Determination of Pentachlorophenol and Its Oil Solvent in Wood Pole

tions, better control of the initial loading of the PCP in impregnated wood poles is needed. Monitoring of oil solvents is also important since excess...
1 downloads 0 Views 116KB Size
Download PDF
Anal. Chem. 1999, 71, 78-85

Determination of Pentachlorophenol and Its Oil Solvent in Wood Pole Samples by SFE and GC with Postcolumn Flow Splitting for Simultaneous Detection of the Species Yves G. Leblanc,†,§ Roland Gilbert,*,† and Joseph Hubert‡

Institut de recherche d’Hydro-Que´ bec (IREQ), 1800, boulevard Lionel-Boulet, Varennes, Que´ bec, Canada J3X 1S1, and De´ partement de Chimie, Universite´ de Montre´ al, P.O. Box 6128, Que´ bec, Canada H3C 3J7

An alternative approach is described for the measurement of pentachlorophenol (PCP) and its oil solvent in wood samples by supercritical fluid extraction (SFE) and gas chromatography (GC). The determination is achieved over a single chromatographic run using postcolumn flow splitting for simultaneous ECD/FID detection of the SFE extracted species. First, PCP and oil components are quantitatively extracted from a 0.3-g wood sample using 10% MeOH/CO2 supercritical fluid at 0.65 g/mL and 120 °C. An aliquot of the SFE solution is then mixed with 10 mL of a buffered aqueous phase at pH 9.4. After PCP is acetylated by the addition of 500 µL of acetic anhydride, it is followed by its extraction with 2.00 mL of hexane along with oil. Then, 0.5 µL of supernatant organic phase is injected into the GC for a selective and simultaneous determination of the species. The method has a linear response over 3 orders of magnitude for both species with a linear regression correlation coefficient higher than 0.98 (95% confidence limit) and an absolute detection limit of 60 ng of PCP and 80 µg of oil per 0.1-g wood sample. The precision (relative standard deviation) is 4% for PCP and 1% for oil as established for a typical average concentration sample. The accuracy of the SFE GC-ECD/ FID combined technique for PCP and oil was assessed by analyzing wood samples collected from newly and inservice PCP/oil-impregnated red pine poles. Pentachlorophenol (PCP), which was commercially introduced in Canada in the early 1950s, is still widely used by many public utilities as a biocide for wood pole preservation in new installations.1 It is dissolved in a diesel-type oil and pressure-injected into wood poles. To limit the environmental impact of such installations, better control of the initial loading of the PCP in impregnated wood poles is needed. Monitoring of oil solvents is also important since excessive swelling of the wood poles in service can †

Institut de recherche d’Hydro-Que´bec. Universite´ de Montre´al. § Present address: Institute for Marine Biosciences, National Research Council of Canada, 1411 Oxford St., Halifax, Nova Scotia, Canada B3H 3Z1. (1) Stephens, R. W.; Brudermann, G. E.; Morris, P. I.; Hollick, M. S.; Chalmers, J. D. Value Assessment of the Canadian Pressure Treated Wood Industry, SSC Contract No. 4Y002-3-0187/01-SQ; Carroll-Hatch (International) Ltd.: North Vancouver, BC, Canada. ‡

78 Analytical Chemistry, Vol. 71, No. 1, January 1, 1999

accelerate PCP losses to the environment. Also, given the number of aging PCP/oil-impregnated wood poles already installed and that many are close to the end of their service life, their disposal is becoming a subject of concern, and the residual quantity of PCP and oil in the pole has to be assessed. Various methods can be applied to determine PCP in wood samples. Those usually used by the wood preservative industry, such as X-ray spectrometry,2 are nonspecific to the PCP molecule and subject to matrix interference as they are based only on chlorine atom determination. Other, more selective methods developed to assess occupational exposure and the environmental contamination of water and soil can also be applied on wood extracts. These methods include electromigration techniques such as micellar electrokinetic chromatography (MEKC),3 highperformance liquid chromatography (HPLC),4-6 gas chromatography (GC),7-9 and flow injection analysis (FIA).10,11 Although these applications allow a selective determination of PCP from different phenol mixtures, none of them consider the PCP solvent as part of their assessment. Similarly, the existing gas chromatography/mass spectrometry methods (GC/MS)12,13 used to determine the oil components do not address the biocide component of the solution. Recently, a method was reported for the analysis of both PCP and the organic solvent from a singlesample treatment.14 PCP and oil extracted from powdered wood (2) Kressbach, J. N.; Tang, J.; Walker, J. E. Proceedings of the 83rd Annual Meeting of the American Wood-Preservers’ Association; AWPA: Woodstock, MD, 1987. (3) Van Bruijnsvoort, M.; Sanghi, S. K.; Poppe, H.; Kok, W. Th. J. Chromatogr. 1996, 757 (1/2), 203-213. (4) Puig, D.; Barcelo´, D. Chromatographia 1995, 40 (7/8), 435-444. (5) Baiocchi, C.; Roggero, M. A.; Giacosa, D.; Marengo, E. J. Chromatogr. Sci. 1995, 33, 338-346. (6) Fre´bortova´, J. Biosci. Biotechnol. Biochem. 1995, 59 (10), 1930-1932. (7) Rodrı´guez, I.; Turnes, M. I.; Mejuto, M. C.; Cela, R. J. Chromatogr. 1996, 721, 297-304. (8) Kontsas, H.; Rosenberg, C.; Pfa¨ffli, P.; Ja¨ppinen, P. Analyst 1995, 120, 17451749. (9) Wang, Y.-J.; Lin, J.-K. Arch. Environ. Contam. Toxicol. 1995, 28, 537-542. (10) Navarro-Villoslada, F.; Pe´rez-Arribas, L. V.; Leo´n-Gonza´lez, M. E.; Dı´ez, L. M. Anal. Chim. Acta 1995, 308, 238-245. (11) Moreno-Roma´n, C.; Montero-Escolar, M. R.; Le´on-Gonza´lez, M. E.; Pe´rezArribas, L. V.; Polo-Dı´ez, L. M. Anal. Chim. Acta 1994, 288, 259-264. (12) Sepic, E.; Leskovsek, H.; Trier, C. J. Chromatogr. 1995, 697, 515-523. (13) Wang, Z.; Fingas, M. J. Chromatogr. 1995, 712, 321-343. (14) Besner, A.; Gilbert, R.; Te´treault, P.; Le´pine, L.; Archambault, J. F. Anal. Chem. 1995, 67, 442-446. 10.1021/ac9803683 CCC: $18.00

© 1998 American Chemical Society Published on Web 11/25/1998

Figure 2. Typical chromatograms of PCP (ECD) and oil (FID) recorded simultaneously.

Figure 1. Basic steps for treatment of wood extracts.

were analyzed by gas chromatography-electron capture detection (CG-ECD) and Fourier transform infrared spectroscopy (FT-IR), respectively. However, the method requires large quantities of Freon 113 for solvent extraction/sonication and FT-IR measurement, which is problematic since its availability has been drastically reduced in recent years according to the Montreal Protocol.15 Also, it requires two different analytical instruments which would be difficult to combine and automate for a single PCP and oil instrumental determination. Supercritical fluid extraction (SFE) has recently evolved as an interesting alternative to solvent extraction prior to instrumental analysis for many environmental and industrial applications.16,17 To date, only a few applications of SFE in PCP determinations have been reported, and none consider the oil content of the sample. This paper covers the development of an analytical method for achieving oil determination along with PCP by using a pretreatment of the wood sample by supercritical fluid and simultaneous ECD/FID detection of the species by gas chromatography with postcolumn splitting. A validation of the combined improvementssSFE wood sample treatment and a simultaneous GC determination of the speciesswill be presented by comparing the results obtained with those of a previous method on samples collected from newly and in-service PCP/oil-impregnated red pine poles. (15) Ko, M. K. W.; Sze, N. D.; Molnar, G.; Prather, J. Atmos. Environ. 1993, 27A (4), 581-587. (16) Lee, M. L.; Markides, K. E. Analytical supercritical fluid chromatography and extraction; Chromatography Conferences, Inc., c/o Milton L. Lee, Department of Chemistry, Brigham Young University, Provo, UT 84602; 1990. (17) Hawthorne, S. B. Anal. Chem. 1990, 62, 633-642.

EXPERIMENTAL SECTION Apparatus. All supercritical fluid extractions were performed off-line with a Hewlett-Packard system, model 7680A (HewlettPackard, Palo Alto, CA), using a 1.5-mL extraction cartridge. Analytes were collected with a cryogenically cooled stainless steel bead trap (“solid trap”). The trap was then washed with a recovery solvent in a 1.5-mL septum-screwcap-top bottle (Supelco, Mississauga, Ontario, Canada) and used as sample stock solution for the derivatization procedure and/or oil separation from PCP. When methanol-modified supercritical CO2 was used, a second trap (“solvent trap”) had to be installed to avoid losses of extracted species through leaching of the stainless steel trap by the modifier that remains liquid upon depressurization. This was done by coupling the outlet of the solid trap to an 18-gauge needle inserted into an empty 15-mL septum-screwcap-top bottle (Supelco). A second needle was also inserted to allow the overpressure from the expanding supercritical fluid to be released. Combination of all the extracted species into a single fraction was achieved by washing the first trap into the second one with methanol. The determination of PCP and oil was performed on a HewlettPackard 5890 Series II gas chromatograph equipped with an oncolumn injector, 63Ni electron capture detector (ECD), flame ionization detector (FID), and HP 7673B autosampler. The injector temperature was set on a track mode at 3 °C above oven temperature. A fused silica capillary column was supplied by Hewlett-Packard: 30-m × 0.32-mm-i.d. × 0.17-µm HP-Ultra-1 (100% cross-linked methyl silicone), programmed at 55 °C for 1.75 min, then from 55 to 280 °C at 20 °C/min, with the final temperature maintained for 8 min (total analysis time of 21 min). A 1-m guard column (0.32-mm i.d.) was connected at the head of the chromatographic column with a glass union (Supelco). The carrier gas was helium at a flow rate of 1.5 mL/min. The flow was split to both detectors with a ratio of 1:10 in favor of the FID using a glass “Y” connector (Supelco) and two 1-m deactivated fused silica Analytical Chemistry, Vol. 71, No. 1, January 1, 1999

79

Table 1. Calibration of Analytical Method

number of data points working concentration range (µg/L) regression coefficient slope intercept

PCP

oil

10 0.1-30 0.997 1.71 -0.0337

10 5-1000 0.998 2.42 × 103 1 × 105

Table 2. Analytical Performance previous wood extract treatment,14 1-g sample size relative detection limits PCP (mg/L) oil (mg/L) absolute detection limits PCP (ng) oil (µg)

0.2 2 28 1000

Table 3. Effect of Oil-Standard Chemical Composition on Oil Assessment GC-FIDa,b (mg/L)

FT-IRa,b (mg/L)

difference (%)

shell label 645-900 CCLGO HGO

523 ( 6 520 ( 6 535 ( 6

741 ( 8 949 ( 10 376 ( 4

-29 -45 +42

average concn (mg/L) standard deviation (%)

526 8

700 40

oil standards

a Average of three determinations. measurement.

b

Standard deviation on the

this work, 0.1-g sample size

1.5 2 60 80

columns of different inside diameter (0.110 and 0.210 mm). The injection volume was set at 0.5 µL. Instrument control, data acquisition, and analyses were performed using the HP 3365 Chemstation software. For the purposes of method comparison, a second setup identical to the one described in ref 14 (GC and FT-IR) was used. Chemicals. The SFE-certified CO2 (99.9999%) and premixed methanol-modified CO2 were obtained from Scott Specialty Gases (Plumbsteadville, PA). Helium (99.999%) carrier gas and 5% argon/ methane ECD auxiliary gas were from Union Carbide (Toronto, Ontario, Canada). 2,3,6-Trichlorophenol (TCP) was provided by Fluka (Ronkonkoma, NY), PCP (99%) by Aldrich (Milwaukee, WI), and acetic anhydride (98%) by BDH (Montre´al, Que´bec, Canada). All the chemicals were used as received. The methanol, Freon 113, toluene, hexane (all 99.8% and more), and anhydrous sodium phosphate (99%) were supplied by Fisher Scientific (Fairlawn, NJ). Blanks of these solvents were verified by GC-ECD/FID before being used for sample preparation and analysis. The PCP solvent used is manufactured by Shell Canada (Sarnia, Ontario, Canada) under the label 645-900 and complies with industry standards for oil-borne preservatives.18 A heavy gas oil (HGO) and a catalytic cracking light gas oil (CCLGO), also obtained from Shell Canada, were used to compare system calibration. Pipetted volumes of PCP and oil in methanol solution were used for instrument calibration. TCP was added as an internal standard for samples prior to derivatization and/or solvent extraction. Wood Sample Preparation. Supercritical fluid and solvent extractions were performed on two stocks of powdered wood, with their preparation described elsewhere.14 The first, used as a standard sample for evaluating extraction efficiencies, was obtained from a PCP-impregnated red pine pole. Its pentachlorophenol and oil average contents were assessed using neutron activation analysis (NAA) for PCP and the method in ref 14 for both PCP and oil: 5.8 ( 0.3 mg/g PCP and 75 ( 1 mg/g oil. The second, used for blank extractions, was from a nonimpregnated red pine pole. Spiked samples were prepared by adding known (18) Canadian Standards Association. Wood Preservation; Series CAN/CSA-O80M89; CSA: Toronto, Ontario, Canada, 1990.

80 Analytical Chemistry, Vol. 71, No. 1, January 1, 1999

amounts of a 5% PCP-in-oil solution to the blank wood powder. The wood borings from a newly PCP/oil-impregnated red pine pole used to establish the performance of the technique were prepared according to a procedure described elsewhere.19 Supercritical Fluid Extraction. SFE experiments were achieved dynamically on 0.3 g of powdered wood using CO2 or methanol-modified CO2. All extractions were done with a pump flow rate of 1 mL/min and the extraction time adjusted so that 10 thimble volumes (Nv ) 10) were extracted for all the experiments. The extraction time was varied from 1.75 to 16 min, and the cartridge temperature was fixed between 40 and 120 °C. The variable restrictor temperature was held constant at 45 °C throughout the extraction. The stainless steel trap was rinsed with 1.2-mL portions of organic solvent (Freon or methanol) following the extraction step. METHOD DEVELOPMENT Treatment of Wood Extracts. An attempt to simplify the solvent fractionation step of the existing instrumental method by finding selective SFE conditions to successively extract the two types of compounds (PCP and oil) was first performed. Results showed that the species were always coextracted but in different relative amounts.20 The possibility of using bonded-phase solid traps for selective postextraction recovery was also evaluated by other authors21 and proved to be incompatible with solventmodified supercritical fluids. A new selective and simultaneous chromatographic determination of PCP and oil was then considered. The basic steps of the treatment of wood powder extracts developed in order to optimize this approach are shown in Figure 1. The method was adapted from one reported earlier that required an isolation step of PCP and oil prior to their analysis by GC and FT-IR.14 It can be applied to samples extracted by both SFE and solvent sonication. A 200-µL aliquot of the organic solution is mixed with 10 mL of a buffered aqueous solution at pH 9.4. Most of the PCP is in its ionic form at this pH and therefore remains in the aqueous phase. Upon adding 500 µL of acetic anhydride, the PCP is acetylated and extracted with 2.00 mL of hexane along with oil. Then, 0.5 µL of supernatant organic phase is injected into the GC for the determination of both PCP and oil species. (19) Gilbert, R.; Besner, A.; Octeau, P. For. Prod. J. 1997, 47 (3), 81-88. (20) Leblanc, Y.; Gilbert, R.; Hubert, J. Proceedings of the 15th Annual Meeting of the Canadian Wood Preservation Association (CWPA); CWPA: Vancouver, BC, Canada, 1994; pp 49-59. (21) Mulcahey, L. J.; Hedrick, J. L.; Taylor, L. T. Anal. Chem. 1991, 63, 22252232.

Chromatographic Determination of PCP and Oil. Figure 2 shows typical chromatograms of PCP and oil recorded simultaneously under an oven temperature program slightly different from the final conditions given in the Experimental Section (total analysis time extended to 30 min). The ECD signal shows trichlorophenol and pentachlorophenol in their acetylated forms. On the FID channel, the chromatogram is more complex, and conditions do not provide a baseline resolution for all the constituents. The low retention peaks (5-10 min) are associated with the most volatile aromatic compounds in oil such as benzene, toluene, and xylenes (BTEX). A large number of fairly wellresolved peaks are also present in the retention time range of 8-14 min which are all naphthalene alkyl derivatives (naphthalene being the highest peak) such as methyl-1 naphthalene, dimethyl1,1 naphthalene and ethyl-1 naphthalene. Among other features observed, a series of fine, well-resolved peaks appears in the retention time range of 10-25 min that is associated with the n-alkane series. The chromatographic profile also shows an important baseline drift, probably as a result of the presence of a large number of aromatic derivatives in the oil. A better resolution could possibly be achieved with a longer column or a smaller bore but at the expense of analysis time. Under these conditions, it shows the different constituents of the oil solvent used for oilborne preservatives, which is a level of information not accessible when using the FT-IR technique. Oil determination is achieved by integrating the area under the total of the peaks and drawing a baseline from the beginning of the oil pattern up to the end. This chromatography was slightly compressed for the rest of the study (total analysis time of 21 min) by using the instrumental conditions given in the Experimental Section. Different calibration levels showed the analytical method to have working ranges extending over 3 orders of magnitude for PCP and oil. An example of the regression parameters obtained for a typical calibration is given in Table 1. The precision of the method was obtained by performing 10 consecutive injections of a typical average concentration sample. A relative standard deviation of 1% was observed for oil and 4% for PCP. To compare the performance of the new method with the previous technique, aliquots of the same wood extracts were analyzed using both approaches with their respective wood extract treatments (see Figure 1 of ref 14 for FT-IR/GC-ECD and Figure 1 in this work for GC-ECD/FID). Although only 0.1 g of wood powder was extracted in the second case, the extract solution concentration is the same since all the quantities have been scaled down by a factor of 10, i.e., 0.1 g in 4.00 mL with an aliquot of 0.200 mL. The results compared in Table 2 show a relative detection limit for PCP almost 10 times higher than for the FT-IR/GC-ECD method. This is explained by the fact that only 1/10th of the flow rate at the outlet of the column actually goes to the ECD, the remaining portion passing through the FID. Also, half the sample volume is injected in the new system. For oil, the inherent higher sensitivity of the FID when compared to an FT-IR based method is counterbalanced by the fact that the amount of sample injected is actually spread over a 10-min detection period. When considering the absolute detection limit, the GC-ECD/FID method has a higher limit for PCP, not because of the smaller sample size but because a smaller aliquot is transferred for derivatization and solvent extraction. As a matter of fact, 0.200 mL of wood extract

Figure 3. SFE recoveries for PCP under different experimental conditions.

is transferred in the derivatization bottle and then recovered into 2.00 mL of hexane, giving a dilution factor of 2.5 when compared with the GC-ECD and FT-IR methods. However, this absolute detection limit for PCP is still 1000 times lower than the Canadian Standards Association (CAN/CSA-O80.4-M89)22 required concentration in the impregnated portion of the wood. It can also be noted that oil has a lower absolute detection limit since it does not have to be extracted successively with solvent, which accounts for the limiting dilution of oil prior to analysis. The accuracy of the GC-ECD/FID and the FT-IR/GC-ECD methods for oil determination is equivalent when analyzing oil standards or spiked samples of blank wood powder or of filter paper. However, when an extract of the PCP/oil-impregnated wood standard is used, the net result in this case is that the FID detection of the GC-ECD/FID method gives 30% lower oil content when compared to the FT-IR determination. The FID chromato(22) Canadian Standards Association. Wood preservation; CAN/CSA-O80.4-M89; CSA: Rexdale, Ontario, Canada, 1989.

Analytical Chemistry, Vol. 71, No. 1, January 1, 1999

81

Figure 5. Partition of the oil components of SFE extracts between the solid trap and solvent trap as established by the ECD chromatograms (120 °C, fluid density of 0.65 mg/L, and Nv ) 10).

Figure 4. SFE recoveries for oil under different experimental conditions.

gram obtained from the analysis of this extract clearly showed that the volatile aromatics appearing in Figure 2 (BTEX) were missing, possibly due to their loss during the grinding process. The FID response in terms of the total area is an average of all molecular species response factors to the detector. However, these factors do not seem to vary enough to affect the overall accuracy of the results. On the other hand, the accuracy of the FT-IR detection mode is expected to be affected by these changes since the measurement is predominantly on the intensity of the CH vibration band around 2930 cm-1 of the saturated hydrocarbons; the net result in this case for the impregnated wood standard sample would be an overestimation of the oil content when compared to the FID determination. To verify this, an extract of the impregnated wood standard sample was analyzed against different calibration curves obtained from five standard solutions of three different oils using the GC-ECD/FID and FT-IR/GC-ECD techniques. The first calibration was done with the Shell label 645900 used for oil-borne preservatives. The second calibration was 82 Analytical Chemistry, Vol. 71, No. 1, January 1, 1999

done with a catalytically cracked light gas oil (CCLGO, hence rich in aromatics), and the third calibration was done with a heavy gas oil fraction (HGO), which comprises mainly n-alkane homologues. The oil assessment results are given in Table 3. With the FID mode of detection, the three calibrations lead to a similar oil concentration with only 8% RSD. On the other hand, using FT-IR and the Shell standard, an overestimation of about 29% oil is obtained due to the fact that the volatile aromatics present in the calibration samples are not considered by this mode of detection. If the same sample is analyzed by FT-IR using the CCLGO calibration that has a higher aromatics content than the Shell 645900, this method tends to accentuate the overestimation of the oil content in the wood sample. The opposite can be observed when calibration is achieved with the alkane-rich HGO oil. This shows that the accuracy of the analysis of the samples by FT-IR is effectively dependent on the aromatic contents, 40% RSD in the case of the three oil standards that were used for calibration. Therefore, the FID method proves to be less susceptible to variation in the oil sample composition, which can be an important issue when it comes to measuring weathered samples. At this point, the GC-ECD/FID method for oil determination not only proves to be simpler in sample preparation but also gives more information on the constituents of the mixture and is less susceptible to variation in the final composition of samples collected from in-service poles. Optimization of the SFE Parameters. Preliminary results have shown that quantitative extraction using pure CO2 is not possible, as some degree of interaction could be taking place

Figure 6. Overall PCP and oil recovery from SFE at 120 °C, fluid density of 0.65 mg/L, and Nv ) 10.

between the wood and analytes, especially PCP.20 Wood is a complex matrix composed of lignin and cellulose, the former being a long polymeric chain with many polar functional groups at its surface, such as phenols, carbonyls, and aliphatic hydroxyls. These functions are susceptible to form hydrogen bonding with PCP that needs to be broken up by a competitive solvent modifier in the extraction fluid. Methanol was chosen for that purpose since it can be mixed at reasonable proportions with CO2 and showed good solubility properties for PCP and oil components. Considering this, the SFE parameters were optimized for methanol-CO2 composition, temperature, and pressure. A two-trap system was used in order to take into account the losses observed in preliminary tests with a single trap.20 It was found that, when methanol is used in CO2, droplets tend to form in the solid trap during CO2 depressurization. After a certain time, the liquid methanol covering the solid phase is pushed through and sputters at the end of the trap (even if cryogenically cooled), resulting in apparent losses of the analytes. The effect is more dramatic for PCP because this species is more soluble than oil in the evacuated solvent. Figures 3 and 4 show PCP and oil extraction recoveries respectively as a function of density for different fluid compositions (CO2 with 2% and 10% MeOH) at two temperatures (40 and 120 °C). Other temperatures were also investigated, but results ranged between the extreme cases presented here. Each graph presents recovery plots for the solid trap and the solvent trap containing hexane or Freon 113 and the total amount recovered by combining

Figure 7. Correlation between SFE GC-ECD/FID and solvent/ sonication FT-IR/GC-ECD results for PCP and oil of in-service red pine poles.

both traps. Clearly, as the extraction time is increased along with the extraction densityswhich gives more time for methanol to accumulate and ultimately break throughsmore PCP gets into the second trap. Ultimately, almost all the PCP is washed through the solid trap when 10% MeOH is used. For oil, increasing the temperature from 40 to 120 °C actually increased the recovery on the solid trap. The simple reason for this was that the oil extraction efficiency is increased markedly under these conditions, before methanol accumulates and washes it away from the trap. Figure 5 shows some examples of the partition of the oil components between the two traps when SFE is performed on wood samples under the given conditions. It is interesting to note that selectivity in trapping the n-alkanes in the solid trap increases with the amount of modifier in the fluid. Figure 6 summarizes the PCP and oil recoveries achieved by SFE on wood samples when the contents of both traps in series are combined. Whereas it is evident that PCP is quantitatively extracted from wood at 120 °C and 0.65 g/mL with 10% MeOH in CO2, the oil appears to reach a maximum of about 70% for all fluid compositions. However, this is 70% of the nominal oil content as determined by FT-IR on the powdered wood standard sample used for establishing the extraction efficiencies (75 ( 1 mg of oil/g of wood powder). It was demonstrated previously that analyzing the oil by GC gave 30% lower values than the FT-IR determination, so that the oil extraction is considered complete when it reaches 70% of the FT-IR nominal value. The final SFE conditions for PCP/ oil-impregnated wood samples were then set to 10% MeOH/CO2, 120 °C, and 0.65 g/mL. Analytical Chemistry, Vol. 71, No. 1, January 1, 1999

83

Figure 8. FID chromatograms resulting from the analysis of (a) PCP/oil-impregnated wood sample at radial position of 4 cm, (b) blank wood sample, and (c) PCP solvent under the same instrumental conditions.

VALIDATION WITH FIELD SAMPLES Newly PCP/Oil-Impregnated Pole. Various samples collected from a newly impregnated red pine pole were analyzed using the SFE GC-ECD/FID combined method. The accuracy of the results was established by comparing them with those obtained with the previous method (solvent/sonication followed by FT-IR/GC-ECD) on a different project,23 using aliquots of the very same samples that were stored in the freezer. Figure 7 shows the radial distribution of PCP and oil obtained at one longitudinal position of the pole. PCP concentration decreases regularly from the surface (16.8 mg/g of wood) to the heart of the pole, with no traces found at depths of 8-10 cm. This is a normal distribution since the impregnation process generally limits the penetration of the biocide into the sapwood. A good correlation of values can be observed between the two curves (SFE GC-ECD/FID versus solvent/sonication GC-ECD) with a maximum deviation of 7%. On (23) Besner, A.; Gilbert, R. Unpublished results, 1994.

84 Analytical Chemistry, Vol. 71, No. 1, January 1, 1999

Figure 9. PCP and oil radial distributions in a newly PCP/oilimpregnated red pine pole as established by SFE GC-ECD/FID and solvent/sonication FT-IR/GC-ECD.

the other hand, the highest oil concentration is also found at the wood pole surface (107 mg/g of wood) and then decreases to 24 mg/g. The radial distribution of oil also generally follows the trend previously established but with a degree of correlation less than that for PCP. However, the overestimation characterizing the FTIR values needs to be taken into account, as explained previously in this paper. On the other hand, it is surprising that the SFE GC-ECD/FID method gives higher values for the position of 4 cm and above. The FID chromatogram obtained for the sample at the 4-cm radial position and the chromatograms recorded for an extract of a blank wood sample and for the PCP solvent (Shell Canada, label 645-900) under the same instrumental conditions are presented in Figure 8. By looking more closely at the chromatograms, which show the strength of this approach as compared with FT-IR, it can be seen that compounds not seen previously were extracted and interfere with the oil components at the right end side of the chromatographic pattern. The majority of these compounds in the retention time range of 11-15 min are attributed to coextraction of natural compounds in wood. In some cases, these wood extractives contribute to more than 84%

of the total FID area and remain practically unseen by the ECD. These compounds have not been identified yet. By integrating the chromatograms in the retention time range of 0-11 min to exclude the wood extractives, a better correlation is obtained for samples at these depths (see dotted line in Figure 7). In-Service PCP/Oil-Impregnated Poles. Further validation of the SFE GC-ECD/FID combined technique was obtained by analyzing samples from in-service wood poles. Some of these poles had been subject to weathering for more than 15 years. The results were compared to those obtained by solvent/sonication FT-IR/ GC-ECD in a previous project 2 years earlier.23 Aliquots of the very same powdered samples kept continuously frozen in airtight bags were used. Figure 9 shows the correlation between the two methods for PCP and oil, respectively. The perfect correlation is given by the dotted line of each graph. The deviation of the results from the perfect match was estimated by performing a linear fit through zero over the data points after exclusion of the three double-dotted circle points shown in Figure 9. The curve from the linear fit shows that the SFE values are 13% lower than those obtained previously for PCP. To ensure that quantitative SFE extraction is still achieved on these aged wood powdered samples, postextraction residues were analyzed by NAA for residual chlorine. Residual PCP quantities found after SFE extraction were generally less than 5% of the total and could not completely account for lower PCP recovery. As for the oil content, the

concentrations obtained were 19% lower than the former FT-IR values. The observed bias between the two methods for oil is somewhat consistent with the previous values and explains the difference observed in Figure 9. There are no apparent losses of oil due to extraction or storage of the sample. Given the results obtained for newly and in-service impregnated wood pole samples, the combined SFE GC-ECD/FID method proves to be a valuable tool for PCP and oil determination. ACKNOWLEDGMENT The authors thank Hydro-Que´bec’s research institute (IREQ) for its contribution to the funding of the project. They are grateful to Que´bec’s Fonds pour la Formation des Chercheurs et l’aide a` la recherche (FCAR) and the Natural Sciences and Engineering Research Council of Canada (NSERC) for a postgraduate scholarship awarded to Y.G.L. and to the support of the research. They express their gratitude to G. Arsenault for his assistance in some of the experimental work. A special word of appreciation goes to A. Besner, P. Te´treault, and J.-F. Labrecque for their support in the application of the GC/FT-IR method and for supplying newly and in-service PCP/oil impregnated red pine pole samples. Received for review April 1, 1998. Accepted October 15, 1998. AC9803683

Analytical Chemistry, Vol. 71, No. 1, January 1, 1999

85