Anal. Chem. 1883, 65, 1728-1735
1728
Quantitative Determination of Macromolecular Chlorolignosulfonic Acids in Water by Pyrolysis-Gas Chromatography/Mass Spectrometry with Single Ion Monitoring Willem M. G. M. van Loon. and Jaap J. Boon Unit for Mass Spectrometry of Macromolecular Systems, FOM-Zmtitute for Atomic and Molecular Physics, Kruislaan 407, 1098 SJ Amsterdam, The Netherlands
Bob de Groot Laboratory for Special Research, Water Transport Company Rijn-Kennemerland, Groenendael6, 3439 LV Nieuwegein, The Netherlands
A specific, highly sensitive and quantitative pyrolysis-gas chromatography/mass spectrometrysingle ion monitoring (Py-GC/MS-SIM)procedure has been developed for dissolved high molecular weight (MW >1000) chlorolignosulfonic acids in sulfite pulp mill effluents, river water, and drinking water. Satisfying reproducibility (average RSD 5 % ) was obtained using polystyrene as an internal standard. Maximum accuracy was obtained using a standard addition quantification procedure. The detection limit is 16-32 pg/L chlorolignosulfonic acids, corresponding to 0.10.2 pg/L adsorbableorganic halogen (AOX) and to 10-20 ng/L chloroguaiacyl substructures. The distribution of the chloroguaiacyl isomeric substructures in pure chlorolignosulfonicacids could be determined accurately by Py-GC/MS-SIM. In this paper, macromolecular chlorolignosulfonic acids have been identified and quantified in river water and in drinking water for the first time. The amounts of detected chlorolignosulfonicacids are 180-310 pg/L in the Rhine and 60 pg/L in derived drinking water, corresponding to 1-2 and 0.4 pg of AOX/L, respectively.
INTRODUCTION In the past decade, the analysis of the total amount of dissolved organic halogen (DOX) compounds by means of the adsorbable organic halogen (AOX)procedure has become very important.1-3 The interest in AOX is primarily fueled by the potential toxicity of DOX compounds and by its use as a general anthropogenic pollution i n d i ~ a t o r .It~ has been demonstrated by Stachel et al. that, in 1984, the pulp mills situated on the Rhine river were by far the major source of anthropogenic AOX in this river system.6 The AOX discharges by pulp mills can be estimated using their pulp production data and their average AOX discharge/tonne of (1) K h , W. Ph.D. Dissertation, University of Karlsruhe, 1974. (2) Summrische Wirkungs- und Stoffkenngrksen; Gruppe, H., Ed.; Beuth-Verlag: Berlin, 1985; Vol. 14, DIN 38409. (3) Stevens, A. A.; Dreesman, R. C.; Sorrell, R. K.; Brass, H. J.-Am. Water Works Aesoc. 1986, 77, 146-154. (4) Von Keller, M. Dtsch. Gewasserkd. Mitt. 1987, 31, 38-42. (5) Stachel, B.; Lahl, U.; Zeschmar, B. Sci. Total Enuiron. 1984, 40, 103-113. 0003-2700/93/0365-1728$04.00/0
pulp.B-8 These calculations show that approximately 1100 tonnes of AOX was discharged by pulp mills into the Rhine in 1990. Assuming that no losses of pulp mill AOX have occurred during the river transport, these AOX discharges would account for 76 % of the organohalogenload of the Rhine at Lobith, NL.9 The composition of the low molecular weight AOX discharged by pulp mills has been investigated extensively.lOJ1 It appears that only0.5-15% of the AOX of pulp mill effluent can be extracted by diethyl ether.12Js Lindstrtim et al. reported that 5-30 % of the AOX of pulp mill bleach effluents is of low molecular weight (MW lOOO), and lyophilized. The isolationefficiencyof this procedure for chlorolignosulfonicacids from pulp mill effluent is 39%of the DOC.S2 Aqueous solutions of the isolated materials (8mg/mL) were prepared. The sample solutionswere homogenized by magnetic stirring before analysis. A standard solution of polystyrene in toluene (23.6 ng/pL) was prepared. Standard solutions of the four chloroguaiacylisomers (-lOng/L),eachspikedwithguaiacol (-lOng/L), were prepared in diethyl ether. Py-GC/MS-SIM. Two sample aliquots of 5 pL (2 X 40 pg) were deposited onto the pyrolysis wire (distance two droplets, 0.5 cm) and were dried under rotation and in vacuo. Pyrolysis conditions used were as follows: Curie point temperature, 510 OC; pyrolysis time, 5 s; glass liner temperature, 180 OC; pyrolysis injector temperature, 240 OC. The volatile pyrolysis products were swept by the carrier gas onto the GC column. GC conditions were as follows: column, DB-1 (J&W,Folsom, CA);length 60 m X i.d. 0.32 mm; film thickness, 1pm; carrier gas, helium; linear gas velocity, 27 cm/s; temperature program, 30 "C (5min) to 290 OC (10min) at 8 OC/min. MS conditions: ionsource temperature, 180 O C ; ionization mode, EI; ionization current, 250 pA; electron energy, 70 eV; scan range, m/z 102-105 + 157-161 (SIM mode) or 45-350 (full-scanmode) and cycle time, 0.6 (SIM mode) or 1.0 s (full-scan mode). The standard solutions, the reference pulp mill effluent sample, and the Rhine water sample (inlet WRK) were measured by Py-GC/MS (full-scanmode). All the samples were measured by Py-GUMS-SIM. Identification of Chloroguaiacyl Isomers. The elution order and the relative retention times of the four chloroguaiacyl isomers compared to guaiacol were determined by splitless injection (1pL) of the standard solutions and GC/MS analysis (full-scanmode). The chloroguaiacyl isomers in effluent, river, and drinking water samples were identified by their retention times ( f 2 scans), and coelution (scan defect f l scan) of the molecular ions m/z 158 and 160, and the correct m / z 160/158 chlorine isotope ratio range (2635%). Quantification of Chloroguaiacyl Isomers. Polystyrene (118 ng) was used as an internal standard and was introduced into the procedure in the pyrolysis step. The reproducibility of the Py-GC/MS-SIM analysis of the internal standard was investigated by five replicate runs of polystyrene (295 ng) and by copyrolysis of a mixture of polystyrene (295 ng) and sodium lignosulfonate (50 pg). In addition, separate pyrolysis followed by the combined separation and detection of polystyrene (295 ng) and sodium lignosulfonate (50 pg) was performed. This separate pyrolysis procedure was performed as follows. First, the polystyrene sample was introduced and pyrolyzed; second, a transfer time of 60 s was allowed during which the produced styrene entered the GC column quantitatively; third, the polystyrene pyrolysis liner was removed; and finally the sodium lignosulfonate pyrolysis sample was introduced and pyrolyzed. In order to minimize syringe volumeerrors (RSD
Z7.m
n**
chlorinatedguaiacyl and syringyl derivatives, the distribution of chlorinated substructures in chlorolignosulfonicacids and in chlorothiolignins probably can be determined quantitatively. Figure 4 shows the Py-GUMS-SIM trace of macromolecular material isolated from the Rhine. The internal standard (styrene) is the base peak in this pyrogram, and the scan range from 2500 to 3000 is the monochloroguaiacol search area. The identification of 5-chloroguaiacol in material isolated from Rhine water is demonstrated in Figure 5. As described above, confident identification of the peak at scan 2766 as 5-chloroguaiacol is based on the retention time (*2 scans) derived from the reference pulp mill effluent sample, on the coelution of mlz 158/160 ( h l scan), and on the correct mlz 1601158 chlorine isotope ratio range (20-35%). The identification data for 5-chloroguaiacol are as follows: scan 2766; scan defect 1scan; rnlz 1601158 ratio 26%. No other chloroguaiacyl isomers could be identified in this sample. To our knowledge, this ia the first report on the detection of a chlorinated pyrolysis product from macromolecular chlorolignosulfonic acids in river water. Figure 6 demonstrates the detection of 5-chloroguaiacol in material isolated from drinking water of the city of Amsterdam, NL, according to the same identification criteria as described above. The identification data for 5-chloroguaiacol are as follows: scan 2766; scan defect 1scan; rnlz 160/158 ratio 165%. This is the first time that a pyrolysis product from macromolecular chlorolignosulfonicacids has been detected in drinking water. Several small interfering peaks are present in the mass
ANALYTICAL CHEMISTRY, VOL. 65, NO. 13, JULY 1, 1903
'I 4
A
SCIG
I\
Flgurr 6. Identificationof the 5thioroguaiacyi(5Cffi)substructure of chlorolig~uifonlcacids k, drinking water of Amsterdam, NL, by PyGClMSSIM (mlz 158, 160; TIC mlz 102-105 157-161). The ldentiflcation data for 5-chloroguaiacoi are as follows: scan 2768; scan defect 1 scan: mlz 160l158 ratio 16%.
+
chromatograms near the 5-chloroguaiacol peak. Therefore, it is of crucial importance for the identification of chloroguaiacyl isomers in this complex matrix that the chlorine isotope peak (mlz 160) coelutes ( f l scan) with the molecular ion peak (mlz 158) within a correct mlz 1601158 chlorine isotope ratio range (2&35 % ). SIM at m/z 143/145 shows that these masses are more subject to interference than m/z 158/160. Obviously, the lower selectivity of m/z 143/145 is the result of the large amount of fragmentation induced by 70-eV E1 ionization, leading to a high abundance of odd-fragment ions. It is observed that the m/z 160/158 ratio decreases as the signal intensity approaches the detection limit. This can be explained by the larger loss of the weaker m/z 160 in the baseline signal compared to m/z 158, and this effect becomes more noticeable at lower concentrations. Py-GC/MS/MSSIM may further improve the selectivity and the S/N ratio of this analytical procedure. For example, LindstrGm and Schubert reported the specific GC/MS/MS analysis of 1,ldichlorodimethyl sulfone in aquatic organisms.M Quantificationof ChloroguaiacylIsomers. During the development and evaluation of a quantitative analytical procedure for chloroguaiacyl isomers, the following aspecta must be considered. (a) Reference Material. A representative reference material is essential for quantitative analysis of chlorolignosulfonic acids. The composition of the discharged chlorolignosulfonic acids depends on, for example, the type of wood, the pulping process, the bleaching processes, and the waste water treatment used at the time of sampling. Van Loon et al. demonstrated large differencesin the composition of these pulp mill effluentsusing Py-GUMS and Py-MS.18 Therefore, no reference material is commercially available and systematic sampling and analysis is necessary to obtain accurate information on the macromolecular compositionand its variations of a specific pulp mill effluent. Since this was not possible within the time frame of our research project, we composed a volume-averaged pulp mill effluent reference sample using nonsystematic samples.22 We determined DOC and AOX fraction factors to be as follows: FDOC= 0.53; FAOX= 0.0061. Probably, such a reference sample gives the most representative information on the composition of the discharged chlorolignosulfonic acids at the time of sampling. It is expected that the composition of the reference material is the most important factor which determines the accuracy of our procedure. The fact that the discharged chlorolignosulfonic acids cannot be defined precisely precludes a quanti(40) Lindetr6m, K.;Schubert, R. J. High Resolut. Chromatogr. Chromatogr. Cornrnun. 1984, 7,68-'73.
1799
tative procedure that is completely accurate. Furthermore, consideringthe rapid developmentsin the bleaching processea employed by pulp mille, annual samplingand composition of a reference pulp mill effluent sample followed by isolation and characterization of reference material is recommended. Synthetic polymers, e.g., poly(4-chlorostyrene),are structurally not representative for chlorolignosulfonicacids and probably give significantdifferencesin pyrolysis and standard addition behavior; thus they were not considered for use. The synthesis of reference chlorolignosulfonicacids has been attempted?2 but the resulting material was found not to be similar to chlorolignosulfonic acids isolated from pulp mill effluent@ (see also the quantitative results paragraph). (b) Reproducibility. It is well-established that Curie point pyrolysis is a reproducible sample introduction technique.mJ1 However, compared to liquid injection, the pyrolysis step introduces an additional source of experimental variation. In order to minimize the variation of the sample size, introduced when the dissolved sample is applied onto the pyrolysis wire using a syringe, relatively large sample volumes of 5 pL (RSD 4%) were used routinely. The pyrolysis reproducibility (RSD) of 5-chloroguaiacol from several samples is shown in Table I. The variation of the 5-chloroguaiacol signal produced by pyrolysis of the chlorolignosulfonic acid samples is high (RSD 19-36%). This phenomenon is discussed in more detail below and in the quantitative results paragraph. However, it does not present an analytical problem, since the high molecular weight fractions isolated from pulp mill effluent are virtually pure chlorolignosulfonicacids32 and consequentlyweight and DOC and AOX determinations are adequate for this sample type. On the other hand, the pyrolytic production of S-chloroguaiacol from material isolated from Rhine river and drinking water samples is quite reproducible (RSD
