 0.5 mm. and 1 mm. thick, and the areas inserted in the condenser were varied from 4 to 8 sq. cm. The results of the measurements of C , were 5 f 0.5 and 14 f 1.5 ppf. for the thin and thick plates, respectively, being insensitive to the variation of the inserted area. Determination of C and R. Tune the circuit when the sample is not loaded in the condenser and read the tuning condenser, C,, and the meter as CN and iN, respectively. Mount the sample and retune the circuit. The measured values of C , and i are referred to as CI and id, respectively. Then,
*b-:-*-y,L IO
~. ~.
._
' f
5
0
time (rnin) Figure 5. Results of measurements of current for photochemical reactions in living leaves mounted in tuning condenser Ordinate refers to current, i, measured b y meter, M, and abscissa refers to time of measurement
High-Frequency Analysis of Photochemical Reactions in Plants
(Resonant frequency, 2.1 Mc.) Type
8
Samplea
Exptl.
di,
C,,
R=I KQ
:J.
- h A', p./Q
x
10-11
pa.
a
Light Blue
60
10
14
0.74
b
I1
Normal
30
2
5
3.8
17.5
8.3
b
I11
Red
30
2
5
3.8
17.5
8.3
a, Fagopyrum esculentum. b, Glycine max.
1588
AC,,
Curve I
ANALYTICAL CHEMISTRY
WMf.
ppf.
100
33.92
d i / d C . ( X 10')
-
Exptl. 5 0 ~ 6 5 50
380
N
430
Calcd. Ionic 37 Dielectric 2500 Ionic 24 Dielectric 1160 Ionic 2.4 Dielectric 1160
18)
Substituting Equation 22 into 21, we get R, = 2/(w2C R C,) (23)
Solving this with respect to R,,
R,
=
ACKNOWLEDGMENT
(21)
R
=
.4’RoZ/(+” - i’) - Ro
(20)
Yumerical valum of R and C are obtained from Equations 19 and 20 b j ~ using the values of A’ and R,. Quantities of the sample, R, and C,, are then obtained by multiplying Equation 4 by Equation 5 to give CR
c, + dR,2 =
C,C,(C, w2CC2 R,
+ c,
+ C,) R z 2 w2C, C R, R
w4
CcCZR2- 4wz/C= (Cz
+ C,)
0
c,
=
-
The authors wish to express their gratitude for the support by the Grant in Aid of the Ministry of Education in Japan. LITERATURE CITED
Solving this,
I
c, * dC2 + W ~ C C 2 C ~ X Z 2
With the condition that C, must be positive,
lr-hichcan be rewritten [ts w’C,(C,
As Equation 21 refers to two positive real roots for R,, it is necessary to have
(1) Blaedel, W. J., Malmstadt, N. V.,
Petitjean, D. L., Anderson, W. K., ANAL.CHEM.24, 1240 (1952). (2) Fujiwara, S., Hayashi, S., Zbid., 26, 239 (1954). (3) Reilley, C. N., McCurdy, W. H., Jr., Ibid., 25, 86 (1953).
RECEIVED for review September 10, 1958 hccepted hpril 17, 1959.
+1=0
Arsenic in Naphthas GEORGE W. POWERS, Jr., RONALD L. MARTIN, and FRANK J. PIEHL Research and Development Depwtment, Standard Oil Co. (Indiana), Whiting, Ind. J. MARCUS GRIFFIN Research Department, Utah Oil Refining Co., Salt Lake City, Utah
,Arsenic in naphthas is determined a t the parts per billion level b y a combination of chromatography and colorimetry. It is first quantitatively adsorbed from naphtha onto silica gel impregnated with sulfuric acid. The adsorbent is then’digested with sulfuric, nitric, and perchloric acids to remove organic matter and to dissolve the arsenic, which is subsequently converted to arsine and determined colorimetrically with silver diethyldithiocarbamate in pyridine. Precision and accuracy ore about 3% relative or 0.3 p.p.b., whichever is greater. Four to six samples are analyzed in 8 hours. Periodic analyses of naphthas during storage show that an inherent source of error in any method is adsorption of arsenic on the sample container. Although of the arsenic present as much as in a typical naphtha can b e lost in 4 hours, simple techniques eliminate the loss. The arsenic compounds in virgin naphtha seem to b e alkyl arsines.
T
constituents in petroleum profoundly affect catalytic processes. I n catalytic reforming over platinum, a few parts per billion of arsenic in the naphtha feed gradually deactivate the catalyst (8) and eventually destroy its activity. Because naphthas contain varying small amounts of RACE
arsenic (7, l a ) , they must be monitored for arsenic content. The problem is twofold; the arsenic must be concentrated and then determined. Three ways of concentrating the arsenic in naphtha have been described. The usual one is digestion with sulfuric acid, nitric arid, and hydrogen peroxide (10, 12, IS) to oxidize organic matter and dissolve the arsenic. Combustion of the naphtha in a Beckman oxyhydrogen burner and absorption of the combustion products in sodium hydroxide solution have been described recently (1). Chromatographic adsorption on alumina (9, 22) or silica gel (4) has been suggested but has not been adequately tested. The simplicity and speed of the chromatographic approach make it potentially the most useful. Four methods of determining arsenic have been used. Neutron activation is the most sensitive (16, 18), but few laboratories have the needed equipment; moreover, errors as large as 20 p.p.b. a t the level of 50 p.p.b. have been encountered in the method as used a t Oak Ridge (11). Emission spectrosropy (22) lacks precision and sensitivity. The Gutzeit method (15), in which arsine reacts with mercuric bromide to form a yellow spot on paper, is the most sensitive chemical method; although most procedures (9, 10, 13) use it, obtaining reproducible spots
and measuring the intensity of them quantitatively is difficult. The molybdenum blue method (6, 19) avoids these difficulties (12) because the color is formed in solution; however, it is only one tenth as sensitive as the Gutzeit method. A new method combines the advantages of chromatographic concentration and colorimetric determination. The arsenic is concentrated on a new adsorbent-silica gel impregnated with sulfuric acid-and determined with silver diethyldithiocarbamate in pyridine, which gives a soluble red complex with arsine (5, 20, 21). This reagent is somewhat simpler to use than molybdenum blue and is equally sensitive. DEVELOPMENT
OF
METHOD
I n developing the new method, four problems were: how to collect samples for analysis, to concentrate the arsenic, to convert it to arsine, and to determine the arsine. Two special sampling techniques were devised when it was discovered that as much as 30% of the arsenic can be adsorbed on the sample container in 4 hours. I n one, the naphtha is sampled into a bottle that has been prewashed with acid; after the naphtha has been withdrawn, the arsenic is desorbed from the glass by rinsing x i t h concentrated sulfuric acid, and the acid and naphtha VOL. 31, NO. 9, SEPTEMBER 1959
1589