'
01 ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' 0 I 2 3 4 5 6 7 8 9 1 0 1 2 3 4 5 6 7 8 9
NOMINAL
pH
Figure 5. Solvent extraction curves of metal oxinates Salting-out agent, 65 wt % sucrose: extracting agent, 50 pprn oxine
In
acetone
adjusted with acetic acid to the p H of maximum extraction of the desired metal. Three milliliters of freshly prepared 1% APCD solution were added and solvent extraction was performed with two 50-ml portions of acetone. The results are summarized in Table III. The concentrations were calculated assuming complete extraction of the impurities. Significant quantities of all the elements tested were present and would interfere in many analyses. However, purification of the calcium chloride solution is sufficiently and easily performed by pre-extraction. Other compounds potentially useful as salting-out agents (4) also contained impurities. For example, a saturated ACS reagent grade aluminum chloride solution was observed to contain the following concentrations of impurities: iron, 110 ppm; nickel, 1.5 ppm; cobalt, 0.005 ppm; copper, 0.3 ppm; and manganese, 0.2 ppm.
Solvent Extractions from Sucrose Solutions. Sucrose was chosen for further study as a salting-out agent because of its lack of transition metal impurities and the absence of a complexing cation. Extractions were performed using 65 wt % sucrose as the salting-out agent. The results are shown in Figure 5 . In all cases, the solvent extraction curves of metal oxinates using acetone were similar to those found using chloroform (9). Titanium(1V) is completely extracted by 0.1N oxine in chloroform in the pH range 2.5-9.0. Similarly, vanadium(V) is quantitatively extracted into chloroform at pH 2-6 (9). It is not extracted above pH 9. The extraction curve observed here for vanadium is very similar to the one reported by Talvitie (15). Iron(I1) is not extracted below a pH of 4 (9). In neutral and basic solutions, it is oxidized to iron(II1) and extracted as the iron(II1)-oxine chelate. Iron(II1) extracts with 0.01 to 0.1M oxine in chloroform in the pH range 2 to 10. Copper(I1) extracts into chloroform with 0.1M oxine in the pH range 2-12. Solvent extraction with sucrose as the salting-out agent is not as convenient as with calcium chloride solutions because of the high viscosity of the solutions. In addition, the sucrose concentration, pH, temperature. and volume have to be very carefully controlled (4). However, the use of sucrose as a salting-out agent does permit the use of chelating agents that would react with calcium. Applications. Since the atomic absorption measurement of manganese exhibits minimum detectability in acetone (2), the solvent extraction of the manganese-dithizone chelate into acetone is being evaluated for the determination of physiological levels of manganese in digested blood samples. Preliminary results are more precise than with other procedures and the method requires less sample. This study is being continued.
Received for review May 7, 1973. Accepted July 26, 1973. The work on which this report is based was supported in part by the Office of Water Resources Research, Department of the Interior, under provisions of Public Law 89379, as Project No. A-013-KY. 115) N . A . Talvitie, Ana/. Chem., 25, 604 (1953)
Stable Isotope Dilution Applied to the Determination of Zirconium in Geological and Lunar Samples Shin Tsuge, J. J. L e a r y , and T. L. lsenhour
Department of Chemistry, University of North Caroiina. Chapei H i / / . N . C . 27514 The method of metal labeled stable isotope dilution (MLSID) is developed in detail. This technique is then applied to the microdetermination of zirconium in standard geological and lunar samples as the volatile chelate tetrakis (1,1,1,2,2,3,3-heptafluoro-7,7-dimethyl-4,6-octanedionato) Z r ( l V ) [ Z r ( f ~ d ) ~by ] means of a mass spectrometer equipped with a voltage peak switching facility. Samples analyzed include a USGS standard ( A G V - l ) , a CAAS standard (SY-l),and five Apollo 14 and 15 samples (1431 0,135; 14321,188: 15021,96: 15301,78: and 106
15471,30). This method of zirconium determination requires only small samples (10-40 mg), can be performed using an ordinary electron impact mass spectrometer, and is essentially free from matrix interferences.
Zirconium determination in geological samples has always been notoriously difficult. Large variations in the reported zirconium concentration of lunar samples indicate the need to improve existing methods and/or develop new methods of analysis to determine this element.
A N A L Y T I C A L CHEMISTRY, VOL. 46, NO. 1, J A N U A R Y 1974
Zirconium in nature occurs in the form of extremely refractory oxides such as baddeleyite (ZrO2) and zircon (ZrOz SiOz); therefore, if prudent dissolution processes are not used, the zirconium compounds often remain as an insoluble residue. Once a zirconium-containing sample has been dissolved, emission spectrometry and atomic absorption spectrometry can be used for the analysis, but these methods are often complicated by the formation of very refractory oxides in the arc or flame. Although the sensitivity of atomic absorption for zirconium has recently been improved (1, Z) , the technique is still not generally applicable to geological samples. X-Ray fluorescence methods are complicated by the overlap of interfering lines, the absorption-excitation effect, and the preparation of a suitable target. Spectrophotometric ( 3 ) and spectrofluorometric ( 4 ) methods are applicable to the analysis but the matrix often introduces serious interferences. Neutron activation analysis has very poor sensitivity for zirconium ( 5 ) , because of the zirconium's extremely small neutron cross section. Recently, a method for the preparation of volatile zirconium chelates from geological samples has been published (6). The reported mass spectral detection limit for one of these chelates, Zr(1V)-benzoyltrifluoroacetonate (Zr(bta)4), is 5 X lO-I4g ( 7 ) .The stable isotope dilution technique has been successfully applied to the chromium analysis in geological samples using mass spectrometry of Cr(II1)-trifluoroacetylacetonate (Cr(tfa)s) (8). This paper describes an application of the isotope dilution technique to the determination of zirconium in standard geological and lunar samples using Zr(1V)1,1,1,2,2,3,3-heptafluoro-7,7-dimethyl-4,6-octanedionate (Zr(fod)c). Important advantages of this method are the following: matrix interferences are essentially eliminated since other metal chelates give different mass fragments than the zirconium chelate; sample sizes are small which is particularly important for lunar samples; and the final analysis can be performed on a medium-resolution organic mass spectrometer. Metal Labeled Stable Isotope Dilution. Metal labeled stable isotope dilution (ML-SID) is rapidly becoming a standard technique in spark source mass spectrometry (9). However, the technique also has great potential when used in conjunction with volatile metal chelates. All metals which have more than one naturally occurring stable isotope are now available with altered isotope distributions (Isotope Development Center, Oak Ridge National Laboratory, P.O. Box X, Oak Ridge, Tenn. 37830) a t very reasonable costs, often less that $l.OO/mg. Stable isotopes are ideal internal standards. Except in the case of the very lightest elements, all isotopes of an element exhibit virtually identical chemistries and, furthermore, there are none of the safety problems associated with radioactive isotopes. This is particularly desirable when biological systems are involved. Since most metals can be converted A. M. Bond, Anal. Chem., 42, 932 (1970) R . M . Dagnall, G. F. Kirkbright, T. S. West, and R. Wood, Anal. Chem., 42, 1029 (1970). R . Villarreal, J. W. Young, and J. R. Krsul, Anal. Chem.. 42, 1419 (1970). T. D. Filer, Ana/. Chem., 43, 469 (1971) W. S. Lyon, "Guide to Activation Analysis," D. Van Nostrand Co., Inc., New York, N.Y.. 1964, p 167. S. Tsuge, J. J. Leary, and T. L. Isenhour, Anal. Chem., 45, 198 (1973), M. G. Allcock, R. Belcher, J. R . Majer, and R. Perry, Anal. Chem., 42, 776 (1970). N. M. Frew, J. J. Leary, and T. L. Isenhour. Anal. Chem., 44, 665 (1972). H. Farrar, "Relating the Mass Spectrum to the Solid Sample Composition" in "Trace Analysis by Mass Spectrometry,'' A. J. Ahearn. Ed., Academic Press, New York. N.Y., 1972.
into metal chelates, the problems of volatility, separation, and detection are, for the most part, solved simultaneously by using a mass spectrometer. With the increasing availability of medium-resolution mass spectrometers ( M I A M 5 5000) in chemistry departments, this method can become a very practical analytical technique. A general discussion of the stable isotope dilution technique was given by Webster in 1960 (10). Until recently, however, the application of this method was mostly restricted to mass spectrometers with thermal or spark sources for nonvolatile elements and electron impact sources only for gases. More recently, the development of methods for the synthesis of volatile metal chelates has permitted the application of this technique to the determination of many metals using an ordinary electron impact mass spectrometer. Although applications of ML-SID have appeared in the recent chemical literature (8, 11-13), a thorough development of the underlying principles has not. Most analyses can be performed using the lightest two isotopes, and many complications and interferences are eliminated. Therefore, the next few paragraphs will be devoted to a systematic development of the case where the mass positions being monitored contain the two lightest stable isotopes of the metal ( e . g . , 50Cr,52Cr; S4Fe,56Fe; 90Zr,91Zr, etc.). The discussion will be generalized to cover the effects of heavy isotopes in the ligand part of the molecule ( e . g . , W ) ; however, the initial part of the development is much more straightforward if other isotopes are neglected. When a sample mixture containing both the normal (N) and enriched ( E ) isotopes of element M is admitted to the source of the mass spectrometer, the total signal a t the detector due to the lightest isotope of the metal (LM) is given by Equation 1. T o t a l ' M s i g n a l = ' M y signal + ' M E signal (1) where the signal due to 'MN is proportional to the number of naturally occurring atoms of 'M present and the superscript 1 refers to the lightest isotope. This relationship is given by Equation 2.
'MN signal a number of naturally occurring 'M atoms = 'ANX (2) 'AN is the isotopic abundance of the lightest naturally occurring isotope of the element M, and X is the total number of natural atoms of the element in the sample. Now writing the ratio ( R ) of the intensity of the two isotopes 1 and h gives Equation 3. h is the second stable isotope of element M.
R=
hAzX 'AsX
+ hA~Y + 'AEY
(3)
X and Y are the number of atoms of the natural and enriched isotopes, respectively. A is the isotopic abundance, with the superscript referring to the isotope and the subscript defining whether the isotope came from the natural or enriched part of the sample. Solving Equation 3 for X gives
(10) R. K. Webster, "Mass Spectrometric Isotope Dilution Analysis" in "Methods in Geochemistry." A. A. Smales and L. R. Wagner, Ed., Interscience, New York, N.Y. 1960. (11) J . K. Terlouw and J. J. deRidder. Fresenius' Z. Ana/. Chem.. 250, 166 (1970). (12) J. K. Terlouw and J. J. deRidder, Fresenius' Z. Anal. Chem.. 250, 377 (1970). (13) J. K. Terlouw and G. Dijkstra, Reprint of International Conference on Mass Spectrometry, Brussels, August 1970.
A N A L Y T I C A L C H E M I S T R Y , VOL. 46, NO. 1, J A N U A R Y 1 9 7 4
107
Using the definition of the mole, Equation 3a can be expressed in terms of weights rather than numbers of atoms. This result is given in Equation 4.
x and y are the weights of the pure natural and enriched metal, respectively. Equation 4 could, in theory, be used for an analysis if the weight of the enriched metal added to a sample is known, and if the isotopic abundances of the pure natural and enriched metals are known. The general procedure for such an analysis is to add a known weight of the enriched element to the sample containing an unknown weight of the natural element. The sample is then subjected to whatever chemical steps are necessary to give a homogenous mixture of the isotopes from both sources. The ratio ( R ) of the abundance of the two isotopes 1 and h is measured, and Equation 4 is used to determine the weight of the metal initially present in the sample. It is convenient to rearrange Equation 4 by multiplying the numerator by 'AE/'AEand the denominator by 'ANwhich gives Equation 4a
or the corrected ratios ( R ) can be used with Equation 4a because if uncorrected ratios are used, F is subtracted out of both the numerator and denominator, giving Equation 4a as written. A more difficult problem lies in experimental discrimination. This means that the ratio being measured is DRU rather than RU where D is the experimental discrimination factor. D might be due to nonlinearity in the detection system, changes in instrument sensitivity as a function of mass, systematic errors in the measurement of peak areas, etc. Equation 6 is a modification of Equation 4a, taking experimental discrimination into account
In general D, DE, and DN are not the same and it might appear that Equation 4a has been reduced to a useless state by practical considerations; however, back calibration holds the solution to this problem. If an accurate standard of the natural metal can be made and mixed with a known amount of the enriched metal, the apparent amount of the enriched metal (APP) can be experimentally determined by rearranging Equation 6.
where where st = standard. Now if a known amount of the enriched metal is added to an unknown sample containing this metal, Equation 6 again applies 1
K = ,
AE X at. wt MN A s X at. w t ME
If an organic chelate is used to increase the volatility of the metal, one might worry about isotopic effects due to the atoms in the ligand portion of the chelate ( e . g . 13C, 1 8 0 , etc.). However, it can be shown that as long as 1 is the lowest mass position in the metal containing cluster of the peaks and h is the first peak containing the next heavier metal isotope, no correction for the isotopic distribution of the atoms in the ligand is needed, when RE, R, and RN are determined experimentally. For example, in the analysis of zirconium using Zr(fod)d (see Experimental section) the s0Zr(fod)3+ peak is 1 and the g1Zr(fod)3+ peak is h. The measured (uncorrected) ratio is defined by Equation 5 all c o n t r i b u t i o n s t o p e a k a t m a s s 976 R" = all c o n t r i b u t i o n s t o p e a k a t m a s s 975
- 'lAF,
+
+
'OAF, 'OAF, "AF,
+
'OAF,
(5)
where 91A and 9OA refer to the abundances of slZr and 9OZr, respectively. The F's refer to the probabilities of the following atom combinations:
F , = F 5 = 12C3016061H30 F 2 = 12C2913C116061H30 F 3 = 12Cj0160j1i011H30 F 4 = 12C3016061H2:H1 Note that fluorine is monoisotopic and therefore need not be considered. Simplifying Equation 5 gives Equation 5a.
+
+
+
R" = "A/"A + ( F 2 F 3 F , ) / F , = R F (5a) where R is the corrected ratio and F is the term due to the heavier isotopes in the ligand. Thus, if all ratios are determined experimentally, either the uncorrected ratios ( Ru) 108
where s = sample. Note: there will be a simple proportionality between ystAPPand ysAPP(y, = By,,). Substituting Equation 6a into 6b and rearranging gives Equation 7 xs
TRUE
=
BXStTRUE x (DR"),, - DNRX" DERE" - (DR"), ( m u ) , - DrRNu} L R E U - (DRU),,
}
i
(7)
Now if the experiment can be arranged such that (DRU),, (DRU),,the terms inside each of the braces in Equation 7 will approximately cancel and Equation 7a results X,TRUE
=
BxStTRUEC
(74
where C is a correction factor, approximately one. needed because in general (DRU),t is not exactly equal to (DRU),. This extremely simple result shows that the amount of the metal in an unknown sample is proportional to the amount of the standard used in the back calibration times a small correction term.
EXPERIMENTAL Reagents
and Samples. 1,1,1,2,2,3,3-Heptafluoro-7,7di-
methyl-4,6-octanedione [H(fod)] was obtained from Pierce Chemical Company and was used after distillation at reduced pressure. Reagent grade sodium borate (decahydrate) was obtained from J. T. Baker Chemical Company. Reagent grade carbon tetrachloride was dried with freshly activated Molecular Sieve 5-A (50,450 mesh). Specpure zirconium oxide was obtained from Johnson Matthey Chemicals Ltd. Zirconium oxide enriched in 91Zr was obtained from Oak Ridge National Laboratory. Canadian Association f w Applied Spectroscopy Standard SY-1 and U.S. Geological Survey Standard AGV-1 were kindly supplied by P. D. Fullagar (Geology Department, University of Korth Carolina, Chapel Hill) and by Francis J. Flanagan of the U.S. Geological Survey, respectively. The lunar samples received from NASA were a crystalline rock chip (14310,135); a coarse grained breccia (14321,188), and three fines (15021,96; 15301,78; and 15471,30). These samples were used after grinding to a fine powder in agate mortars and drying at 110 "C for 2 hours.
A N A L Y T I C A L CHEMISTRY, VOL. 46, NO. 1, J A N U A R Y 1974
Procedure. An enriched zirconium standard is prepared by borax fusion of about 5 mg of the glZr-enriched ZrOz followed by 1M HC1 digestion of the flux and dilution to an approximate concentration of 1 mg of Zr/ml. A portion of the sample t o be analyzed is weighed into a small platinum crucible (volume, 1 ml). The enriched zirconium standard solution containing approximately the same weight of zirconium estimated t o be in the sample is weighed into the crucible. After gently heating the crucible to dryness on a hot plate at 80 "C, the sample is decomposed t)ia borax fusion at about 1250 "C. The chelate compound is then prepared using the previously published method ( 6 ) ,with the exception that during the evaporation following the CC14 reaction step (Zrsoxides ZrCld, a heat lamp is used to raise the temperature to between 40 and 50 "C. The instrument used for this work was an Associated Electronics MS-902 mass spectrometer equipped with a direct insertion probe and peak switching facility. The latter allows two different mass positions to be scanned by rapidly switching the accelerating voltage and the potential across the electrostatic analyzer. About 5 p1 of a CC14 solution containing about 1 pg of the zirconium chelate is deposited on the glass capillary of the direct insertion probe. The CC14 is then evaporated under a steam of nitrogen and the probe is introduced into the mass spectrometer. The operating conditions for the mass spectrometer and computerized data acquisition system are the same as those previously reported (8, 14) with the following exceptions: the ion chamber temperature is 190 "C; and the peaks being measured are g1Zr(fod)3+and gOZr(fod)3+.Therefore, the mass ratio used is 1.001026.
-
Table I . Determination of Zr in A G V - 1 Zr, ppm
Method of analysis
218 220 207 221 198 216 175
Mass spec-stable isotope dilution Emission spec Emission spec Emission spec Emission spec X-ray fluorescence Neutron activation analysis
This work
Other work 1 Other work 2 Other work 3 Other work 4 Other work 5 Other work 6
Table I I . Determination of Zr in Lunar Samples Individual analysis, pprn
L u n a r sample
2190 2040
2115
15-301
606
600
593 15-471
345 31 3
329
1 4 - 3 10
1350 1080
1215
14-321
1200 1160
1180
RESULTS AND DISCUSSION Because the experimental discrimination factors listed in Equation 6 of the introduction are unknown, it is necessary to determine the experimental zirconium ratios (S1A/90A) for both the pure enriched and pure natural zirconium. The average experimentally determined value for the natural ratio is 0.547 (seven determinations) with a relative standard deviation of 1.6%. This is in good agreement with the theoretical value for this ratio (0.555), calculated using the isotope abundances given in the CRC handbook. The average experimental value for the enriched ratio is 14.0 (eight determinations) with a relative standard deviation of 3.6%. Once DNRN and DERE have been determined, it is possible t o use these values for the analysis of the zirconium content of real samples. The zirconium contents of two geological standards (AGV-1 and SY-I) were determined. A summary of obtained and reported values (1.5) for the zirconium content of AGV-1 is given in Table I. Reported values for the zirconium content of SY-1 range from 1500 ppm t o 7000 ppm (16); furthermore, two values by the same authors (17) using the same method vary by 750 ppm. These inconsistencies raise serious questions concerning the homogeneity of zirconium in this sample. The zirconium concentration of SY-1 found in this work is 4210 ppm with a standard deviation of 2.8% obtained on five individual analyses done on two different days. When using the stable isotope dilution technique, it is necessary to know the relative isotopic abundances of the element being determined. It was expected that the relative abundances of the zirconium isotopes in the lunar samples would be the same as those in earth samples, but no information was available on lunar zirconium isotope ratios. It was therefore necessary to measure the ratio of slZr/sOZr in lunar sample 14,321. The slZr/sOZr ratio was in agreement with the earth value t o within 1% which is within the precision of individual ratio measurements. The zirconium concentrations in five lunar samples were determined and these values are listed in Table 11. (14) J. J. Leary. S. Tsuge. a n d T. L. Isenhour. Ana/. Chem.. 45, 1269 (1973). (15) G. Thompson, D. C. Bankston, a n d S. M. Pasley. Chem G e o / . . 5, 21 5 (1969/1970). (16) G. R. Webber. Geochirn Cosmochim. Acta , 29, 229 (1965). (17) N. M . Stne. W. G . Taylor. G . R. W e b b e r , and C. L. Lewis, Geochim Cosmochim. Acta.. 33, 121 (1969).
Average
15-021
Throughout this work, the gOZr(fod)3+ and "Zr(fod)3+ peaks were used for all ratio measurements. To determine what elements might interfere with these measurements, a spark source mass spectrometer was used for the determination of all metals present in the zirconium containing portion of the effluent from the ion exchange column. Yttrium was the only metal found capable of interfering with the zirconium ratio measurements. Although yttrium is monoisotopic (SSY), the P + 1 and P + 2 peaks of SSY(fod)3~could interfere. Therefore, a quantitative study of the magnitude of the interference due to yttrium was conducted and it was found t h a t the intensity ratio of ~sY(f0d)3+/~0Zr(fod)3+ was approximately 0.3%; a n yttrium interference of this size is negligible. A competitive chemical reaction involving C12 is a far more serious problem than the yttrium interference. As mentioned in the Experimental section, one of the steps in the syntheses of the chelate is conversion of the various oxides of' zirconium to ZrC14 using a CC14 reaction in a sealed borosilicate glass tube a t 400 "C. During this reaction, excess CC14 will react with itself, giving CzC16 and Cl2 as the major byproducts; therefore, a t the end of this step, the reaction mixture consists of ZrC14, cC14, C2Cl6, Clz, and trace amounts of other byproducts. If H(fod) is added to this mixture without removing the Clz, a chlorinated chelating agent [H(fod-Cl)] is formed by the following reaction R /
o=c\ H-0-C
C-H /
\
R'
H(fod)
+
CI,
-
R /
o=c,
C-CI
/ H-0-C
+
HCI
(8)
\
R'
H(fod-Cl)
The presence of both H(fod) and H(fod-C1) can result in the formation of five different chelates [Zr(fod)N(fod-
ANALYTICAL CHEMISTRY, VOL. 46, NO. 1, JANUARY 1974
109
Cl)4-N;0 5 N 141. In the presence of Clz, the Zr(fod)4 chelate appears to undergo a ring opening reaction rather than a simple substitution. Since each chlorine-containing chelate reduces the intensity of the Zr(fod)3+ peaks used in the analysis, this interference must be eliminated. Elimination of the formation of H(fod-C1) is most easily accomplished by removing the Clz prior to the addition of the ligand. Gently heating the reaction tube with a heat lamp (40-50 "C) during the evaporation step prior to adding the ligand essentially removes all Clz. Evaporation should be stopped before dryness, otherwise a portion of ZrC14 might be lost.
ACKNOWLEDGMENT The authors gratefully acknowledge the assistance of David Rosenthal for his cooperation and for the use of the facilities of the Research Triangle Center for Mass Spectrometry which is supported by the Biotechnology Resources Branch of the NIH Grant Number PR-330. The authors also thank Arthur Fitchett for running the spark source sample. Received for review May 18, 1973. Accepted July 20, 1973. Work supported by Materials Research Center, University of North Carolina under Contract Number DAHCl5-67C-0223 with the Advanced Research Projects Agency.
Esterification of (2,4-Dichlorophenoxy)acetic AcidA Quantitative Comparison of Esterification Techniques Jesse Horner, Shane S. Q u e Hee, a n d Ronald G. Sutherland Department of Chemistry and Chemical Engineering, University of Saskatchewan, Saskatoon, Saskatchewan. Canada-S7N
Esterification of 2,4-D with BF3/alcohol mixtures produces esters in greater than 90% yield in 20 minutes with few byproducts. Diazoalkylation is also efficient but produces many' impurities. In addition, the reagent is more difficult to make and is more labile than the BF3/alcohol mixtures. Mineral acid catalysis of 2,4-D/alcohol mixtures is time consuming and inefficient. The esters produced by silylation are unstable.
Esterification is invariably the final step before quantitation of free herbicidal chlorophenoxyacetic acids by gas chromatographic methods. Although many papers on esterification techniques exist ( I , 2 ) , no previous workers have compared the efficiencies of the different methods used for (2,4-dichlorophenoxy)aceticacid (2,4-D)analysis. Metcalfe and Schmitz ( 2 ) showed that the BFs/alcohol reagent gives yields varying from 60 to 95% for a series of fatty acids, and that the method compared well with the yields from diazomethylation and acid catalysis. Vorbeck et a1 ( I ) confirmed that the above methods were comparable only for higher molecular weight fatty acids. Diazomethylation produced 100% esterification while for low molecular weight fatty acids, BF3/MeOH gave 75%, and methanol/HCl produced 49%. This paper critically compares the various methods of esterification with respect to yield, purity, concentration, and stability of esterifying reagent.
EXPERIMENTAL Four methods are commonly used for esterification: mineral acid catalysis of alcohol/acid mixtures; Lewis acid catalysis of alcohol/acid mixtures; diazoalkylation of acids: and silylation. All reagents were reagent grade, and solvents were dried, then distilled. Mineral Acid Catalysis of AlcoholiAcid Mixtures. Pure (2,4-dichlorophenoxyjaceticacid (2,4-D) (1.7 g) was dissolved in methanol (10 ml). Varying amounts (10 ~1 to 2 ml) of concentrated strong mineral acids (sulfuric or hydrochloric acids) were (1) M . L. Vorbeck, L. R . Mattick, F. A . Lee, and C . S. Pederson, Anal. Chem., 33, 1512 (1961). (2) L. D. Metcalfe and A. A . Schmitz. Anal. Chem.. 33, 363 (1961).
110
OW0
added to the reaction mixture before refluxing. Ten-microliter samples of the reaction mixture were then injected onto a GLC column until five successive injections produced peaks of approximately the same area. The column used was a 6-ft long y,-in. i.d. stainless steel tube packed with 10% polyamide 103a (Applied Science Laboratories, Inc.) on 60/80 mesh Chromosorb W(DMCS-AW). The injector, column, and thermal conductivity detector were maintained at temperatures of 242, 192, and 250 "C, respectively. The flow of helium carrier was (25 1) ml/min; the filament current was 200 mA. Five identical reaction mixtures at reaction equilibrium from which no GLC samples had been removed were titrated with NaOH solution, using potentiometric and indicator (phenolphthalein) methods. The mean acidity content of mineral acid methanol controls ( 5 replicates) was subtracted from the mean total proton concentration about to give the mean amount of unreacted 2,4-D a t reaction equilibrium. The amount of ester at reaction equilibrium was found by first adding water (5 ml), evaporating the excess methanol at room temperature, and extracting the aqueous layer with hexane (3 X 5 ml); the hexane was evaporated, and the weight of the white residue was ascertained. The solid was subjected to spectroscopic, melting point, and GLC analyses. The weight of ester was found by the external standard method, using GLC analysis of a known fraction of the sample. Equilibrium constants and yields were computed, and compared with those of the uncatalyzed reaction. Yields were expressed as percentages of expected ester weight, assuming complete reaction. The above method was advantageous when the alcohol is easily removed under vacuum without heating. Distillation at high temperatures and high vacuum may be performed to produce pure esters, but pyrolysis and the temperature dependence of the position of equilibrium may change the yield originally obtained at reflux temperature. Hence, the above esterification procedure was repeated with a slight excess of methanol in methylene dichloride (10 ml), and mineral acid catalyst. At reaction equilibrium, two phases were present, the upper layer containing 90% of the total ester. The bottom aqueous layer was extracted by hexane to remove the remaining ester. The organic extract and methylene dichloride layer were combined, and evaporated to produce the ester. Yields and equilibrium constants were computed as above. Diazoalkylation. Diazomethane/ether reagent was synthesized by the method of de Boer ( 3 ) . 2,4-D (1.7 gj was dissolved in dried methanol (4 ml), diazomethane reagent was added ( 4 ml), and the reaction was allowed to proceed at room temperature for 10
*
(3) T. J. deBoer and H. J. Backer, R e d Tfav. Chem. Pays-Bas. 73, 229 (1954) (In E n g ) ; Aldrich Chemical Catalog, 14, 484 (1968)
A N A L Y T I C A L CHEMISTRY, VOL. 46, NO. 1, J A N U A R Y 1974