Detection of nanogram quantitites of hexachlorophene by ultraviolet

Small , Timothy S. Stevens , and William C. Bauman. Analytical Chemistry 1975 47 (11), ... Lyle D. Bighley , Dale E. Wurster , Diana Cruden-Loeb , Rob...
0 downloads 0 Views 320KB Size
were recorded over long periods of time. The change in potential was of the same order as the drift of the blank. The pH of the solutions used in the liquid membrane studies was taken immediately following the completion of the study because the pH changed so rapidly. Hence no comparison could be made with the solid electrode studies. The results obtained with the solid membrane electrode in phosphate and borate buffers did not differ significantly. Therefore, it was concluded that phosphate must interfere in the alkylation step rather than in the cyclization reaction.

CONCLUSIONS

The ion-selective electrodes used in this study offered a rapid, continuous, and simple method for the measurement of rate constants without a flow-through system. It is suggested that other electrodes could similarly be applied in reaction mechanism studies and in the determination of rate constants. RECEIVED for review February 23, 1972. Accepted May 1,1972.

Detection of Nanogram Quantities of Hexachlorophene by Ultraviolet Liquid Chromatography Peter J. Porcaro and Peter Shubiak Research Department, Givaudan Corporation, Clifton, N.J., 07014

GASCHROMATOGRAPHY with electron capture has been extensively applied to the detection of subnanogram amounts of hexachlorophene [2,2'-methylenebis(3,4,6-trichlorophenol)]. The detection is usually made after derivatizing, and the most common derivatives are the trimethylsilyl ether ( I , 2), diacetate ( 3 ) ,and dimethyl ether ( 4 , 5 ) . Each has met with particular favor for various reasons. The problem of derivatizing, however, is not a major one for the eventual detection at the low level capabilities of electron capture detectors. Whether these detectors use radioactive sources (3H, e3Ni, 226Ra)or an activated rare gas (He, Ar), they are all prone to easy contamination, sensitivity losses, and a variety of other peculiar behaviors. A new approach has been made for the separation and detection of this industrially important material using liquid chromatographic techniques and ultraviolet detection. The method has shown itself to be sensitive, reliable, and troublefree compared to the use of electron capture detection using gas chromatography. Ultraviolet detection has thus far been used only for hexachlorophene in the macro region in applications such as drugs and cosmetics (6, 7). EXPERIMENTAL

Reagents. A 55/45 V/V mixture of hexane and n-butylchloride is used. Spectro or reagent grade material is adequate with no further purification necessary. Procedure. The UV detector is used at its maximum sensitivity, 0.02 absorbance full scale. A 2-ft stainless steel 2.3-mm i.d. column is packed with Sil-X silica 36-40 p particle size (Nester-Faust) with no prior conditioning or activation. A constant flow rate of 0.7 ml/min is maintained, generating a pressure of approximately 200 psi for the system which has two restrictors and pulsation dampers in-line as supplied. Septum used is Viton with no pre-leaching required. Buna-N and EPR are not suitable. Derivative Preparation. The derivative used is the di-pmethoxy benzoate or dianisate ester of hexachlorophene (HCP-DA). CI

0

-

Ee

OCH,

CI

Apparatus. A Waters Model ALC 202/R-401 Liquid Chromatograph was used. It incorporates a differential refractometer and an ultraviolet detector. The UV detector is made by Laboratory Data Control (Riviera Beach, Fla.) which employs a low pressure mercury lamp source emitting its strongest radiation at 254 nm. (An alternate optional bandpass is available at 280 nm, if desired.) The recorder used was a Honeywell Electronic Model 194. UV spectra were recorded on a Beckman ACTA I11 spectrophotometer. (1) J. Wisniewski, Facts Metlzods, 8, 10 (1967). (2) P. J. Porcaro and P. Shubiak, ANAL.CHEM., 40, 1232 (1968). ( 3 ) R. S. Browning, Jr., J. Grego, and H. P. Warrington, Jr., J. Pharm. Sci., 57, 2165 (1968). (4) A. Curley, R. E. Hawk, and R. Kimbrough, Lancet, 2, (7719)

296 (1971). ( 5 ) W. H. Gutenmann and D. J. Lisk, J . Ass. Ofic.Anal. Chern.,

53, 522 (1970). (6) D. A. Elvidge and P. Peatrell, J. Pharm. Pharmacol., 13, l l l T (1961). (7) R. W . Daisley and C . J. Olliff, ibid., 22, 202 (1970).

2,2 '-Methylenebis(3,4,6-trichlorophenol)di-p-methoxybenzoate

Dissolve 2.0 grams of HCP in 60 ml of 10% NaOH contained in a 125-ml erlenmeyer. Add 8.0 ml of anisoyl chloride (Aldrich Chemical Co.) and mix contents for 1 hr. Collect crystals on a coarse sintered glass funnel using slight vacuum. Wash the crystals with 100 ml of distilled water and allow vacuum to air dry, Transfer to 50 ml of hexane, and add diethyl ether until solid dissolves. Filter and heat solution gently on a steam bath until it starts to boil. Remove and place in an ice bath to induce crystallization. Recrystallize two additional times. Dry at room temperature. The yield is approximately 500 mg of pure diester, mp 209.5-10 "C. Structure was verified by NMR, IR, and MS. The UV spectrum of the diester (HCP-DA) is shown in Figure 1, and the UV scan of underivatized HCP is contrasted as shown in Figure 2. Calibration Curve Using Pure Diester. Ten milligrams of prepared HCP-DA are dissolved in exactly 100 ml of n-butylchloride. Aliquots of 0.4, 0.8, 1.2, 1.6, 2.0, 2.4, and 2.8 ml

ANALYTICAL CHEMISTRY, VOL. 44, NO. 11, SEPTEMBER 1972

1865

-1.0

'"

9

8

N

7

I

6

5

w

5m

: 9

-.4

w

z 0

I

8 a m

80

-.3 2 I

0 5

Ib

I

.

,

'

I

'

I

'

I

'

I

'

,

'

,

'

,

TIME (MIN.) 0

PA0

4

3bO

950

350

WAVELENGTH(NM)

Figure 1. Ultraviolet spectrum of HCPDA in n-butylchloride 1-cm cell, A,, = 31,725;

€2:4

262.5 nm. = 23,188

€263

=

43,260;

€280

Figure 3. Series of typical liquid chromatograms. Calibration of HCP-DA Chart speed 5 min/in., 225 psi, 0.7 ml/min flow rate, 1O-pl injections, 0.02 A full scale.

on a steam bath under a N, stream. Add 1.00 ml of nbutylchloride to each tube. Inject using 10-pl volume. Plot peak heights and/or area us. weights which range from 40-200 ng.

RESULTS AND DISCUSSION

WAVELENGTH (NM)

Figure 2. Ultraviolet spectrum of HCP in n-butylchloride 1-cm cell, A,, 296.5 nm. €254 = 509; EZEO= 3,050

= 5,800;

are withdrawn and diluted to exactly 10 ml with n-butylchloride. Ten microliters of each dilution are successively injected (Pressure-Lok series "B" Syringe Precision Sampling Corp., Baton Rouge, La.) and peak heights and/or areas are plotted L.S. weights from 40-280 ng. Calibration Curve by Direct Derivatization of Microgram Quantities of HCP. Ten milligrams of HCP are dissolved in exactly 100 ml of acetone. Aliquots of 40, 80, 120, 160, and 200 pl are placed in 40-ml round bottom centrifuge tubes (previously cleaned in chromic acid). The solvent is evaporated under a N, stream and 1 ml of 5 NaOH is added to each tube followed by 30 pl of anisoyl chloride. Place tubes successively on a Vortex mixer, or some other suitable device, for 1 minute. Allow to remain at room temperature for 20 minutes to complete reaction. Quench with 9 ml of distilled water and Vortex-mix for 2 minutes. Extract each solution three times with 10 ml of hexane using the mixer. After centrifuging, draw off the hexane with a capillary pipet and transfer to clean 40-nil tubes. Evaporate to dryness 1866

A series of typical chromatograms is shown in Figure 3, depicting the linear response of HCP-DA dilutions as calibrated. The peak height or peak area plots are linear over the concentration range measured. The calibration curve prepared using HCP in nanogram concentrations, derivatized as described, is also linear. Peak heights are very reproducible and are dependent only on concentrations and not on volumes injected in the range of 5 to 20 p1 tested. The variance for repetitive 10-p1injections containing 20 ng is 0.175, with a mean peak height of 41.6 chart units. The standard deviation is 0.418 giving a relative standard deviation of 1.o %. Two methods are presented for HCP detection; one employs a previously prepared derivative and the other describes the derivative prepared in situ. The former was used to check the efficiency of the latter. The rate of derivatization is related to both the time and quantity of anisoyl chloride added as illustrated in Table I. Using the calibration made of peak area to known weights of HCP-DA, the most suitable reagent concentration and reaction times were investigated. A constant addition of 10 p1 of anisoyl chloride alone had little effect on increasing the conversion of a constant amount of HCP. Almost complete conversion was accomplished by tripling the reagent concentration and allowing the reaction time to proceed for 20 minutes. At the 254 nm analytical wavelength HCP underivatized has a molar absorptivity of 509 making only milligram quantities detectable. However, the HCP-DA derivative shows a molar absorptivity of 31,725 at this point, or a 64-fold increase in detection level. The dicinnamate ester, for example, would be suggested for those detectors set at 280 nm, although this has not been investigated. Its molar absorptivities are: €286 = 50,077; €280 = 49,628 ; € 2 5 4 = 18,608. This coupled with the fact that injection can be as high as 20 pl into the liquid chromatograph with no adverse effects, allows for some interesting comparisons.

ANALYTICAL CHEMISTRY, VOL. 44, NO. 11, SEPTEMBER 1972

Present methods for electron capture detection require sample injections restricted to approximately 5 p1 or less to avoid the possibility of detecting solvents or other source contaminants, interferences from carrier vehicles, e.g., blood, serum, and sewage. With the LC-UV method, many of these problems are avoided because of the specificity of UV. An example is given in the less stringent requirements for the solvents used. This technique can serve as a complementary method to GC-EC, achieving a comparable range of sensitivity. The UV detector was found to be very reproducible, requiring much less time for stabilization. The cleanup procedures for HCP extracted from blood serum, or other body fluids (3,4,8,9) are good starting points for specific problems. The authors have not made any extensive studies on the suitability of this approach to the various cleanup methods referred to. However, it is interesting to postulate. In the case of HCP added to and recovered as described (2, 8), approximately 0.03 ppm was found to be the lowest detectable range. Although calibrations are presented by LC with lower limits of 40 ng detectable per 10-111 injections, it becomes possible with the larger 20-111 injections permissible and after suitable concentrating, to detect at the 0.03-ppm level also. If, as an example, 100 ng of HCP were added to 3 ml of blood, the level would be 0.033 ppm or 33 ppb. After suitable cleanup, using the method described (8) as a guide, assume this amount is isolated and exists in a dry residue. It can then be derivatized as described, the caution being to allow the anisoyl

Table I. Dianisate Derivatization of HCP Anisoyl Conversion, chloride, pl HCP, pg Time, min 10 10 10 77.1 10 10 20 77.1 10 10 30 76.0 20 10 20 89.2 30 10 20 99.0

z

chloride reagent to contact HCP without interference from residual fatty materials acting as a barrier when the diester hexane extraction is evaporated to dryness. One hundred microliters of n-butylchloride can be used for solution. This volume will contain a stoichiometric amount of HCP-DA, which in this case is 166 ng. A 2 0 4 aliquot injection will contain 33 ng. Therefore, the original 100 ng of HCP added to 3 ml of blood is ultimately detected as 33 ng of HCP-DA in 20 pl of butylchloride. This novel approach is presented to those actively engaged in similar work to evaluate as an alternate detection method. It demonstrates once again the broad capabilities and utility of LC as a complement to GC. ACKNOWLEDGMENT

The authors express their appreciation for the cooperation and helpful suggestions given by Julian Dorsky and Gary Shaffer and to Axel Kiesslich, who contributed much to the LC work.

(8) P. J. Porcaro, P. Shubiak, and M. Manowitz, J. Pharm. Sci., 58, 251 (1969). (9) R. C. Bachman and M. R. Shetler, Biochem. Med., 2, 313 (1969).

RECEIVED for review February 23, 1972. Accepted May 1, 1972.

Hamming Type Codes Applied to Learning Machine Determinations of Molecular Formulas F. E. Lytle Department of Chemistry, Purdue University, Lafayette, Ind. 47907

MOSTOF THE CURRENT CHEMICAL RESEARCH involving machine intelligence has been devoted to new data base applications and faster or more reliable training procedures ( I ) . In contrast, relatively little attention has been paid to the problem of class construction with respect to both minimizing the number of necessary threshold logic units (TLU’s) and maximizing their reliability. The purpose of this communication is to demonstrate how arrays of TLU’s might be constructed so as to introduce a degree of redundancy into the decision making process. In particular, it is shown theoretically, how Hamming type self-correcting error codes can improve classification schemes used for the determination of molecular formulas. The proposed method can not only detect the fact that an error has been made in the overall procedure, but can also indicate which TLU was wrong and in what direction!

RESULTS AND DISCUSSION

Assume for the sake of discussion that the molecular formula questions asked of the machine involve carbon number, with the only possibilities being zero through fifteen. Historically, two approaches have been used. The first one might be called the “linear” method. Sixteen TLU’s are trained, corresponding to all possible carbon numbers. This scheme proceeds as follows: /Ta(l) = {Co] uO\T~(0) = {CI,CS,C3. , , .c14, C I ~ ]

ANALYTICAL CHEMISTRY, VOL. 44, NO, 11, SEPTEMBER 1972

1867