Anal. Chem. 1999, 71, 4030-4033
Determination of Submicromolar Concentrations of Formaldehyde by Liquid Chromatography S. Bart Jones,* Christopher M. Terry, Tedd E. Lister, and David C. Johnson
Department of Chemistry, University of North Carolina at Wilmington, Wilmington, North Carolina 28403
Dissolved formaldehyde in aqueous samples was determined at submicromolar levels by derivatization with Nash’s reagent followed by liquid chromatography. The method requires little sample preparation, and the chromatogram is simple even in the presence of other aldehydes and ketones. The isocratic HPLC analysis is rapid with a low limit of detection (0.1 µM) and a precision of 1% RSD at 150 µM. The derivative is stable for at least 3 days at room temperature. Accuracy of the method was verified by intercomparison with an alternate, completely independent method, which utilizes another derivatizing agent. Formaldehyde is a common carcinogen and contaminant in occupational and environmental atmospheres. The effect of formaldehyde on people exposed to this compound is wellknown: irritation of the eyes and upper respiratory tract, headache, nausea, drowsiness, and allergic skin reactions.1 Potential health hazards from formaldehyde derived from various consumer products have been a matter of concern. Levels of outgassed formaldehyde need to be checked regularly in order to maintain safe levels.2 Aside from measurement of formaldehyde in atmospheric water or aerosols, gaseous formaldehyde concentrations are often measured by collection into a suitable absorbent.3 Even when solid sorbents are used,4 analysis of an aqueous extract of the sorbent is typical. Most methods for detection of formaldehyde require chemical reaction of the formaldehyde with various reagents to form colored derivatives, which can be observed spectrophotometrically. Recently, high-performance liquid chromatography (HPLC) has been used to isolate these derivatives from possible interferences and thereby improve the detection limit. Most methods use the reaction of formaldehyde with 2,4-dinitrophenylhydrazine (DNP) to form the hydrazone.5-10 One of these methods5 has a very low * Corresponding author: (fax) (910) 962-3013; (e-mail)
[email protected]. (1) Rietz, E. B. Anal. Lett. 1980, 13 (A12), 1073-1084. (2) Consumer Products Safety Commission. Fed. Regist. 1980, 45, 3403134033. (3) Miksch, R. R.; Anthon, D. W.; Fanning, L. Z.; Hollowell, C. D.; Glanville, J. Anal. Chem. 1981, 53, 2118-2123. (4) Matthews, T. G.; Howell, T. C. Anal. Chem. 1982, 54, 1495-1498. (5) Kieber, R. J.; Mopper, K. Environ. Sci. Technol. 1990, 24, 1477-1481. (6) Puputti, E.; Lehtonen, P. J. Chromatogr. 1986, 353, 163-168. (7) Raymer, J.; Holland, M. L.; Wiesler, D. P.; Novotny, M. Anal. Chem. 1984, 56, 962-966. (8) Van Hoof, F.; Wittocx, A.; Van Buggenhout, E.; Janssens, J. Anal. Chim. Acta 1985, 169, 419-424.
4030 Analytical Chemistry, Vol. 71, No. 18, September 15, 1999
(3 nM) limit of detection (LOD), but they all have some drawbacks if formaldehyde is the only analyte of interest. DNP reacts with most aldehydes and ketones. This results in longer HPLC analysis times and could require a mobile-phase gradient if other aldehydes and ketones are present in addition to formaldehyde. Another potential problem for pH-sensitive samples can be the very acidic (pH ∼1.5) reaction conditions of the DNP methods. If formaldehyde is the only component of interest, an attractive alternative is to react it with a more specific reagent. The condensation reaction of ammonia and 2,4-pentanedione in Nash’s reagent11 with formaldehyde forms yellow 3,5-diacetyl-1,4-dihydrolutidine (DDL) at a pH of ∼6.3.
This reagent does not react with ketones, and the only aldehydes it reacts with are formaldehyde and acetaldehyde.11 This would lead to less complicated HPLC chromatograms. This reagent is well-known and has been used previously to determine formaldehyde quantitatively by absorption spectroscopy12 and by fluorescence spectroscopy13-15 of DDL. These spectrophotometric methods have only been coupled with HPLC in one instance using postcolumn derivatization.16 However, the 13 µM limit of detection is relatively high and depending on the analyte sample the method can require a ternary gradient. We have developed a rapid method with a low LOD for the quantification of formaldehyde. This method uses the reaction of (9) Tanner, R. L.; Meng, Z. Environ. Sci. Technol. 1984, 18, 723-726. (10) Reindl, B.; Stan, H. J. J. Agric. Food Chem. 1982, 30, 849-854. (11) Nash, T. Biochem. J. 1953, 55, 416-421. (12) Official Methods of Analysis, 14th ed.; AOAC: Arlington, VA, 1984; Sections 1.203-31.208. (13) Dong, S.; Dasgupta, P. K. Environ. Sci. Technol. 1987, 21, 581-588. (14) Belman, S. Anal. Chim. Acta 1963, 29, 120-126. (15) Sawicki, E.; Carnes, R. A. Mickrochim. Acta 1968, 148-159. (16) Summers, W. R. Anal. Chem. 1990, 62, 1397-1402. 10.1021/ac990266s CCC: $18.00
© 1999 American Chemical Society Published on Web 08/12/1999
formaldehyde with Nash’s reagent to form a derivative (DDL) which is separated from interferences by liquid chromatography (HPLC). The derivative is formed in 12 min at 51 °C and a pH of 6.3 and is stable for up to 72 h at room temperature. The DNP derivatization reaction takes 1 h, and the derivatives formed are stable for 4 h at room temperature. The 72-h stability allows many derivatized samples to be analyzed sequentially using an autosampler without compromising the analyte. While the method presented here requires no extensive cleanup of reagent, carbon tetrachloride extraction of reagent is required for the DNP method5 with the 3 nM LOD. Due to the small number of peaks, the HPLC analysis is rapid and isocratic unlike that used in many DNP methods.5-7 The accuracy of results using the Nash reagent, which has never been studied before, is verified by intermethod comparison in this paper. EXPERIMENTAL SECTION Reagents and Standards. HPLC solvents used were acetonitrile (HPLC grade, Fisher Scientific Co., Pittsburgh, PA) and water purified by a MilliQ system (Millipore, Bedford, MA). Ammonium acetate, acetic acid, and 2,4-pentanedione (distilled under nitrogen before use) were all reagent grade and purchased from Fisher. Nash reagent was prepared by dissolving 7.5 g of ammonium acetate, 0.150 mL of acetic acid, and 0.100 mL of freshly distilled 2,4-pentanedione in water to make 50.0 mL of reagent solution. Formaldehyde obtained from Wright Chemical Corp. (Wilmington, NC) as a freshly made and analyzed 37.31% aqueous solution was used to prepare a 1000 µM formaldehyde stock solution. This solution was used to prepare all other formaldehyde solutions used in the experiments. All work with formaldehyde should be performed in a well-ventilated area such as a fume hood. Pure derivatives of formaldehyde and acetaldehyde were synthesized to determine HPLC conditions for optimum separation and for spectral measurements. The formaldehyde derivative was synthesized by mixing 2 g of 2,4-pentanedione with 0.75 mL of 40% CH2O and 7.6 g of ammonium acetate in 100 mL of water. The acetaldehyde derivative was formed by mixing 2 g of 2,4-pentanedione with 0.44 g of acetaldehyde and 7.6 g of ammonium acetate in 100 mL of water. These solutions were allowed to sit at room temperature for 48 h. The products were then filtered and rinsed and allowed to air-dry. HPLC Apparatus. The liquid chromatograph system consists of two Waters model M-45 pumps (Waters Assoc., Milford, MA), a Valco six-port injector (Valco Instruments, Houston, TX) with a 50-µL sample loop, an octadecyl reversed-phase (C18) 100 mm × 8.0 mm column (although other column diameters would also work) with 5-µm particle size (Resolve type) (Waters), a Kratos Spectroflow 757 variable-wavelength absorbance detector (Applied Biosystems Inc., Foster City, CA.), and a Hewlett-Packard 3396A integrator (Hewlett-Packard Co., Palo Alto, CA). HPLC Conditions. The mobile phase was a 20:80 acetonitrile/ water isocratic mixture, which completely resolved the formaldehyde and acetaldehyde derivative peaks (Figure 1). This was determined by making up solutions of formaldehyde and acetaldehyde derivatives and varying the mobile-phase composition until baseline resolution was achieved. The operating conditions were as follows: column temperature, ambient; mobile-phase flow rate, 2.0 mL/min; detection wavelength, 412 nm.
Figure 1. HPLC chromatogram of 100 µM CH2O (1) and 100 µM CH3CHO (2) derivatives in 20:80 acetonitrile/water.
Reaction Conditions. Nash reagent (0.200 mL) was added to 1.00 mL of sample in a 2-dr glass vial with a polypropylene cap. The vial was tightly capped and placed in a 51 °C water bath (Precision water bath, model 182, Precision Scientific Co., Chicago, IL.) for 12 min to form the derivative. The samples were cooled to room temperature, and portions of each solution were analyzed by HPLC. Blanks were quantified by using 1.00 mL of MilliQ water in place of the sample or known. RESULTS Structure of DDL. In previous work,11,14 the structure of DDL has been proposed and supported by elemental, infrared, and UVvisible analysis. In addition to UV-visible analysis, which agreed with that of previous work, proton and carbon NMR spectra of DDL were measured. Chemical shifts and coupling constants were also consistent with the accepted structure. The purified synthetic DDL derivative coeluted with derivatives formed in actual samples. Derivatization Reaction. The time required to derivatize formaldehyde by the Nash reagent was evaluated by following the derivatization to DDL as a function of time. Two concentrations of formaldehyde (250 and 10.0 µM) were used as samples and were reacted at 51° C. Separate vials with the same amount of sample and reagent were heated for increasing periods of time (0, 2, 4, 6, etc., min up to 28 min). The solutions were analyzed by HPLC. The results are presented in panels A (250 µM) and B (10.0 µM) of Figure 2, which show complete derivatization for both concentrations in 12 min. Other temperatures were also evaluated. At 40 °C, the derivatization process was complete in 30 min. Temperatures higher than 51 °C were not suitable because the signal never reached a constant value. Specificity of the Derivatization Reaction and Interferences. Previous results11,12 reported possible interferences and how they may be removed. In this work, a concentration of 1.0 µM sulfite had no effect on signals of 10 and 100 µM formaldehyde samples. Although acetaldehyde does react with Nash reagent to form a derivative that also absorbs at 412 nm, it does so much more slowly and the peak for the acetaldehyde derivative is well separated from that of DDL in this method. A sample containing 100 µM each of formaldehyde, acetaldehyde, propanal, acetone, and methyl-2-propanone was derivatized as described above. There were no other peaks aside from that of the formaldehyde derivative and the early-eluting peaks from the reagent. Stability of Formaldehyde Derivative (DDL). Studies were conducted to see how stable the blank and DDL derivative were Analytical Chemistry, Vol. 71, No. 18, September 15, 1999
4031
Figure 2. Formation of DDL from Nash reagent and (A) 250 µM CH2O; (B) 10 µM CH2O as a function of time.
Figure 4. Chromatograms of (A) blank and (B) rainwater sample containing 0.3 µM formaldehyde. (R) reagent; (1) DDL.
Figure 3. Test of stability for (A) blank, (B) 10 µM DDL, and (C) 100 µM DDL.
at room temperature and ambient room light. To determine stability, samples were prepared by adding 0.2 mL of Nash reagent to 1.00 mL of Milli Q water or formaldehyde sample (10 and 100 µM). These were then heated for 12 min at 51° C and analyzed by HPLC. Samples were left on the benchtop at room temperature and analyzed every 2 h for 12 h and every 24 h out to 7 days. The results of these experiments are shown in Figure 3. There was 4032 Analytical Chemistry, Vol. 71, No. 18, September 15, 1999
no change in concentration (determined using F statistic, R ) 0.05) in the 10 (3B) and 100 µM (3C) samples up to 72 h and up to 168 h in the blank (3A). Limit of Detection. A typical blank chromatogram is presented in Figure 4A. In addition to early-eluting peaks from the reagent, a small reproducible formaldehyde signal is present. Various extraction and adsorption methods were used unsuccessfully to try and lower the background signal from the reagent. The concentration and volume of Nash reagent were also varied to give the minimum background signal while providing enough reagent to ensure complete reaction over the entire concentration range that was studied. The concentration and volume specified above are the results of these experiments. The limit of detection was determined by measuring nine blanks. Three times the standard deviation of the blanks divided by the slope of the calibration curve gave a limit of detection of 0.1 µM CH2O. Linearity and Precision. The calibration curve was linear from 0.500 to 250 µM (solubility limit of the DDL derivative in water) with a correlation coefficient over this range of 0.9999 (n ) 15). The precision of the method was determined by multiple injections (n ) 3) for each of seven rainwater samples and numerous samples of known concentration. The average relative
standard deviation (RSD) for the rain samples (all of whose formaldehyde concentrations were less than 10 µM) was 4%. A chromatogram of a rain sample containing ∼0.3 µM is presented in Figure 4B. Triplicate analyses of concentrations ranging from 10 to 250 µM gave an average RSD of 1%. Intermethod Comparison. An intermethod comparison using rainwater samples was performed in order to verify the accuracy of the results from the Nash reagent analysis. This is the first time an intermethod comparison with independent derivatization methods has been done for formaldehyde on complex environmental samples that contain a host of possible interferences. Rain samples (n ) 7) were collected for 1 month and analyzed by both the Nash and DNP5 methods. Plotting the concentrations obtained by the DNP method versus the concentrations obtained by the Nash method produced a line with a slope of 1.00 ( 0.03 (95% confidence level). This is not statistically different from a line with a slope of unity. This represents perfect agreement between the methods. The lack of statistically significant differences between the results from the two independent methods verifies the accuracy of analytical results obtained by the Nash reagent method. DISCUSSION In summary, we present a rapid, mild, and specific method that can be used to analyze formaldehyde at concentrations down to 0.1 µM. This limit of detection is considerably lower than those
seen in previous work using the Nash reagent and absorbance detection because the HPLC removes the significant background signal from the reagent as early-eluting peaks (see Figure 4A and B). This method would be sufficient for all but the lowest levels of formaldehyde such as those seen in some natural water samples.5 We are currently investigating the use of fluorescence detection to increase the sensitivity of the method. The accuracy of the method has been verified by comparison with a completely independent method.5 The precision of the method is excellent with relative standard deviations of 4% for concentrations less than 10 µM and 1% for concentrations between 10 and 250 µM. The ease of the method is noteworthy. The derivatization reaction is rapid and requires no difficult reagent purification procedures. The derivatives can be stored for up to 72 h at room temperature before chromatographic analysis, which is isocratic and can be performed on a very simple chromatograph. ACKNOWLEDGMENT The financial support of the University of North Carolina at Wilmington in the form of an Undergraduate Research Fellowship for C.M.T. is gratefully acknowledged.
Received for review March 9, 1999. Accepted June 25, 1999. AC990266S
Analytical Chemistry, Vol. 71, No. 18, September 15, 1999
4033
