A Method for Calculating Gas Chromatographic Relative Retention Values for High Boiling Phenothiazine Derivatives H. F. MARTIN, J. 1. DI;:ISCOLL, and B. J. GUDZINOWICZ Boston laboratory, Monsanfo Research Corporation, Everett, Mass.
bA
method applying established relationships is presemnted for determining the relative gas chromatographic retention values of some high boiling phenothiazine drugs and their derivatives. The assumption is made that a relationship between the log of the retention time and the boiling point holds for phenothicizines chromatographed at high temperatures on a nonpolar liquid stationary phase such as SE-30. The unknown boiling points, from which the retention times are predicted, are calculated from reported boiling points at reduced pressures and from boiling point numbers, which are summations of atomic and/or molecirlar increments.
U
isothermal or programmed tempwature chromatographic column conditions, many investigators have mtablished that a direct relationship exists between a compound’s retention time and its boiling point on nonpolar liquid stationary phases. ‘This relationship holds true for materials whose boiling points have been established, but because of limited or nonexistent boiling point data, fcw correlations in the field of high temperature gas chromatography have been reported. The phenothiazine drugs, which are solids a t room temperature, fall into this category. This paper describes a method whereby the boiling points and relative retention times for a series of high boiling, high ;nolecular weight related compounds c m be calculated, with fair accuracy, from a minimum of reported data. Good agreement exists between the predicted and observed retention times. This method has been applied by Martin (8) to estrogens and by Gudzinowicz and Martin ( 4 ) to organo- and organobromoarsines. Here, the method is described as it has been applied to the various phenothiazine derivatives shown in Table I. Assuming a simple kinetic model ( I I ) , the retention time of a compound varies with its he:tt of vaporization. For members of a related series of
Table 1.
Structure and Nomenclature of Phenothiazine Derivatives Investigated 2 3 1 Promazine Desdimethylchiorpromazine Triflupromrtzine
@ 0 %I F 3 (CH2)3
I
HaC/NbH3 5
4 Desmethylch1orprom:rzine
Chlorpromazine
6
RlethoxyprornaziIie
SINQ EITHER
HnC
A,
H 8 Deadirnethylchlorpromazine
7 Trifluoperazine
acetamide
9
Prochlorperazirie
QCXZCF3 I
‘N-N-
u
CH3
10
Desmethylchlorprornahe acetamide
a:Jac* I
(C”2)3
I
o
H3CyN\C- // CH3
compounds, such as the phenothiazine derivatives chromatographed at high temperatures on a nonpolar liquid phase, it is expected that the corn-
pounds will elute from the column approximately in the order of their boiling points. Since the heat of vaporization for each compound is approximately VOL. 35, NO. 12, NOVEMBER 1963
1901
equal to the product of Trouton’s constant ( K T )and its boiling point (BP), it follows that
+ In 12
KTBP In t, = - __ RT
where t, = retention time
R T
= gas constant = absolute temperature
k
=
column constant
Equation 1 thus shows the linear relationship between retention times and boiling points. This is a limited version of an equation derived and tested by Grant and Vaughn ( I ) . To determine a compound’s boiling point a t atmospheric pressure, two methods are applicable. The method of Haas and Newton (6) is based on a known boiling point value for a compound a t a reduced pressure from which its boiling point a t 760 mm. Hg is computed. The second method proposed by Kinney (7) is based on numerical boiling point number ( B P N ) values assigned to atomic and/or molecular groups, the summation of which is the compound’s boiling point number. This summated value is related to boiling point by Equation 2. BP
=
230.14 q/spy - 543
Table II.
-H
The instrument used was the Aerograph A-600 Hy-Fi (Wilkins Instrument Co., Walnut Creek, Calif.) with a hydrogen flame ionization detector. The column, made of 4 feet of 1/8-inch 0.d. copper tubing, was packed with 5% by weight SE-30 (General Electric Silicon Gum Rubber) on silanized 60to 80-mesh Diatoport. The silanization, which was carried out for 18 days with hesamethyldisilazane, is necessary to minimize the adsorption of the solid support. The column and injector port temperatures were both maintained a t 270” C. The nitrogen carrier gas inlet pressure was 16 p.s.i. and the hydrogen flow was maintained a t 23 cc. per minute by the Aerograph Model A-650 generator.
B PA’
increments
Origin of B P S increment
1.00 8.40 3.05
Lit. ref. (4) Lit. ref. (4) Lit. ref. (4)
group
2.00
Lit. ref. (4)
group
5.00
Lit. ref. (4)
7.30 5.50
Lit. ref. ( 4 ) Calc’d. from
8.10
Calc’d. from
2.10
Calc’d. from
-(cH2)3-
-CHI
C1 substituted for H
EXPERIMENTAL
Boiling Point Number Increments for Various Atomic and/or Molecular Groupings
Atomic/molecular groups
NH2 in -(CH2-NH2)
(2)
The various boiling point number increments attributed to the atomic and/or molecular groupings have been established by Kinney (7) and Herbrandson and Nachod (6). Table I1 lists the values used in our present investigations. Although the original method described in the literature was not designed for aromatics and fused ring systems, it was our contention that if a BPN could be determined for the phenothiazine nucleus, then by adding increments t o this stem, the BPN and thus the B P for different derivatives could be estimated.
group
0
@CI
0
@CFz
and
and
21 .00
Calc’d. from six acetamides
23.00
Calc’d. from three methyl piperazine compounds
6 1 . GO
Calc’d. from phenothiazine
~~
1902
ANALYTICAL CHEMISTRY
0
2
4 6 8 IO RETENTlOd TIMEMINUTES)
12
Figure 1. Gas chromatogram of a 10-component mixture of phenothiazine derivatives. Numbered peaks correspond to compounds identified in Tables I and IV Operating conditions: Aerograph Hy-Fi with 4-foot b y ‘/s-inch 0.d. column packed with 5% SE-30 on 60- 80-mesh Diatoport (silanized for 18 days wiih HMDS). Column temperature, 27OOC.; inicction block, 27OOC.i hydrogen flow rate, 23 cc. per minute; nitrogen pressure, 16 p.s.i.
With the exception of the acetamides of desmethyl- and desdimethylchlorpromazines, the majority of the compounds shown in Table I were received as the hydrochloride salts from the Pharmacology Unit, Psychopharmacology Service Center, National Institute of Mental Health, Bethesda, Md. These were all converted to their free bases prior to injection into the chromatograph since hydrochloride salts per se are decomposed on the column. The stability of the free bases and the degradation of hydrochloride salts have been verified by comparing the infrared spectra before and after chromatographing. The thermally stable desmethyl- and desdimethylchlorpromazine acetamides were prepared by acetylating the free bases with acetic anhydride in a pyridine solution. After the removal of the excess anhydride and pyridine by evaporation, reagent grade carbon disulfide was used t o dissolve the residual acetamides for subsequent injection into the instrument. Carbon disulfide was the solvent selected because of its low response with the flame detector. Under constant chromatographic operating conditions, the observed retention time for each compound was determined individually, and these have been used to identify each peak in the gas chromatogram for the 10component mixture shown in Figure 1.
Boiling Point, Boiling Point Number, and Relative :Retention Time Calculations. With the exception of
the boiling point of phenothiazine a t 760 mm. Hg and the boiling points at reduced pressure of triflupromazine (IO) and the chlorpromazine (9) shown in Table I, very little boiling point data for the phenothiazines have been reported in the literasure. Using the available reduced prescnre-boiling point data, the boiling poini,s a t atmospheric pressure for trifluproniazine and chlorpromazine can be cdculated by the method of Haas and Newton assuming that entropy of vaporization of the phenothiazines is bes i represented by the value assigned to Group 2 compounds (5). The boiling point results are shown in Table I11 for these materials together with their BPN values calculated using Equation 2. To calculate the L1P and BPN of chlorpromazine by incbrement addition, m e starts with the BPN of phenothiazine (Table 111). To this phenothiazine nucleus, the BPN for one hydrogen atom is subtracted and the BPN values for chlorine substitution on the ring and the dimethylaminopropyl chain are added. These values and their origin are shown in Table 11. The BPN value for chlo,.promazine using this method is 83.6, which is in good agreement with the calculated 83.0 value obtained from reported B P data shown in Table 111. The latter is taken as the true value. The BPN values for the other phenothiazine derivatives shown in Table IV were also determined in this manner. For trifluoperazine, the methyl piperazine increment was added to the triflupromazine BPN obtained from B P data, from which the contribution of --N(CH& had been subtracted. When the BPN for either triflupromazine or trifluoperazine (com~~ounds1A and 7A, respectively, in Figure 2) is determined by the substitution of a CF3 group value for hydrofgen on the phenothiazine nucleus, the values obtained are not in agreement with the rest of the boiling point data. This is shown in Figure 2, which is a plot of the observed retention time as a function of the calculated boiling point for each compound listed in Table IV. A similar plot is obtained for boiling point numbers. This deviation for the fluorine-containing compounds is not too surprising since the momalous boiling point behavior of fluoro-compounds is well known (2, 3, 12) Therefore, i t is our opinion that the BPN value for CF3 in Table I1 is in error. Assuming that all other values o i BPN increments listed are correct, the CF3 increment relative to other increments would be negative in order to bring the calculated boiling points for triflupromazine and trifluoperazine into line with the other phenothiazine data. Gsing Equation 1 and the calculated boiling points (Table IV) and observed retention times (Figure 1) for triflupromazine and desmethylchlorpromazine acetamide, which should represent the average slope of the series, the constants KTIRT and In k are determined
IO 9 8
'
Figure 2. Plot of logarithm of retention time vs. calculated boiling points. Numbers correspond to compounds identified in Tables I and IV
I
I
I
io
60
I
7
-g5 5 4
w
E, F z z W
h 2
'420
eb
SA0
do
4b
20
BOILING POINT, *C.
by solving simultaneous equations. The retention times of the remaining eight compounds are then determined using these constants. Their retention times relative to trifluuromazine are noted in Table I V in addition to their predicted values and order of elution. There is another method of calculation that can be used. Once the boiling point increments have been calculated as boiling point numbers or as calculated boiling points, then the relative retention times can be estimated by either of two methods with just one experimental point. I n general, the retention time
Table 111.
.-A
0
=
+ K (BP,) + In k
log tr,
=
K ( B P I - BP,?
where = retention time B P = boiling point Subscripts 1, 2, and n = experim ental
t,
Calculated Boiling Points and Boiling Point Numbers for Phenothiazine, Triflupromazine and Chlorpromazine Caldd - ..~....
Compound Phenothiazine Triflupromazine Chlorpromazine a Lit. ref. ( 9 ) . b Lit. ref. (10). Table IV.
corresponding to the boiling point or the boiling point number of the nucleus serves as the arbitrating unit (one) in the ordinate of a plot of the log of retention time us. the boiling Doint. log t i , = K ( B P I ) In k
Reported b.p., ' C.
b.p. at 76,O mm. Hg, C.
Calc'd. BPN
371 176 at 0 . 7 0 mm." 200 at 0.80 mm.*
371 432 461
62.6 75.8 83.0
Calculated Physical Constants and Observed Chromatographic Data for Phenothiazine Derivatives
Compound Triflupromazine Promazine Desdimethylchlorpromazine Desmethylchlorpromazine Chlorpromazine Methoxypromazine Trifluoperazine Desdimethylchlorpromazine acetamide Prochlorperazine Desmeth;lchlorpromazine acetamide
Numerical designa- Calc'd. tion b.p.
Calc'd.
2
43 1 438
BPAI7 75.8 77.5
3
456
4 5 6 7
Relative Elution sequence retention time PreObCalc'd. Observed dicted served 1.0 1.1
1.0 1.3
1 2
1 2
82.2
1.6
1.9
3
3
461 461 470 490
82.9 83.0 85.6 90.7
1.7 1.7 2.0 2.9
1.9 1.9 2.1 2.9
4 5 6
7
3 3 4 5
8 9
513 520
97.0 98.0
4.5 5.1
4.6 5.3
8 9
6 7
10
523
99.0
5.3
5.3
10
7
1
VOL. 35, NO. 12, NOVEMBER 1963
1903
point, unknown, and nucleus of the phenothiazines, respectively.
R
= slope
k = column constant ,4nd for any subsequent point log t,,
=
(E*)
log L1
Similarly, the same may be done for the boiling point numbers except that it will be noted that i t is the cube root of the boiling point that is related to the log of the retention times. DISCUSSION
OF
RESULTS
Based on the data presented, the feasibility of predicting the order of elution of certain phenothiazines, their separation on gas chromatographic columns, and the type of derivative most desirable to resolve a nonseparable pair, when derivatives are possible, in addition to presenting a method for calculating their relative retention values, have been shown and established. In Table IV one notes that the predicted and observed order of elution of the compounds is in good agreement but that compounds 3, 4, and 5, and 9 and 10, as expected, cannot be separated. From the boiling point data, it would also be questionable to predict the separation of compounds 1 and 2, which, however, do separate. For identification purposes, the compounds shown numerically in Figure 1 cor-
respond to those numbered in Tables I and IV. In contrast to predicting the elution sequence, prediction of the relative retention values is subject to greater deviation. Nevertheless, the average deviation between the observed and calculated relative retention times is 8%. Although compounds 3, 4, and 5 could not be separated on a SE-30 column as predicted, acetylation of desdimethylchlorpromazine (compound 3) and desmethylchlorpromasine (compound 4) yields acetamides (compounds 8 and 10, respectively) that are separable. Despite the method’s possible applications and obvious advantages, there are several disadvantages. First, the equation AH, = KrBP holds true only for a small number of related compounds. Second, there is a sparsity of boiling point data available to establish valid boiling point number increments for specific atomic and/or molecular groupings. Third, this method cannot be used to make any predictions for isomeric structures. Nevertheless, it is our opinion that the method as described can be applied to various types of organic molecules when boiling point information is limited or nonexistent. ACKNOWLEDGMENT
The authors thank C. J. Carr and G. Cosmides of the Pharmacology Unit, Psychopharmacology Service Center,
National Institute of Mental Health, Bethesda, Md., for the phenothiazine samples used for this investigation. LITERATURE CITED
(1) Grant, D. W., Vaughn, G. A., J. Appl. Chem. London 6,145 (1956). (2) Gudzinowice, B. J., Alm, J., Driscoll,
J. L., Smith, W. R., Abstracts, Tenth Detroit Anachems Conference, Wayne State University October 1962. (3) Gudeinowicz, b. J., Driscoll, J. L., J. Gas Chromatog. 1, No. 5, 25 (1963). (4) Gudzinowicz, B. J., Martin, H. F., ANAL.CHEM.34, 648 (1962). (5) Haas, H. B., Newton, R. F., in “Handbook of Chemistry and Physics,” 38th ed., pp. 2121-2, Chemical Rubber Publishing Co., Cleveland, Ohio, 1956. (6) Herbrandson, H. F., Nachod, F. C., in “Determination of Orgamc Compounds by Physical Methods,” p. 16, Academic Press, New York, 1955. (7) Kinney, C. R., J. Am. Chem. SOC.
60,3032 (1939). (8) Martin, H. F., Ph.D. thesis, Boston University, Boston, Maas., 1961. (9) “Merck Index of Chemicals and Drugs,” 7th ed., p. 250, Merck and Go., Inc., 1960. (10) Ibid., p. 1065. (11) Pecsok, R. L., “Principles and Practices of Gas Chromatography,” pp. 847, Wiley, New York, 1959. (12) Sievers, R. E., Moshier, R. W., Ponder, B. W., Abstracts, p. 35M, 141st Meeting, ACS, Washington, D. C., March 1962. RECEIVEDfor review April 23, 1963. Accepted August 28, 1963. Division of Analytical Chemistry, 145th Meeting, ACS New York, N. Y., September 1963.
Work sup orted by the National Institute of Mentay Health under Contract No. PH 43-62-195.
Determination of Iron with Diethylacetic Isotope Dilution Analysis
Acid by
INARA MENCIS and THOMAS R. SWEET McPherson Chemical laboratory, The Ohio Stafe University, Columbus I 0, Ohio
b Solvent extraction, isotope dilution, and spectrophotometry are combined to determine small amounts of iron in the presence of copper. Diethylacetic acid is used as the reagent and solvent for the formation and extraction of the iron complex. Fe69is used as the radioactive tracer.
A
isotope dilution, spectrophotometry, and solvent extraction have been suggested for the determination of cobalt (3, 4). A study of the extraction of iron in diethylacetic acid indicates that the iron complex with diethylacetic acid is extracted into diethylacetic acid at p H values rn low aa 1. Before the iron is completely extracted, copper begins to NALYSES INVOLVING
1904
0
ANALYTICAL CHEMISTRY
extract. Since complete extraction of iron would lead to larger copper corrections, especially for high copper content samples, it was found convenient to use isotope dilution analysis. The absorption curves of iron and copper are shown in Figure 1. On the basis of these curves, it was decided to measure the absorbance at 345 and a t 685 mp. The amount of copper extracted can be corrected for by using a working curve which relates the absorbance of copper at 685 and at 345 mp. EXPERIMENTAL
Apparatus. The absorption curves shown in Figure 1 were made on a Cary Model 14 recording spectrophotometer. All other absorbance meas-
urements were made with a Beckman DU spectrophotometer using I-cm. matched rectangular silica cells equipped with ground glass stoppers. Solvent extractions were performed with A-8312 Duraglass bottles with 20-400 polyseal caps (Owens-Illinois Glass Co.). All radioactivity measurements were made with a RIDL Model 49-54 scaler. The detector used was a 5- X 4-inch, thallium-activated sodium iodide well crystal. Reagents. Diethylacetic Acid. Practical grade diethylacetic acid (Matheson, Coleman, and Bell) was purified by distillation. Infrared spectral analysis indicated t h a t the acid is stable for a t least 2 months. Standard Iron Solutions. High purity iron wire (99.9% Fe) was cleaned by immersion in dilute nitric acid and was
