Anal. Chem. 2007, 79, 106-112
Separation of C2-Naphthalenes by Gas Chromatography × Fourier Transform Infrared Spectroscopy (GC×FT-IR): Two-Dimensional Separation Approach Frank Cheng-Yu Wang* and Kathleen E. Edwards
Analytical Sciences Laboratory, Corporate Strategic Research, ExxonMobil Research and Engineering Company, 1545 Route 22 East, Annandale, New Jersey 08801
A two-dimensional separation approach involving gas chromatography (GC) and Fourier transform infrared spectroscopy (FT-IR) is used to separate C2-naphthalene isomers at or near baseline resolution. In addition to GC separation, the FT-IR also plays an important role in the separation, as well as its traditional role of detection and identification. This two-dimensional separation approach for the analysis of a C2-naphthalene isomeric mixture is a good example of separation design based on molecular difference and the characteristics of an analytical instrument. The details of this two-dimensional separation are discussed, along with the advantages and limitations of this approach. While two-(or multiple-)dimensional separations have demonstrated superior capabilities in the characterization of complex, largely unknown mixtures, the development of GC×FT-IR illustrates the applicability of this analytical approach in the separation of simpler but still challenging, mixtures. The GC×FT-IR results have extended this approach toward its application to the analysis of samples of more complicated composition. Two-(multiple-)dimensional separation is one of the major, recent advances in analytical science.1 Most investigations have focused on two-dimensional chromatographic techniques that offer superior advantages in the separation and quantification of complex mixtures.2,3 However, multidimensional separations can be applied readily to solve certain problems that involve mixtures that are not that overly complex. Many such analytical problems appear to be straightforward when, in fact, they are very difficult to implement in routine practice. The concept of multidimensional chromatographic separation can be extended to other types of separation techniques, such as GC×MS.4 Ideally, when a detector is utilized as a separation tool, each component in a mixture will produce one, and only one, * To whom correspondence should be addressed. E-mail: frank.c.wang@ exxonmobil.com. (1) Liu, Z.; Phillips. J. B. J. Chromatogr. Sci. 1991, 29 (6), 227-231. (2) Kinghorn, R. M.; Marriott, P. J. J. High Resolut. Chromatogr. 2000, 23 (3), 245-252. (3) Blomberg, J. J. Chromatogr,, A 2003, 985 (1-2), 29-38. (4) Wang, F. C.-Y.; Qian, K.; and Green, L. A. Anal. Chem. 2005, 77 (9), 27772785.
106 Analytical Chemistry, Vol. 79, No. 1, January 1, 2007
identity by that technique. If the separation technique produces more than one possible identity or if the detector response is either unknown or impossible to predict theoretically, then the separation technique may not be suitable for unknown mixtures. However, the technique may still be applicable to specific analytical problems and may, in fact, have certain advantages over other, more universal, techniques. In contrast to the exploration of complex, largely unknown mixtures, routine and consistent quantitative analyses may be required for simple mixtures comprised mostly of known components. Experiments that monitor the relative concentration change of each component during a reaction process are such examples. Here, the separation-detection-identification system, while not suitable for unknown component exploration, may be ideal for quantitative analysis of the target components. Fourier transform infrared spectroscopy (FT-IR) is a candidate for such two-dimensional separations designs as long as FT-IR participates in the “spectroscopic” type of separation and detection. GC hyphenate with FT-IR (GC-FT-IR) has been used as a separation, detection, identification, and quantitative analysis tool for many applications.5,6 Most applications utilize the GC as the separation device and the FT-IR as the detection and identification unit. FT-IR also can be used as a discrimination/separation device based on different absorption band attributes from different functional groups or different configurations of the components in the mixture.7,8 Although the spectroscopic separation is not a physical separation as performed by multidimensional chromatographic separations, it can perform the same purpose as a chromatographic separation in certain analytical procedures. In this paper, we show that a two-dimensional GC×FT-IR technique can be used to effectively study the catalytic isomerization of C2-naphthalene isomeric mixtures. C2-Naphthalene isomers, produced mainly by fractional distillation and other petroleum separation processes, have wide applications in the (5) Herres, W. HRGC-FTIR: Capillary Gas Chromatography-Fourier Transform Infrared Spectroscopy. Theory and Application; Marcel Dekker Inc.: New York, 1990. (6) White, R. Chromatography/Fourier Transform Infrared Spectroscopy and Its Application; Verlag: Heidelberg, Germany, 1987. (7) Ragunathan, B. N.; Krock, K. A.; Klawun, C. Sasaki, T. A.; Wilkins, C. L. J. Chromatogr., A 1999, 856 (1-2), 349-397. (8) Ragunathan, B. N.; Krock, K. A.; Klawun, C. Sasaki, T. A.; Wilkins, C. L. J. Chromatogr., A 1995, 703 (1-2), 335-382. 10.1021/ac061149h CCC: $37.00
© 2007 American Chemical Society Published on Web 11/24/2006
chemical industry. One example is 2,6-dimethylnaphthalene (2,6DMN), which can be used as a precursor for 2,6-naphthalenedicarboxylic acid, which then can be used as the monomer for polyalkylnaphthalate. Efficient production requires the qualitative and semiquantitative measurement of the concentrations of each C2-naphthalene isomer to monitor and determine (1) the mechanism of catalytic isomerization reaction, (2) the catalytic reaction kinetics, and (3) the most economic point (the degree of the reaction) that the product (2,6-DMN) should be removed from the reaction mixture. The separation of C2-naphthalene isomers has a long history. Common nonpolar or polar stationary phases do not provide baseline resolution of all isomers. Mass spectrometry is of little discriminating value as the mass spectra of all C2-naphthalene isomers are very similar under various ionization techniques. Comprehensive two-dimensional gas chromatography (GC×GC) may offer another possible approach to separate this mixture, but the complete separation of C2-naphthalene isomers has not been achieved successfully using various stationary-phase capillary column combinations. Other approaches to the separation of the C2-naphthalene isomers include both gas9 and liquid chromatographic methods.10 However; these efforts have focused on the development of unique stationary phases to attempt to achieve baseline resolution.11 While these separation methods work, the stationary phase may not be commercially available, amenable to being chemically bonded to silica, or suffer from rapid degradation. The more desirable approach would be to develop a method based on commonly available stationary phases. In this study, a separation method for C2-naphthalene isomers was developed with a two-dimensional separation approach based on gas chromatographic separation and FT-IR spectroscopic separation/detection. This alternate approach utilizes a more general purpose stationary phase (such as nonpolar methylsilicon) as a boiling point separation capillary column in the GC. The FTIR detector provides a second dimension of separation by differentiating the various ring-position, alkyl substitutions for isomers that coelute in the boiling point separation. The details of this two-dimensional separation are discussed as well as the advantages and disadvantages compared with the traditional GC separation with common stationary phases. The qualitative and quantitative analysis aspects of this approach also are discussed. EXPERIMENTAL SECTION (a) C2-Naphthalenes. C2-Naphthalene isomers were purchased from Aldrich Chemical Co. (Milwaukee, WI). A mixture of C2-naphthalenes was prepared by mixing the isomers in approximately equal amount. (b) GC×FT-IR Conditions. The GC×FT-IR system consists of a Hewlett-Packard model 5890 gas chromatograph (Agilent Technology, Wilmington, DE) configured with inlet, columns, and detectors. A split/splitless (S/S) inlet system with an Agilent 7673B autosampler was used. The detector is a Hewlett-Packard model 5965B Fourier transform infrared spectrometer (Agilent (9) Zhang, H.; Dai, R.; Ling, Y.; Wen, Y.; Zhang, S.; Fu, R.; Gu, J. J. Chromatogr., A 1997, 787, 161-169. (10) Toshio, S.; Yoshihito, S.; Toshiyuki, K.; Takasi, I.; Yoshiobu, N.; Kazuhisa. J. Chromatogr. A 2000, 877, 61-69. (11) Dai, R.; Ling, Fu, R.; Zhou, W. J. Microcolumn Sep. 1995, 7 (5), 455-460. (12) Andrews, A. R. J.; Wu, Z.; Zlatkis, A. Chromatographyia 1992, 34 (3/4), 163-165.
Technology) The GC column used was a nonpolar column, Quadrex 007-1, 30 m, 0.25-mm i.d., 5.0-µm film, (Quadrex Corp. Woodbridge, CT). This column separates hydrocarbon molecules mostly by boiling point. About 1.0 µL of the sample was injected via a split/splitless (S/S) injector with a split ratio 25:1 at 300 °C in constant-pressure mode at 20 psi at an oven temperature of 230 °C. The oven was held at constant temperature at 230 °C for a 30 min total run time. The FT-IR spectrometer used a mercury cadmium telluride narrow-band (4000-750 cm-1) (nominal D* ) 1 × 1010 cm Hz 0.5/W) detector. The temperatures of the light pipe and the interface were maintained at 275 °C.The spectral resolution was set for 8 wavenumbers (cm-1) with six interferograms coadded for a scan rate of 1.5 spectra/s. FT-IR data were acquired and processed using Hewlett-Packard Petro-IRD GC-FT-IR software. After the GC×FT-IR data acquisition, the data set was further processed by GRAMS software (Galactic Industries. Salem, NH) to convert to ASCII format and exported to Microsoft Excel. The Excel file was processed and imported to a commercial program Transform (Research Systems Inc., Boulder, CO), to plot the twodimensional/three-dimensional (2D/3D) images, which were rendered in PhotoShop (Adobe System Inc., San Jose, CA) to generate publication-quality images. (c) 2DGC or GC×GC Conditions. The GC×GC system consists of an Agilent 6890 gas chromatograph (Agilent Technology) configured with autosampler, S/S inlet, columns, and detectors. The first dimensional separation was performed by a weak polar capillary column (SPB-5, 30 m, 0.25-mm i.d., 1.0-µm film), (Supelco Inc. Bellefonte, PA) and second dimensional separation by a midpolar column (BPX-50, 3 m, 0.25-mm i.d., 0.25µm film), (SGE Inc., Austin, TX). A dual jet thermal modulation assembly13 based on Zoex technology (Zoex Corp., Lincoln, NE) was installed between the columns. A flame ionization detector was used for this study. A 0.2-µL mid-distillate refinery stream sample was injected via a S/S injector with 75:1 split at 300 °C at constant-pressure mode at 45 psi. The oven was programmed from 60 °C with 0 min hold and 3 °C/min increment to 300 °C with 0-min hold. The total run time was 80 min. The modulation period was 10 s. The sampling rate for the detector was 100 Hz. After data were acquired, the data set was processed for qualitative analysis. The data were first converted to a twodimensional image by a commercial program Transform (Research Systems Inc.) and rendered further in PhotoShop (Adobe System Inc.) to obtain publication quality. RESULTS AND DISCUSSION A comprehensive GC×GC of a sample from a mid-distillate refinery stream is illustrated in Figure 1. The yellow box in the chromatogram indicates the elution time windows for the C2naphthalene isomers with the inset showing an enlarged view that identifies the elution position of individual C2-naphthalene isomers. The GC×GC chromatogram shows that traditional gas chromatographic separation with one capillary column is not able to completely separate all isomers of C2-naphthalene. The black circle in the inset of the Figure 1 indicates the position where 2,6-DMN (13) Ledford, E. B.; Billesbach, C. J. High Resolut. Chromatogr. 2000, 23 (3), 202-204.
Analytical Chemistry, Vol. 79, No. 1, January 1, 2007
107
Figure 1. GC×GC chromatogram of mid-distillate petroleum refinery stream. The yellow box in the chromatogram marks the elution position of the C2-naphthalene isomers. This region is expanded in the inset box, and the elution positions of individual C2-naphthalene isomers are labeled. The black circle in the inset chromatogram indicates that 2,6-DMN and 2,7-DMN coeluted in both separation dimensions.
and 2,7-DMN coelute regardless of the polarity of the stationary phase (nonpolar, boiling point type, or polar, wax type). In addition to the coelution of 2,6-DMN and 2,7-DMN, other C2-naphthalene isomers also coelute using single-column chromatographic separation. Consequently, even the advanced two-dimensional GC×GC separation with a nonpolar/polar capillary column set is not able to completely separate the C2-naphthalene isomers. Hyphenated GC and MS techniques, regardless of the form of combination (GC/MS, GC×MS, GC×GC/MS, or GC×GC×MS), offer little help as the mass spectra of the C2-naphthalene isomers have very analogous fragments under electron ionization conditions and the same parent mass under soft ionization conditions. Except for the two ethylnaphthalenes (ENs), the C2-naphthalene isomers differ only by the positions of the methyl groups. IR is one possible approach to distinguish these isomers by looking for the absorption related to their out-of-plane ring bending. The absorption frequency will shift depending on the position of ring substitution and the symmetry or nonsymmetry of the ring position substitution. This out-of-plane ring bending occurs between 600 and 1000 cm-1 in the infrared absorption. By coupling the GC with a FT-IR, the separation and semiquantitative analysis of all isomers of C2-naphthalene becomes possible. Instead of using FT-IR as a detection and identification device, one can take advantage of the concept of two-dimensional separation approach by utilizing the full or partial range of a FTIR spectrum to obtain a GC×FT-IR two-dimensional separation. In the two-dimensional separation by GC×FT-IR, the first dimension is conducted by GC with a nonpolar capillary column (boiling point-type) separation. The second dimension is a 108 Analytical Chemistry, Vol. 79, No. 1, January 1, 2007
Figure 2. Reconstructed GC-FT-IR chromatogram of a C2naphthalene mixture.
spectroscopic separation based on the difference of infrared absorption bands. In this case, FT-IR acts as a separation device for coeluted components using different absorption bands, in addition to its traditional role as a detection and identification device. Spectroscopic separation is not a real physical separation; however, it may be treated as such. Figure 2 demonstrates a reconstructed GC-FT-IR chromatogram of a C2-naphthalene mixture. As in previous work,11 a nonpolar capillary column (similar to boiling point) separation produces six major peak groups with four groups of coeluted isomers. Peak 1 is a coeluted mixture of 2-EN and 1-EN. Peak 2 is a coeluted mixture of 2,6-DMN and 2,7-DMN. Peak 3 is a coeluted mixture of 1,3-DMN, 1,7-DMN, and 1,6-DMN. Peak 4 is
Figure 3. GC×FT-IR 2D/3D chromatogram plot of C2-naphthalenes. The yellow box in the chromatogram indicates the area of interest.
a coeluted mixture of 2,3-DMN, 1,4-DMN, and 1,5-DMN. Peaks 5 and 6 are pure compounds of 1,2-DMN and 1,8-DMN, respectively. In GC-FT-IR, each data point along the retention time axis of the reconstructed chromatogram is represented by a FT-IR spectrum. Instead of plotting the summed absorbance as a data point, one can plot the whole FT-IR spectrum for each data point and align the spectra vertically along the retention time axis, with the intensity sticking out from the plane. This newly constructed chromatogram becomes a 3D plot, where the X-axis is retention time, Y-axis is FT-IR wavenumber (cm-1), and Z-axis is the absorbance (A). The chromatogram may be plotted in a 2D plane by converting the Z-axis absorbance intensity to a predefined color table/scale. Figure 3 is the 2D/3D plot of a GC×FT-IR chromatogram of the same experiment as in Figure 2. Although the gas chromatographic separation is not able to completely separate all the isomers of C2-naphthalene, the various bands of absorption (especially between 750 and 940 cm-1) indicate the opportunity for spectroscopic separation. The inset of Figure 3 illustrates the amplified chromatogram of the coeluted isomers at retention time between 10.0 and 13.5 min and FT-IR wavenumber between 750 and 940 cm-1. One way to judge the degree of separation among those coeluted isomers in this twodimensional separation approach is to examine FT-IR spectra of coeluted components along with the chromatographic separation. That is, if sufficient differences exist between their FT-IR spectra, the coeluting peaks can be separated. The power of this approach is that while individual separation/detection methods are inadequate, their combination may be sufficient such that all components of interest are fully resolved in two-dimensional space. FTIR spectra of the four coeluting peaks are processed separately to achieve the needed resolution. The separation of peak 1 into 1-EN and 2-EN isomers is very challenging, since there is little chromatographic separation and
the major absorption bands of 1-EN and 2-EN isomers are too close to be used to achieve complete baseline resolution. Fortunately, 2-EN has more than one absorption band that can be used to improve the separation between 1-EN and 2-EN in the second dimension. Figure 4 shows the overlapping pure compound reference FT-IR spectra and the amplified GC×FT-IR chromatogram of 1-EN and 2-EN in the coeluted peak 1 region. The figure is from retention time between 10.0 and 10.5 min as well as the FT-IR wavenumber between 750 and 940 cm-1. The separation of peak 2, consisting of 2,6-DMN and 2,7-DMN, would appear to be challenging as these isomers are almost superimposed on each other in both polar and nonpolar GC separations. Figure 5 shows the overlapping pure compound reference FT-IR spectra and the amplified GC×FT-IR chromatogram of 2,6-DMN and 2,7-DMN in the coeluted peak 2 region. The figure is from retention time between 10.5 and 11.0 min as well as the FT-IR wavenumber between 750 and 940 cm-1. As in the case of the separation of 1-EN and 2-EN isomers, the major absorption feature is not distinguishing, but FT-IR spectroscopic separation can reach near-baseline resolution when one of the minor absorption bands of 2,6-DMN is used. The separation between 2,6-DMN and 2,7-DMN is achieved with this twodimensional approach. The separation of peak 3 of 1,3-DMN, 1,6-DMN, and 1,7-DMN is not as straightforward as the 1-EN and 2-EN or 2,6-DMN and 2,7-DMN isomers, because two of the isomers in peak 3 are almost fully unresolved in the chromatographic separation dimension as well as in the FT-IR absorption. Figure 6 shows the overlapped pure compound reference FT-IR spectra and the amplified GC×FT-IR chromatogram of 1,3-DMN, 1,6-DMN, and 1,7-DMN in the coeluted peak 3 region. The figure is from retention time between 10.9 and 11.4 min as well as FT-IR wavenumber between 750 and 940 cm-1. Through careful examination of the separation Analytical Chemistry, Vol. 79, No. 1, January 1, 2007
109
Figure 4. Amplified GC×FT-IR chromatogram. The 2D/3D display is from retention time region between 10.0 and 10.5 min as well as the FT-IR wavenumber region between 750 and 940 cm-1.
Figure 5. Amplified GC×FT-IR chromatogram. The 2D/3D display is from retention time region between 10.5 and 11.0 min as well as the FT-IR wavenumber region between 750 and 940 cm-1.
in both dimensions (chromatographic and spectroscopic), a degree of separation was found to exist between the three coeluted components. Similar to last two cases, the separation does not depend on the major absorption band; rather, the second major absorption band provides better resolution with manageable peak intensity. The GC×FT-IR two-dimensional separation is demonstrated to be clearly superior to one-dimensional separation methods and GC×GC. 110 Analytical Chemistry, Vol. 79, No. 1, January 1, 2007
Another example of FT-IR two-dimensional separation is in resolving peak 4. In the GC separation shown in Figure 2 (Xaxis, retention time separation), 2,3-DMN, 1,4-DMN, and 1,5-DMN are almost superimposed as a single peak. Figure 7 shows the overlapped pure compound reference FT-IR spectra and the amplified GC×FT-IR chromatogram of 2,3-DMN, 1,4-DMN, and 1,5-DMN in the coeluted peak 4 region. The figure is from retention time between 11.6 and 12.1 min and FT-IR wavenumber
Figure 6. Amplified GC×FT-IR chromatogram. The 2D/3D display is from retention time region between 10.9 and 11.4 min as well as the FT-IR wavenumber region between 750 and 940 cm-1.
Figure 7. Amplified GC×FT-IR chromatogram. The 2D/3D display is from retention time region between 11.6 and 12.1 min as well as the FT-IR wavenumber region between 750 and 940 cm-1.
between 750 and 940 cm-1. The FT-IR separation can reach nearbaseline resolution for all three components and provides a perfect example of two-dimensional separation. When migrating from oneto two-dimensional separation, the resolution between 2,3-DMN, 1,5-DMN, and 1,4-DMN is enhanced beyond that achieved with either dimension separation alone. In this two-dimensional separation, the FT-IR functions not only in its traditional role as a detection and identification unit but as
a separation device as well. Hence, it is necessary to discuss resolution, qualitative, and quantitative analysis considering separation, detection, and identification in a collective fashion. The resolution of coeluting or adjacent peaks in both chromatographic and spectroscopic separations heavily depends on the concentration of components. High-concentration components not only exhibit extra length in peak height but also in peak width. An extremely high concentration component may further overload Analytical Chemistry, Vol. 79, No. 1, January 1, 2007
111
the chromatographic separation column, resulting in a nonsymmetric peak, or may saturate the spectroscopic detector causing a nonlinear detector response. A combination of high and low concentrations of coeluting or adjacent peaks will create more difficulty in the quantitative analysis. Detector sensitivity and dynamic range limitations are issues in all spectroscopic analyses, including FT-IR. The linear range of detection will further define the applicable range of quantitative analysis. The quantitative analysis of a FT-IR spectrum is mainly based on Beer’s law (A ) bc). Because the IR light path length b is fixed due to the dimension of a light pipe in GC×FT-IR configuration, the variables that determine signal response are absorbance (A), absorption coefficient (), and concentration (c). Absorbance (A) is normally calibrated based on the range of linear response and is a characteristic property of the specific detector. The absorption coefficient () depends on the strength of the absorption; however, once the specific band has been chosen for the quantitative analysis, the value will be fixed. The most accurate way to obtain the absorption coefficient is to calibrate with a pure reference compound. The concentration (c) can be adjusted by varying the sample size. However, no matter how A, , and c are optimized, the dynamic range of quantitative analysis for most FT-IR absorption measurements will not exceed 2 orders of magnitude. That is, the typical sensitivity limit for FT-IR detector for a component is ∼1% concentration, and the dynamic range of linear detection normally is no more than 2 orders of the magnitude (from 1 to 100%). Quantitative analysis of a FT-IR spectrum also involves the degree of spectroscopic separation of the component. Occasionally, even if the spectroscopic separation is not baseline resolved, one can still manipulate the spectra based on other absorption bands to perform the spectrum manipulation (addition, subtraction, multiplication, and division) to generate a spectrum that is unique to the component of interest. Within these restrictions, FT-IR is a good technique for qualitative and semiquantitative analysis. While GC×FT-IR may be very useful for some component separations, the lack of a unique absorption spectrum for many individual compounds limits its use in the separation and identification of unknowns in complex mixtures. GC×FT-IR may improve qualitative analysis, because the two-dimensional separation approach conceptually enhances its capability in separation. However, even combined with retention time information, the identification of an unknown component via IR spectroscopy is still a challenging task. The availability of spectral libraries associated with sophisticated and effective search and match software routines and automatic data processing of the digital absorption band/structure assignment have greatly broadened the application of this technique. Nevertheless, a positive identification in many cases still relies on the analysis of a pure reference
112
Analytical Chemistry, Vol. 79, No. 1, January 1, 2007
compound. Hence, the approach of two-dimensional separation by GC×FT-IR is most useful when facing a mixture that is not completely unknown, the difference among the components is unique in the FT-IR absorption, and the purpose of separation is to monitor the change of relative concentration. CONCLUSION C2-Naphthalene isomers may be separated by traditional onedimensional chromatographic only with special and not readily available stationary phases. However, separation can be achieved by a two-dimensional separation approach using commercial columns and FT-IR spectrometric detection. With the GC providing partial separation based largely on boiling point differences, FT-IR separation focuses on the structural differences within coeluting isomers with different infrared absorption bands. Without further hardware modification, the GC-FT-IR can be turned into a two-dimensional separation technique, GC×FT-IR, with a change in data processing. The advantage of using GC×FTIR methodology is shown in the baseline separation of C2naphthalene isomers. This type of separation is unique when compared to other separation approaches. The GC×FT-IR technique is suitable for qualitative analysis and for semiquantitative analysis over the limited linear dynamic range of detection (2 orders of magnitude). From GC×GC to GC×MS to GC×FT-IR, the power of two(multiple-)dimensional separations continues to demonstrate advantages not only for the analysis of complex, unknown mixtures but also for less complicated, largely known mixtures that are still challenging by conventional methods. In order to take advantage of multidimensional separations, one must have a good understanding of the purpose of analysis and the specific capability of each separation technique in order to maximize the benefit from the integrated multidimensional separation system. The GC×FT-IR experience has pushed this effort one step ahead to a new application. ACKNOWLEDGMENT The authors express their appreciation to Dr. Fred Lo, who provided the background knowledge of C2-naphthalene isomers mixture. The authors thank Dr. Clifford C. Walters for helpful discussion about the multidimension (heart-cut) separation and comprehensive two-dimensional separation. The authors also thank Dr. John S. Szobota for his help on the FT-IR instrumentation and data interpretation. The help of Norman E. Hoosain, who operated the instrument and collected the data, also is greatly appreciated. Received for review June 26, 2006. Accepted October 22, 2006. AC061149H
