trends in analytical chemistry, vol. 10, no. IO,1991 337

Vibrational spectroscopy for chromatographic detection in environmental analysis

Yuko Soma Tsukuba, Japan

Most environmental samples are complex mixtures and the molecular structural information provided by vibrational spectroscopy is therefore important in environmental analy- sis. The application of matrix isolation spectroscopy and re- sonance Raman spectroscopy for the detection of eluents in gas and liquid chromatography is reviewed.

Introduction Vibrational spectra provide molecular structural

information which is often not obtainable from other techniques e.g. the identification of functional groups. In environmental analysis structural infor- mation is essential, especially for the identification of molecular species, as environmental samples are always complex mixtures. The Raman microprobe method has been used for the characterization of in- dividual particles in various types of environmental particulate samples i. Major components in airborne particles of 1 pm and larger have been detected and identified. Pesticides adsorbed on soil surfaces have been characterized by infrared and Raman spectros- copy, where the accumulation behaviour in soils has been studied. The infrared technique has been par- ticularly successful in the study of the upper atmo- sphere, where spectrometers in an aircraft or a bal- loon measure the atmospheric concentrations of many compounds. Solar radiation is used as the in- frared source. Fourier transform infrared spectros- copy is important in the analysis of atmospheric trace gases, not only in the upper atmosphere but also in urban smo$.

However, the relative insensitivity of vibrational spectroscopy compared to mass spectrometry has limited its application as a detection method in chro- matography, although gas chromatography-Fourier transform spectrometry (GC-FTIR) with the light- pipe interface has now been widely used for complex mixture analysis. In this article, recent develop- ments in the application of vibrational spectroscopic methods for detection in GC or liquid chromatogra- phy (LC) are reviewed.

0165-9936191/$03.00.

Matrix isolation GC-FTIR spectroscopy An important development in the GC-FTIR tech-

nique has been the light-pipe interface, especially the ‘gold-coated light-pipe’, and the small focal area MCT detector (liquid nitrogen cooled mercury cad- mium telluride photoconductive detector). These improvements have helped minimize the loss in sig- nal intensity and made FTIR adaptable to capillary columns for environmental analysis3J4. However, the light-pipe must be heated to a temperature higher than that of the GC oven to avoid the condensation of analytes and the high temperature required for the analysis of complex mixtures creates a high back- ground signal in the IR spectrum. Therefore, only spectra of major components separated by capillary GC columns can be measured using the light-pipe in- terface. Gurka et ~1.~ determined the minimum quantities needed to yield identifiable spectra for 52 typical environmental contaminants. These ranged from 20 to 120 ng.

GC-matrix-isolation-FTIR (GC-MI-FTIR) uses matrix isolation to trap the GC effluent prior to measuring the IR spectrum of each component. He- lium containing a small amount of argon is used as the GC carrier gas. The separated component eluted from the GC column is deposited in an argon matrix on the rotating gold-plated disc which is cooled to 12 K in a vacuum chamber. IR spectra can be obtained by positioning the desired portion of the deposit at the focus of the IR beam. Small area deposition of the GC solute on a 0.25mm diameter spot and the narrow spectral bandwidth of the matrix isolated compounds increase peak absorbances. This method is roughly 100 times more sensitive than GC-FTIR with the light-pipe interface3T4. The spectra of the 22 isomers of tetrachlorodibenzo-p-dioxin (TCDD) could be distinguished and gram leve16. Mossoba et al. 9

uantitated at the nano- quantified the level of

2,3,7&TCDD in fish extracts in the 170-220 pg range. The S/N ratio of the spectra of minor GC peaks can be increased by extensive signal-averag- ing, because the solid argon matrix is stable for hours. The signal-averaging is useful in environ- mental analysis, because minor components can of- ten be important and should be identified. However,

OElsevier Science Publishers B.V.

338 trends in analytical chemistry, vol. IO, no. IO,1991

the spectra cannot be measured in real time because of optical arrangements. The spectra are usually measured after the chromatography has been com- pleted, whereas the light-pipe interface provides real time data.

Resonance Raman spectroscopy for detection in liquid chromatography Visible resonance Raman spectroscopy

Although remarkable progress has been made in the instrumentation for liquid chromatography in re- cent years, further progress can be expected in the separatory power of columns and detector perfor- mance .

Several conditions for the ideal LC detector should be considered: Yeung’ has listed six desirable characteristics, in which sensitivity (or detectability) and selectivity are the most important. The difficulty in improving detectors is mainly due to the presence of large amounts of the eluent (solvent molecules). Detection by resonance Raman spectroscopy has some advantages, especially for selectivity which de- pends on the difference between the intensity of re- sonance and that of normal Raman scattering. Inten- sities of resonance Raman scattering are ld to lo6 times higher compared with those of normal Raman scattering. When the excitation wavelength of the laser is selected properly at or near an allowed elec- tronic transition of the component of interest, con- siderable enhancement of the intensity of resonance Raman scattering of the component is observed, whereas the intensities of the background spectrum from the eluent remain unenhanced. Because there is no need to remove the eluent at the detector, the optical arrangement for the observation of Raman spectra is simple and similar to detection by fluores- cence.

The identification by resonance Raman spectros- copic detection is selective, because it utilizes infor- mation from both the vibrational spectrum and the electronic transition for the excitation. One of the difficulties in applying resonance Raman spectros- copy to LC detectors is that the excitation wave- lengths needed are generally in the UV region, whereas strong and stable lasers commonly used for Raman spectroscopy have excitation wavelengths in the visible or near IR region. Koizumi and Suzuki’ used coloured derivatives to analyse aliphatic amines and aldehydes by high-performance liquid chromatography (HPLC), monitoring the intensities of a resonance Raman band enhanced by the 488.0 nm line of the Ar ion laser. The detection limit for IZ- propylamine was 2 ng, using a semi-micro column (250 x 1.5 mm).

D’Orazio and co-workerslO,ll used a multichannel

Raman spectrometer for real time detection in HPLC, and described the advantages of this tech- nique in detail. With a multichannel Raman spec- trometer the measuring time of a spectrum was shortened and real time detection of fractions in HPLC was possible.

We have tried to improve detection in HPLC- resonance Raman. The Raman spectrometer con- sists of a double-monochrometer (f = 25 cm), equipped with an intensified photodiode array de- tector (700 channels). The observable frequency range was about 500 cm-’ with 488 or 514.5 nm exci- tation. The laser power at 488 or 514.5 nm was varied from 200 to 700 mW, depending on the inten- sity of the background spectrum from the eluent. The quartz Raman flow cell was 1 x 1 x 10 mm and the volume was 10 ~1. The longer side of the cell was held parallel to the entrance slit of the spectrometer to collect the scattered radiation efficiently. The laser beam was focused in the cell from the upper side and the injected laser beam was reflected by a small concave mirror located on the opposite side to increase Raman signals. The improvement of the cell shape and the double pass of the laser beam in the cell increased Ramen signals 20 to 30 times com- pared with a conventional Raman measurement using a micro solution cell, which enabled the signal collection every few seconds. The Raman signal was collected for 5 or 6 s at 0.1-s intervals. Afterwards, the spectrum of the solvent was subtracted out and the Raman spectra of analyte fractions were ob- tained.

Phenol and cresols in airborne dust were analyzed by this method. They were derivatized to the azo- compounds, which had an absorbance around 510 nm, to induce resonant enhancement for the excita- tion by an argon ion laser. Fig. 1 shows a three di- mensional description of Raman spectra with time, of a sample containing azo-derivatives of phenol, o- and m-cresols. Raman signals were accumulated for 5 s at intervals of 0.1 s. In the frequency range be- tween 1500 and 1100 cm-‘, N = N and C-N stretch- ing vibration, which is characteristic for azo-com- pounds, appears and the distinction between these compounds seems clear. The eluent was 80% aq. methanol in this analysis and methanol had a Raman peak around 1454 cm-‘, which limited the detectabil- ity at this frequency. This limitation can be over- come by an appropriate selection of the eluent. High scattering cross-sections of resonance Raman spec- tra, as compared to those of normal Raman spectra, simplify the detection and identification of LC frac- tions.

The chromatogram of methanol extract from dust collected in the ventilation system of an animal labo-

trenak in analytical chemistry, vol. 10, no. 10, 1991 339

m-cresol 80 ng

1500 IL00 1300 1200 cn-1

Fig. 1. Three dimensional Raman spectra of phenols in HPLC. Eluent: 80% aq. methanol, Excitation wavelength: 514.5 S&al collection: every 5 s at the interval of 0. I s.

mW

ratory and the corresponding Raman spectra are shown in Fig. 2. Peak absorbances in the chromato- gram were detected at 488 nm, which corresponds to the excitation wavelength of Raman spectra. The dust sample was characteristic for animal laborato- ries in that m-cresol was abundant, and presumably originated from the cresol solution used as a steril- izer in this laboratory. o-Cresol was predominant in cresols of airborne dust collected from the ventila- tion systems of a highway tunnel12 and it was con- sidered to be present in automobile exhaust; o-cresol is dominant in cresols contained in petroleum and cresols have been detected in exhaust gases13. o-Cre- sol is also a photoxidation product of toluene14.

The application of resonance Raman spectros- copy for detection purposes in HPLC has some limi-

Raman 488.0nm SJOmW

13.2

’

Fig. 2. Chromatogram of azo-derivatives of phenols contained in the airbone dust from an animal laboratory and Raman spectra of these fractions. Wavelength 488.0 nm, 500 mW. Eluent: 80% aq. methanol, Excitation wavelength: 488.0 nm, 500 m W.

tations, especially in the case of resonance enhance- ment in the visible wavelength region. The extent of the resonance enhancement and the form of a reso- nance Raman spectrum depend on the character of an electronic absorption band close to the laser fre- quency. When the coloured derivatives were used to analyze certain compounds, strong resonance Ra- man enhancement was observed in the vibrations as- sociated with a chromophore (part of a molecule which gives rise to the electronic transition) of the derivatives. In such cases, resonance Raman spectra of each derivative look similar and identification may become difficult when a chromophore is not coupled with the key vibrations of parent molecules.

UV resonance Raman spectroscopy To extend the applicability of resonance Raman

spectroscopy as an LC detector, the use of UV reso- nance Raman spectrosco y should be considered. Asher and co-workers’s3’gused UV resonance Ra- man spectroscopy for the determination of trace polycyclic aromatic hydrocarbons. Raman spectra of polycyclic aromatic hydrocarbons enhanced by visi- ble laser lines are disturbed by fluorescence emis- sion, but no fluorescence is evident in the 200-300 nm UV spectral region. UV resonance enhancement was sufficient to study trace levels of naphthalene, phenanthrene, and pyrene down to 20 ppb levels in complex mixtures of coal liquid distillates. However, Raman saturation induced by the high peak pulse powers and the high pulse energies of Nd:YAG lasers as well as the formation of photochemical in- termediates complicated resonance Raman mea- surement. The use of a high repetition-rate UV ex- timer pumped frequency doubled dye laser (200 Hz, 16 ns pulses) avoids these optical phenomenal7 and may improve the UV resonance Raman spectrome- tric detection method for HPLC in the near future.

trends in analytical chemistry, vol. 10, no. 10, 1991 340

References 1 J.J. Blaha, E.S. Etz and K.F.J. Heinrich, Raman Micro-

probe Analysis of Stationary Particulate Pollutants, EPA 600/2-80-173, US EPA, Research Triangle Park, NC, 1980. P.L. Hanst, Fresenius’ Z. Anal. Chem., 324 (1986) 579. J.F. Schneider, J.C. Demirgian and J.C. Stickler, J. Chro- matogr. Sci., 24 (1986) 330. P.R. Griffiths and D.E. Henry, Prog. Anal. Spectrosc., 9 (1986) 455. D.F. Gurka, R. Titus, P.R. Griffiths, D. Henry and A. Gior- getti, Anal. Chem., 59 (1987) 2362. T.T. Holloway, B.J. Fairless, C.E. Freidline, H.E. Kimball, R.D. Kloepfer, C.J. Wurrey, L.A. Jonooby and H.G. Palm- er, Appl. Spctrosc., 42 (1988) 359. M.M. Mossoba, R.A. Niemann and J.T. Chen, Anal Chem., 61 (1989) 1678.

7

8 E.S. Yeung, Adv. Chromatogr. (N.Y.), 23 (1984) 1. 9 H. Koizumi and Y. Suzuki, Bunseki Kagaku, 37 (1988) 190.

10 M. D’Orazio and U. Schimpf, Anal. Chem., 53 (1981) 809.

11 12

13

14

15 16

17

M. D’Orazio and R. Hirschberger, Opt. Eng., 22 (1983) 308. Y. Soma, National Institute for Environmental Studies, Tsu- kuba, unpublished data. P.H. Howard, Handbook of Environmental Fate and Expo- sure Data for Organic Chemicals, Vol. 1, Lewis, Chelsea, MI, 1989. P.B. Shepson, T.E. Kleindienst, E.O. Edney, G.R. Namie, J.H. Pittman, L.T. Cupitt and L.D. Claxton, Environ. Sci. Technol., 19 (1985) 249. S.A. Asher, Anal. Chem., 56 (1984) 720. CM. Jones, T.A. Naim, M. Ludwig, J. Murtaugh, P.L. Flaugh, J.M. Dudik, C.R. Johnson and S.A. Asher, Treruis Anal. Chem., 4 (1985) 75. C.M. Jones, V.L. Devito, P.L. Harmon and S.A. Asher, Appl. Spectrosc., 41(1987) 1268.

Dr. Yuko Soma is at the National Institute for Environmental Studies, Tsukuba, Japan. Her current interest is the exposure as- sesment of environmental hazardous compounds.

Certified reference materials for the quality control of measurements in environmental monitorina

E. A. Maier Brussels, Belgium

The monitoring and protection of the environment are based on measurement campaigns which cover long peri- ods of time and large geographical areas. Only accurate analytical results allow valid conclusions to be drawn about a situation and its evolution. The use of certified reference materials (CRMs) permits verification of the ac- curacy of the measurements. The importance of accuracy and the way CRMs may be used are presented in this review. The production and properties required of a good CRM are discussed. An overview of the types of CRMs in the non-nuclear field, available for the monitoring of the environment, is also given.

Particular aspects of environmental analysis The protection of the environment is currently a

major priority in many countries. Public concern and, consequently, the economic and political impact of environmental protection has led to the develop- ment of a number of projects. Regional, national or international regulations (e.g. EC Directives; interna- tional conventions such as the Paris and Oslo Con- ventions for the North Sea or the Barcelona Conven- tion for the Mediterranean) require monitoring programmes and sometimes establish maximum per- missible concentrations for certain contaminants.

These monitoring programmes have one common prerequisite: measurements of various parameters are necessary to evaluate the situation and its de- velopment. The results of the determinations are the basis for the decisions taken by the authorities and for possible ensuing actions. The effect of such ac- tions is again evaluated on the basis of measurements which are conducted over long periods of time. Trends and even the kinetics of decontamination processes may be established and actions can be modified on the basis of these. The economic impact of the decisions (closing factories, changes in work- ing practice, waste management, etc.) and the human effect (unemployment, displacement of populations, e.g. Seveso in Italy) can be enormous. Therefore, the analyses have to be the most accurate possible.

Need for accuracy An enormous number of analyses is performed for

the purpose of monitoring the environment. In addi- tion to the diversity in the number of the analyses to be determined and the wide range of concentrations, the complexity of the matrices to be studied must also be considered. Current analytical techniques with powerful instruments (often to a large extent au- tomated), have made it possible to determine concen- trations of lo-‘* g/g or quantities of lo-‘* g routinely

0165-9936/911$03.00. OElsevier Science Pub1ishersB.V.