Chemosphere 83 (2011) 131–136 Contents lists available at ScienceDirect Chemosphere journal homepage: www.elsevier.com/locate/chemosphere Determination of n-octanol/water partition coefficient for DDT-related compounds by RP-HPLC with a novel dual-point retention time correction ⇑ ⇑ Shu-ying Han a, Jun-qin Qiao a, Yun-yang Zhang a, Li-li Yang b, Hong-zhen Lian a, , Xin Ge a, , Hong-yuan Chen a a Key Laboratory of Analytical Chemistry for Life Science (Ministry of Education of China), School of Chemistry & Chemical Engineering and Center of Materials Analysis, Nanjing University, 22 Hankou Road, Nanjing 210093, China b Nanjing Environmental Monitoring Center, 175 Huju Road, Nanjing 210013, China article info abstract Article history: n-Octanol/water partition coefficients (P) for DDTs and dicofol were determined by reversed-phase high Received 22 June 2010 performance liquid chromatography (RP-HPLC) on a C18 column using methanol–water mixture as Received in revised form 18 December 2010 mobile phase. A dual-point retention time correction (DP-RTC) was proposed to rectify chromatographic Accepted 5 January 2011 retention time (tR) shift resulted from stationary phase aging. Based on this correction, the relationship between log P and log kw, the logarithm of the retention factor extrapolated to pure water, was investi- gated for a set of 12 benzene homologues and DDT-related compounds with reliable experimental P as Keywords: model compounds. A linear regression log P = (1.10 ± 0.04) log k – (0.60 ± 0.17) was established with n-Octanol/water partition coefficient (P) w correlation coefficient R2 of 0.988, cross-validated correlation coefficient R2 of 0.983 and standard devi- RP-HPLC cv Dual-point retention time correction (DP- ation (SD) of 0.156. This model was further validated using four verification compounds, naphthalene, 0 RTC) biphenyl, 2,2-bis(4-chlorophenyl)-1,1-dichloroethane (p,p -DDD) and 2,2-bis(4-chlorophenyl)-1,1- DDTs dichloroethene (p,p0-DDE) with similar structure to DDT. The RP-HPLC-determined P values showed good Dicofol consistency with shake-flask (SFM) or slow-stirring (SSM) results, especially for highly hydrophobic POPs compounds with log P in the range of 4–7. Then, the P values for five DDT-related compounds, 2-(2-chlo- rophenyl)-2-(4-chlorophenyl)-1,1,1-trichloroethane (o,p0-DDT), 2-(2-chlorophenyl)-2-(4-chlorophenyl)- 1,1-dichloroethane (o,p0-DDD), 2-(2-chlorophenyl)-2-(4-chlorophenyl)-1,1-dichloroethene (o,p0-DDE), and 2,2,2-trichloro-1,1-bis(4-chlorophenyl)ethanol (dicofol) and its main degradation product 4,40-dichlorobenzophenone (p,p0-DBP) were evaluated by the improved RP-HPLC method for the first time. The excellent precision with SD less than 0.03 proved that the novel DP-RTC protocol can significantly increases the determination accuracy and reliability of P by RP-HPLC. Ó 2011 Elsevier Ltd. All rights reserved. 1. Introduction et al., 1974; Platts et al., 2006). Therefore, there is a continuing de- mand for acquiring accurate and reliable P values. n-Octanol/water partition coefficient (P) provides direct infor- By now, a variety of methods have been in use to obtain P, mation on hydrophobicity that describes the tendency of distribu- which are divided into two major classes: experimental measure- tion of a solute from aqueous phase into organic constituents of ment and theoretical calculation. Traditional experimental meth- environmental compartments (Kenaga, 1980; Karickhoff, 1981; ods are shake-flask method (SFM) (OECD, 1981) and slow-stirring Bodor et al., 1989), and even into biological membranes (Kubinyi, method (SSM) (Brooke et al., 1986). Although basically correct, 1979; Sangster, 1989; Bechalany et al., 1991). This has made it SFM and SSM have some drawbacks concerning the consumption one of the most commonly reported physico-chemical properties of time and labor. Moreover, it is unreliable by using SFM for sub- of drugs, pesticides and other chemicals (Leo, 1991; Lambert, stances having high lipophilicity (log P > 5) due to the formation of 1993). Furthermore, P is frequently used to predict water solubil- emulsions between octanol and water phases during the shaking ity, soil (sediment)/water partition coefficient, bioconcentration (Finizio et al., 1997). A large number of theoretical methods also factor and nonreactive toxicity, etc. (Gobas et al., 1988; Neely have been developed for calculation of P. Among them, the most fa- mous one is fragmental constant approach proposed by Leo et al. ⇑ (1971). However, theoretical calculation is inclined to be inaccu- Corresponding authors. Tel.: +86 25 83686075; fax: +86 25 83325180 (H.-z. rate for compounds possessing several functional groups with Lian), tel.: +86 25 83592466; fax: +86 25 83592723 (X. Ge). E-mail addresses: [email protected] (H.-z. Lian), [email protected] (X. Ge). interactions not sufficiently defined. For relatively complex 0045-6535/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.chemosphere.2011.01.013 132 S.-y. Han et al. / Chemosphere 83 (2011) 131–136 molecules, such as some pesticides, the breakdown of the molecule where tR and tR represent the retention time and average retention can produce, in a few cases, large or unusual fragments for which time, respectively, of oligonucleotides. In fact, it is a dual-point fragmental values are not available in the list provided by the (DP)-RTC method. The essence of this DP-RTC is considered as rec- method or in the software library (Finizio et al., 1997). Reversed- tifying tR measured at different column situations to the one ob- phase high performance liquid chromatography (RP-HPLC) (Garst tained on an ‘‘imaginary average column’’ that is expressed as and Wilson, 1984) is another commonly adopted method for deter- tR,normalized. However, this normalization procedure have to be taken mining P for a wide range of compounds (OECD, 1989; Kurz and out after all experiments have been finished for the acquisition of Ballschmiter; 1999; Paschke et al., 2001), which is proved to show final tR data only by a selected mobile phase composition at which more reliable results than theoretical calculation (Finizio et al., all investigated oligonucleotides were completely separated for 1997). It profits from the relationship between the chromato- retention behavior study. In this present work, we proposed a novel graphic retention on a reversed-phase column and P of analytes. DP-RTC protocol to obtain reliable tR by assigning a real LC column Compared to SFM and SSM, RP-HPLC is more rapid and easy-to- used on a certain day (usually the day a column first used) as ‘‘stan- operate because it does not require any quantification of concen- dard column’’ instead of ‘‘imaginary average column’’ suggested by tration, and only retention time (tR) needs to be measured. Notice- Kohlbacher et al. Two specific standard compounds are employed as ably, determination of P by RP-HPLC is highly dependent on ‘‘anchor compounds’’. This normalization is able to rectify tR ob- accurate solute retention. Even though retention factor (k) is re- tained every day to that obtained on the certain day, thus reflects lated to P, for a given solute and a given stationary phase, it will de- the truest retention behavior of analytes. pend on the composition of the hydroorganic mobile phase Persistent organic pollutants (POPs) have drawn more and more (Sangster, 1997). Therefore, kw, the hypothetical retention factor attentions because of their highly bioaccumulative, carcinogenic, corresponding to 100% water, instead of k at a specific organic mutagenic and environmental endocrine disruptor effects. P is a modifier content, is employed (Snyder et al., 1979; Braumann, key physico-chemical property of POPs not only for evaluating their 1986). In this case, the retention is independent of any organic fate and distribution in the environment (Cohen et al., 1990; modifier effects, and the determined P is more similar to SFM/ Mackay, 1991), but also for assessing their accumulation and toxic- SSM-measured one. Some papers pointed out that the reliability ity in organism tissues (Mackay, 1982;USEPA, 1992). Nevertheless, for P determination of RP-HPLC is in the range of log P 0–6 (OECD, till now P data reported for POPs are still deficient and sometimes 1989). Problematic with more hydrophobic substances is the long- incorrect, and even discrepant each other for a compound. This is er retention during chromatographic elution, which may cause because their strong hydrophobicity is unsuitable for SFM measure- uncertainty of tR owing to peak broadening. Gradient elution has ment, or because their high structure complexity is difficult for soft- been tried to solve this problem (Krass et al., 1997; Verbruggen ware calculation. Therefore, RP-HPLC has been considered as a et al., 1999), but it actually takes more run time for system equilib- promising alternative for estimating P values of POPs in recent years rium before next injection, and normally gives poorer reproducibil- (Ayouni et al., 2005; Demare et al., 2007). DDT, one of the most well- ity in tR measurement than isocratic elution. In addition, whatever known synthetic organochlorine pesticides, belongs to POPs. The isocratic or gradient mode is used, some P data obtained by RP- term DDTs (total DDT) is often used to refer to the sum of all DDT HPLC still have large discrepancy with those of SFM/SSM. (dichlorodiphenyltrichloroethane), DDD (dichlorodiphenyldichlo- As we know, silica-based stationary phase collapses unavoid- roethane) and DDE (dichlorodiphenyldichloroethylene), as well as ably by continual flushing of aqueous mobile phase during the their isomers. DDE and DDD are also the major metabolites and course of usage. This column aging causes shift of tR for an analyte breakdown products of DDT in the environment. In addition, under the same chromatographic conditions in different measuring 2,2,2-trichloro-1,1-bis(4-chlorophenyl)ethanol (dicofol) is an orga- periods.