摘要:

ANALYST, JUNE 1993, VOL. 118 601 Attomole Detection of Nitroaromatic Vapours Using Resonance Enhanced Multiphoton Ionization Mass Spectrometry* Alastair Clark, Kenneth W. D. Ledingham, Archibald Marshall, Joseph Sander and Ravi P. Singhal Department of Physics and Astronomy, University of Glasgow, Glasgow, UK G 12 80Q A very sensitive and selective procedure has been developed for the detection of nitrobenzene (C6H5N02) and o-nitrotoluene (C6H4cH3No2) vapours using resonance enhanced multiphoton ionization mass spectrometry. The time-of-flight mass spectra of these two nitroaromatic molecules are characterized by a prominent NO+ ion signal (mlz 30) together with a characteristic pattern of hydrocarbon fragment ions. The intense NO+ ion signal arises via efficient two-photon resonant ionization of neutral nitrogen monoxide (NO) molecules produced by dissociation of the nitroaromatic species.In the wavelength range studied to date, 224-260 nm, NO+ ion generation is observed to be strongly dependent on laser wavelength, with an intensity maximum occurring at 226.3 nm. At this particular wavelength, NO+ ion signals have been detected with less than 1 amol (<10-'* mol) of nitrobenzene vapour in the laser beam. The two aromatics can be distinguished by observing differences in the laser induced mass spectra and in the wavelength dependence of fragment ion production. Furthermore, it is possible to distinguish NO+ ion formation from NO and NO2 gases and NO+ ion formation from nitroaromatic molecules in vacuum by studying the wavelength dependence of the NO+ ion signal in the range 245-250 nm and it is hoped that this procedure can be used to make similar distinctions with atmospheric samples. Keywords: Resonance enhanced multiphoton ionization spectroscopy; nitroaromatic vapours; mass spectrometry; wavelength dependence One of the most strategically important problems in analytical science today is the sensitive detection of explosive com- pounds.In particular, the detection of explosive compounds concealed in airline luggage is of extreme importance. For this particular problem, the detection method employed must necessarily respond to any characteristic vapours given off by the concealed compounds, and therefore must possess certain characteristics. Firstly, high detection sensitivity is essential as most commonly used explosive compounds such as 2,4,6- trinitrotoluene (2,4,6-TNT), 2,4-dinitrotoluene (2,4-DNT), pentaerythritol tetranitrate (PETN), 173,5-trinitro-1,3,5-tri- azacyclohexane (RDX) and 1,3,5,7-tetranitro-1,3,5,7-tetra- azacyclooctane (HMX) have very low vapour pressures at ambient temperatures. Secondly, a selective technique is required to discriminate the explosive compounds of interest from any innocent vapours present in the sampled volume. A series of experimental investigations has recently been undertaken at the University of Glasgow to develop a laser- based procedure for the detection of nitroaromatic vapours.Nitrobenzene and o-nitrotoluene are regarded as two of the simplest explosive compounds and have been chosen for initial study because of their relatively high vapour pressures at room temperature [0.477 mbar (47.7 Pa) for nitrobenzene1 and 0.315 mbar (31.5 Pa) for o-nitrotohenel at 300 K]. Resonance enhanced multiphoton ionization spectroscopy (REMPI) is a laser-based technique that possesses the necessary attributes of high sensivitity and selectivity, and has been used extensively over the last decade by many research laboratories2-8 to study the spectroscopy of organic molecules.The technique requires the use of powerful pulsed lasers to effect a multiple photon absorption process in a gas phase molecule, which leads to ion formation once the ionization energy is exceeded. The ionization process may exhibit molecular selectivity if the multiphoton absorption proceeds through characteristic quantum states of the molecule of interest.This is manifest in a series of peaks in the ionization spectrum corresponding to resonances in the absorption process. The technique is, therefore, most suited for use with tunable laser sources to exploit the selectivity aspect, although * Presented at SAC '92, an International Conference on Analytical Chemistry, Reading, UK, September 20-26, 1992. a common approach involves the use of fixed-wavelength sources such as excimer lasers9,10 or the fourth and fifth harmonics of an Nd : YAG laser.11.12 With respect to sensi- tivity, it is possible to ionizc with high efficiency under appropriate operating conditions of wavelength and laser flux. 7,13,14 Experimental A Lumonics TE-860M XeCl pulsed excimer laser was used to pump a Lumonics EPD-330 dye laser, which was operated with two laser dyes, Coumarin 47 and Coumarin 102, to span the wavelength range 448-524 nm.The dye laser output was frequency doubled by a BBO (P-barium borate) 'B' cut crystal to cover the range 224-262 nm, and wavelength tracking at maximum fluence was accomplished using an Inrad Model 5- 12 SHG autotracking system. Typical pulse energies of -8 and -1 mJ were readily generated in the fundamental and second harmonic beams, respectively. The pulse repetition rate was 10 Hz with a pulse length of 6 ns. In all experiments both fundamental and second harmonic laser beams were focused into a high vacuum chamber using a 30 cm quartz lens. The spot size is estimated to be 200 vm in diameter at the interaction region, giving maximum ultraviolet (UV) laser fluences of 25 mJ mm-2.No ionization signals were observed when the fundamental laser beam was focused into the chamber on its own, hence there was no need to separate from the second harmonic beam before passing through the chamber. On passing through the chamber, the beams were dispersed in a prism arrangement and the pulse energy of the second harmonic beam was measured using a Molectron 33-09 pyroelectric joulemeter. The laser pulse energies of both beams were varied using a Newport 935-5 attenuator. No differential attenuation in the fundamental and second harmonic beams is incurred when using this devicc. The vacuum chamber was evacuated by a BaIzers TPU 240 turbomolecular pump to a base pressure of typically 1 x mbar (1 x Pa) , measured by an Ionivac TM210 ion gauge.A diagram of the experimental arrangement is shown in Fig. 1. Sample vapour was admitted to the acceleration region of the time-of-flight (TOF) mass spectrometer through a 0.5 mm diameter hole in the centre of the sample stub using a precision602 ANALYST, JUNE 1993, VOL. 118 LeCroy 9410 digital oscilloscope or CAMAC ADC system Lumonics r excimer laser TE-860M pyroelectri'c joulemeter Second harmonic beaa multiplier 30 cm C> quartz Optical lens Pump beam Y attenuator I Lumonics EPD-330 Second harmonic dye laser generation unit Fig. 1 Block diagram of the experimental set-up leak valve and capillary tube arrangement. The laser-sample interaction occurred at a distance of 1 mm from the surface of the sample stub, and ions passed into a 1.20 m field-free drift space where they were detected by a standard Thorn-EM1 18- dynode electron multiplier.The TOF mass spectrometer was operated in the Wiley-McLaren configuration15 at a resolu- tion of 220 at rnlz = 77. For wavelength-dependent measurements, signals from both the electron multiplier and the joulemeter were amplified and recorded simultaneously by a CAMAC-based 11-bit analogue-to-digital converter (ADC) system.16 In fixed- wavelength measurements, TOF mass spectra were accumu- lated by taking the output of the electron multiplier directly to a LeCroy 9410 100 MHz digitizing oscilloscope where signals were averaged typically over 600-1000 laser shots. All samples were used as received. The samples of nitrobenzene and o-nitrotoluene were obtained from Aldrich, high-purity N O gas from UCAR Speciality Gases and high- purity NO2 gas from Gas & Equipment.Results and Discussion In order to obtain some spectral information on the interme- diate states of nitrobenzene and o-nitrotoluene vapours, absorption spectra of both compounds were recorded at a resolution of 0.1 nm on a Beckman UV 5270 spectropho- tometer, and are shown in Fig. 2(a) and (b), respectively. Both spectra exhibit broad structureless absorption bands with intensities peaking at about 230-240 nm. In many instances, it has been observed that the resonant two-photon ionization (R2PI) spectrum of a molecule matches closely to its electronic absorption spectrum.3,6.17,18 In both R2PI and UV absorption spectroscopy, absorption of a single photon couples the electronic ground state with excited states of the molecule.Provided the cross-section for ionization from the excited state does not depend strongly on laser wavelength, the R2PI and absorption spectra are expected to be similar. On this basis, the absorption spectra shown in Fig. 2 suggest that obtaining easily identifiable ionization spectra of these nitroaromatic molecules would be unlikely. However, as will be seen later, the broad absorption bands are linked to predissociative states which give rise to characteristic frag- ment ion spectra. Nitrobenzene vapour (C6H5NO2, rnlz = 123) was admitted to the vacuum chamber to a partial pressure of 2 X mbar (2 X Pa), and a typical TOF mass spectrum, recorded at a laser wavelength of 247.3 nm and at a UV laser fluence of 8.5 mJ mm-2, is shown in Fig.3(a). The spectrum consists of a series of C,H,+ (n = 1,...,6; m = 1,...,6) type fragments owing to extensive fragmentation of the aromatic ring, and an intense fragment at mlz = 30, which is attributed to NO+. 210 230 250 270 290 310 330 Wavele ng t h/n rn Fig. 2 Single photon absorption spectra of ( a ) nitrobenzene vapour and (b) a-nitrotoluene vapour, recorded at 0.1 nm resolution on a Beckman UV 5270 spectrophotometer NO lcsl C U I Ion mass Fig. 3 and (b) a-nitrotoluene at 247.5 nm Time-of-flight mass spectra of ( a ) nitrobenzene at 247.3 nm Although two photons in the available wavelength range have sufficient energy to ionize nitrobenzene, no ion signal corresponding to the molecular ion at rnlz = 123 was observed at any fluence. The laser induced mass spectrum of nitroben- zene has been compared with the 70 eV electron impact (EI) mass spectrum,lg and several differences were observed. Firstly, in the EI mass spectrum, an ion of mlz = 93 and a parent ion were observed, whereas no corresponding peaks were observed in the laser induced spectra. Also, the NO+ ion in the EI spectrum is of small intensity relative to other fragment ions that appear, whereas at discrete wavelengths the NO+ ion was observed to dominate the laser induced spectra.However, it is well known that nitroaromatic species have a tendency to predissociate when irradiated by UV light in vacuum conditions,l2 hence the absence of a molecular ion was not unexpected.The highest rnlz signal observed in the mass spectrum is 77, which corresponds to the C6Hs+ ion, ie., the loss of NO2 from the parent molecule. A similar procedure was carried out for o-nitrotoluene vapour (C6H4CH3NO2, mlz = 137). A heating tape arrange- ment was used to raise the temperature of the sample inlet lineANALYST, JUNE 1993, VOL. 118 603 to 50 "C to obtain the same partial pressure in the chamber as with nitrobenzene. A TOF mass spectrum recorded at 247.5 nm and at a UV laser fluence of 20 mJ mm-2 is shown for comparison purposes in Fig. 3(b). As with nitrobenzene, no molecular ion was observed, but a series of C,H,+ type fragments were detected together with an intense NO+ ion signal. However, in this instance, no fragment ion correspond- ing to the loss of NO2 from the parent molecule was observed.The highest mass fragment ion observed was 52, correspond- ing to the C4H4+ ion. The mass spectra shown in Fig. 3 were recorded at the same amplifier gain. On comparing the laser induced mass spectra of o-nitrotoluene with the 70 eV EI mass spectra,lg several differences are observed. As with the EI mass spectrum of nitrobenzene, a parent ion signal was observed together with a series of ion fragments of m/z 352. Also, the signal corresponding to the NO+ ion in the EI spectrum was relatively weak, whereas the NO+ ion is observed to dominate the laser induced mass spectrum at several positions in the wavelength range studied. In general, the signal-to-noise (S/N) ratios of the various fragment ions were smaller for o-nitrotoluene than for nitrobenzene, and this is the reason why the mass spectrum of o-nitrotoluene shown in Fig.3(a) was recorded at a higher fluence. It can be seen that each spectrum has an intense NO+ ion peak but distinguishable hydrocarbon fragmentation patterns. In order to investigate the possibility of distinguishing between the two compounds on the basis of spectral charac- teristics, the wavelength dependence of C,H,+ type fragment production was recorded for samples of both nitrobenzene (Fig. 4) and o-nitrotoluene in the range 245-250 nm (Fig. 5 ) . f I0 1 I $ 1 I I I I I I 246 247 248 249 250 1 0 245 k CsH5' The laser fluence profiles, which were recorded simul- taneously with the ionization signals, are also shown in Figs.4 and 5. For nitrobenzene, all hydrocarbon fragment ions show similar resonant structure in this wavelength range, with ionization enhancements observed at several points in the 5 nm range studied. However, apart from the C+ spectrum, which shows ionization features attributable to resonant ionization of carbon atoms,20 no characteristic structure was observed in the spectra of the hydrocarbon fragments produced from o-nitrotoluene (Fig. 5 ) . The structure asso- ciated with the hydrocarbon fragment ions from nitrobenzene is thought to arise from some intermediate fragment ion or neutral species, as the absorption spectrum of nitrobenzene vapour shows no such structure, as mentioned earlier. It is, therefore, possible to distinguish between these particular compounds by utilizing information from both mass spectra and wavelength-dependent fragmentation spectra.In the remainder of this paper, the formation of the NO+ ion is considered in detail. The wavelength dependence of the NO+ ion in the range 245-250 nm was recorded for samples of nitrobenzene and o-nitrotoiuene, and are shown in Figs. 6 and 7, respectively, where the laser fluence profiles are also shown. Clearly, the spectra shown are similar in that they both exhibit a series of narrow resonances with two distinct edges at 247.2 and 247.8 nm. In fact, formation of the NO+ ion signal in this wavelength range has been identified as resonant two-photon ionization (one photon to excite and one to ionize) of neutral nitrogen monoxide (NO) molecules. The rcsonances in the spectra of Figs.6 and 7 actually correspond to transitions between rotational levels of the X2111,2,3/2 (v" = 2) state and i i , , l l j -1 0 245 246 247 248 249 2 50 245 246 247 248 249 250 Wavelengthlnm Fig. 4 Wavelength dependence of a variety of nitrobenzene frag- ment ions in the range 245-250 nm I I I I - R l I I I I I 245 246 247 248 249 250 Wavelengthlnm Fig. 5 Wavelength dependence of a variety of o-nitrotoluene fragment ions in the range 245-250 nm604 / / ANALYST, JUNE 1993, VOL. 118 A // //,///// / / //// A v‘ = 3 v’ = 2 v r = 1 B * n v‘ = 0 4 N , L $ 0 I I I I I -1 245 246 247 248 249 2 50 Wavelengthlnrn Fig. 6 produced from a nitrobenzene sample Ionization spectrum of the NO+ ion in the range 245-250 nm N 245 246 247 248 249 250 Wavelengthlnrn Fig.7 produced from an o-nitrotoluene sample Ionization spectrum of the NO+ ion in the range 245-250 nm the A22 (v’ = 0) excited state. Here, v’ and V” represent the quantum numbers of the relevant vibrational levels of the excited and ground electronic states, respectively. The distinc- tive double edge feature observed arises because of transitions originating from both members of the ground electronic state, X2111/2,3/2, which is split owing to spin-orbit interactions. A diagram showing the relevant energies21 of the states involved in this two-photon ionization is shown in Fig. 8. The very distinct structure and the large intensity of the NO+ ion signal in the range 245-250 nm prompted a search for even stronger absorption bands of the NO molecule.For this reason, an absorption spectrum (Fig. 9) of high-purity NO gas was recorded at 0.5 nm resolution using a Perkin-Elmer Lambda 9 spectrophotometer in order to identify the positions of the strongest absorption wavelengths. The conditions under which this spectrum was recorded were such that pressure broadening of the absorption lines allowed only the gross features of the absorption spectrum to be observed. No absorption intensity corresponding to the A22(v’ = 0) t X2n(v” = 2) transitions at 247.2 and 247.8 nm was observed, Ionization limit v“ = 2; t‘, = 3724 crn-l v” = 1; E, = 1875 crn l v” = 0; E, = 0 crn-l Fig. 8 =PI scheme for generation of NO+ ions. The diagram also gives the energies21 of the relevant vibrational levels of the ground electronic state and the wavelengths of the transitions observed I I 1 200 210 220 230 240 250 Wavelengthlnm Fig.9 Single photon absorption spectrum of NO gas recorded at 0.5 nm resolution on a Perkin-Elmer Lambda 9 spectrophotometer. Peaks 1 and 2 correspond to the A?Z(v’=O) t XW(v” = 1) and A2X(v’ = 0) t X2II(v” = 0) transitions, respectively. Peaks 3 and 4 correspond to the A2C(v’ = 1) t X2R(v” = 0) and A*C(v‘ = 2) t X2II(v” = 0) transitions, respectively but several absorption lines were observed in the range 220-240 nm. Peak 1 at 236.3 nm corresponds to the A?Z(v’ = 0) t X2II(v” = 1) transition, and is noticeably weaker than the absorption at 226.3 nm (peak 2), which corresponds to the A2C(v’ = 0) t X2lX(v” = 0) transition. In fact, the transition originating from the v” = 0 level of the ground state is more intense by a factor of approximately 1000 than the absorption originating from the v” = 1 level.This is in good agreement with a Boltzmann distribution of population amongst the vibrational energy states of the electronic ground state, assuming that the absorption cross-sections for the two transitions are approximately equal. Peaks 3 and 4 are of similar intensity to the 226.3 nm band, and have also been identified as absorptions into the A22 state. The absorption lines situated at 204 and 214 nm correspond to the A2C(v’ = 2) t X2II(v” = 0) and A22(v’ = 1) t X2Il(v” = 0) transitions, respectively. Once information on the positions of strong absorption bands in NO had been accumulated, the wavelength depen- dence of the NO+ ion arising from nitrobenzene was recordedANALYST, JUNE 1993, VOL.118 605 0-0 0 - 1 ~ 0 - 2 I 234 236 238 245 247 249 J 224 228 232 236 240 244 248 Wavelengt hln m Fig. 10 with the A2Z(vr = 0) t X2II(v” = 1) and A2Z(vr = 0) t X211(vrr = 2) bands in greater detail Ionization spectrum of the NO+ ion, produced from nitrobenzcnc, in the range 224-250 nm. The insets show the structure associated p.’ . 12 2 *.- 8 a;% .. ... . 0 200 400 600 800 1000 Pulse energyIkJ Fig. 11 with laser fluence at 226.3 nm Variation in NO+ ionization intensity from nitrobenzene in the range 224-250 nm to investigate the relative intensity of the ionization signal at 226.3 nm compared with those at 247.2 and 236.6 nm. The resulting spectrum is shown in Fig. 10. As can be seen from the graph, the ionization intensity of the 226.3 nm resonance is larger by a factor of approximately 30 than the 236.3 nm band, suggesting a non-thermal population distribution. The spectrum covers the A?Z(v’ = 0) t X2rI(v” = 0), A2X(v’ = 0) t X2II(v” = 1) and A2C(v’ = 0) c X2rI(v” = 2) transitions and is a composite of three 9 nm scans, a procedure adopted owing to the difficulty of obtaining a constant fluence over a large wavelength range.The ionization intensity has been normalized linearly to laser fluence, as the NO+ ion signal was observed to vary linearly with laser fluence at the three band head positions (Fig. 11). This spectrum clearly shows that, of the wavelengths so far investigated, an irradiation wavelength of 226.3 nm should be chosen for maximum detection efficiency. If this approach is to be used to detect nitroaromatic vapours selectively and sensitively in the atmosphere, it is essential that we are able to distinguish between NO+ ion signals arising from the nitroaromatic molecules and from atmospheric NO and NO2 gases.Fortunately, a distinction can be made by comparing the ionization spectra of the nitroaro- matics with high-purity NO nd NO2 gases in the wavelength range 245-250 nm, as shown in Fig. 12. When NO+ is produced from nitrobenzene and o-nitrotoluene [Fig. 12(a) and (b), respectively], the familiar double edge feature is dominant in both instances. In the spectrum of NO gas [Fig. 12(c)], no characteristic structure is observed whatsoever, illustrating a lack of population in the v” = 2 level of the X2II state. This is not unexpected as the X2II (v” = 2) state lies at 3724 cm-l (see Fig. 8) above the lowest lying vibrational level, and therefore is not expected to be populated significantly at room temperature.In the NO+ spectrum from NO2 gas [Fig. 12(d)], an ionization enhancement is observed at 249.2 nm. The NO2 molecule has been the subject of considerable spectroscopic study by a number of workers using a variety of different multiphoton procedures.21-27 From the literature,21 it is well known that NO2 readily predissociates by single photon absorption at wavelengths less than 397.9 nm, forming N O and atomic oxygen. Depending on irradiation wavelength, the dissociation products are produced in a range of energy states24 as shown below NO2 + hv + NO(v = 0,1,2,3,..) + O(3P), if 243.9 < h <397.9 nm (1) or NO2 + hv + NO(v = 0,1,2,3,..) + O(3P or ID), if h <243.9 nm (2) The 249.2 nm resonance has been identified as a transition to the B2II(v’ = 2) state from the v” = 4 level of the X2rI state in NO.The ionization spectrum of nitrobenzene shown in Fig. 10 suggests that the population of the vibrational levels of the X2n state decrease as the vibrational energy increases. However, in the NO+ ion spectrum of NO2 gas, the 249.2 nm resonance is stronger than the A2C(v’ = 0) t X2n(v” = 2) transition, which is thought to be due to an increase in the v” = 4 population of the ground state. At 249.2 nm, and provided that the NO2 molecules irradiated are in their ground state, scheme (1) will be followed.This has been reported to result in an inverted population distribution among the ground state vibrational states of the NO molecule ,21923 and could account for the enhancement observed at 249.2 nm in the NO2 spectrum. However, scheme (2) may be accessible at 249.2 nm if the NO2 molecules irradiated have some initial internal energy. This may be the situation when the NO2 molecules are produced in a photodissociation process, as for the nitroaro- matic samples. In this instance, no population inversion is expected in the vibrational state distribution of the NO molecules formed as most of the excess dissociation energy is channelled into producing ID oxygen atoms, and the relative intensitities would be as observed in Fig. 12(a) and (b). As discussed earlier, the NO+ signal dominates the nitrobenzene TOF spectrum when the laser is tuned to theANALYST, JUNE 1993, VOL.118 245 246 247 248 249 250 Wavel e ngt hln m Fig. 12 Ionization spectra of the NO+ ion in the range 245-250 nm from ( a ) nitrobenzene; (b) o-nitrotoluene; (c) NO gas; and ( d ) NOz gas A2C+(v” = 0) t X211(v’ = 0) transition at 226.3 nm in NO. At this particular wavelength, the detection sensitivity of ni- trobenzene vapour in the TOF system was measured in the following way. The vacuum chamber was baked thoroughly for several days at 200 “C and pumped to a base pressure of 2 x lo-* mbar (2 X Pa). No ionization signals were observed at 226.3 nm under these conditions. Nitrobenzene vapour was then admitted to the chamber, the sample inlet line was shut off and the chamber pressure was allowed to fall until the NO+ S/N ratio was 2 : 1.This occurred at a pressure of 6 x mbar (6 x Pa). Signal averaging over 750 laser shots was necessary to achieve this S/N ratio. The ionizing beam was focused to a spot size of 200 pm at the interaction region, which gives an interaction volume of 6.3 x cm3. At the operating pressure, it was estimated that approximately 68 000 molecules were present in the interaction volume. These measurements demonstrate detection of nitrobenzene vapour present at sub-attomole (<lO-lX mol) levels. As the NO+ ion signal varied linearly with UV laser fluence at 226.3 nm and as the ionization signal was not optically saturated, the possibility exists for a further increase in detection efficiency by using higher laser Auences.In this paper, a highly sensitive laser based approach for the detection of two simple nitroaromatic molecules with only one NO2 group per molecule has been described, and sub-atto- mole sensitivity for nitrobenzene via the generation of intense NO+ ion signals has been demonstrated. Table 1 shows four of the most commonly used explosive molecules, namely 2,4,6- TNT, 2,4-DNT, RDX and HMX, and it can be seen that all of these compounds have more than one NO2 group. It is hoped that the presence of more than one NO2 group per molecule will lead to a further increase in the sensitivity, which is of great importance with respect to the sensitive detection of the more important di- and trinitroaromatic compounds. How- ever, if the approach described here for the selective and sensitive detection of explosive compounds is to be used as a practical detection method, it is essential that we are able to discriminate explosive compounds from other materials com- Table 1 Structures of some common explosive materials Species Formula Structure NO, RDX CH, ,CH, ‘N I N 0 2 No, monly found in luggage, such as perfumes or after-shaves.A series of experiments are planned in order to investigate the signatures of some common organic compounds likely to be present in aircraft luggage. This work was carried out with the support of the Procurement Executive, DRA. J. S. is indebted to the SERC for a postgraduate studentship. The authors thank Dr. R. D. Peacock, Department of Chemistry, University of Glasgow, for granting access to absorption spectrometers.1 2 3 4 5 6 7 8 9 10 11 References CRC Handbook of Chemistry and Physics, ed. Weast, R. C., CRC Press, Cleveland, OH, 53rd edn., 1972, D151. B o d , U., Neusser, H. J., and Schlag, E. W., J. Chem. Phys., 1980, 72, 4327. Fisanick, G. J . , Eichelberger, T. S . , IV, Heath, B. A., and Robin, M. B., J. Chem. Phys., 1980, 72, 5571. Hager, J. W., and Wallace, S . C., Anal. Chem., 1988, 60,5. Lubman, D. M., Naaman, R., and Zare, R. N., J. Chem. Phys., 1980, 72,3034. Marshall, A., Clark, A., Jennings, R., Ledingham, K. W. D., and Singhal, R. P., Meas. Sci. Technol., 1991,2, 1078. Rettner, C. T., and Brophy, J . H., Chem. Phys., 1981, 56, 53. Zandee, L., and Bernstein, R. B . , 1. Chem. Phys., 1979, 71, 1359. Hodges, R. V.. Lee, L. C., and Moseley, J. T., Int. J. Mass Spectrom. Ion Phys., 1981, 39, 133. Apel, E. C., and Nogar, N. S., Int. J. Mass Spectrom. Ion Processes., 1986, 70, 243. Kolaitis, L., and Lubman, D. M., Anal. Chem., 1986,58,1993.ANALYST, JUNE 1993, VOL. 118 607 12 13 14 15 16 17 18 19 20 Zhu, J., Lustig, D., Sofer, I., and Lubman, D. M., Anal. Chem., 1990, 62, 2225. Boesl, U., Neusser, H. J., and Schlag, E. W., Chem. Phys., 1981, 55, 193. Frucholz, R., Wessel, J., and Wheatley, E., Anal. Chem., 1980, 52,281. Wiley, W. C., and McLaren, I. H., Rev. Sci. Instrum., 1955,26, 11 50. Raine, C.. Ledingham, K. W. D., and Smith, K. M., Nucl. Instrum. Methods, 1983, 217, 305. Towrie, M., Cahill, J. W., Lcdingham, K. W. D., Rainc, C., Smith, K. M., Smyth, M. H. C., Stewart, D. T., and Houston, C. M., J. Phys. B , 1986, 19, 1989. Zandee, L., and Bernstein, R. B., J. Chem. Phys., 1979, 70, 2574. A n Eight Peak Index of Muss Spectra of Compounds of Forensic Interest, Scottish Academic Press, Edinburgh, 1983, p. 50. Clark, A . , Ledingham, K. W. D., Marshall, A., and Singhal, R. P., Spectrochim. Acta, Part R , 1992, 47, 799. 21 22 23 24 25 26 27 Slanger, T. G., Bischel. W. K . , and Dyer, M. J . , J. Chem. Phys., 1983, 79, 2231. Busch, G. E., and Wilson, K. R., J. Chem. Phys., 1972, 56, 3626. McFarlane, J., Polanyi, J. C., and Shapter, J. G., J. Photochem. Photobiol. A , 1991, 58, 139. Morrison, R. J . S., and Grant, E. R., J. Chem. Phys., 1982.77. 5994. Morrison, R. J . S . , Rockney, B. H., and Grant, E. R., J. Chem. Phys., 1981, 75, 2643. Uselman, W. M . , and Lee, E. K. C., J. Chem. Phys., 1976,65. 1948. Zacharias, H., Geilhaupt, M., Mcicr, K., and Welge, K. H., J. Chem. Phys., 1981, 74, 218. Paper 2106599A Received December 14, 1992 Accepted February 22, 1993

ISSN:0003-2654    

DOI:10.1039/AN9931800601

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ANALYST JUNE 1993 VOL. 118 609 Analysis of Steroids* Part 46.t Qualitative and Quantitative Characterization of Bulk Cholesterol by Gas Chromatography and Gas Chromatography-Mass Spectrometry Anna Lauko Eva Csizer and Sandor Gorog Chemical Works of Gedeon Richter Ltd. P.O.B. 27 H-1475 Budapest 10 Hungary Methods are described for the determination of the cholesterol (cholest-5-en-3P-ol) content of crude and bulk cholesterol samples and the determination of their impurity profile. For both purposes capillary gas chromatography was used with phenylmethylsilicone gum as the stationary phase. The cholesterol contents of bulk samples obtained from ten different manufacturers ranged between 90.8 and 98.5%. The relative standard deviation of the method was 0.62%. In the course of the impurity profiling by gas chromatography-mass spectrometry two major impurities were found in all samples desmosterol (cholesta-5,24-dien-3P-ol) and lathosterol (5a-cholest-7-en-3(3-ol).Some minor steroidal impurities were also detected. Keywords Cholesterol; capillary gas chromatography; gas chromatograph y-mass spectrometry; impurity pro filing; steroids Cholesterol (cholest-5-en-3P-ol)1-3 is an important constituent of the organism of mammals. The starting materials for its industrial-scale isolation are the spinal cord and brain of cattle and wool grease. Large amounts of cholesterol are used by the pharmaceutical and cosmetic industries as an excipient in the production of ointments and creams. At the beginnings of the semi-synthetic production of steroid hormones cholesterol was the most important starting material.Although plant sterols and sapogenins have largely superseded cholesterol in this respect it is sometimes still used. It is also a starting material for the syntheses of vitamin D3 and liquid crystals. As a consequence of its great importance in biology and pathology the determination of cholesterol in biological samples mainly blood serum is very important and there have been many reports of methods for its spectrophoto-metric chromatographic enzymic electrochemical etc., determinations many of them being adopted in routine clinical analysis.4.5 However the analytical investigation of bulk cholesterol for its active ingredient content and impurity profile making use of the possibilities of modern instrumenta-tion does not seem to be documented.The aim of this study was to develop methods for these purposes and with the aid of these to compare commercially available bulk cholesterol samples. Experimental Reagents and Materials The cholesterol samples investigated were partly products and intermediates in the production of cholesterol from spinal cord at the Chemical Works of Gedeon Richter Budapest, Hungary and partly samples obtained from various manufac-turers. The solvents used were of analytical-reagent grade from Reanal (Budapest Hungary). Digitonin reagent was purchased from Merck (Darmstadt Germany). * Presented at SAC '92 an International Conference on Analytical Chemistry Reading UK September 20-26 1992. + For Part 45 of this series see Gazdag M.Szepcsi G. Mihalyfi, K. Kemcnes-Bakos P. Horvath P. Vegh Z . . RCnyei M., Trischler F. and Gorog S . A d a Pharm. Hung. 1992 62 88. This paper is Part 10 in a series of papers 'Estimation of Impurity Profiles of Drugs and Related Materials'; for Part 9 see Gorog S . HerCnyi B., and Renyei M. J. Pharm. Biomed. A n d . 1992 10 831. Instruments Gas chromatography (GC) was performed with a Model 5890A Series I1 instrument (Hewlett-Packard Waldbronn, Germany) equipped with a flame-ionization detector. The column used was 25 m X 0.32 mm i.d. fused silica containing an immobilized film of 5% phenylmethylsilicone gum of thickness 0.52 pm (Ultra-2 Hewlett-Packard). For GC-mass spectrometry (GC-MS) and for impurity profiling the instru-ment used was a VG-TRIO-2 quadrupole mass spectrometer (VG Altrincharn UK) attached to a Hewlett-Packard Model 5890A gas chromatograph.Procedures Assay method Determination of the active ingredient content of bulk cholesterol samples was carried out by injecting into the chromatograph 1 yl volumes of the test solution containing 1 g 1-1 of each of the test samples and the internal standard (testosterone propionate) in chloroform. The carrier gas was nitrogen at an average linear velocity of 20 cm s-1. The split-injection technique was used with a split ratio of 1 50. The oven temperature was 280°C and the injector and detector temperatures were 300 "C. Digitonide precipitation method The gravimetric determination of the total 3P-hydroxysteroids was performed as described by Gorog and Szasz.6 Impurity profiling by GC-MS A 1 yl volume of a 50 g 1-1 chloroform solution was injected.The carrier gas was helium at an average linear velocity of 15 cm s-l. The split ratio was 1 10. The mass spectrometer was used in the positive-ion electron impact mode with the following temperatures oven 255 "C; injector 300 "C; trans-fer line 230 "C; and ion source 200 "C. The ionization energy used was 70 eV. Determination of impurities by GC The conditions were as under Assay method; the concentra-tion was 10 g I-' in chloroform 610 ANALYST JUNE 1993 VOL. 118 Results and Discussion Determination of the Active Ingredient Content of Bulk Cholesterol None of the existing pharmacopoeias or handbooks on cosmetics describe the determination of the active ingredient content of bulk cholesterol samples.Its characterization is restricted to the determination of the melting-point and optical rotation. As can be seen in Table 1 these data provide only a poor characterization of the quality of the product. Because of the lack of suitable functional groups in the cholesterol molecule and the presence of structurally closely related impurities classical methods (titrimetry spectropho-tometry polarography etc.) cannot be used and only high-resolution chromatography can be considered. Because of the poor detectability of cholesterol using an ultraviolet (UV) detector in high-performance liquid chromatography (HPLC) and the considerable success achieved by capillary GC with underivatized sterols,4 this technique was selected for the t -m C 0 to .-S L I I I 5 10 15 tlmin Fig.1 Gas chromatographic assay of a bulk cholesterol sample (manufacturer H in Table I). IS internal standard (testosterone propionate); and peak 1 cholesterol determination. (It should be mentioned that in catalogues for fine chemicals the manufacturers usually indicate GC as the assay method for cholesterol but no details are given.) The gas chromatogram of a bulk cholesterol sample obtained using the assay method described under Experimen-tal is shown in Fig. 1. The moderately polar phenylmethylsili-cone gum stationary phase permits the separation of the impurities from cholesterol with a relatively short analysis time. Testosterone propionate proved to be a suitable internal standard.The results in Table 1 indicate that the rec-ommended method is suitable for the determination of the active ingredient content of bulk cholesterol samples and the intermediates in their isolation and purification. It is interesting to compare the results obtained by the proposed method with those obtained by the classical gravimetric method based on the precipitation of cholesterol digitonide.6 The latter method is by no means specific for cholesterol; it measures the sum of 3P-hydroxysterols. As most of the steroidal impurities of cholesterol belong to this t -m C 0 z 1 i 5 10 15 20 tlmin Fig. 2 Gas chromatogram for the determination of the impurity profile of cholesterol (manufacturer H in Table 1). Peak 1 choles-terol; peak 2 desmosterol; and peak 3 lathosterol Table 1 Analytical data for various cholesterol samples Melting-Cholesterol sample pointPC* [.I@* Munufucturer-A 148.0 B 147.5 C I45 .O D 145.5 E 146.5 F 147.5 G 145.5 H 146.0 I 149.5 G.Richter 149.0 Crude sample 135.0 * US Pharmacopeia XXII range 147-150°C. * IS = internal standard. In dioxane (US Pharmacopeia XXII). -32.0 -35.2 -34.5 -34.5 -35.1 -35.2 -34.9 -34.4 -33.5 -34.3 -29.2 Assay (%) Impurities (YO) GC (area Gravimetry GC (IS)$ normalization) Desmosterol Lathosterol 98.3 98.8 96.2 97.9 ---97.1 97.7 86.6 -98.5 97.2 90.8 94.1 96.1 93.6 93.1 92.9 94.8 97.0 82.3 98.5 96.4 93.6 98.5 96.5 93.5 93.3 93.8 98.3 98.4 92.0 0.3 2.3 3.4 0.3 2.2 4.0 4.0 3.4 0.4 0.2 0.6 0.7 0.9 1.9 0.7 0.8 1.3 1.4 1.7 1 .o 1.1 1 .ANALYST JUNE 1993 VOL. 118 61 1 group the results obtained for bulk cholesterol are higher than those obtained by capillary GC. This time-consuming but surprisingly precise method however afforded acceptable results for crude cholesterol samples (intermediates in indus-trial-scale purification). The reason for this is that most of the impurities are non-steroidal lipids that do not react with digitonide. It is also interesting to compare the GC results obtained by the internal standard method and area normalization. The agreement between the results is acceptable for the commer-cial samples. For crude cholesterol samples however the results obtained by the area normalization method are higher, because some of the non-steroidal lipids are not eluted under the given conditions and their detector responses are differ-ent.The regression equation for the internal standard method is: cholesterol peak area IS peak area = a -k bccholesterol where TS = internal standard and c is in g 1-1 (range 0.2-1.3), giving a = -0.019 b = 1.961; Y = 0.997; relative standard deviation = 0.62% ( n = 9). Determination of the Impurity Profile For cholesterol GC-MS has proved to be an idcal tool for separation and identification of impurities. The moderately polar stationary phase separates the impurities thus permit-ting their identification by GC-MS with the aid of spectrum collection.7 A typical chromatogram is shown in Fig.2 and the results are presented in Table 1. As can be seen two major impurities occur in all samples desmosterol (cholesta-5,24-dien-3f3-01) and lathosterol (5a-cholest-7-en-3~-01). These are well known accompanying sterols of cholesterol in the spinal cord. In addition to these major impurities some minor impurities were also identified in some of the batches e.g. Sa-cholestan-3p-01 cholesta-3,5-dien-7-one and cholest-4-en-3-one (rela-tive retention times with respect to cholesterol = 1.023 1.177 and 1.214 respectively). In the crude samples volatile lipids were also found such as ethyl palmitate and ethyl stearate (relative retention times = 0.154 and 0.190 respectively). References Ficser L. and Fieser M Steroids. Reinhold New York 1959. Cook R. P. Cholesterol Academic Press New York 1958. Kritchevsky D. Cholesterol Wiley New York 1958. Gorog S. Quuntitutive Anulysis of Steroids Elsevier Amster-dam 1983 pp. 247-289. Copeland. B. E. Ann. Cfin. Lab. Sci. 1990 20 1. Gorog S. and SZ~SZ Gy. ~ Analysis of Steroid Hormone Drugs, Elsevier Amsterdam 1978 pp. 313-314. EPAINIH Muss Spectral Data Base National Bureau of Standards Washington DC. 1978. Paper 2105159A Received September 25 1992 Accepted October 20 199

ISSN:0003-2654    

DOI:10.1039/AN9931800609

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ANALYST, JUNE 1993, VOL. 118 613 Photobleaching of Methylene Blue in Continuous Wave Thermal Lens Spectrometry" Roger D. Lowe and Richard D. Snook+ Department of Instrumentation and Analytical Science, UMIST, P.O. Box 88, Manchester, UK M60 ?OD In this study continuous wave thermal lens spectrometric measurements are made using solutions of Methylene Blue in short pathlength sample cells (0.1 cm). Under the conditions described, rapid photobleaching of Methylene Blue is observed, which degrades the thermal lens signal with potentially serious consequences in terms of detection limits. The phenomenon has been studied and an empirical relationship developed to describe the behaviour of the thermal lens signal when photobleaching occurs. Keywords: Continuous wave thermal lens spectrometry; photobleaching; Methylene Blue; flow injection Thermal lens spectrometry is not considered by many to be a routine analytical technique, even though it has been shown to yield limits of detection in the ppb range, and to have a volumetric resolution of approximately 50 nl.1-3 The main reason for this is the high cost of the necessary equipment.Similar detection limits to those published previously4 for aluminium using thermal lens spectrometry have been achieved using an inexpensive fluorimetric method.5 There- fore, this technique is now being applied to more novel applications, such as the construction of circular dichroism thermal lens spectrometers6 or for monitoring photolytic events7 rather than purely for the determination of the concentration of a trace analyte species.If the absorption maximum of an analyte species coincides with the wavelength of some incident light radiation, absorp- tion will occur. In order to induce a photothermal effect, a fraction of the absorbed energy must follow the collisional or vibrational de-excitation pathway to produce heat. The generation of heat will consequently result in a change of refractive index. Lasers used in this technique are operated in the transverse electromagnetic mode (TEMoo), which gives the beam a Gaussian intensity distribution and results in the sample being more strongly heated along the axis of propaga- tion than in the wings of the beam. This difference in intensity across the beam profile produces a gradient of refractive index change between the beam axis and the bulk solution.In most common solvents at ambient temperatures the change of refractive index with increasing temperature is negative, so that the optical pathway is shorter at the beam centre, making the solution locally equivalent to a diverging lens. When a weak probe beam passes through the pump-irradiated region, the lens element causes it to diverge. This defocusing effect is generally detected as a change in intensity {[Z(eqm) - Z(O)]/Z(O)} at the beam centre in the far field. The intensity change is greatest when the centre of the sample cell is located 31/2ZC [see eqn. ( 5 ) ] beyond a waist in the beam. Under continuous illumination, the thermal lens reaches a steady state (S), which can be given by:8 S = 1 - - tan-1 where in this equation, Wlm) - I @ ) = [ ; ( 2mV )I*- (1) 1+2m+v2 * Presented at SAC '92, an International Conference on Analytical Chemistry, Reading, UK, September 20-26, 1992.To whom correspondence should be addressed. where Z(eqm) is the equilibrium intensity at the beam centre and Z(0) is the intensity at time, t = 0 and rn is the degree of mode-mismatching of the probe beam and the excitation beam, and is given by: m = (:I2 (3) V is the ratio of the distance from the cell to the probe beam waist over the confocal distance 2,. Where, ( 4 ) wp and LO, are the probe beam spot size, and excitation beam spot size in the sample cell, respectively, hp is the wavelength of probe laser radiation, P, is the excitation laser power (W), A is the absorption coefficient of the sample (cm-I), I is the pathlength (cm), dn/dT is the refractive index change with temperature (K-I), k is the thermal conductivity of the solvent (W cm-1 K-1) and w&, is the probe beam waist radius.The time-resolved thermal lens signal can be expressed as:8 I(t) Z(0) I = - - - = 0 2mV [ 1 -;tan-l( where Z(t) is the intensity of the beam centre at time, t > 0 and tc is a characteristic time constant (s), which is given by [(l + 2 4 2 + v2]/(2t/tc) + 1 + 2m + v2 and p is the density of the solvent (g ml-l) and Cp is the specific heat capacity (J g-* K-1). Methylene Blue Methylene Blue was discovered by Car09 in 1877, and is the most common representative of the thiazine group of dyes. The visible absorption spectrum of Methylene Blue in water at room temperature is shown in Fig.1. The spectrum is the superposition of two bands, one with a maximum at 665 nm, and the second with a maximum at 620 nm. As part of the early characterization of Methylene Blue,lo it was noted that it does not obey Beer's law, even at concentrations as low as 1 x 10-7 mol 1-1. Deviations from Beer's law are usually indicative of association, and two forms of association are possible for Methylene Blue (MB), polymerization, i. e . , self association of614 0.09 0.08 g 0.07 6 0.06 0.1 I 1 - - - - ANALYST, JUNE 1993, VOL. 118 1.9 1.8 2 - Baseline t ? / \ I L Methylene Blue 400 450 500 550 600 650 700 Wavelengthhm Fig. 1 Absorbance spectrum of Methylene Blue in water 1 output Krypton ion pump I I Fig. 2 Schematic plan of the thermal lens spectrometer.A, Bandpass filter and B, pinhole the dye cations, or the association of the dye cations with anions. In solvents of high relative permittivity, such as water, where the coulombic repulsion is not sufficient to overcome the addition forces, the deviation from Beer's law was explained quantitatively11 by the formation of dimeric ions (MB)2+2. The two bands observed in Fig. 1 were, therefore, attributed to the monomeric (A,,, = 665 nm), and to the dimeric (Amax = 620 nm) forms of the dye. Early work on the development of photographic bleach-out processes showed that thiazine dyes are photoreduced readily in the presence of mild reducing agents such as allylthiourea, and iron(I1) sulfate.12 Since then, many studies13.14 have been made in the absence, and in the presence of a number of different reductants, using a wide range of solvents and experimental parameters such as temperature, pH, light intensity and dissolved oxygen content.This photobleaching reaction was initially observed accidentally whilst performing thermal lens spectrometric experiments using short pathlength cells (0.1 cm), similar to those used in flow injection. Further research was then carried out to examine this phenomenon and this paper describes the results of a thermal lens spectrometric investigation of the photochemical reduction of plain aqueous solutions of Methylene Blue. Experimental Reagents Methylene Blue. A dark green powder with a metallic lustre. Chemical Abstracts Service (CAS) No. 61-73-4. Relative molecular mass (M,) = 373.90.Source, BDH (Poole, Dorset, UK). Control of Substances Hazardous to Health (COSHH) assessment: harmful by ingestion; stains and may irritate eyes and skin. Ethylenediaminetetraacetic acid disodium salt dihydrate (EDTA). A white crystalline powder. CAS No. 6381-92-6. M , = 372.24. Source, Aldrich (Gillingham, Dorset, UK). COSHH assessment: harmful if ingested in quantity; may be irritating to eyes. All solutions were prepared using doubly distilled, de-ionized water. 1.3 H-l I 1.1 r 1 1 I ' 1 1 ' 1 1 1 ' 1 1 0 2 4 6 8 10 12 14 16 18 20 22 Time after shutter opens/s Fig. 3 Preliminary experimental data. Methylene Blue concentra- tion = 5 x 10-6 moll-1 Procedure A Methylene Blue stock standard solution (1 x 10-4 moll-1) was prepared daily, and working solutions in the range 1 x 10-5-1 x 10-7 mol 1-1 were made immediately before each experiment.The EDTA solutions were freshly prepared each week. Apparatus The thermal lens spectrometer is similar to the design published previously4 and is shown in Fig. 2. The pump radiation was provided by a krypton ion laser (Innova 90, Coherent, Cambridge, UK), operating at 647.1 nm and at powers of between 10 and 100 mW. The probe beam was supplied by a 10 mW He-Ne laser (Melles Griot, Aldershot, Hampshire, UK) operating at 632.8 nm. The pump beam was focused by an achromatic doublet converging lens ( F , = 30.9 cm). A 1.5 mm pathlength precision flow-through cell (Hellma, Westcliff-on-Sea, UK), or a 10 or 1 mm pathlength cuvette was placed at the focal point of the pump beam. Exposure of the cell to the pump radiation was controlled by means of a camera shutter.The He-Ne probe laser was focused by a lens (F2 = 16.1 cm); the resulting beam waist was positioned at a distance 31/22, from the cell. The two beams crossed in the cell at an angle of 1.49". Three adjustable mirrors (Ealing, Watford, Hertfordshire, UK) were used in order to achieve a long optical path (about 5 m) for the probe beam from the cell to a pinhole mounted in front of a photomultiplier tube (PMT). Although the pinhole was relatively large (2 mm), the aperture was significantly smaller than the far-field diameter of the probe beam, which was about 5 cm. A neutral density filter, [absorbance = 1.0 (Ealing)], was placed close to the probe laser output for intensity attenuation, and a bandpass filter at the He-Ne wavelength (Ealing) was placed over the pinhole to prevent stray light from entering the PMT.The output of the PMT was then coupled directly to either a digital storage oscilloscope (LeCroy, Abingdon, UK) or a chart recorder (Siemens, Manchester, UK). Results and Discussion On exposing the sample to continuous wave (CW) laser radiation, rather than obtaining an almost step change in intensity, the signal resembled that shown in Fig. 3. After opening the shutter at t = 0, the expected decrease in intensity is initially observed with the formation of the thermal lens, but then after approximately 250 ms the intensity begins to increase. On closing the shutter the signal does not immediately return to the baseline, but overshoots, and eventually returns after a period of approximately 60 s.This phenomenon can be explained in terms of the disappearanceANALYST, JUNE 1993, VOL. 118 615 I Methylene Blue 1 Semiquinone 2 H leuco- M et h y I e n e B I u e 3 Fig. 4 Blue Simplified reaction scheme for photobleaching of Methylene of the absorbing species from the interaction volume of the excitation laser with the sample, which is likely to be due to photochemical reaction. Removal of the absorption chromo- phore in this way removes the precursive excitation mechan- ism for the generation of the thermal gradient in the sample solution. Thus the thermal lens collapses. A scheme for the photoreduction of an aqueous solution of Methylene Blue is discussed below. Methylene Blue undergoes a two-step reaction that ultimately results in the elimination of the chromophoric properties of the molecule.The nature of the reaction has been extensively studied14,15 and the reaction scheme shown is a simplified version of the proceedings (see Fig. 4). Kato et al.14 proposed that the semiquinone species was created solely by the reaction of the excited monomer triplet with the ground-state monomer, and assumed that the reaction was diffusion controlled. Danziger et al. ,*5 however, showed that this was not the case. They denoted species 2 shown above, as a loosely combined pair consisting of a semiquinone, and a half oxidized state of the dye, i . e . , (MB+----MB-) or probably more correctly (MBZ+----MB*), if MB+ represents the dye- cation. The ground-state dimer is considered to be held together by multipole-multipole interactions of the Van der Waals type.11.16 In relation to the dimer, species 2 can, therefore, be formulated as a charge-transfer state that results from an internal electron transfer. From this point of view the binding between dye molecules might well be stronger in the intermediate species than in the ground-state dimer .leuco- Methylene Blue shows no significant absorbance in the visible region of the electromagnetic spectrum. The chromophoric properties of the molecule result from the high degree of conjugation that is present in Methylene Blue but absent in leuco-Methylene Blue. The conversion to leuco-Methylene Blue is the likely cause of the deviation from predicted thermal lens behaviour, and is potentially a serious drawback for the application of thermal lens detection to flow injection. Small volumes or thin films of photosensitive reagents could be easily photodegraded at even modest laser power levels.In order to quantify the departure from predicted behaviour , time-resolved experiments have been carried out to elucidate the proposed mechanism further. Fig. 5 compares a plot of the normalized CW signal with time for the experimental data with that of the theoretically Isample absorbance = 0.017 I I I I I I I 0 100 200 360 400 500 600 700 800 Z 0.75 Time/ms Fig. 5 Comparison of the experimental and theoretical data for an aqueous solution of Methylene Blue (5 x mol 1-1). A, Theoretical and B, experimental 2 0.998 !! 0.997 0.996 0.994 .: 0.993 0.992 w 0.991 ; 0.990 0.989 Z 0.988 0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 Time/ms Comparison of the experimental and theoretical data for the Fig.6 first 3 ms * 1 ( k , ) 111 ( k 3 ) (MB):' --t (MB):' (MB2 + ---- MB-) *--- - MB2+ + MB- 11 (k2) IV Fig. 7 aqueous solution of Methylene Blue Predicted reaction scheme for the photobleaching of a plain predicted curve [using eqn. ( 5 ) ] , and as can be seen a considerable difference exists between the two. However, for the first 3 ms the theoretical curve and the experimental data correlate, see Fig. 6 ( r = 0.9999). From this initial period a value for 8 was calculated using eqn. (4), 8 = 0.20 k 0.01. This corresponds to an initial absorbance of 1.0783 X 10-3, which is in good agreement with the measured value of 9.79 x 10-4 (error = 10%).For the next =250 ms the formation of the thermal lens is the dominant feature in the signal. After this period the experimental signal can be seen to be decreasing, and at this point the photochemical reaction dominates the formation of the signal. It is from this section of the experimental data that some useful information may be gained about the kinetics of the reaction. From the work of Danziger et al.,15 a probable reaction scheme can be deduced, and is shown in Fig. 7. Assuming that contributions from process IV are negligible under these reaction conditions, then an expression can be derived to describe the change in absorbance with time.17 The equation is given by: A Absorbance = ( A X 1U-(k+~)') + ( B x 10-ck-s)t) (7) where t is time in ms, k = - 1 ( k , + k2 + k3) 2616 ANALYST, JUNE 1993, VOL.118 V) -0.097 I Y .- C 3 2- -0.102 c? 5 c .- m -0.107 Y - m C .o, -0.112 B E -0.117 z" -0.122 T 1 I I I 300 400 500 600 700 800 Time/ms Fig. 8 Correlation of the modified steady-state expression with the experimental data and c is the initial absorbance. After approximately 300 ms the theoretical curve has reached 90% of its maximum and the thermal lens can be said to have reached a steady state. Eqn. (7) can then be substituted into the expression for the steady-state thermal lens [eqn. (l)], to give: S = (1 - 117.779[(A X 10-(k+s)t) + ( B X 10-(k-s)t)]}2 - 1 (11) where 117.779 is a constant derived from 8, rn and V . This equation is then fitted to the experimental values of S with times between 300 and 800 ms; the results are shown in Fig.8. The theoretically generated curve correlates with the experimental data very well ( r = 0.9996). From the curve fit, values of A , B , (k + s) and (k - s) were obtained. From these values the rate constants kl, k2 and k3 can be calculated using eqns. (8), (9) and (lo), and were found to be 1.538 X 10-2, 1.167 x 10-2 and 2.987 x 10-4, respectively. Eqn. (10) indicates that A + B should be equivalent to c , the initial absorbance. The sum of A and B is 1.0467 X 10-3, which correlates well with the previously stated values (error <lo%). Assuming that the excitation step is virtually instan- taneous, the values of kl and k2 indicate that the intermediate state is almost in equilibrium with the excited Methylene Blue dimer, and, therefore, has a fairly long lifetime.This corresponds to the measurements of Danziger et a1.15 who report a lifetime for this charge transfer state of 140 ps. This intermediate state shows no absorbance at 647.1 nm, and it is the formation of this species that causes the thermal gradient to collapse. As the value of kl is slightly larger than k2 then the formation of a small amount of the final product is possible. This is reflected by the value of k3. The thermal lens theory can, therefore, be modified to account for the effects of such a photochemical reaction. Conclusion It has, therefore, been shown that raw data resulting from the formation of a thermally induced lens, which has been affected by the action of a photochemical change, can be fitted to a theoretically predicted curve.From the curve fit, values were calculated for the rates of the steps involved in the photo- induced reaction, kl = 1.538 x 10-2, k2 = 1.167 x 10-2 and k3 = 2.987 x 10-4. In order to obtain information about the reaction quantum yields, more data are needed in the form of experimentation using different concentrations of Methylene Blue. Other future work would include the investigation of the laser power dependency of the reaction, as well as the effect of temperature and dissolved oxygen content upon the reaction characteristics. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 References Miyaishi, K., Imasaka, T., and Ishibashi, N., Anal. Chim. Acta, 1981, 124, 381. Nolan, T. G., Bornhop, D . J . , and Dovichi, N. J., J. Chromatogr., 1987, 384, 189. Sepaniak, M. J., Vargo, J. D., Kettler, C. N., and Maskarinec, M. P., Anal. Chem., 1984, 56, 1252. Lowe, R. D., and Snook, R. D., Anal. Chim. Acta, 1991, 250, 95. Uehard, N., Kanbayashi, M., Hoshino, H., and Yotsuyanagi, T., Talanta, 1989, 36, 1031. Tran, C. D., and Xu, M., Rev. Sci. Instrum., 1989, 60, 3207. Villanueva Camanas, R. M., Sanchis Mollols, J. M., Simo Alfonso, E. F., and Ramis Ramos, G., Anal. Chim. Acta, 1992, 257, 217. Shen, J . , Lowe, R. D., and Snook, R. D., Chem. Phys., 1992, 165, 385. Caro, Br. Pat. 3751177, 1877. Michaelis, L., Schubert, M. P., and Granick, S., J. Am. Chem. Soc., 1940, 62, 204. Rabinowitch, E., and Epstein, L. F., J . Am. Chem. Soc., 1941, 63, 69. Weber, K., Z . Phys. Chem., Abt. B , 1931, 15, 18. Koizumi, M., Obata, H., and Hayashi, S . , Bull. Chem. SOC. Jpn., 1964, 37, 108. Kato, S., Morita, M., and Koizumi, M., Bull. Chem. Soc. Jpn., 1964, 37, 117. Danziger, R. M., Bar-Eli, K. H., and Weiss, K., J. Phys. Chem., 1967, 71, 2633. Bergmann, K., and O'Konski, C. T., 1. Phys. Chem., 1963,67, 2169. Matsumoto, S . , Bull. Chem. SOC. Jpn., 1962, 35, 1860. Paper 21064.521 Received December 3, 1992 Accepted February 11, 1993

ISSN:0003-2654    

DOI:10.1039/AN9931800613

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ANALYST, JUNE 1993, VOL. 118 617 Partial Least Squares Resolution of Multianalyte Flow Injection Data* Paul MacLaurint and Paul J. Worsfold* Department of Environmental Sciences, University of Plymouth, Drake Circus, Plymouth, UK PL4 8AA Philip Norman ICI Chemicals and Polymers Ltd., The Heath, Runcorn, Cheshire, UK WA7 4QD Michael Crane ICI Engineering Department, Winnington, North wich, Cheshire, UK CW8 4DJ The development of a multivariate calibration model to resolve the complex spectra produced by a mixed chemistry flow-injection manifold is presented. The combination of established spectrophotometric reactions for the simultaneous determination of diverse analytes is discussed and the value of applying partial least squares regression, particularly in the interpretation of reaction com bination effects, is considered.The approach is demonstrated with reference to the simultaneous determination of phosphate and chlorine using a combined reaction system that incorporates a mixed molybdate-o-tolidine reagent. This has been achieved using a physically simple flow-injection manifold with photodiode array detection. Calibration is discussed in terms of the experimental design, pre-processing and the number of wavelengths used in modelling. Finally, results of the prediction of a new and independent test set are presented with relative root mean square values of 4.0% for phosphate and 2.4% for chlorine. Keywords: Flow injection; photodiode array spectrophotometry; partial least squares regression; process analysis; simultaneous determination The desire to acquire multianalyte information from a single instrument is particularly important in areas such as industrial and environmental monitoring. Flow injection (FI) has been shown to provide accurate and reliable on-line single analyte information on a near real time basis for a number of water quality parameters (phosphate ,I nitrate ,2 ammonia3 and aluminium4) and a chemical process parameter (sulfites) .These applications demonstrate the capability of process FI to operate continuously in remote locations for long periods6 and deal with aggressive matrices.7 The ability of FI to determine several analytes simul- taneously is the best demonstration of its capacity and versatility for on-line monitoring8 The various approaches to FI multi-determinations were classified in a 1984 review,9 but much of the reported methodology involves complex proce- dures.This conflicts with one of the most salient features of FI for process analysis, that of simplicity. Of the numerous approaches to FI multi-determinations those that utilize multidetection systems offer the greatest potential, particularly for process monitoring.10 Multidetec- tion systems, notably photodiode array spectrophotometers (PDAs) , are routinely used to enhance chromatographic selectivity and this approach has also been shown to minimize matrix interferences and enable multicomponent resolution in F1.11312 The PDAs can acquire and store full ultraviolet- visible (UVNIS) spectra in less than 1 s, thus complementing the rapid and reproducible sample manipulation characteris- tics of automated FI.Some of the early work published using the FI-PDA combination for simultaneous determinations used specific wavelength monitoring (see, for example, refs. 13 and 14) and whilst this proved to be adequate for the applications discussed, more complex analyses often require more advanced data processing. * Presented at SAC '92, an International Conference on Analytical Chemistry. Reading, UK, September 20-26, 1992. 1 Present address: Process Technology Department, Zeneca Fine Chemicals Manufacturing Organisation, P.O. Box A38, Leeds Road, Huddersfield, West Yorkshire, UK HD2 1FF. t To whom correspondence should be addressed. The development of a statistical model that relates the concentration of one or more analytes to the signals from a multichannel instrument is known as multivariate calibration.A number of different multivariate calibration approaches are available and are discussed in detail elsewhere.15-18 Direct multicomponent analysis (DMA) is a statistical technique analogous to the Beer-Lambert model and has been applied to a number of FI multi-determinations (see, for example, refs. 19 and 20). The model assumes that absorbance (or response) at a particular wavelength (or sensor) is a linear function of the concentration of the absorbing species present in the solution under examination. The pure component spectra of all absorbing species present in the samples are, therefore , required in advance for successful calibration. Any inter-analyte interaction or non-modelled interferences that influence the spectral data can yield erroneous information on prediction. Multiple linear regression (MLR), in contrast, can only use a selection of wavelength variables due to collinearity and, therefore, does not share the same signal averaging properties.However, MLR is an indirect calibration method and as such does not require pure component spectra but all expected phenomena must be spanned in the calibration set. Principal components regression (PCR) and partial least squares regression (PLSR) are examples of indirect calibra- tion methods that do offer full spectrum advantages. They overcome problems of collinearity by rank reduction, concen- trating information onto a few latent factors. Principal components regression derives its name from principal components analysis, which is used to decompose the spectral matrix into its most dominant factors.The first principal component, therefore, corresponds to the linear combination of original variables that best describe the spectral variance. The second, and subsequent factors, successively describe the remaining variance. These factors can then be used in an MLR equation to complete the PCR model. Just as PCR utilizes the most dominant factors in the spectral data, PLSR attempts to define the factors that are most relevant to the concentration of the analyte in question. This is achieved by simultaneously estimating the factors in both the spectral and the concentration data, and actively using the concentration data in the bilinear decomposition of618 ANALYST, JUNE 1993, VOL.118 the spectral data. In this way PLSR can reduce the influence of dominant but irrelevant factors and in some cases yields models of lower dimensionality, which are subsequently easier to interpret. Partial least squares regression also has the advantage of being able to model a number of analytes simultaneously, the so-called PLS-2 approach. Both of these bilinear modelling techniques have been successfully applied in FI2'-25 for multi-wavelength data, time- selective data and unfolded, time versus wavelength data matrices. A gradient system for the simultaneous determina- tion of up to five organic acids21 and the determination of a single analyte in a complex matrix,22 have both been reported using PLSR.The simultaneous FT determination of free acid and hydrolysable metals suitable for at-line analysis23 has been presented, again using PLSR, and a comparison of calibration methods has been reported for the simultaneous determina- tion of nickel and iron using a double-injection FI approach.24 In a recent study of multivariate calibration techniques for the multicomponent resolution of UV/VIS data26 it was demonstrated that in ideal circumstances DMA performed no worse than PCR and PLSR. However, when physical and chemical interferences were incorporated into the experi- ments, the bilinear modelling techniques consistently pro- duced significantly better predictions. The same experiments revealed no significant difference between the predictions made by PCR and PLSR.These findings agree with those of other workers24727-29 but if one considers its theoretical advantages and optimal performance over a wide range of conditions,3" PLSR can be considered to be the general method of choice. This paper describes the development of a combined reaction Fl system with photodiode array detection and direct data treatment using PLSR. The primary objective is to demonstrate the feasibility of this integrated approach for simultaneous mu1 tianalyte determinations in a process en- vironment. With this in mind, the emphasis is on the investigation of a number of calibration criteria using a physically simple manifold. Nonetheless, the model system considered is a real one. Zinc phosphate and chlorine are added to industrial cooling waters as a corrosion inhibitor and biocide , respectively, and on-line chemical information is desirable for control purposes.The practical implications of combining established spectrophotometric methods for analy- tes of a diverse nature are considered and the influences of the calibration parameters are presented in detail. Experimental Reagents All solutions were prepared in Milli-Q water (Millipore) and all reagents were of AnalaR grade (Merck) unless otherwise indicated. A stock phosphate solution containing 1000 mg 1-1 phosphorus (PO4-P) was prepared by dissolving 4.390 g of potassium dihydrogen orthophosphate (dried for 2 h at 105 "C) in 1 1 of water. A stock hypochlorite solution containing 1000 mg I- free chlorine as chlorine was prepared by dilution of an appropriate volume of iodimetrically standardized sodium hypochlorite solution (Merck; general purpose reagent).Calibration and test set solutions were prepared by serial dilution of these stock solutions and are subsequently referred to as phosphate and chlorine solutions. A solution of N,N- diethyl-174-phenylenediamine sulfate (DPD, Aldrich; 4- N , N, -diethylaminoaniline sulfate) was prepared by dissolving 1.5 g of DPD in 1 1 of water. The acid-molybdate reagent was prepared by dissolving 10 g of ammonium hepta- molybdate in 1 1 of 0.4 mol 1-1 nitric acid and the ascorbic acid solution was prepared by dissolving 80 g in 1 1 of water. A solution of o-tolidine was prepared by dissolving 0.86 g of o- tolidine dihydrochloride (Fluka; purum) in 2 mol I-1 HCI. [Caution: o-tolidine is highly toxic and should be handled with extreme care.] Instrumentation A schematic diagram of the automated FI-PDA arrangement is shown in Fig.1. The FI manifold was constructed from 0.8 mm i.d. poly(tetrafluoroethy1ene) (PTFE) tubing and in- house PTFE T-pieces. Absorbance spectra were measured using a Hewlett-Packard HP 8451A photodiode array spectro- photometer fitted with an 18 p1 glass flow cell with a pathlength of 1 cm (Hellma). Raw data were stored using an HP 9121 disk drive and output in ASCII format using an HP 82939A serial interface to a Viglen 386 DX personal computer with 8 Mbytes of RAM. All subsequent data processing was carried out using this computer. Sample injections (150 pl) were made using a pneumatic valve control unit (PS Analytical) and all solutions were propelled by two peristaltic pumps (Ismatec Mini S-820) with poly(viny1 chloride) (PVC) pump tubing (Labsystems) . Control of the valves and pumps was maintained via an HP 82940A GPTO interface. Software A general purpose program was written in HP BASIC to control the FI components, to measure and record spectra, and to carry out some basic data processing. A further program was used to transmit spectral data to the personal computer via the serial interface.Kermit serial interface software version 3.01 was used to collect and store data in ASCII format on the personal computer. All multivariate data analysis was carried out using Unscrambler IT Extended version 4.00 (Camo N S ) , which incorporates matrix handling routines allowing manipulation of the ASCII files.Procedures Batch experiments In order to evaluate the compatibility of the phosphate and chlorine reaction chemistries, a number of preliminary experiments were carried out. The visible spectra of combi- nations of the molybdate, DPD and o-tolidine reagents were recorded after addition of combinations of water and solutions of 10 mg 1-1 phosphate and 10 mg 1-l chlorine. Flo w-injection experiments The Fl manifold used in all experiments is shown in Fig. 2. The absorbance was measured every 2 nm over the wavelength range 352-550 nm yielding 100 data points per spectrum and one spectrum was recorded every second between 1 and 60 s after injection (with an integration time of 0.5 s). This resulted in a total of 6000 data points for each injection.All spectra were measured against a reagent blank. The control software was designed to calculate and store to disk the mean spectrum of the three spectra nearest to the peak maximum for each Pri n te r/ computer interface I I 1 I I - Pump - Photodiode module array Injection valve Fig. 1 Schematic diagram of the automated FI-PDA systemANALYST, JUNE 1993, VOL. 118 619 SAM 150 1.11 1 PQ- 550 nrn rnl rnin-l Fig. 2 Flow-injection manifold for the simultaneous determination of phosphate and chlorine: SAM, sample; TOL, o-tolidine; and MOL, acid-molybdate Table 1 Concentration data of the calibration set and test set (results given in mg 1-') Sample number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Calibration set Test set PO4 c1 PO4 CI 2 1 3 1 2 2 3 2 2 3 3 3 2 4 3 4 2 5 3 5 4 1 5 1 4 2 5 2 4 3 5 3 4 4 5 4 4 5 5 5 6 1 7 1 6 2 7 2 6 3 7 3 6 4 7 4 6 5 7 5 8 1 9 1 8 2 9 2 8 3 9 3 8 4 9 4 8 5 9 5 - - 10 1 10 2 10 3 10 4 10 5 - - - - - - - - injection.All solutions were measured in triplicate and the overall mean spectrum of the three injections was also stored to disk. This overall mean spectrum was used for all subsequent data processing unless otherwise stated. Calibration set solutions (training set) were prepared to cover the ranges 2-10 mg 1-I phosphate and 1-5 mg 1-I chlorine in a five-level factorial design. A further 20 samples were prepared and analysed 48 h later as an independent test set. The concentration details are given in Table 1. Both the calibration and test sets were analysed in random order to reduce any drift effects.Results and Discussion Batch Experiments One reason for the widespread use of FI techniques is the breadth of established spectrophotometric procedures that can be implemented. The analytes under investigation in this study are routinely determined by spectrophotometric pro- cedures; e.g., the Molybdenum Blue method for phosphate and the DPD method for chlorine.31 Furthermore, both reaction chemistries have been successfully used in FI methods for phosphate1.32.33 and chlorine.34.35 Initial experi- ments were conducted to combine these reaction chemistries to enable simultaneous determinations. However, batch experiments revealed that the two procedures were incompat- ible due to differing pH requirements; acid media for the molybdate reaction and pH 6.2-6.5 for the DPD reaction.31 Another method for the spectrophotometric determination of chlorine uses o-tolidine and can be carried out over a wide pH range.36 This reaction has also been used in an FI method.37 However, when the o-tolidine reaction was combined with the Molybdenum Blue reaction the chlorine response was lost completely.This was found to occur instantaneously upon addition of the ascorbic acid solution. Ascorbic acid is added in the determination of phosphate to reduce phosphomolybdic acid to the Molybdenum Blue complex. The monitorand for chlorine, in contrast, is a haloquinone; the product of chlorine oxidizing the o-tolidine. Tin(I1) chloride was found to have the same effect on the chlorine reaction, suggesting that the haloquinone is being reduced by the ascorbic acid.Neverthe- less, the yellow phosphomolybdic acid can also be monitored spectrophotometrically, thus eliminating the need for ascorbic acid reduction. This approach, while less sensitive than the Molybdenum Blue approach, has also been used in FT.38 Flow-injection Experiments The successful implementation of process analytical methods requires the fulfilment of a number of important criteria, one of the most important of which concerns reliability. Any instrumentation to be installed in a manufacturing environ- ment needs to be robust and the overall procedure must be dependable, especially if the information is going to be used for process control. In FI terms, the manifold design must be kept as physically simple as is permissible with the analytical requirements.This would be a single-line manifold in ideal situations. In this work, a single injection, two-line manifold with one detector and a second pump for sample loop filling was used throughout (Fig. 2). This configuration was required for two reasons. Firstly, injection of sample into a molybdate stream caused a large negative response due to reagent dilution, and secondly, a mixed molybdate-o-tolidine reagent was found to be unstable. A typical 3-D FI response profile obtained from this manifold is shown in Fig. 3, representing 60 spectra measured at 2 nm intervals over the 352-550 nm range. The univariate procedures on which this work was based used 362 nm for phosphate38 and 438 nm for chlorine37 and it can be seen that the same spectral regions are active after combination of the reaction chemistries.It is obvious from Fig. 3, however, that when phosphate and chlorine are present in the same solution a more complex picture arises. Most noticeable is the emergence of a shoulder at wavelengths greater than 460 nm in the chlorine active region of the spectrum. This is more distinct in Fig. 4, which shows the mean spectra recorded at the FI peak maximum for each of the 25 calibration standards of the 52 experimental design. For reasons of clarity the spectra have not been labelled. Variance in the chlorine active region of the spectra is particularly evident and grouping of equal chlorine concentration samples is noticeable, especially at lower concentrations.Calibration All PLSR models were developed in the PLS-2 mode and the optimal dimensionality was defined as the first local minimum of the PRESS (prediction error sum of squares) relative to the number of factors included. The PRESS is defined as: I PRESS = (yj - j i ) 2 where I = the number of samples, yi = the true concentration of sample i and ji = the predicted concentration of sample i. The PRESS was calculated in all cases using leave-one-out i= 1620 ANALYST, JUNE 1993, VOL. 1 118 Absorbance I Fig. 3 mg 1-l and phosphate at 10 mg I-' 3-D FI response profile for a solution containing chlorine at 5 0.6 360 400 440 480 520 Wavelengthlnm Fig. 4 the 25 solutions of the 52 experimental design Mean spectra recorded at the FI peak maximum for each of internal cross-validation; thus the concentration of the sample left out was predicted using the I - 1 model for all I samples.The prediction ability of the models for both analytes is expressed in terms of the relative root-mean-square (RRMSE).24 R R M S E = F $7 RESS Y where 7 = the true concentration mean. The RRMSE is used for comparisons of both the cross-validation models (RRMSECV) and the prediction of the independent test set (RRMSEP). In both cases no degrees of freedom are lost. The first three PLSR loading vectors for the 52 calibration model are shown in Fig. 5 . In the process of PLSR modelling, the covariance between the spectral scores and a single analyte is maximized. This often leads to the loadings of the first PLSR factor approximating to the pure component spectrum of the analyte under examination. The PLS-2, however, maximizes the covariance between the spectral scores and a linear combination of a number of variables (2 in this case).The physical significance of the loadings , therefore , becomes less clear. Inspection of the plot of the scores of the first PLSR factor versus the second factor reveals a very interesting structure (Fig. 6). The samples are aligned, as expected, in the order of the 52 experimental design but not in an equidistant fashion. This is particularly noticeable between the samples containing 2 and 4 mg 1-l phosphate, where the distance between pairs of samples of equal chlorine concentration increases with chlorine concentration. This would suggest some type of non-linear relationship caused by the combina- tion of the phosphate and chlorine reaction chemistries .39,40 Although PLSR is a linear method, it can handle non- linearities by the inclusion of additional factors's and this 0.2 0.1 0, C 0 -I .- g o -0.1 -0.2 350 400 450 500 550 Wavelengthlnm Overlay of the loading vectors of the first three PLS-2 factors Fig.5 as a function of wavelength I? 0- y. u 0 -4 rn I -81 03 - O4 05 -12 I I I I - 20 -10 0 10 20 Scores of PLSR factor 1 Fig. 6 PLS-2 scores of factor 1 versus factor 2 Table 2 Effect of a number of pre-processing techniques on the relative prediction errors of PLSR and PCR models PLSR PCR RRMSECV RRMSECV Pre-processing No. of No. of techniques factors PO4 CI factors PO4 CI None 3 11.9 1.9 3 12.0 1.9 Mean-centring(M-C) 3 5.4 2.0 3 5.6 1.9 M-C and autoscaling (AS) 3 4.0 2.4 3 4.0 2.4 M-C, AS and normalization 6 12.8 14.7 5 15.1 15.1 M-C, AS and first derivative 3 6.5 1.8 3 6.5 1.8 M-C AS and second derivative 4 11.8 4.3 4 14.9 4.5 could explain the need for three factors to describe a two- component system.Pre -processing The effect of a number of pre-processing techniques on the RRMSECV of the 52 experimental design is shown in Table 2. Mean-centring41 is traditionally applied in PCR and PLSR and, as the name suggests, involves the subtraction of the variable mean from the individual variable values. Whilst the model dimensionality has not been reduced by mean-centring in this case, the phosphate predictions are significantly improved. Setting all variables to equal variance by dividing the mean-centred values by their standard deviation is known as autoscaling and it can be seen that autoscaling has had a small but beneficial effect on this data set.Normalization on the other hand, which sets all spectra to unit length, has had a grossly detrimental effect. Spectral derivatives that can enhance resolution generally lead to a depreciation in the signal-to-noise ratio with eachANALYST, JUNE 1993, VOL. 118 62 1 Table 3 Effect of wavelength selection on the relative prediction errors RRMSECV RRMSEP Number of wavelengths selected PO4 c1 PO4 c1 100 4.0 2.4 4.0 2.9 50 4.1 2.4 4.2 2.9 25 4.3 2.4 4.9 2.8 10 4.3 2.5 7.2 3.3 5 5.2 2.2 7.0 2.3 Table 4 Effect of wavelength averaging on the relative prediction errors RRMSECV RRMSEP Number of wavelengths averaged PO4 c1 PO4 C1 0 (loo)* 4.0 2.4 4.0 2.9 4.0 2.4 4.0 2.9 4.0 2.4 4.0 2.9 10 (10) 4.0 2.4 4.1 2.9 20 (5) 4.1 2.3 4.2 2.9 2 (50) 4 (25) * Values in parentheses are the number of data points used.derivatization. Both the first and second derivatizations had a detrimental overall effect on the RRMSECV of this data set. The PCR models were also built using the pre-processed data and, as expected, resulted in dimensionality and RRMSECV values very similar to those for PLSR.26 Wavelength selection and averaging The effect of the size of the spectral data matrix on the prediction ability was studied in two ways. Firstly, wavelength variables were simply selected from the original data set and used to build PLS-2 models after mean-centring and autoscal- ing.Selection was made by taking every second variable to reduce the number from 100 to 50 and the same approach was taken for the selection of the 25 and 10 point data sets. The 5 point data set was selected according to the perceived importance of the variables; 360, 400, 440, 470 and 510 nm were used. The RRMSECV values for the five models are given in Table 3 together with the RRMSEP values for the independent test set. The prediction error for chlorine is very stable with decreasing data set size but that for phosphate increases. in the second case, the data set was reduced by averaging the spectral variables before mean-centring and autoscaling. Inspection of Table 4 reveals that both the phosphate and chlorine predictions are stable to the data set averaging.In averaging the spectral variables the original data are largely retained, albeit in a modified form, whereas informa- tion is lost in wavelength selection. This could explain the small increase in RRMSEP for phosphate using the selected variable data sets. The practical implications of these findings are that full spectra should be collected and stored at the measurement stage and that some wavelength averaging could be carried out before model building. However, the only advantage of wavelength averaging is a reduction in the time taken for model building, which for data sets of this size is not problematic and, given the loss of qualitative information associated with reducing the data set size, it would be provident to use the full spectra.Calibration design The effect of the size of the calibration set on the RRMSEP of the independent test set was determined by reducing the number of levels of the experimental design. The four level design includes the samples at 2 , 4 , 8 and 10 mg 1-l phosphate and 1, 2, 4 and 5 mg I-' chlorine, and the three level design was constructed from the 2,6 and 10 mg 1- phosphate and 1 , 3 Table 5 Effect of reducing the size of the calibration set on the prediction errors for an independent test set RRMSEP Size of calibration set PO4 c1 5 level; 25 4.0 2.4 4 level; 16 4.5 3.5 3 level; 9 4.7 3.3 2 level; 4 6.8 3.7 2level+ 1;5 6.3 3.2 Sample number 13 8 17 18 6 20 19 5 15 2 9 7 1 11 3 10 4 16 12 14 Added/ mg 1-1 7.0 5.0 9.0 9.0 5.0 9.0 9.0 3.0 7.0 3.0 5 .O 5.0 3.0 7.0 3.0 5.0 3.0 9.0 7.0 7.0 Found/ mg 1-1 7.1 5.1 9.4 9.4 5.2 9.1 9.2 2.6 6.7 2.9 4.8 5.1 3.0 7.0 2.8 4.5 2.7 9.0 7.0 6.7 Difference -6.7 -6.7 -5.0 (% - - +2.0 -2.5 -2.0 -5.0 -2.5 -5.0 - - - -3.3 +2.0 -2.5 -5.0 - - Table 6 Predictions of the independent test set Phosphate Chlorine 1 Difference Added/ Found/ (Yo) +1.4 +2.0 +4.4 +4.4 +4.0 +1.1 +2.2 -13.0 -4.3 -3.3 -4.0 +2.0 - - -6.7 -10.0 -10.0 - - -4.3 mg 1-1 mg 1-1 3.0 2.8 3.0 2.8 2.0 1.9 3.0 3.0 1.0 1.0 5.0 5.1 4.0 3.9 5.0 ' 4.9 5.0 5.0 2.0 1.9 4.0 3.9 2.0 1.9 1.0 1.0 1.0 1.0 3.0 2.9 5.0 5.1 4.0 3.9 1.0 1.0 2.0 1.9 4.0 4.0 and 5 mg 1-l chlorine samples.The samples containing 2 and 10 mg I-' phosphate and 1 and 5 mg 1-I chlorine made up the two level design and finally the 6 mg I-1 phosphate and 3 mg 1-1 chlorine sample was included to give a calibration set of five samples. The results, given in Table 5 , show a general increase in the RRMSEP as the number of calibration samples is reduced.Nevertheless, this increase is not dramatic, and in a situation where analysis time is an important consideration, the use of a 9 sample calibration set requires only a small compromise in prediction error. Predictions The predicted values for the independent test set and the percentage difference between the added and calculated concentrations of phosphate and chlorine are given in Table 6. Predictions were made using the model built from the five level experimental design after mean-centring and autoscaling the data. The repeatability ( n = 3) for the test solutions ranged from 0.1 to 2.9%.Conclusions A physically simple, combined reaction FI manifold with photodiode array detection integrated with PLSR of the data has been shown to be a feasible approach to simultaneous multianalyte determinations. The combination of established spectrophotometric meth- ods is a non-trivial matter and judicious choice of reaction chemistries is required to avoid gross interference. Visual inspection of the scores and loadings of the multivariate calibration model has been shown to reveal some of the underlying effects of the reaction combination. Mean-centring and autoscaling of the data sets was found to be profitable, whilst selection and averaging of the spectral622 variables had no beneficial effect. Reducing the number of calibration standards used in modelling increased the error of prediction, but not prohibitively so.A procedure has been developed for the simultaneous detcrmination of phosphate and chlorine and the prediction of analyte concentrations for an independent test set, prepared and analysed 48 h after calibration, yielded RRMSEP values of 4.0% for phosphate and 2.4% for chlorine. The authors thank ICI Chemicals & Polymers Ltd. for financial support. P. M. also thanks the SERC for support under the CASE scheme. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 References Worsfold, P. J . , Clinch, J. R., and Casey, H., Anal. Chim. Acta, 1987, 197, 43. Clinch, J . R., Worsfold, P. J., and Casey, H., Anal. Chim. Acta, 1987, 200, 523. Clinch, J. R. Worsfold, P. J . , and Sweeting, F., Anal.Chim. Acta, 1988, 214, 401. Benson, R. L., Worsfold, P. J., and Sweeting, F., Anal. Chim. Acta, 1990, 238, 177. MacLaurin, P . , Worsfold, P. J., Townshend, A., Barnett, N. W., and Crane, M., Analyst. 1991, 116, 701. Casey, H., Clarke, R. E., Smith, S. M., Clinch, J. R., and Worsfold. P. J., Anal. Chim. Acta, 1989, 227, 379. MacLaurin, P., Parker, K. S . , Worsfold, P. J., Townshend, A., Barnett, N. W., and Crane, M., Anal. Chim. Acta, 1990, 238, 171. Luque de Castro, M. D., Talanta, 1989, 36, 591. Luque de Castro, M. D., and Valcarcel Cases, M., Analyst, 1984, 109, 413. MacLaurin, P., and Worsfold, P. J., Microchem. J., 1992, 45, 178. Wolf, K., and Worsfold, P. J.. Anal. Proc., 1986, 23, 365. Lazaro, F., Rios, A., Luque de Castro, M. D., and Valcarcel, M., Analusis, 1986, 14, 378.Lazaro, F., Rios, A., Luque de Castro, M. D., and Valcarcel, M., Anal. Chim. Acta, 1986, 179,279. Wada, H., Murakawa, T., and Nakagawa, G., Anal. Chim. Acta, 1987, 200, 515. Geladi, P., and Kowalski, B. R., Anal. Chim. Acta, 1986, 185, 1. Beebe, K. R., and Kowalski, B. R., Anal. Chem., 1987, 59, 1007A. Sanchez, E., and Kowalski, B. R., J. Chemometr., 1988,2,247. 18 19 20 21 22 23 24 25 26 27 28 2Y 30 31 32 33 34 35 36 37 38 39 40 41 Paper 21060476 Received November 13, 1992 Accepted March 25, 1993 ANALYST, JUNE 1993, VOL. 118 Martens, H., and Naes, T., Multivariate Calibration, Wiley, Chichester, 1989. Blanco, M., Genk, J., Itturriaga, H., and Maspoch, S., Analyst, 1987, 112, 619. Blanco, M., Gene, J . , Itturriaga, H., Maspoch, S . , and Riba. J., Talanta, 1987, 34, 987. Lukkari, I . , and Lindberg, W., Anal. Chim. Acta, 1988,211, 1. Gerritsen, M. J. P., Kateman, G., van Opstal, M. A. J., van Bennekom, W. P., and Vandeginste, B. G. M., Anal. Chim. Acta, 1990, 241, 23. Lindberg, W., Clark, G. D., Hanna, C. P., Whitman, D. A . , Christian, G. D., and RGiiEka, J . , Anal. Chem., 1990,62,849. Whitman, D. A., Seasholtz, M. B., Christian, G. D., RfiiiEka, J., and Kowalski, B. R., Anal. Chem., 1991, 63, 775. Blanco, M., Coello, J., Itturriaga, H . , Maspoch, S . , Redon, M., and Riba, J., Anal. Chim. Acta, 1992, 259, 219. MacLaurin, P., Worsfold, P. J., Crane, M., and Norman, P., Anal. Proc., 1992, 29, 65. Haaland, D. M., and Thomas, E. V., Anal. Chem., 1988, 60, 1193. Haaland, D. M., and Thomas, E. V.. Anal. Chem., 1988, 60, 1203. Haaland, D. M., Anal. Chem., 1988, 60, 1208. Thomas, E. V., and Haaland, D. M., Anal. Chem., 1990, 62, 1091. Standard Methods for the Examination of Water and Wastewater, American Public Health Association, American Water Works Association, Water Pollution Control Federation, Washington, DC, 17th edn., 1989. Johnson, K. S., and Petty, R. L., Anal. Chem., 1982,54,1185. Lacy, N., Christian, G. D., and RfiiiEka, J . R., Quim. Anal., Leggett, D. J., Chen, N. H., and Mahadevappa, D. S., Fresenius’ 2. Anal. Chem., 1983, 315, 47. Gordon, G., Sweetin, D. L., Smith, K., and Pacey, G. E., Talanta, 1991, 38, 145. Johnson, J. D., and Overby, R., Anal. Chem., 1969,41, 1744. Leggett. D. J., Chen, N. H., and Mahadevappa, D. S . , Analyst, 1982,107,433. RGiiEka, J . , and Hansen, E. H., Anal. Chim. Acta, 1975, 78, 145. Isaksson, T., and Naes, T., Appl. Spectrosc., 1988, 42, 1273. Kowalski, B. R., and Seasholtz, M. B . , J. Chemometr., 1991, 5, 129. Seasholtz, M. B . , and Kowalski, B. R., J. Chemometr., 1992,6, 103. 1989, a, 201.

ISSN:0003-2654    

DOI:10.1039/AN9931800617

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ANALYST, JUNE 1993, VOL. 118 623 Determination of Iodide Ion in Impregnated Charcoals by Flow Injection" Cheryl D. Monks, Duangjai Nacapricha and Colin G. Tavlort School of Chemical and Ph ysicaj Sciences, Liverpool John Moores University, Byrom Street, Liverpool, UK L3 3AF A procedure for the determination of iodide in charcoals impregnated with potassium iodide is described. An aqueous extract is prepared and the iodide ion concentration in this extract (up to 4 x rnol dm-3) is determined by injection into a water stream that is merged with an aqueous solution containing potassium iodate and sulfuric acid, followed by absorptiometric measurement at 460 nm. The analysis is simple and rapid, with a limit of detection of 6 x rnol dm-3 iodide and a relative standard deviation ( n = 10) of 1.2% for 1 x 10-3 rnol dm-3 iodide.The procedure was used to measure the levels and the distribution of potassium iodide in impregnated charcoals. The flow method is discussed with reference to the chemistry of elementary iodine in aqueous solution. Keywords: Flow injection; iodide; charcoal Activated charcoals, impregnated with potassium iodide (1.5%) or triethylenediamine ( 5 % ) or both, are used in the nuclear industry as filter materials for the trapping of volatile radioactive substances, especially those containing radio- iodine. The trapping efficiency of these charcoals is evaluated by the K-value test1 in which a bed of the charcoal is 'challenged' under standard conditions with iodomethane (131l), one of the least efficiently trapped compounds of iodine.As part of a current programme of research on factors that affect the performance of potassium iodide-impregnated charcoals, a simple, rapid procedure for the determination of iodide ion in aqueous extracts of these impregnated charcoals was required. Complete extraction of iodide ion from charcoal can be achieved with the use of potassium hydroxide solution,2 but the determination of iodide in the extract by the recommended method (neutralization with acetic acid, fol- lowed by titration with silver nitrate and use of dichloroflu- oroscein as indicator) is tedious and was found to yield indistinct end-points. Flow injection (FI) was, therefore, examined as an alternative. Iodide ion has been determined by injection into a carrier stream containing the iodate ion and acid, the free iodine produced by the reaction: 51- + IO3- + 6H+ -+ 312 + 3H2O being measured by spectrophotometry at 350 nm.3 A working range from 2 x to 8 x rnol dm-3 iodide was reported.Iodide has also been determined spectrophoto- metrically by injection into aqueous bromine, followed by reaction of the iodate produced with more iodide to form triiodide .4 The working range of this sensitive amplification method is from 0.4 X to 12 X mol dm-3. In the present work, FI by the iodate-acid reaction has been further investigated and a procedure with a sensitivity adequate for the analysis of charcoals, impregnated with 1.5% of potassium iodide, has been developed. The procedure was used to measure the distribution of potassium iodide on charcoal impregnated in rotary drums by spraying and on particles of different sizes within a charcoal batch, and to measure the effect of ageing on the iodide content of potassium iodide-impregnated charcoals.The effects of con- centration of extractant and of the presence of common anions on the FT signal were studied. Some observations on the speciation of iodine in aqueous solution were carried out. * Presented at SAC '92, an International Conference on Analytical + To whom correspondence should be addressed. Chcrnistry, Reading, UK, September 20-26, 1992. Experimental Apparatus A Tecator (Herndon, VA, USA) FTAstar SO10 analyser with a 5017 autosampler and a 5032 controller, incorporating an automatic wavelength scanner and a 1 cm pathlength flow- through cell, was used.Four different diameters of pump tubing were used to provide flow rates of 1.2, 1.5,2.0 and 2.8 cm3 min-I, together with four different mixing coils of length and i.d. 30 cm X 0.7 mm, 60 cm X 0.7 mm, 30 cm x 0.5 mm and 60 cm x 0.5 mm, and a sample loop of volume 0.03 cm3. Spectra were recorded with a Pye Unicam (Cambridge, UK) SP 8/100 ultraviolet-visible (UViVIS) spectrophotometer. Reagents Potassium hydroxide solution, 5% miv. Potassium iodate-sulfuric acid carrier solution, 0.025 rnol dm-3 K I 0 3 and 0.1 rnol dm-3 H2S04. Standard potassium iodide solution, 0.100 rnol dm-3. Working standards were prepared from this solution by dilution with potassium hydroxide solution. Procedure Impregnated charcoal (2.5 g) was heated under reflux with SO cm3 of potassium hydroxide solution for 3 h.The mixture was allowed to cool and then filtered. The residue was washed with water, the washings being used to dilute the filtrate to 100.0 cm3. Portions of the diluted extract were injected via the autosampler into a carrier stream of water, which was merged with the iodate-acid reagent stream. The merged stream was passed through a mixing coil before measurement of the peak height (Fig. 1). A calibration was carried out by injecting under the same conditions standard iodide solutions (0.0000, 0.0010, 0.0015, 0.0020 and 0.0025 mol dm-3 in 2.5% potassium hydroxide solution). The concentration of iodide ion in a diluted extract and, hence, the percentage of potassium iodide in the charcoal sample were obtained by interpolating peak heights on a calibration graph. Results and Discussion Choice of Wavelength The absorption spectrum of a standard solution of iodine in water alone [Fig.2(a>] indicates that iodine in the absence of iodide ion exhibits an absorption maximum at 460 nm. This624 ANALYST, JUNE 1993. VOL. 118 2.8 ctn3 min-1 A-Fl-l Fig. 1 FI manifold for the determination of iodide ion in extracts 1 2 3 from impregnated charcoals. A, Autosampler; C, carrier stream, 0.025 mol dm-3 KI03, 0.1 mol dm-3 H,S04; P, pump, flow rate 2.8 crn3 min-'; I, injection valve, 0.03 cm3 loop; M, mixing coil, 30 cm long, 0.5 mm i.d.; D, detector 460 nm; and W, waste Flow ratekm3 min 1 Fig. 3 FI for iodide ion by injection into water followed by merging with acid-iodate. Influence on peak height of flow rate and coil dimensions.A volume of 30 mm- of 2 x lo-' mol dm-3 iodide was injected. Coil dimensions (lengthkm x i.d./mm): A, 30 x 0.5; B, 30 x 0.7; C, 60 x 0.5; and D, 60 x 0.7 Wavelength - 0.2 ' I I I I 1 0 1 2 3 4 5 KOH (% m/v) Fig. 4 response for 2 x 10 - 3 mol dm-3 iodide by FI with acid-iodate Effect of potassium hydroxide in the test solution on the Reagent Concentrations Concentrations of iodate and sulfuric acid of 0.025 and 0.1 rnol dm-3, respectively, were selected to maintain excesses of reagents over iodide up to and above an iodide concentration of 4 x rnol dm-3, the maximum concentration antici- pated in applications. Limit of Detection and Precision A limit of detection was calculated from the responses of ten blank injections of 2.5% potassium hydroxide solution.The limit ( L ) , in absorbance units, was obtained from the expression: 5 L = mean absorbance of blank + (2.33 x standard deviation) Fig. 2 Abso tion spectra of aqueous iodine. (a) In water (3.0 X and was found to be equivalent to 6 x 10-S mol dm-3 iodide. mol dm- T 12); (b) in aqueous 0.05 mol dm-3 potassium iodide (3.0 wavelengths in nm) mol dm-3 Iz). shown on figure are for Ten injections of 1 x 10-3 and of 2 x 10-3 rnol dm-3 iodide yielded relative standard deviations of 1.2 and 0.370, respect- ively . condition is approximated to at the point of measurement in the FI system. Hence, all measurements were carried out at 460 nm rather than at 350 nm, the wavelength previously used.3 Optimization of Flow Conditions Repeated injections of 0.002 rnol dm-3 potassium iodide were effected with variations of flow rate and mixing-coil length and diameter.Peak-height measurements (Fig. 3) indicate that, of the coils examined, the one that was 30 cm long with 0.5 mm i.d. yielded the highest responses. A flow rate of 2.8 cm3 min- * was chosen as the optimum: although the responses obtained with this flow rate are not the highest measured, it permits a high sample throughput, which is important for applications of the method such as those described here. Injections were made into a water stream rather than directly into the reagent stream in order to improve sensitivity. Interference Hydroxide ion is present at a relatively high concentration (0.5 mol dm-3, 2.5% KOH) in the test solutions from the potassium hydroxide extracts of charcoal samples, and par- tially neutralizes the acid in the reagent, which could lead to a significant reduction in the yield of iodine produced.Results obtained from the injection of iodide standards (2 x rnol dm-3), containing hydroxide ion in the concentration range 0-1.0 mol dm-3 (Fig. 4), indicate that the presence of 2.5% of potassium hydroxide in the test solution reduces the response by about 7%, and emphasize the need to rim standards that are matched with the sample. Interference effects resulting from the presence of common anions as their potassium salts, which can be present in extracts of charcoal samples or similar test solutions, are summarized in Table 1. Relatively high concentrations of chloride and nitrate ions can be tolerated.The presence of bromide ions causes serious interference owing to competition between bromide andANALYST, JUNE 1993, VOL. 118 625 Table 1 Concentrations of common anions, which can be tolerated in the determination of iodide ion (2 X mol dm-3) by FI. Reagents: K I 0 3 (0.025 mol dm-3); H2S04 (0.1 mol dm-3) Tolerance/ Ion mol dm-3* Chloride 0.08 Bromide 0.01 Nitrate 0.1 Sulfate 0.01t * Maximum examined concentration producing a reduction in t Absorbance decreases only slightly with increasing concentration absorbance of <2a (0.6%). above this level of sulfate ion. (a) 16.5 cm \ \ \ v 1.60 1 0.80 5 0 2.46 ( C) 1.60 @ 0 80 n I--- ------ 1 2 3 4 5 6 7 8 9 101112131415161718 1920 Segment Fig. 5 Distribution of potassium iodide sprayed onto charcoal in drums.Spray volume, 20 cm3; 1.5 g KI; 100 g charcoal. (a) Diagram of drum exteriors; ( b ) distribution through drum with straight vanes; and (c) distribution through drum with axial vanes iodide for the iodate reagent. The slight interference from sulfate, which is present at a low concentration, is probably an ionic strength effect resulting from the double charge on the sulfate ion. Applications The distribution of potassium iodide on spray-impregnated charcoal was studied by using this method. Two 100 g portions of the same charcoal were placed in drums of similar dimensions, one drum being fitted with straight mixing vanes parallel to the axis and the other with helical vanes. The drums were rotated on rollers about their axes while potassium iodide solution was sprayed through the lid apertures to Table 2 Potassium iodide content of sized fractions of some commercial charcoals of various ages, initially impregnated to a nominal level of 1.5% potassium iodide KI (Yo) Sample 1 2 3 4 5 6 * US sieve size.t Aperturedmm. &lo* Agetyears 2.36-2.OOt 0.3 1.16 2.0 1.23 3.5 0.85 4.5 1.18 5.2 0.52 7.9 1.17 10-12* 2.00-1.70f 1.56 1.43 0.91 1.30 0.87 1.27 produce a nominal level of potassium iodide on the charcoal of 1.5% [Fig. 5(a)]. The contents of each drum were separated, after spraying, into 20 approximately equal segments along the length of the drum, and the potassium iodide level in each segment was determined by extraction followed by FI. The distribution produced by spraying into a drum fitted with helical vanes was found to be more uniform than that in a drum with straight vanes [Fig.5(b) and (c)]. Also, the loss of potassium iodide during spraying was less in the helical drum than in the axial drum; the levels in all segments of both drums were less than the nominal 1.5%, except at the extremities of the helical drum. The method is particularly suited to such studies where relatively large numbers of samples are generated. A number of commercial spray-impregnated charcoals of various ages were separated by sieving into fractions of different particles sizes. The principal fractions were then analysed by the proposed method. The results presented in Table 2 indicate that the potassium iodide content is inversely related to particle size and that all of the charcoal fractions analysed, except the smaller fraction of the most recent carbon, have potassium iodide contents less than the nominal value of 1.5%, but usually still greater than 1%, suggesting that some loss of iodide could have taken place by oxidation. After initial measurement of the K values, these commercial charcoals had been resealed in polyethylene bags with adhesive tape and stored .in the dark at ambient temperature over periods of up to 8 years.Ageing was accompanied by substantial reductions in K values. Speciation of iodine in aqueous solution It is evident from Fig. 2 that the principal elementary iodine species in water differs from that in aqueous solutions containing an excess of potassium iodide. The reaction: I2 + I- + 13- leads to the formation of a species exhibiting two absorption maxima, one at 285 nm and the other at 350 nm, with high molar absorptivities (E) (3.2 x 104 and 2.1 X 104 dm3 mol-I cm-l, respectively). Iodine in water alone exhibits three absorption peaks, the highest at 460 nm (E = 490 dm3 mol-l cm-I) and the others at the same wavelengths as those observed for the 13- species.It seems that the dominant iodine species in water alone is an hydrated I2 molecule, but that a disproportionation reaction such as 212 + H20 + 13- + 01- + 2H' takes place to yield a very small amount of the strongly absorbing T3- species. Information on the formation constant of 13- { K = [I3-]/[12][1-]} is scarce, but titrimetric studies6 and absorptiometric studies by Monks7 indicate a rather low value for K of 103 dm3 mol-I.Therefore, if either of the intense ultraviolet absorption peaks are to be used with626 ANALYST, JUNE 1993, VOL. 118 advantage for the absorptiometric determination of iodine, a large excess of iodide ion must be present. Under the conditions for F1 reported here, the excess is large only at or near the point of injection and diminishes nearly to zero at the point of measurement, Hence, the concentration of iodine produced is measured more effectively by using the less intense peak at 460 nm. Previous work3 has shown that when the iodide ion is determined by injection into an acid-iodate stream with absorbance measurement of iodine at 350 nm, an increase in the concentration of iodate causes a shortening of the region of linear response, absorbance becoming constant with increasing iodide concentration. This observation is consistent with a significant decrease in the extent of reaction to form the triiodide ion, the species being determined, as the extent of oxidation from iodide to iodine increases. This work is part of a programme of testing and development supported by Sutcliffe Speakman Carbons, Leigh, Lancashire, UK. References 1 Taylor, C. G., and Griffiths, J . G., Carbon, 1991, 29, 101. 2 Wildman, J . , Internal Report 74/52/3.01, 1974, Sutcliffe Speak- man, Leigh, Lancashire. 3 Kamson, 0. F., Anal. Chim. Acta, 1988, 211, 299. 4 Al-Wehaid, A., and Townshend, A., Anal. Chim. Acta, 1987, 198, 45. 5 Fifield, F. W., and Kealey, D., Principles and Practice of Analytical Chemistry, ITC, London, 2nd edn., 1983, p. 26. 6 Davies, D. G., and Kelly, T. V. G., Experimental Physical Chemistry, Mills and Boon, London, 1967, p. 15. 7 Monks, C. D., B.Sc. Project Report, The Liverpool Poly- technic, 1992. Paper 2105495G Received October 14, 1992 Accepted December 3, I992

ISSN:0003-2654    

DOI:10.1039/AN9931800623

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ANALYST, JUNE 1993, VOL. 118 627 Flow Injection Chemiluminometric Determination of Epinephrine, Norepinephrine, Dopamine and L-Dopa" Nikolaos T. Deftereos, Antony C. Calokerinost and Constantinos E. Efstathiou Laboratory of Analytical Chemistry, University of Athens, Panepistimiopolis, 757 77 Athens, Greece A method is proposed for the determination of 0.0500-1 .OO pg ml-l of epinephrine and L-dopa and 0.100-1 .OO pg ml-l of norepinephrine and dopamine by their chemiluminogenic oxidation with potassium permanga- nate in acidic medium, in the presence of formaldehyde, which greatly improves the sensitivity. Flow injection allows the measurement of 80 solutions per hour. The method was also optimized for a continuous-flow system. Comparative results from numerous organic compounds proved the necessity for electron-donating groups on the benzene ring for sensitive chemiluminescent characteristics.Keywords: Chemiluminescence; flow injection; continuous flow; epinephrine; catecholamines Epinephrine (EP) , norepinephrine (NE), L-dopa (LD) and dopamine (DP) are the best known catecholamines, which constitute a group of compounds with an alkylamine chain attached to a benzene ring bearing two hydroxy groups. Epinephrine is biosynthesized in the adrenal medulla and sympathetic nerve terminals. Tyrosine is the starting com- pound and LD, DP and NE are the intermediates of this formation route. Epinephrine and NE stimulate glycogen breakdown in muscle and liver by increasing the amount of glucose released into the blood by the liver and decreasing the consumption of glucose by the muscles.Catecholamines are also important neurotransmitters. Catecholamines are being used for the treatment of neural disorders such as Parkinson's disease. Numerous methods have been developed for the determi- nation of these biologically important compounds. Most are based on oxidation prior to the analytical measurement. Typical oxidants are sodium periodate,2 potassium bromate,3 N-bromosuccinimide and other brominating agents,4 ammo- nium metavanadate5 and the 1 ,lo-phenanthroline-iron(1u) complex,6 and the final products are determined by spectro- photometry. Spectrofluorimetry has also been used after oxidation by 2-cyanoacetamide7 or by post-column derivatiza- tion with glycylglycine after liquid chromatographic separa- tion .s High-performance liquid chromatography (HPLC) with electrochemical detection has been proposed for the determi- nation of catecholamines in human cerebrospinal fluid9 and plasma.10 Few chemiluminometric methods have been investigated for the determination of catecholamines. Micelle-enhanced lucigenin chemiluminescence (CL) has been proposed for the determination of EP, NE and DP.11 The oxidation of EP by dissolved oxygen in alkaline solutions has been found to exhibit CL, which is enhanced by the use of dioctadecyl- dimethylammonium chloride bilayer membrane vesicles in the presence of manganese(I1) as catalyst.12 Yamada and co- workers investigated various common inorganic oxidants as potential CL reagents for catecholamines and related com- pounds'3 and proposed permanganate for the chemilumi- nometric determination of catecholamines.14 Epinephrine, NE and DP have been determined after liquid chromato- graphic separation, derivatization with dansyl chloride and measurement of the CL generated by mixing the eluate with bis(2,4,5trichlorophenyl) oxalate-hydrogen peroxide reagent.15 In this work, the effect of various experimental parameters on the emission intensity from the oxidation of catecholamines * Presented at SAC '92, an International Conference on Analytical + To whom correspondence should be addressed. Chemistry, Reading, UK, September 20-26, 1992. by potassium permanganate was investigated. The intensity is dramatically increased by the presence of formaldehyde in the final reaction mixture, allowing the determination of very low levels of catecholamines.An attempt is made to correlate the relative emission intensities from numerous aromatic com- pounds with structures similar to that of catecholamines. Experimental Reagents All solutions were prepared from analytical-reagent grade materials with distilled, de-ionized water. A stock solution of 0.1OOO mol 1-l permanganate was prepared by dissolving 1.580 g of potassium permanganate in water and diluting to 100 ml with water. The solution was kept in the dark. Stock solutions (100 pg ml-l) of EP (Serva) were prepared by dissolving 0.100 g in 1 1 of 0.20 moll-' sulfuric acid. Stock solutions of LD (Serva), DP hydrochloride (Serva) and NE (Sigma) were prepared by dissolving 0.100 g of each catechol- amine in water and diluting to 1 1 with water.Dissolution of NE was aided by sonication. More dilute solutions were prepared daily by the minimum number of dilution steps possible. All other chemicals were of the best grade available and were used without further purification. Apparatus A schematic diagram of the automated flow injection (FI) analyser used is shown in Fig. 1. It consisted of two basic units, the detector housing and the flow system. The flow manifold included a peristaltic pump (Technicon proportioning pump 111), which pumped both reagent solu- tions at different flow rates through poly(tetrafluoroethy1ene) (PTFE) flow tubes. The two streams were mixed in a Y-shaped mixing element positioned 20 mm before the flow-cell inlet. The sample was injected into the carrier stream by a 500 pI loop of a low-pressure PTFE injection valve operated by a Rheodyne Model 5701 pneumatic actuator controlled by two three-way solenoid air valves.The detector flow cell is a flat spiral consisting of 3.5 turns of glass tubing (i.d. 2 mm, total length 90 mm) with a total height of 22 mm and a volume of 300 pl. The flow cell is placed in front of the photomultiplier tube (PMT) at a distance of 2 mm from the photocathode and is backed by a self-adhesive plane mirror for maximum light collection by the PMT. The PMT (EM1 9783R, S-5 response) and the coil were housed in a laboratory-made light-tight unit. High voltage (-750 V) was supplied to the PMT (cathode luminous sensitivity S = 67628 ANALYST, JUNE 1993, VOL. 118 PMT j I -Recorder Hig h-voltage unit Fig.1 tion of catecholamincs Schematic diagram of the FI manifold used for the determina- FA lm-I) by the high-voltage unit of a Hitachi Perkin-Elmer Model 139 ultraviolet (UV)/visible spectrophotometer. The current output of the PMT is fed to an ilV converter circuit consisting of a CA3 140 operational amplifier, equipped with offset and damping control. The currcnt output (in the range 0-1 V) of this circuit is fed to an analogue-to-digital converter (ADC) circuit (National ADC0820 8-bit converter) interfaced to an Amstrad CPC-6128 microcomputer (microprocessor: Zilog 280 at 4 MHz). Under computer control are also both air valves of the pneumatic actuator, which operates the injection valve. A program written in BASIC supervises the whole measurement sequence, undertakes the data acqui- sition and the presentation of analytical signals.The i/V converter was also connected to a multi-speed variable-span recorder (Knauer, Model 73341; full-scale deflection 0.35 s). Damping was provided by inserting an RC circuit between the converter and the recorder. The FI analyser was also used as a continuous-flow (CF) analyser by de-activating the pneumatic actuator and intro- ducing the analyte solution through the carrier stream until the steady state was reached. A Hitachi U-2000 spectrophotometer with 10 mm quartz cells was used for scanning absorption spectra, when required. Procedure The instrument was set at the optimized conditions shown in Fig. 1, except that the injection valve was kept in the 'wash' position until the baseline on the recorder had been estab- lished.For CF measurements the analyte solution was introduced into the manifold until the steady state was reached. For F1 measurements, the time intervals selected for sampling and washing with pure carrier solution were 10 and 35 s, respectively. The injection valve was activated and the analysis proceeded automatically. A calibration graph of emission intensity ( I , mV) versus2oncentration of the desired catecholamine (c, pg ml-I) was constructed and the content of each compound in the sample solution was determined. A standard solution should be included after every 12 sample solutions. Results and Discussion Preliminary Work By using the CF system, various preliminary experiments were performed.No emission was detected after mixing aqueous EP solutions with solutions of cerium(Iv), dichromate, bro- mate, peroxodisulfate and ammonium metavanadate dis- solved in sulfuric or perchloric acid. There was also no emission observed after mixing alkaline solutions of EP with alkaline solutions of hexacyanoferrate(r1r) , hydrogen peroxide. and hypochlorite. N-Bromosuccinimide was also tried as its chemiluminogenic properties have been established recently,l6 but no emission was recorded. No radiation was detected even after stopping the pump and allowing each reaction mixture to stand for 5-10 min in front of the PMT. All preliminary observations were also confirmed with the FI system. When EP was mixed with acidic solutions of potassium permanganate, an intense emission was recorded, confirming the observation of Ikkai et af.14 Therefore, the only oxidant that generated CL with EP under the experimental conditions used was potassium permanganate.Potassium permanganate is a strong inorganic oxidant widely used in CL studies despite the disadvantage of giving coloured solutions that absorb the emitted radiation. The chemiluminogenic superiority of potassium permanganate over other common oxidants has been demonstrated for the determination of sulfite'7 and morphine's and can be attributed to the energy released by the transfer of five electrons per molecule of oxidant during the redox reaction. By using potassium permanganate, the method was evalu- ated with the CF and FI systems. The parameters studied were potassium permanganate and E P flow rates, nature and concentration of the acid and oxidant concentration.Effect of Flow Rate The effect of flow rate on the emission intensity from 0.0010 rnol I-' potassium permanganate in 1.4 rnol 1-1 perchloric acid was studied with the CF system. The optimum values for reagent and sample (10.0 pg ml-l of EP) flow rates were 0.60 and 3.40 ml min-l, respectively, and allow about a 1 s time interval between mixing of solutions and measurement of the emitted radiation. The optimum values for reagent (7.0 X rnol I-' potassium permanganate) and carrier (0.50 mol 1- sulfuric acid) flow rates with the FI system were 2.90 and 3.90 ml min-I, respectively, and allow about a 0.6 s time interval between mixing of reagents and measurement of the emission intensity.Effect of Potassium Permanganate and Acid Concentration The effect of the concentration of potassium permanganate dissolved in sulfuric acid on the emission intensity from 10.0 pg ml-1 of EP by the CF system is shown in Fig. 2. Similar results were obtained by using 5.00 pg ml-' of EP and also with perchloric acid in the range 0.010-2.0 mol I-' instead of sulfuric acid. The optimum concentrations of oxidant in 0.50 mol I-' sulfuric acid and 1.0 rnol I-' perchloric acid do not differ significantly (7.0 x and 5.0 x lo-'? rnol I-', respectively). Sulfuric acid was chosen for all further work. In the F1 system, the oxidant and the acid solution are introduced separately into the manifold. The optimum time intervals for sample and wash were established as 10 and 35 s, respectively, and these time intervals allow the measurement of 80 solutions per hour.The results for the effect of the sulfuric acid concentration on the emission intensity from EP using the FI system are shown in Fig. 3. The optimum concentrations are 5.0 x 10- rnol I-' for permanganate when the carrier is 1.0 rnol I-' sulfuric acid. The intensity is reduced by 8,20,50 and 45% if the same concentrations of perchloric, orthophosphoric, hydrochloric or ace tic acid are used instead of sulfuric acid. Effect of Emission Enhancers It is well known that the CL intensity from various reactions can be enhanced by energy transfer procedures or by some sensitizers by an unknown me~hanism.1~ Therefore, it was decided to investigate the effect of common emission en-ANALYST, JUNE 1993, VOL.118 620 -4 -3 -2 Log([KMn041/mol I-') Fig. 2 Effcct of potassium pcrmanganate concentration on the emission intensity from 10.0 pg ml- of EP at: A, 0.0050; B , 0.050; C, 0.50; D, 1 .O; and E, 2.0 rnol 1 - l of sulfuric acid by the CF technique 100 1 In C a, C C 0 In *-' .- +- .- .- .- : a, ,z 50 m a, I1I - B I -x-x- c I 0 1 .o 2.0 [H,SO,]/mol 1-1 Fig. 3 Effect of sulfuric acid concentration on the emission intensity from 20.0 pg ml-' of EP at: A, 1.0 x C, 7.0 x 10V; D, 1.0 x E. 2.0 x and F, 5.0 x lo-? rnol I-' of potassium permanganate by the F1 tcchnique B, 5.0 x hancers on the CL oxidation of EP by potassium permanga- nate using the Fl and CF systems. Initially 0.0010 mol 1-1 of 3- cyclohexylaminopropanesulfonie acid (CAPS) was tried, as this compound amplifies the radiation from the CL oxidation of sulfite by potassium permanganate20 and cerium(iv).21 However, no effect from CAPS on the CL reaction studied was detected (Table 1).This observation indicates that CAPS can only be used as an enhancer when the CL reaction involves inorganic reactants. If at least one reactant is an organic molecule, then CAPS has no effect, and this observation has been verified for other CL reactions. On the other hand, 0.0010 rnol I-' of enhancers acting by an energy-transfer mechanism, such as quinine,22 were found to react with potassium permanganate immediately after mixing the solu- tions. By using the FI system and adding the potential enhancers to the acid solution, no significant increase in the emission intensity was observed (Table 1).The effect of common micellar systems on the emission intensity was also investigated, but cetyltrimethylammonium bromide, Tween-80 and cetylpyridinium chloride were found to react with permanganate in acidic solutions and sodium lauryl sulfate had no effect on the emission intensity under the experimental conditions used. Cyelodextrins have also been used for enhancing the intensity from various CL reactions23 but were found to have no effect on the reaction studied (Table 1). Effect of Formaldehyde as an Enhancer During the study of the effect of common pharmaceutical additives on the emission intensity from EP by the action of permanganate, it was realized that injections contain sulfite as antioxidant and therefore the anion severely interferes with the determination of EP.An attempt was made to reduce the interference by adding an aldehyde or a cyclic ketone to form the corresponding 0-hydro-C-sulfonato addition product. Cyelohexanone was initially tried but as the reaction is reversible in acidic medium,24 sulfite still interfered with the determination. When 0.50 rnol I-' of formaldehyde was added to the EP solution, it was noticed that the intensity was about ten times greater than that from pure EP of the same concentration. The reaction of permanganate with formaldehyde exhibited weak CL but the emission from the mixed EP-formaldehyde solutions was more intense than the sum of the individual emission intensities. However, formaldehyde was not capable o f eliminating the positive interference of sulfite on the determination of EP.The effect of formaldehyde on the determination of EP was studied further in order to investigate the mechanism of the enhancement. lnitially it was thought that the effect might be due to energy release during formation of carbon dioxide by oxidation of formaldehyde, in a mechanism similar to that for dioxetanone formation .25 Other organic compounds that react with permanganate and produce carbon dioxide were also studied. The results (Table 1) show that formic acid enhances the emission intensity but not as strongly as formaldehyde, while the effects of methanol and oxalic acid are much weaker. Although the enhancing effect of formaldehyde and formic acid on the permanganate-EP CL reaction is not fully understood, it was decided to investigate these enhancers further in order to improve the sensitivity of the measurement. The effect of formaldehyde and formic acid concentrations on the determination of EP is shown in Fig.4. The optimum formaldehyde and formic acid concentrations were 0.80 and 1.00 rnol 1-I, respectively. Formaldehyde was used in all further studies, as its enhancement of the emission intensity is greater than that given by formic acid. Analytical Parameters Fig. 5 shows typical recordings for a series of EP standards obtained using the proposed procedure. All other catechol- amines tested gave similar recordings. Table 2 summarizes the analytical parameters for EP, NE, LD and DP with and without formaldehyde as CL enhancer.The relative standard deviations for 0.500 vg ml-l of EP and 0.300 vg ml-I of NE, LD and DP were 1.2, 1.1, 1.2 and 0.5%, respectively (n = 6). Interference Studies In order to assess the possible analytical applications of the CL method described above, the effect of some common exci- pients used in pharmaceutical preparations was studied by analysing synthetic sample solutions containing 2.00 and 20.0 pg ml-I of E P and excess amounts of each excipient, without formaldehyde as enhancer. The undissolved material, if any, was filtered before measurement. The results (Tablc 3) show that sulfite is the only severe interferent. When 0.80 moll - I of formaldehyde was used as an enhancer, no interfering effect was observed from 10-fold excesses of sodium sulfite on 0.100 and 1.00 pg ml-I of EP.The interference was reduced mainly because the presence of formaldehyde in the final solution increases the sensitivity and allows extended dilution of the analyte solution so that the concentration of interferent is reduced below the tolerable level.630 ANALYST, JUNE 1993, VOL. 118 Table 1 Effect of various compounds commonly used as potential CL enhancers on the relative emission intensity from 1 0 . 0 or 1.00 pg ml-' of EP (taken as 100) Compound Cyclomaltohexaose (wcyclodextrin) Cyclomaltoheptaose (p-cyclodextrin) Quinine CAPS+ CTAB* SDSS CPCT Tween-80 Formaldehyde Formic acid Methanol Acetic acid Oxalic acid Concentration/ mol I-1 0.010 0.010 0.010 0.020 0.010 0.010 0.010 0.020 0.0010 0.0010 0.0130 0.0810 0.0120 0.0130 0.80 1.40 0.50 0.10 0.40-1.6011 Added to the stream of" Acid Oxidant Both Acid Acid Oxidant Both Acid Acid Acid Acid Acid Acid Acid Analyte Analyte Anal yte Analyte Analytc Relative emission intensity (%) 10.0 pg ml-' 1.00 pgml-I 107 SO4 so1 118 107 103 114 105 101 109 114 99.6 99.6 85.4 - - - - - * 5.0 X 10k4 mol 1-l potassium permanganate and 1.0 mol 1-' sulfuric acid are the oxidant and carrier stream, respectively t 3-Cyclohexylaminopropanesulfonic acid.* Cetyltrimethylammonium bromide. 3 Sodium lauryl sulfate. 11 Cetylpyridinium chloride. 11 Emission was independent of concentration within this range. 100 >- v) a, 4- .- .I- .- C 0 v) ." 50 .- : a, .- .I- - a, U 0 0.5 1 .o 1.5 [HCHOI or [HCOOH]/mol I-' Fig. 4 Effect of formaldehyde (solid lines) and formic acid broken and C, the corresponding blank emission by the FI technique.(The concentrations of potassium perman anate and sulfuric acid were 5.0 x 10k4 and 1.0 mol 1-I, respectively7 lines) on the emission intensity from: A, 0.500; B, 1.00 pg ml- (I of EP; Recovery of EP and LD From Pharmaceutical Formulations Recovery experiments on solutions from EP ampoules (Chropi, Athens) and the LD formulation Madopar (Roche) were performed using the proposed FI CL method. The recoveries of 2.00 and 4.00 pg ml-l of EP from solutions made from EP injections (1 mg of EP per injection) were 90.5 and 92.5%, respectively. The recovery of 0.200 and 0.400 pg ml-l of LD from solutions made from Madopar tablets (200 mg of LD and 50 mg of benserazide per tablet) was 115%. Benserazide does not interfere severely owing to the extended dilutioa of the solution.Chemiluminogenic Characteristics of Related Compounds The results obtained from other organic compounds studied by using the experimental conditions for catecholamines t > v) c C C 0 fn .I- .- .I- .- .- .- E w 3.00 I Time - Fig. 5 Typical recorder outputs for a series of EP standards under the proposed conditions by the FI technique (numbers above the peaks arc concentrations in pg ml-l) without the addition of formaldehyde are summarized in Table 4. From the results, some general conclusions can be drawn, as follows. The introduction of electron-withdrawing groups in phenol (o-nitrophenol, salicylic acid) reduces the CL intensity. The observation is confirmed by the presence of a carboxyl group in gallic acid, which therefore chemiluminesces less than pyrogallol upon oxidation.The introduction of electron-donating groups in phenol (o- aminophenol) increases the CL intensity. This has also been observed during the investigation of the CL oxidation of acetaminophen by cerium(1v) .26 The introduction of a second hydroxy group at the para (hydroquinone) and ortho (pyrocatechol) position generates more intense emission than that at the meta (resorcinol) position owing to activation of the benzene ring. The sensitivity does not change significantly when the hydroxy groups of pyrocatechol are replaced with amino groups (o- phenylenediamine). In fact, when an organic compoundANALYST, JUNE 1993. VOL. 118 63 1 Table 2 Analytical characteristics for the determination of EP, NE, LD and DP in the presence of 0.80 moll-' of formaldehyde by the proposed method (numbers in parentheses are corresponding values without formaldehyde) Regression line7 Linear range/ LOD*/ Compound pg ml-' pgml-l Slope Intercept rt EP NE LD DP 0.050&1.00 0.100-1.00 0.050cL1.00 0.100-1 .oo (3.00-50. 0) (2.00-60.0) (2.0060.0) (2.0S30.0) 0.0300 (1.47) 0.0500 (1.95) 0.0300 (1.17) 0.0400 (2.90) 73.2 (3.20) 94.8 (2.68) 79.3 (1.75) 105 (2.89) -0.300 (-0.270) 3.10 -0.600 (-0.765) 1 .so (-2.59) (2.26) 0.9998 (0.9998) 0.9994 (0.9991 ) 0.9991 (0.9997) 0.9998 (0.9993) * Limit of detection (without formaldehyde, analyte concentration giving a signal equal to the intercept plus three times the standard deviation of the regression line; with formaldehyde, analyte concentration giving a signal equal to the blank emission plus three times its standard deviation).t Emission intensity (mV) versus concentration (pg mlkl). * Correlation coefficient. 9 Number of measurements. contains two hydroxy groups or two amino groups or one of each in the ortho or para position, the corresponding benzoquinones are easily formed.27 This observation has been used for the determination of catecholamines and other organic compounds by oxidation with bromine .28 The introduction of a third hydroxy group in resorcinol (phloroglucinol) is accompanied by an increase in CL intensity. The radiation is more intense when the three hydroxy groups are adjacent (pyrogallol) . The introduction of an alkylamine chain on C-5 of pyrocatechol to form the corresponding catecholamine acti- vates the benzene ring considerably.L-Dopa is the least sensitive of the catecholamines studied owing to the presence of the electron-withdrawing group on the alkylamine chain. Nature of Emitting Species The increase in emission intensity observed in the presence of formaldehyde during the oxidation of catecholamines by potassium permanganate is similar to the increase in the CL emission when a polyhydric phenol is added to an alkaline chemiluminogenic mixture of hydrogen peroxide and formal- dehyde, known as the Trautz-Schorigin reaction. The reac- tion has been studied thoroughly with gallic acid and the mechanism proposed proceeds via the formation of a semi- quinone intermediate and excited molecular oxygen species (excimers) are finally generated.29 The emission can arise either from the de-activation of the excimer oxygen species or by energy transfer to a fluorophore.Similarly, pyrogallol has been found to generate CL when oxidized by periodate in the presence of hydroxylamine at pH 7.5-8.2.30 The possibility for the reaction reported in this paper to involve active oxygen was studied by measuring the CL intensity before and after de- aeration of all solutions, but no change was detected. Catecholamines are oxidized to aminochromes by a large number of inorganic oxidants, including potassium permanga- nate. The first stage of the reaction with EP is the oxidation of the catecholamine to the open-chain quinone (adrenoqui- none), followed by cyclization to leucoadrenochrome, which is then oxidized to adrenochrome.31 The reaction is usually very rapid.32 Leucoadrenochrome exhibits green fluores- cence.33 Therefore, the CL reaction probably proceeds through the formation of excited molecular oxygen species, which then excites leucoadrenochrome by energy transfer.Furthermore, bleaching of the final solution also occurs. This was verified by recording the absorption spectrum of potas- sium permanganate with EP with and without formaldehyde at concentrations equal to those in the measuring coil. The results showed that the absorption of permanganate at 525 nm did not change even 5 min after mixing with 0.0600 pg ml-l of Table 3 Recovery of 2.00 and 20.0 pg m1-l of EP from solutions with various additives used as excipients in the absence of formaldehyde Concentration ratio Additive (additive : EP) Galactose 10 5 1 Fructose 10 5 Sorbitol 10 5 1 Sugar 10 5 Carbowax* 10 CAHPt 10 5 Talc 10 Gelatine 10 Starch 10 Sodium lauryl sulfate 10 Magnesium stearate 10 5 NaCl 10 Na2S03 1 Carbopolt 10 0.5 * Polyethylene glycol 4000. t Cellulose acetate hydroxyphthalate.* Carboxypolymethylene. Recovery (%) ( n = 4) 2.00 pg ml-l 100 100 100 104 100 127 109 104 127 100 104 104 118 109 104 104 100 109 100 110 95.6 87.0 95.6 20.0 pgml-l 114 110 101 113 105 127 116 100 104 101 104 100 98.0 95.0 96.7 99.2 1 04 106 106 105 99.1 80.0 89.6 EP. When 0.10 moll-' of formaldehyde is also present, then the absorption is reduced by about 60% in 90 s after mixing. Conclusions The CF technique is simpler than the FI technique and is very useful for preliminary and comparative studies. Optimum concentration values for both systems do not differ signifi- cantly. From the analytical point of view, the FT technique is more useful than the CF technique owing to the higher sample throughput and the ease of automation of the analytical system.The proposed method is fully automated, very simple, fast and sensitive. The results show that it will be possible to extend the method to other similar compounds with more electron-donating groups than the four catecholamines studied. The sensitivity is greatly increased by the presence of632 ANALYST, JUNE 1993, VOL. 118 Table 4 Relative emission intensity and analytical characteristics from various organic compounds by the action of acidic permanganate in the absence of formaldehyde REI* from Regression line$ 2.72 x 5.45 x 10k5 Linearrangel LODl/ Compound moll mol I - 10- 5 moll 1 10-5 moll- 1 Slope ( x 104) Intercept r+ n1l EP NE DP LI> Hydroquinone Pyrogallol Phloroglucinol Pyrocatechol Rcsorcinol o-Phenylenediaminc Gallic acid o-Aminophenol Phenol Salicylic acid o-Nitrophenol Tyrosinc 100 115 131 88.5 165 157 123 85.7 75.0 80.0 42.8 26.9 11.5 8.30 ND ND 100 113 122 162 175 137 88.9 82.1 85.7 82.0 42.8 30.4 9.20 9.60 ND ND 1.10-33.0 1.20-36.0 1.60-26.0 1.00-15.0 1.80-27.0 1.60-32 .0 1.20-18.0 1 .00-36.0 3.00-36.0 1 .00-30.0 0.60-20.0 Weak emission Weak emission Weak emission - - ' Kclative emission intensity (ND = not detected).I Limit of dctection. 1 Emission intensity (mV) versus concentration (mol l - I ) + Correlation coefficient.11 Number o f measurements. 0.80 1.15 1.53 0.60 1.37 1 .oo 0.80 0.50 0.63 0.90 0.50 - - - - - 58.6 45.4 54.9 34.5 71.6 59.5 62.1 38.5 54.0 29.1 18.3 - - - - - -0.273 2.30 -2.60 -0.77 -3.50 0.90 -2.70 - 1.38 -8.10 1 .oo -0.60 - - - - - 0.9998 0.9991 0.9993 0.9997 0.9995 0.9994 0.99990 0.99993 0.99990 0.999 1 0.9996 - - - - - 6 11 9 8 8 7 8 9 6 9 8 - - - - - formaldehyde in the final reaction mixture and it would be possible to increase the selectivity of the proposed preliminary method by HPLC separation before CL detection. This work was supported in part by a research grant from the Royal Society of Chemistry for the purchase of the injection valve with the pneumatic actuator. The authors thank C. Polydorou for invaluable assistance in developing the com- puter interface.1 2 3 4 5 6 7 8 9 10 I I 12 13 References Strycr, L . , Biochemistry, Freeman, San Francisco, New York, 3rd edn., 1988. 61-Kommos, M. E., Mohamed, F. A. and Khedr, A. S . , Taluntu, 1990. 37. 625. Mohamed, W. I.. and Salem, F. B., Anal. Lett., 1984, 17, 191. Abou O u f , A . , Wala\h, M. I . , and Salem, F. B., Analyst, 1981, 106. 949. Salem, F. H . , 7alanta. 1987, 34, 810. Carmona, M., Silva, M., and Pkrez-Bendito, D., Analyst, 1991, 116. 1075. Honda. S . , Araki, Y.. Takahashi, M., and Kakehi, K., Anal. Chim. Acta, 1983, 149, 297. Seki, T.. Yanagihara, Y., and Noguchi, K., J. Chrornatogr., I99O,SlS, 435. McClintock, S . A.. and Purdy, W. C., Anal. C'him. Acta, 1984, 166, 171. Musso, N . R., Vergassola.C.. Pende, A., and Lotti, G.. J . Liq. C'Iiromutogr., 1991. 14. 3695. Kamidate, T., Yoshida, K . . Kaneyasu, T., Scgawa, T., and Watanabe. H., Anal. Sci., 1990, 6, 645. Matsuc, K., Yamada, M., Suzuki, T., and Hobo, T., Anal. Lett. . 1 989, 22, 2445. Nakagama, T., Yamada, M., and Suzuki, S . . Anal. Chim. Acta, 1989, 217, 371. 14 15 16 17 18 19 Ikkai, H., Nakagama, T., Yamada, M., and Hobo. T., Bull. Chem. Soc. Jpn., 1989, 62, 1660. Xie, G.? and Dai, J . , Huaxue Tonghao, 1987, 5, 33; Anal. Abstr.. 1988. 50, 1D172. Halvatzis, S. A., Timotheou-Potamia, M. M., and Calokcrinos, A. C., Analysr, 1990, 115, 1229. Stauff, J . , and Jaeschke, W., Atmos. Environ., 1975, 9, 1038. Abbott, R. W., Townshend, A., and Gill, R., Analyst, 1986, 111, 635. Koukli, I. I . , Sarantonis, E. (3.. and Calokerinos, A. C., Anal. Lett., 1990, 23. 1167. 20 21 22 23 24 25 26 27 28 29 30 31 32 33 Al-Tamrah, S. A . , Townshend, A . , and Wheatley, A. R.. Anulyst, 1987, 112, 883. Koukli. 1. I . , Sarantonis, E. G., and Calokerinos. A. C., Anulyst, 1988, 113, 603. Koukli, I. I . , and Calokerinos, A. C., Anal. Chim. Acfa, 1990, 236, 463. Malehorn, C. L., Riehl, T. E . , and Hinze, W. L . , Analyst, 1986, 111. 941. Young, P. R., and Jcncks, W. P., J . Am. Chem. Soc., 1978,100, 1228. Birks, J. W., Chemiluminescence and Photochemical Reaction Detection in Chromatography, VCH, New York, 1989. Koukli, I. I . , Calokcrinos, A. C., and Hadjiioannou, 'r. P., Analyst, 1989, 114, 71 I . Kolthoff. I. M., and Belcher, R., Volumetric Analysis. Intcr- scicnce, New York, 1957, vol. 111. Amin, D., Analyst, 1986, 111, 255. Slawinska, D., and Slawinski, J., Anal. Chem., 1975, 47, 2101. Evmiridis, N. P., Analyst, 1987, 112, 825. Heacock, R. A . , and Powell, W. S . , Prog. !Wed. Chem., 1973,9. 275. Heacock, R. A., Adv. Heterocycl. Chem., 1965, 5, 206. Harley-Mason, J . , -1. Chem. Soc., 1950, 1276. Paper 2104061 A Received July 29, 1992 Accepted October 8, 1992

ISSN:0003-2654    

DOI:10.1039/AN9931800627

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ANALYST, JUNE 1993. VOL. 118 633 Continuous-flow Chemiluminometric Determination of Tetracyclines in Pharmaceutical Preparations and Honey by Oxidation with A/-Bromosuccinimide" Stergios A. Halvatzis, Meropi M. Timotheou-Potamiat and Antony C. Calokerinos Laboratory of Analytical Chemistry, University of Athens, Panepistimiopolis, Zogra fou, 157 71 Athens, Greece A continuous flow chemiluminometric method for determining 0.050-3.00 pg cm-3 of tetracycline, 0.50-5.00 pg ~ m - ~ of oxytetracycline, 0.50-7.00 pg ~ m - ~ of doxycycline and chlorotetracycline and 0.30-3.00 pg ~ m - ~ of demeclocycline in pharmaceutical preparations and honey is described. The method is based on the chemiluminescence produced by the action of N-bromosuccinimide on tetracyclines in alkaline solution. The emission intensity is greatly enhanced by the presence of ammonia. The procedure is automated and solutions can be analysed at a rate of 130 h-I with a relative error of about 2%.During evaluation of possible interferences of the method, recoveries from solutions with common excipients and other concomitant compounds were in the ranges 97.8-1 04.0 and 96.6-1 09.4%, respectively. Recoveries of various tetracyclines from commercial formulations and honey samples were in the ranges 95.2-1 03.7 and 89.2-1 06.6%, respectively. The results obtained for the assay of commercial pharmaceutical preparations compared well with those obtained by an official method and demonstrated accuracy and precision (<5%). Keywords : Ch e rn ilu m in esce nce; N - 6 rom osuccin irn ide; tetracycline; p h a rrn a ce u tical preparation; honey Tetracyclines are widely used as bacteriostatic and antibiotic drugs.Tetracycline (TC) and its derivatives oxytetracycline (OTC), chlortetracycline (CTC), doxycycline (DC) and demeclocycline (DMCC) have been employed extensively in veterinary medicine, animal nutrition and feed additives. However, many health authorities do not allow antibiotic residues in foods because of allergic reactions, particularly in hypertensive people. The majority of the methods described for the determina- tion of tetracyclines are based on their ability to form coloured metal ion complexes. 1 - 4 Spectrophotometry can also be used after oxidation with ammonium vanadate5 or sodium hexani- trocobaltate(r1i) .h Other methods use derivative spectropho- tome try ,7 fluorime try ,8-10 derivative fluorime try ,I1 ion-selec- tive electrodes,12 differential-pulse polarography,l3 absorptive stripping voltarnmetry14 as well as diode array's and ultraviolet16 detection after high-performance liquid chromatographic separation.Chromatographic techniques in combination with fluorescence17 or electrochemically pre- treated glassy carbon electrodes'* have also been reported. Chemiluminometric (CL) methods can be successfully used for the determination of tetracyclines. Tetracycline has been determined by its chemiluminescent reaction with bromine at pH 10.4'9 or after pre-oxidation with potassium peroxodisul- fate and with further mixing of the reaction products with 1,3- dibromo-S,5-dimethylhydantoin.~() Tetracyclines have also been determined by their chemiluminescent reaction with hexanocyanoferrate(ii1) after acidic degradation to the corre- sponding anhydro derivatives." Some of the above methods7,15.16 have been applied to the determination of tetracyclines in honey.However, no CL method has been proposed for this particular application. This work describes the development of a continuous flow CL method for the determination of tetracyclines in phar- maceutical preparations and honey without any pre-treatment of the samples. * Presented at SAC '92, an lnternational Conference on Analytical + To whom correspondence should be addressed. Chemistry. Rcading. UK, Scptcmbcr 20-26, 1992. Experimental Apparatus The continuous flow CL analyser pump system and sampler used were as previously described.22 The photometric unit of the chemiluminometer included a coiled flow cell of 3.5 turns of glass tubing (2 mm i.d.) and total height and volume equal to 22 mm and 300 pl, respectively.The coil was situated in front of the photomultiplier tube (EM1 9783R, S-5 response) operated at -720 V. The output of the photometric unit was recorded with a Knauer recorder (Model 73341). The solutions of reactants were supplied by a Technicon propor- tioning pump 111 and were mixed at a Y-junction, 20 mm before entering the flow cell. The final solution was carried into the flow cell by a Tygon tube of 2 mm i.d. Samples were supplied to the manifold by a Technicon Sampler TI with a 40- sample capacity. Reagents All solutions were prepared from analytical-reagent grade materials with de-ionized, distilled water.A stock solution of 0.050 mol dm-3 N-bromosuccinimide (NBS) solution was prepared daily by dissolving 2.225 g of NBS (Serva) in water and diluting with water to 250 cm3. Stock tetracycline solutions (100 pg cm-3) were prepared by dissolving 0.100 g of each tetracycline in water and diluting to 1 dm3 with water. Tetracycline, OTC, and CTC were supplied by Serva, and DC (83.8%) and DMCC (91.2%) by Chropi. All tetracyclines were in the form of their hydrochloride salts. All other chemicals used were of the best grade available and were used without further purification. Procedures Measurement procedure Initiate the instrument under the optimized conditions of 3.90 cm3 min-l for the sample and 2.50 cm3 min-' for the NBS reagent.Keep the sampling needle always in the 'wash' position until the baseline on the recorder has been estab- lished. Adjust the sampler to allow 15 s for sample and 18 s for634 ANALYST, JUNE 1993, VOL. 118 wash water to enter the manifold. After activation of the sampler, the analysis proceeds automatically. Construct a calibration graph of emission intensity (UmV) versus concen- tration (c/pg ~ m - ~ ) for TC, OTC, CTC and DMCC or log I versus log c for DC and determine the content of each compound in the sample solution. Include a standard solution after every 12 sample solutions. Procedure for the Determination of Tetracyclines in Pharmaceutical Preparations Sample preparation Tablets. Not less than 20 tablets were weighed and finely powdered.A sample equivalent to approximately 200 mg of analyte was weighed accurately, transferred into a 1 dm3 calibrated flask and diluted to volume with water. The powder was sonicated for 10 min to aid dissolution and then filtered. Working solutions were prepared from this stock solution by appropriate dilution so that the final analyte concentration was within the working range. Capsules. Ten capsules were weighed, emptied and the procedure for tablets followed. Powder for external use. A portion was accurately weighed and diluted with water so that the final tetracycline concentration was within the working range. Procedure for the Determination of Tetracyclines in Honey Transfer an accurately weighed portion of about 2.5 g of tetracycline-free honey sample into a 250 cm3 calibrated flask and dilute to volume with water (aqueous honey solution).Construct the calibration graph for TC or DMCC or OTC by transferring 5.00 cm3 of aqueous honey solution into a calibrated flask with the appropriate volume of stock solution of TC and diluting to 100 cm3 with water. All final solutions for the determination of tetracyclines in pharmaceutical preparations and honey should also contain 0.030 rnol dm-3 of sodium hydroxide and 8.0 X loh4 mol dm-3 of ammonia, except for OTC, which requires 3.0 X rnol dm-3 of ammonia, and DMCC, which requires 0.30 mol dm-3 of sodium hydroxide. Results and Discussion Preliminary Work Tetracyclines possess a hydronaphthacene skeleton with a wide variety of functional groups.These molecules are A 0.05 0.1 cNaOH/mol dm 3 Fig. 1 Effect of sodium hydroxide concentration on the chemi- luminescent emission intensity from 2.00 (broken line) and 5.00 (solid line) pg ~ m - ~ of TC at: A, 5.0 x B, 8.0 X and C, 0.001 mol dm-3 of ammonia. [NBS] = 0.015 mol dm-3 converted to the corresponding iso- or anhydro derivatives by alkaline or acidic degradation, respectively.23.24 The degrada- tion products are fluorescent compounds25 and were found to chemiluminesce during oxidation by NBS in alkaline medium. The nature of the emitting species is not yet clear. No emission was detected on recording the intensity immediately after mixing aqueous acidic TC solutions with NBS. The most intense radiation was obtained when tetracyclines were degraded in 0.050 rnol dm-3 sodium hydroxide.By using this alkaline medium , various approaches were considered in order to increase the emission intensity (Table 1). When the alkaline solutions of the tetracyclines were allowed to stand at 60°C, the emission intensity increased due to completion of degradation. 173-Dibromo-5 ,5-dimethylhydantoin has been used by Owa et a1.20 as a chemiluminogenic reagent after the action of peroxodisulfate on TC. As NBS and 173-dibromo- 5,5-dimethylhydantoin belong to the group of oxidants containing a positively charged bromine atom, it was decided to investigate peroxodisulfate as co-oxidant . The results show that the signal was higher than when only NBS was used. lo mV/ 2.00 11111ll , 5min , -Time Fig. 2 Typical recorder output for the NBS-TC reaction under the recommended conditions (the numbers above each set of peaks are pg ~ m - ~ of TC) Table 1 Chemiluminescent emission intensities (mV) from A, 20.0 and B, 50.0 pg ~ m - ~ of TC and 0.050 mol dm-3 of sodium hydroxide prepared by different procedures Emission intensity/mV Procedure* A, B C D E F G H A 6.8 8.0 34.4 21.6 92.0 72.0 260 B 10.8 12.4 74.0 28.8 88.0 120 500 * Procedures for each solution of TC: A, is measured immediately after addition of NaOH; B, is measured 30 min after addition of NaOH; C, is mixed with NaOH, allowed to stand at 60°C for 15 min, cooled to ambient temperature and measured; D, is mixed with NaOH, allowed to stand at 60°C for 2 h, cooled to ambient temperature and measured; E, F and G, also contains 0.001,O.Ol and 0.10 mol dm-3 peroxodisulfate, respectively; and H, also contains 0.005 mol dm-3 ammonia.ANALYST, JUNE 1993, VOL.118 635 ~ ~~ Table 2 Analytical characteristics of the proposed CL method for the determination of tetracyclines Tetracycline TC TCt TCe TCY OTC OTCt OTCw DCll CTC DMCC DMCC6 Linear range/ pg cm-3 0.050-3.00 10.0-400 - 0.50-5.00 5.0-400 0.50-7.00 0.50-7.00 0.30-3.00 - LOD*/ pg cm -3 0.0049 - - 0.40 - - 0.23 0.22 0.0020 - Slope (f s) 29.33 rt 0.29 0.42 f 0.01 10.40 f 0.32 19.11 f 0.31 5.27 k 0.07 0.70 k 0.01 1.19 f 0.04 0.53 f 0.005 1.56 f 0.02 12.04 k 0.13 3.12 rt 0.10 Intercept 4.40 f 0.44 0.37 rt 0.02 2.19 & 0.16 1.65 f 0.42 -1.18 f 0.19 -0.64 f 0.02 0.68 f 0.06 0.79 f 0.003 2.47 rt_ 0.08 4.54 rt 0.22 3.15 f 0.09 (f s) r 0.9995 0.997 0.9990 0.9997 0.9993 0.9994 0.9991 0.9997 0.9996 0.9995 0.999 nt 12 9 4 4 10 9 4 10 8 10 4 DC* 2.00-6.00 - 1.77 f 0.03 5.66 f 0.12 0.9996 5 13 DMCC**,II 0.80-8.00 - 0.78 f 0.01 1.23 f 0.004 * Limit of detection: concentration of tetracycline that generates signal equal to the blank plus three times its standard deviation (s).+ Number of samples, each measured three times. + Standard solutions do not contain ammonia, optimum cNaOH = 0.050 mol dm-3. 5 Standard solutions are spiked with aqueous honey solution. 7 Standard solutions also contain 0.010 rnol dm-3 sodium citrate. 11 Log Zllog c calibration graph. ** Reported for comparison of sensitivities. CTC"*.ll 2.00-20.0 - 0.68 f 0.01 0.55 -t 0.007 0.9993 11 0.9996 Table 3 Results for the determination of tetracyclines in aqueous solutions Error (YO) Concentration Tetracycline range/pg ~ m - ~ Range Average TC 0.050-3 .oO 0.1-4.6 2.5 OTC 0.50-5 .OO 0.5-4.7 2.1 DC 0.50-7.00 0.4-4.0 1.7 CTC 0.30-7 .00 0.3-3.5 2.0 DMCC 0.30-3.00 0.3-4.0 1.9 * Number of samples, each measured three times.r Relative standard deviation Concentration/ n* (%) (n = 10) pg cm-3 0.9996 10 1.2 0.4 0.9992 9 1.3 0.7 0.999 9 4.0 1.6 0.9997 8 4.2 1.3 0.9997 10 3.1 0.9 0.50 2.00 2.00 5.00 1 .00 5.00 2.00 5.00 0.50 2.00 Table 4 Recovery of 0.500 pg cm-3 of TC from solutions with a 10-fold concentration of various additives used as excipients Additive Glucose Galactose Lactose Sugar Sorbitol Starch Talc Cellulose acetate hydroxyphthalate Ethylenediaminetetraacetic acid Carbowax* Carbopolt Magnesium stearate Sodium lauryl sulphate CaS04* Sodium citrate Recovery (% ) (n = 3) 100.3 100.0 102.0 104.0 100.8 100.0 99.3 102.3 102.0 99.8 102.5 98.5 101.0 54.2 100.5 * Polyethylene glycol 4000.t Carboxypoly(methy1ene). t When present at equal concentration (pg ~1131~) the recovery was 97.8%. Nevertheless, the most intense chemiluminescent emission was observed when ammonia was also added to the solutions of TC. Ammonia chemiluminesces when mixed with NBS22 and it enhances the emission due to a synergistic effect between the two chemiluminescent reactions occurring after mixing the reagents. Optimization of Experimental Parameters The instrumental parameters (flow rate of reagent and analyte solution) and the NBS concentration were the same as previously described.22 The effect of sodium hydroxide and ammonia concentrations was investigated in order to obtain the maximum emission intensity from TC (Fig.1). An 8.0 X rnol dm-3 solution of ammonia and 0.030 mol dm-3 solution of sodium hydroxide were the best concentrations for maximum intensity. Optimization experiments for all tetra- cyclines studied gave similar optimum concentrations for ammonia and sodium hydroxide except OTC (3.0 x mol dm-3 of ammonia) and DMCC (0.30 mol dm-3 of sodium hydroxide). Analytical Parameters Fig. 2 shows a typical recording for a series of TC standards obtained by the proposed method. All other tetracyclines gave similar recordings. The analytical parameters for each tetra- cycline examined are summarized in Table 2. Results for the determination of tetracyclines in aqueous solutions are given in Table 3.Interference Studies In order to assess the possibility of applying the proposed continuous flow CL method to the assay of commercial tetracycline formulations, the effect of some common ex- cipients used in pharmaceutical preparations was studied by636 ANALYST, JUNE 1993, VOL. 118 Table 5 Recovery of 0.500 pg cm-3 of TC from solutions with some concomitant compounds Table 8 Determination of tetracyclines in commercial formulations with the proposed CL method and official mcthods.28 (All formula- tions are capsules unless stated otherwise) Concentration ratio Recovery (YO) Compound (compound to TC) (n = 3) TCs/mg Found difference Relative (CL - Official official) Form u 1 at i o n Amount CL+s* method (YO) Ascorbic acid Thiamine hydrochloride Calcium pantothenate Pyridoxinc hydrochloride 10 10 10 10 1 1 0 1 0.1 10 10 200 200 97.7 98.0 98.6 74.3 99.8 170.4 109.4 97.7 101.1 102.2 99.6 96.6 Riboflavin Tetrucycline- Chropicycline Hostacyclin (tablets) Oxytetrucyclinc- Terramycin Oxacycle (powder) 250 2372 1 500 503 IL 4 250 502 -5.2 0 Saccharin Caffeine NaCl KHZP04 250 270k 1 0.06 3% m/m 3.21 k 272 3.3 -0.7 -2.7 Doxycy cline- Vi bram ycin Lentomyk Ledermycin (tablets) Demeclocycline- Table 6 Recovery of 0.500 pg cmP3 of TC from solutions with some bivalent cations (A) and after addition of 0.010 mol dm-3 sodium citratc ( H ) 100 96.9kO 100 93.3 k 5.0 97.0 94.2 0 -1.0 300 301 k 3 299 Mean: +0.7 +1.5 Concentration ratio Recovery (YO) (n = 3) Cation (cation to TC) A B s = standard deviation ( n = 3).Ca* + Mg2-’ cu2+ Zn*+ 20 2 0.2 20 2 0.2 20 2 0.2 200 20 2 15.7 49.4 81.8 12.7 36.7 92.1 19.6 25.3 58.4 55.4 83.6 98.6 70.2 99.3 72.9 102.2 - analysing synthetic sample solutions contianing 0.500 pg cm-3 of TC and a 10-fold concentration of each excipient. The undissolved material, if any, was filtered off before measurement. The recovery results are given in Table 4. With the exception of calcium sulfate, no interference was observed from any of the excipients tested, which showed recoveries in the range 97.&104.0%. Calcium sulfate did not interfere with the determination of TC when present at equal or lower concentrations. The effect of some common concomitant compounds on the recovery of 0.500 pg cm-3 of TC was studied by analysing synthetic samples, as for the excipient study, but with various amounts of each concomitant compound (Table 5 ) .Pyridox- ine hydrochloride interferes, when present at a concentration ratio >1, mainly due to its non-chemiluminogenic reaction with NBS.26 Riboflavin interferes when present at a concen- tration ratio >0.1, due to its chemiluminescent reaction with NBS. No interference was observed from the other concomi- tant compounds tested even when present in a 10-fold excess. 88.4 98.2 90.9 100.1 - Table 7 Recovery experiments for tetracyclines added to sample solutions of commercial formulations Amount of tetracycline/pg cm-3 Recovery (n = 3) (Yo) 101.9 99.7 97.9 102.1 101.6 101.7 97.3 100.5 103.3 100.5 99.4 100.0 99.1 100.7 98.4 103.7 100.0 95.2 98.7 100.1 98.6 100.0 Initially Re- present Added covered Formulation Chropicycline capsules Tctrucy cline- (250 mg) 0.496 0.500 0.510 1.000 0.997 1.500 1.468 Effect of Metal Ions The ability of tetracyclines to form stable complexes with metal ions20 was investigated to establish whether the CL emission is affected.The effect was studied by analysing synthetic samples, as for excipients, but with various amounts of each cation. The results proved that the formation of these stable complexes severely reduces the chemiluminescent emission (Table 6). In order to minimize the interference, sodium citrate was added to the standard and synthetic sample solutions at a final concentration of 0.010 mol dm-3. In contrast with the common aminopolycarbonic complexing agents, sodium citrate did not affect the emission at this concentration. This observation has been made recently during the determination of ammonium in fertilizers.27 The results of the recovery studies using sodium citrate (Table 6) are significantly improved compared with those obtained without sodium citrate in the solutions.This improvement is probably d u e to t h e effective masking of cations by citrate. Hostacyclin tablets (500 mg) 0.509 0.500 0.510 1.000 1.016 1.500 1.525 Oxytetracycline- Terramycin capsules (250 mg) 2.145 0.500 0.486 1.000 1.005 2.000 2.066 Oxacycle powder for external use (3% m/m) 1.632 0.500 0.503 1.000 0.994 2.000 2.000 Doxycyclinc- Vibramycin capsules (100 mg) 1.911 1.000 0.991 2.000 2.014 4.000 3.936 Lentomyk capsules (100 mg) 1.835 1.000 1.037 2.000 1.999 4.000 3.809 Demechcy cline- Lcdcrmycin tablets (300 ms) Analysis of Tetracyclines in Honey Honey solutions containing one of the tetracyclines studied generated low chemiluminescent intensity due to quenching by the complexity of the solution.Nevertheless, if the standard solutions are spiked with tetracycline-free aqueous honey solutions, then TC, DMCC and OTC can be deter- 1.002 0.500 0.494 1.500 1.502 2.000 1.971 Mean:ANALYST, JUNE 1993, VOL. 118 637 Table 9 Rccovcry experiments for TC, OTC and DMCC added to aqueous solutions of commercial honey samples TC OTC DMCC Recovery Recovery Recovery Sample Added1 Found/ (Yo) Added/ Found (Yo) Addcdl Found/ (Yo) pgccm-3 pgcm-3 ( n = 3 ) pgcm-3 pgcm-3 (n=3) pgcm-3 pgcm-3 ( n = 3 ) 1 0.10 0.50 2 0.10 0.50 3 0.10 0.50 4 0 .I 0 0.50 0. I04 0.520 0.101 0.520 0.097 0.495 0.097 0.504 104.0 104.0 101.0 104.0 97.0 99.0 97.0 100.8 1 .oo 2.00 1 .OO 2.00 1 .OO 2.00 1 .00 2.00 0.988 2.008 1.033 1.924 1.044 1.783 1.066 1.980 101.9 100.4 103.3 96.2 104.4 89.2 106.6 99.0 0.50 0.80 0.50 0.80 0.50 0.80 0.50 0.80 0.497 0.785 0.529 0.801 0.518 0.806 0.518 0.774 101.9 98.1 101.9 100.1 101.9 100.8 101.9 96.8 mined in honey (Table 2). Therefore, the calibration graph must be constructed with standard solutions that contain the appropriate tetracycline and tetracycline-free honey. Tetra- cycline, OTC and DMCC can be determined in honey if present at concentrations at least equal to 0.02, 0.2 and 0.1% m/m, respectively. Accuracy The accuracy of the continuous flow CL method for the determination of tetracyclines in pharmaceutical preparations was examined by performing recovery experiments on solu- tions prepared from various tetracycline formulations.A mean recovery of 100.0% was found (range 95.2-103.7%) (Table 7). The proposed methods were also evaluated by analysing commercial pharmaceutical formulations of tetracyclines. The results were compared with those obtained by the official methods.28 A satisfactory agreement between the results was obtained (Table 8) with a mean relative difference of 1.5% (range &S.2%). The accuracy of the proposed CL method for the determina- tion of tetracyclines in honey was examined by performing recovery experiments on solutions prepared from various tetracyclines (Table 9). Conclusions The proposed automated method is simple, accurate and precise and allows the determination of tetracyclines in pharmaceutical preparations and honey, without any sample pre-treatment. References Sultan, S.M., Alzamil, I. Z., and Alarfaj, N. A., Taluntu, 1988, 35, 375. JelikiC-Stankov, M., and VcsclinoviC, D.. Analyst. 1989, 114, 719. JelikiC-Stankov, M., Stankov, D., Malesev, D.. and Radovic, Z., Microchim. Acta, 1991, I, 65. Alwarthan, A. A., Al-Tamrah, S. A.. and Sultan, S . M., Anulyst, 1991, 116, 183. Abdcl-Khalek, M. M.. and Mahrous, M. S . , Tulunta, 1983. 30, 792. 6 7 8 9 1 0 I 1 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 Mahrous. M. S . , and Abdel-Khalek, M. M., Talanta, 1984,31, 289. Salinas, F., Berzas Nevado, J . J . , and Espinosa, A., AnufyAt.1989. 114, 1141. Poiger, H., and Schlatter, Ch.. Analyvt, 1976, 101, 808. Regosz, A., Pharrnazie, 1977, 32, 681. Abdcl-Hady Elsayed, M., Barary, M. H., and Mahgoub. H., Talunta, 1985, 32, 1153. Salinas, F.. Muijoz de la Pena, A . , and Duran Meras, I . , Anal. Lett., 1990, 23, 863. Shoukry, A. F., and Badawy, S. S . , Microchem. J.. 1987, 36, 107. Sabharwal, S . , Kishore, K., and Moorthly, P. N., J. Pharm. Sci., 1988. 77, 78. Ghandour, M. A., and Ali, A. M. M., Anal. Lett., 1991. 24, 2171. Galeano Diaz, T., Guiberteau Cabalinas, A.. and Salinas, F., Anal. Lett., 1990, 23, 607. Oka, H., Ikai, Y . , Kawamura, N., Uno, K . , and Yamada, M., J . Chromatogr., 1987, 400, 253. Blanchflower, W. J.. McCracken, R. J., and Rice, D. A., Analyst, 1989, 114, 421. Hou, W., and Wang, E., Analyst. 1989, 114, 699. Alwarthan, A. A . , and Townshend, A., Anal. Chim. Acfa, 1988, 205, 261. Owa, T., Masujima. T., Yoshida, H., and Imai, H., Bunseki Kaguku, 1984,33, 568. Syropoulos, A. B., and Calokerino4, A. C., Anal. Chim. Acta, 1991, 255, 403. Halvatzis, S. A . , Timotheou-Potamia, M. M., and Calokerinos, A. C., Analyst, 1990, 115, 1229. Analytical Profiles of Drug Substances, ed. Florey, K., vol. 13, Academic Press, Orlando, 1984. Waller, C. W., Hutchings. B. L., Wolf, C. F., Goldman. A. A., Broschard, R. W., and Williams, J. H., J . Am. Chem. Soc., 1952, 74, 4981. Ahmed, B. M., and Jee. R. D., Anal. Chim. Actu, 1984, 156, 263. Halvat~is, S. A.. and Timotheou-Potamia. M. M., Anal. Chim. Acta, 1989, 227,405. Halvat~is, S. A., and Timotheou-Potamia, M. M., Talanta, in the press. US Pharmacopeial Convention, US Pharrnacopeia, XXI Re- vision, Mack, Easton. PA, 1985. Paper 2105162A Received September 25, I992 Accepted November 25, I992

ISSN:0003-2654    

DOI:10.1039/AN9931800633

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ANALYST, JUNE 1993, VOL. 118 639 Determination of Ascorbic Acid by Flow Injection With Chemiluminescence Detection” Abdulrahman A. Alwarthan Chemistry Department, College of Science, King Saud University, P.O. Box 2455, Riyadh- I 1451, Saudi Arabia Flow injection was used in combination with chemiluminescence detection t o determine 230 amol of ascorbic acid based on a luminol-Fe3+-hydrogen peroxide system. The method was applied t o the determination of ascorbic acid in pure form, pharmaceutical preparations and fruit juices. The method is sensitive, rapid (sampling rate 2-56 samples h-I) and tolerates the presence of common ingredients usually found in fruit juices and drug formulations. Keywords: Ascorbic acid determination; flow injection; chemiluminescence detection; pharmaceutical preparation; fruit juice The determination of ascorbic acid in vitamins and natural products is of great importance.Ascorbic acid occurs naturally in most fruit juices and vegetables. Often, it is added during the manufacture of juices or soft drinks to improve their nutritional value, for attracting consumers or to act as an antioxidant to prolong the shelf-life of the commercial product. However, its concentration in fruit juice requires careful regulation, as too low a concentration will affect its antioxidant properties but too high a concentration can cause certain disadvantages such as accelerated colour fading. 132 Owing to the wide use of ascorbic acid in canned fruits, vegetables, animal foods and drugs, and because of its significance in human physiology, there are numerous conven- tional methods for and reviews on the determination of ascorbic acid.Reported methods include titrimetriq3-5 flu- 0rirnetric~6.7 enzymic,”9 chromatographic,lO~lI electroana- lytica112J3 and spectrophotometric14-17 methods, each with their advantages and disadvantages. The development of extremely sensitive and reliable instrumentation has generated much interest in reactions that produce chemiluminescence (CL) and bioluminescence. Several applications18-21 have shown the high sensitivity and selectivity of CL detection compared with established meth- ods such as ultraviolet-visible and fluorescence spectrometry and electroanalytical detection. The sensitivity of CL methods for the determination of trace metals is well known.22.23 Most such methods are based on the catalysis or inhibition of reactions involving the oxidation of reagents such as luminol, lucigenin, lophine and gallic acid.Iron(ii) ions react with luminol in alkaline hydrogen peroxide and this reaction is sensitive and selective. The detection limit for iron(r1) is of the order of 0.1 pmol.21 This reaction was used as the basis for the CL determination of ascorbic acid. As CL measurements require that the analyte be mixed reproducibly with the reagents necessary for light production, flow injection (FI) is considered to be very suitable for CL methods of analysis, as it readily achieves rapid and reproducible mix- This paper describes a procedure based on the reducing effect of ascorbic acid on iron(ir1) ion and measuring the iron(I1)-catalysed light emission from luminol oxidation by hydrogen peroxide. The procedure is simple, precise, sensi- tive and selective for the accurate determination of ascorbic acid in pure solutions, pharmaceutical preparations and fruit juices.Other compounds normally present along with ascorbic acid do not interfere in the determination. ing.2425 * Presented at SAC ’92, an International Conference on Analytical Chemistry, Reading, UK, September 20-26, 1992. Experimental Reagents Ascorbic acid solution. Prepared daily by direct weighing of ascorbic acid (Merck) in 1% m/v metaphosphoric acid. Luminol stock solution, 0.1 moll-’. Prepared by dissolving 0.1772 g of luminol (general-purpose reagent, Sigma) in 0.1 mol I-’ carbonate buffer. Iron(rr) chloride hexahydrate stock solution, 0.1 mol 1-l.Prepared by dissolving 2.703 g of the compound (BDH, AnalaR) in 100 ml of distilled water and adding a few millilitres of 0.1 mol 1-l sulfuric acid for stabilization. Hydrogen peroxide solution. Prepared daily by diluting a measured amount of 30% mlv H202 (BDH) with distilled water. Metaphosphoric acid solution, 1% mlv. Obtained by dissolving 1 g of the crystals (Merck) in 100 ml of distilled water. Sodium carbonate buffer solution, 0.1 moll-’. Prepared by dissolving 10.6 g of sodium carbonate (BDH, AnalaR) in water and diluting to 1 1. Procedure for the Determination of Ascorbic Acid in Tablets, €apsules, Syrup and Fruit Juices Tablets and capsules Weigh a known amount of the powdered tablets or capsules and dissolve in and dilute to the required volume with 1% rnlv metaphosphoric acid solution.Use the final solution for determining the ascorbic acid content as shown in Fig. 1. syrup Weigh several drops (equal to approximately 100 mg of ascorbic acid) of the drug into a 100 ml calibrated flask and dissolve in and dilute to volume with 1% m/v metaphosphoric acid solution. Use the final solution for determining the ascorbic acid content as shown in Fig. 1. Fruit juices Natural fruits were bought freshly from the local market. Samples were prepared by extracting the juices, the extracts were centrifuged until a clear liquid was obtained and then weighed prior to analysis. Weighed amounts of these liquids were diluted to a known volume with 1% m/v metaphosphoric acid solution.The final solutions were used for determining the ascorbic acid content as shown in Fig. 1.640 1 n Sample I i 5 I I - - - - a rnol I l ) ml min 1 U 9 Fig. 1 Flow injection manifold for ascorbic acid determination: 1 , peristaltic pump; 2, sample injection (30 plj; 3 , reaction coil; 4, Perspex T-piece; 5 . waste; 6, coiled flow cell; 7, PMT; 8, housing; and 9, recorder Instrumentation for the Flow-through CL Detector The flow cell was a coil made of 1.3 mm i.d. glass tubing spiralled to a diameter of 3.5 mm with five turns, enabling the flowing, emitting solution to remain in view of the detector for up to 30 s. The coiled glass flow cell, as used previously,26 was backed by a mirror for maximum light collection and a sensitive photomultiplier tube (PMT) (Thorn EMI, 9789QB) for measurement of the emitted light intensity.The PMT was operated at 400 V, provided by a stable high-voltage power supply (Thorn EMI, Model PM 28BN). No wavelength selection was involved. Flow System for the Determination of Ascorbic Acid by C hemiluminescence A schematic diagram of the flow manifold is shown in Fig. 1. The acidified iron(iI1) solution (3 x rnol I-') was used as the carrier stream for the sample, which was usually acidified with 1% m/v metaphosphoric acid solution, and the luminol solution (1 x 10-4 moll-l) was mixed with the carrier stream at the reaction coil. Each solution was pumped at 2.03 ml min-' by a peristaltic pump (Gilson Minipuls 3MP4). Hydrogen peroxide (3 X rnol I-I) was pumped at 2.03 ml min-l by a peristaltic pump (Ismatec Mini-S-820) and mixed with the other two solutions at a Perspex T-piece.Poly(tetrafluoroethy1ene) tubing (0.8 mm i.d.) was used to connect the reagent streams to the T-piece. Sample injection was effected with a Rheodyne Type RH 5020 rotary valve, fitted with a 30 pl sample loop. The CL cmission was recorded on a chart recorder (Yokogawa Model 3021) and the peak heights were measured. Results and Discussion Optimization of Experimental Variables A series of experiments were conducted to establish optimum analytical variables. The parameters optimized included flow rate, pH, reagent concentrations and reactor length. Effect of lurninol concentration The effect of luminol concentration was studied in the range 1 x 10-3-1 x rnol I-'. It was observed that at concentrations higher and lower than 1 x rnol 1-l there was a sharp decrease in the CL.Therefore, 1 x mol 1-1 was selected for the analysis. Effect of iron(rrr) concentration The effect of the iron(m) concentration on the CL intensity from the ascorbic acid signal was studied in the range 1 x 10-2-1 X mol I-'. The results obtained are shown in Fig. ANALYST, JUNE 1993, VOL. 118 600 > E >. 5 400 . c .- Y 8 C 0 v) ._ .- .- E w 200 n 1 X 10-5 1 x 10-4 1 X 10-3 1 x 10-2 Fell' concentrationhol 1-1 Fig. 2 (1 X 1W6 mol 1-I) in the presence of 1 x 0.025 mol 1-1 H202 Effect of Fe"' concentration on CI, intensity of ascorbic acid mol I-' luminol and 2. A concentration of 3 x mol I-1 was found to be optimum; higher concentrations (up to 1 X rnol I-') gave a significant increase in the signal, but with poorer reproducib- ility.Effect of hydrogen peroxide concentration The optimum hydrogen peroxide concentration was 0.03 moll-'. Higher concentrations (up to 0.1 rnol I-') gave higher emission, but the signal was noisy and irreproducible. The decrease in intensity at lower concentrations (SO.01 rnol I-') might be attributed to an inadequate amount of H202 available for reaction with the luminol. Effect of ascorbic acid pH The pH of the sample solution was found to be very critical. The results of a detailed study of the stability of ascorbic acid at different pH values are shown in Fig. 3. The maximum response was obtained at pH 1.5, indicating the best medium for the reducing activity of a solution of ascorbic acid.At pH >1.5, ascorbic acid became less stable and the destruction rate increased rapidly with increasing pH, causing the reducing activity to decrease. Effect offlow rate The flow rate of the solutions is very important and should be regulated. At flow rates that are too low or too high, CL is not emitted in the flow cell and hence the emitter is not detected. Fig. 4 shows the effect of total flow rate on CL intensity. The intensity increased with increasing flow rate. However, a total flow rate of 6.1 ml min-l (2.03 ml min-I for each channel) is recommended because of the greater precision and economy in the use of reagents. Effect of reactor length The proposed method for the determination of ascorbic acid is based on the reducing effect of the acid on iron(nr) ion, and the chelating ability of luminol with iron(i1) ion (in the presence of H20z as an oxidant) to provide a chemilumi- nescent product.Therefore, the optimum time for the reaction of ascorbic acid with iron(ri1) ion, in order to produce sufficient iron(n), was studied by varying the reactor length from 50 to 400 cm. Maximum emission intensity was obtained when the reactor length was 200 cm (the signal takes about 35 s to reach its maximum). Increasing the reactor length further caused a decrease in the emission intensity, probably owing to sample dispersion. Shortening the reactor length below 200 cm gave a 250% reduction in the emission intensity; probablyANALYST. JUNE 1993. VOL. 118 64 1 1200 1000 > E 5 800 . .- v) 0 4- .- C 600 c 0 v) .- .?? 400 E w 200 0 I I I I I 2 4 6 8 1 0 PH Fig.3 rnol I-') in the prcsence of 3 x Effcct of pH on CL intensity of ascorbic acid ( I X 10-" rnol I-' Fe3+ and 0.03 rnol I-1 H202 1100 1000 . .w > v) .- I 6 900 C C 0 .- .- .- 2 800 E w 700 1 1 I 1 I 2 4 6 8 10 600 Flow rate/ml min -1 Fig. 4 moll-' ascorbic acid, 3 x and 0.03 rnol I-' H202 Effect of total flow rate on CL intensity. Conditions: 1 X rnol I-' Fe3+, 1 X 1 0 P moll-' luminol the iron(iii) ion was in excess and the amount of iron(1r) ion produced was not proportional to the amount of ascorbic acid in the sample. Determination of Ascorbic Acid The optimum values for the CL-FI variables are: flow rate per channel = 2.03 ml min-l, reactor length = 200 cm, pH = 1.5, [H202] = 0.03 rnol I - ' , [luminol] = 1 x rnol I - ] , [Fe3+] = 3 x moll- and voltage supply to the PMT = 400 V.The optimization of the variables ensured the maximum emission intensity from the chemiluminescent product as the sample plug passed through the detector. The applicability studies involved the determination of the practical range, the repeatability in application, the effect of interferents and the reliability of the method for analysing a range of natural samples. Practical Range and Repeatability in Application An aqueous stock standard solution of 0.1 rnol I - ' ascorbic acid was prepared, from which working standard solutions were prepared by serial dilution with 1% m/v metaphosphoric acid solution. A linear calibration graph for ascorbic acid was obtained over the range 1 x 10--'1-1 x rnol I-' using the flow system described in Fig.1 , as shown in Fig. 5 . The slope is 0.52. Above 1 x rnol I - l , the slope decreased because the concentration of the reagents became a limiting factor (i. e., W 200 - 1 x 10 ' 2 1 x 10-8 I x 10-4 Ascorbic acid concentration/mol I-' Fig. 5 four measurcmcnts under the rccommendcd conditions) Calibration graph for ascorbic acid (each point is the mean of Table 1 Effcct of foreign ions (all 1 X I x 10V rnol I-' ascorbic acid rnol I-') on thc signal for Signal height Interfercnt mVq %t - I 060 100 Chloride 1090 102.83 Nitratc 1120 105.66 Nitritc 1100 103.77 Oxalate I200 113.21 Sulfide 1400 132.08 Sulfate 1450 136.79 Tartrate I060 I00 Fructose 1060 100 Glucose 1060 100 Nicotinamidc 1110 104.72 Riboflavin 1070 100.04 Starch 1080 101.89 * Results based on three injections per sample.t Relative to ascorbic acid = 100%. there was not enough sample to react with the luminol). Four replicate injections of ascorbic acid were made per sample. The relative standard deviation for ten replicate sample injections of 1 x rnol I-' ascorbic acid was 1.4%. There was no detectable blank signal. The ascorbic acid concentra- tion that gave a signal-to-noise ratio of 2 was 1 X 10- l2 moll- (30 amol or 5 fg per injection), compared with 1 x 10-*3 rnol 1- iron(ir)21 using a luminol-alkaline hydrogen peroxide system. Interference Studies The specificity of the method for ascorbic acid in the presence of frequently encountered excipients, common reducing agents and inorganic ions was studied, in addition to the effect of other vitamins that are likely to be present along with ascorbic acid in multivitamin formulations. The results obtained are summarized in Table 1.Among the inorganic anions, oxalate, sulfide and sulfate interfered severely in the determination by enhancing thc emission intensity. Nitrate, chloride and nitrite also affected the results but their effects were mild compared with those of oxalate, sulfide and sulfate. Interference by these ions was expected because the proposed method is based directly on the ability of ascorbic acid to reduce iron(i1r). Hence any species that can bring about this reduction would be expected to interfere. However, species such as tartrate, fructose, glucose, riboflavin and starch showed no interference. Nic-642 ANALYST, JUNE 1993, VOL.118 Table 2 Determination of ascorbic acid in pharmaceutical prepara- tions by the proposed method and the official method27 Amount of ascorbic acidmg Proposed Recovery Preparation* Claimed? method* (Yo ) Ascoplex capsules 175 164.44 93.95 Multivitaplex syrup 90 91.94 101.15 Multivitaplex tablets 30 28.24 94.13 Redoxon tablets 1000 908.20 90.82 Cebion tablets lo00 968.80 96.88 * Multivitaplex tablets, Ascoplex capsules and Multivitaplex syrup were obtained from Dumex, Denmark; Redoxon tablets from La Roche, Switzerland; and Ccbion tablets from Merck, Germany. + According to the official meth0d.2~ * Average of six injections per sample. Table 3 Determination of ascorbic acid in fruit juices Amount of ascorbic acidlmg Recovery Juice Claimed* Foundt (Yo) Tomato 980 1035.89 105.70 Orange 1050 1117.89 106.44 Lemon 1033 986.24 95 .47 Banana 950 970.52 102.16 * According to the official method.27 t Average of six injections per sample. otinamide gave a slight enhancement to the emission intensity but can be tolerated as an interferent in 10-fold excess over ascorbic acid.Analysis of Pharmaceutical Preparations The proposed method was applied to the analysis of several pharmaceutical dosage forms containing ascorbic acid either in tablets, capsules or a syrup. The results are summarized in Table 2. Most of the results agree with the reported values. However, non-significant differences were found in the results, which can be attributed to the quenching effect of the coloured solutions of the samples, to the heterogeneity of the samples or the ef€ect of other reducing species in the preparations.Analysis of Some Fresh Fruit Juices Table 3 gives the ascorbic acid content of some fruit juices analysed by the proposed method. The sample matrix is very complex, because a large number of additives are now added to foods for different purposes, and such additives may affect the proposed method either by quenching or enhancing the emission intensity. However, the good recovery of ascorbic acid shown in Table 3 indicates that no significant interference occurred. In conclusion, the determination of ascorbic acid by CL-FI has advantages over conventional methods in that it is faster and simpler, there is minimum manipulation and intervention of the operator, it is suitable for routine control, it can be carried out directly without any pre-treatment and it is extremely sensitive.The author is very grateful to the Islamic Education, Scientific and Culture Organization (ISESCO) for supporting this research. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 References Fogg, A. G., and Summan, A. M., Analyst, 1983, 108, 691. Fung, Y. S., and Luk, S. F., Analyst, 1985, 110, 1439. Evered, D. F., Analyst, 1960,85, 515. Murty, C. N., and Bapat, N. G., Fresenius’ 2. Anal. Chem., 1963, 199, 367. Vallant, H. , Mikrochim. Acta, 1969, 436. Deutsch, M. J., and Weeks, C. E., J. Assoc. Off. Anal. Chem., 1965, 48, 1248. Tevata, T., Hara, S., Yamaguchi, M., Nakamura, M., and Okhura, Y., Chem. Pharm. Bull., 1985, 33, 3499. Lee, M. H., and Dawson, C. R., Methods Enzyrnol., 1979,62. 30. Tono, T., and Fujita, S., Agric. Bid. Chem., 1981, 45,2947. Williams, R. C., Baker, D. R. and Schmidt, J. A., J. Chromatogr. Sci., 1973, 11, 618. Thompson, J . N., Trace Anal. Chem., 1979, 51, 353. Strohl, A. N., and Curran, D. J., Anal. Chem., 1979, 51, 353. Lazaro, F., Rios, A., Luque de Castro, M. D., and Valcarcel, M., Analyst, 1986, 111, 163. Tutem, E., Vlkuseven, B., and Apak, R., Anal. Lett., 1992,25, 471. Backheet, E. Y., Emara, K. M., Askal, H. F., and Saleh, G. A., Analyst, 1991, 116, 861. Anwar, J., Farooqi, M. I., Nagra, S. A., and Khan, A. M., J. Chem. SOC. Pak., 1990, 12, 75. Muralikrishna, U., and Murty, J. A., Analyst, 1989, 114, 407. Yamada, M., Komatsu, T., Nakahara, S., and Suzuki, S., Anal. Chim. Acta, 1983, 155, 259. Abbott, R. W., Townshend, A., and Gill, R., Analysr, 1986, 111, 635. Veazey, R. L., and Nieman, T. A., J. Chromatogr., 1980,200, 153. Alwarthan, A. A., Habib, K. A. J . , and Townshend, A., Fresenius’ J. Anal. Chem., 1990, 337, 848. Kricka, L. J., and Thorpe, G. H. G., Analyst, 1983,108, 1274. Townshend, A., Anal. Proc., 1985, 22,370. Townshend, A., Analyst, 1990, 115, 495. Rule, G., and Seitz, W. R., Clin. Chem. ( Winston-Salem), 1979, 25, 1635. Alwarthan, A. A., and Townshend, A., Anal. Chim. Acta, 1986, 185, 329. British Pharmacopoeia 1980, HM Stationery Office, London, 1980. Paper 2/04657A Received September 1, 1992 Accepted November 23, 1992

ISSN:0003-2654    

DOI:10.1039/AN9931800639

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ANALYST, JUNE 1993, VOL. 118 643 Flow Injection Chemiluminescence Method for the Selective Determination of Chromium(lll)* Rosario Escobar, Qingxiong Lin and Alfonso Guiraum Departamento de Quimica Analitica, Facultad de Quimica, Universidad de Sevilla, 4 10 12-Sevilla, Spain Francisco F. de la Rosa lnstituto de Bioquimica Vegetal y Fotosintesis, Facultad de Biologia, Universidad de Sevilla y CSlC, 4 I 0 12-Se villa, Spa in Flow injection (FI) has been applied t o the determination of Cr"' in water and food samples. The method is based on measurement of the light emitted from the Cr"'-catalysed oxidation of luminol by H202. The apparatus consists of an FI system with a flow cell suitable for chemiluminescence detection. The flow cell, situated near the photodetector, is a coiled tube made from transparent poly(tetrafluoroethy1ene.The typical signal is a narrow peak, the height of which is proportional t o the light emitted and, therefore, t o the concentration of Cr"'. The detection limit is 0.01 ppb and the linear range extends up t o 6 ppb. The concentrations of the reagents, the pH and the flow rates were optimized. Interferences by several metal ions were examined, and the system was shown t o possess a high selectivity. The method was successfully applied t o the determination of CrIll in water and food samples. Keywords: Flow injection; chemiluminescence; chromium (111) determination; water and food samples Chromium, together with insulin, plays an important role in carbohydrate metabolism.1-4 This metal ion dissolved in natural water is present in two different oxidation states: CrlI1 and Crvl.The former is considered to be essential to mammals for the maintenance of glucose, lipid and protein metabolism, whereas the latter is known to be toxic to humans.5," It is also known that Cr"' present in the diet is reduced in the gastrointestinal tract to CrIII.5 Therefore, the determination of Cr in biological materials has been of great interest in clinical research and in elucidating the exact role of this element in human nutrition and health. A number of papers have described analytical procedures for the determination of Cr in biological samples7-8 and for Cr speciation .q7l0 Flow injection (FI) is a technique that was first described by RfiiiCka and Hansenll in 1975. It has developed over the last 15 years into a well established laboratory technique.The scope of FI has been documented in a number of books12,13 and in several monographs.14-2' For example, during 1988 Chemical Abstracts registered 372 new entries on FI related techniques. Chemiluminescence from reactions that usually produce transient emissions is an attractive method for the determina- tion trace metals. Metal ions catalyse the oxidation of luminol by H202 in basic solution, producing chemiluminescence emission. The specificity of this reaction for CrlIi can be achieved in the presence of ethylenediaminetetraacetic acid (EDTA) as the formation of the Cr"'-EDTA complex is kinetically slow, whereas Crvi does not catalyse the reaction at all. Hence, this method can be used for the speciation of both forms of Cr.The development of extremely sensitive and reliable instrumentation has generated much interest in chemiluminescence reactions. Several applications have demonstrated the considerable sensitivity and selectivity of chemiluminescence detection.22-27 Chemiluminescence reactions can be conveniently studied by FI. Chemiluminescence provides sensitivity and selectivity, whereas F1 provides rapid and reproducible sample injection and mixing of the reagents. These factors, together with low cost and simplicity, make the FT-chemiluminescence combina- tion extremely attractive .28-32 * Presented at SAC '92, an International Conference on Analytical Chemistry, Reading, UK, September 20-26, 1992. In this paper, FI with chemiluminescence detection has been applied to the determination of Crlll in water and food samples.The proposed method is rapid, precise and inexpen- sive, and determines Crlll in water directly with a detection limit of 0.01 ppb. Experimental Instrumentation and Procedures The metal-catalysed oxidation of luminol was carried out using the FI apparatus illustrated schematically in Fig. 1. Luminol and H202 were dissolved in 0.2 mol I-' C032-- HC03- buffer with the pH adjusted to a suitable value in each experiment. Luminol and H202 were mixed in the flow system and then with the flow of water in which the sample was injected. A Gilson Minipuls 2 peristaltic pump was used to drive the reactants through the flow cell, which consisted of a laboratory-made coiled transparent poly(tetrafluoroethy1ene) (PTFE) tube positioned near to the photodetector.The coil tubing was of 1 mm i.d. 'The light intensity was continuously measured by a Hamamatsu phototube housed in an Aminco photomultiplier connected to Knauer recorder. The injection valve and tubing system were from Scharlau Science. The reagent streams (see Fig. 1) were water, luminol and H202. Luminol and H202 were first mixed in the flow system and then mixed again with the sample, which was injected and carried by the water stream by manipulating the injection valve. The chemiluminescence was recorded as a function of time over 5 s, the maximum emission intensity being propor- tional to the Cr"' concentration in the sample. Preparation of Food Samples The sample (0.5 g) was placed in a PTFE reactor with 3 ml of HN03.The reactor was transferred into a CEM-MAS-81D microwave oven and heated for 6 min. A 2 ml volume of Hz02 was added to the solution and the mixture was heated in the microwave oven for a further 3 min. The digested sample was diluted with water to 25 ml in a calibrated flask. A 10 ml volume of this solution was placed in a 100 ml beaker and heated to evaporate all the HN03 and H202. Water was then added and the solution transferred into a 50 ml calibrated flask. This solution was used for the determination of Cr as described under Instrumentation and Procedures.644 ANALYST, JUNE 1993, VOL. 118 Pump Sample -1 Injection vaive 8' Flow cell Recorder Q-EI Photodetector Waste Fig. 1 Schematic diagram of the FT assembly for chemiluminescence monitoring used for the determination of Cr"' 4 8 12 16 [Lurninol]/l0-4 rnol 1-1 Fig.2 Dependence of chemiluminescence intensity on luminol concentration. Conditions: luminol in 0.2 rnol 1-1 C032--HC03- buffer ( H 10.87); [H202], 0.14 rnol I-' in 0.2 moll-' C032--HC03 buffer ($I 10.87) and 1 x moll-' EDTA; 1 ppb Crl" in 3 X 10k3 rnol 1-1 H3P03. Flow rates: H20, 11.11 ml min-l; luminol, 2.5 ml min-', H202, 2.11 ml min-I. Injection volume: 0.405 ml Reagents All reagents were of analytical-reagent grade unless stated otherwise. Water obtained from a Milli-Q (Millipore) water purification system was used throughout. Solutions Chromium(II1) standard solution, 1 mg ml-l. A 0.7696 g amount of C T ( N O ~ ) ~ . ~ H ~ O (Merck) was dissolved in water and then diluted to 100 ml with water in a calibrated flask. Working solutions were prepared daily by appropriate dilu- tion with water.The pH was adjusted to about 4 with 0.15 ml of 0.5 mol I-' H3P04 in a 250 ml calibrated flask. Ethylenediaminetetraacetic acid solution, 0.1 mol 1- I . A 3.72 g amount of EDTA was dissolved in 20 ml of 1 mol 1-1 KOH solution and diluted to 100ml with water. A 1 X mol I-' EDTA solution was prepared from this solution by dilution with water. Luminol stock solution, 1 x moll-'. A 0.1772 g amount of luminol was dissolved in 3.5 ml of 1 mol 1-1 KOH solution, diluted to 100 ml with water in a calibrated flask and stored for at least 1 week before use. In order to prepare a 1 x mol 1-I luminol working solution, 10 ml of the solution were diluted to 100 ml in a calibrated flask with 2.7 ml of 0.2 moll-' NaHC03 and 87.3 ml of 0.2 mol 1-1 Na2C03 buffer to adjust the pH to 10.87.Hydrogen peroxide working solution, 0.14 moll-'. A 1.43 ml volume of 30% H202, 1 ml of 1 X mol I-' EDTA, 7.5 ml of water, and 2.7 ml of 0.2 mol 1-1 NaHC03 and 87.3 ml of 0.2 moll-' Na2C03 buffer were mixed. The pH was 10.87. Results and Discussion Optimization of Experimental Conditions A series of experiments were carried out to establish the optimum conditions for the system used. The parameters 0.1 0.2 0.3 [H2021/mol I-' Fig. 3 Variation of chemiluminescence intensity with H202 concen- tration. Conditions: H202 in 0.2 mol 1-1 C03*--HC03- buffer (pH 10.87) and 1 x moll-' in 0.2 mol I-' C032--HC03- buffer (pH 10.87); 1 ppb Cr"' in 3 X moll-' H3P04.Flow rates: H20, 11.11 ml min-l; luminol, 2.5 ml min-'; H202, 2.11 ml min-'. Injection volume: 0.405 ml moll-' EDTA; [luminol], 1 x optimized included flow rates, injection volume, pH and the concentrations of luminol, H202, H3P04 and EDTA. Luminol concentration The chemiluminescence emission was found to increase with increasing luminol concentration. However, the intensity of emitted light increased very slowly after the luminol concen- tration reached 8 x mol 1-l, as shown in Fig. 2; a 1 X rnol 1-1 luminol concentration was used for the analysis. H202 concentration Fig. 3 shows a plot of chemiluminescence intensity as a function of H202 concentration. At low concentrations, light emission was proportional to H202 concentration. At high concentrations, not only did the light emission not increase, but also so many bubbles were generated in the FI system that it was difficult to operate.A concentration of 0.14 rnol I-' H202 was chosen as the most convenient for further studies. PH The effect of pH on chemiluminescence intensity was investi- gated using NaHC03-Na2C03 buffer solution. The maximum response was obtained at pH 10.7-11, as shown in Fig. 4. A pH of 10.87 was selected. Flow rates As shown in Fig. 5 , the optimum flow rate of H202 was 2.1 ml min-l, and light emission remained stable when the flow rate of luminol was greater than 2.5 ml min-l. However, the maximum emission intensity was highly dependent on the rate of mixing of sample and reagents. A rapid mixing of the sample and reagents enhanced the chemiluminescence inten- sity.Flow rates of 2.1, 2.5 and 11.1 ml min-l for H202, luminol and water, respectively, were used for the analysis.ANALYST, JUNE 1993, VOL. 118 64.5 L a, L 0 9.5 10.0 10.5 11.0 11.5 PH Fig. 4 Effect of pH on chemiluminescence intensity. Conditions: [H202], 0.14 mol I-' in 0.2 mol I-' C032---HC03- buffer (pH 10.87) and 1 x 10-5mo11-1 EDTA; [luminol], 1 X 10-3mo11-' in 0.2 rnol I-' C03*--HC03- buffer (pH 10.87); 1 ppb of Cr"' in 3 X moll-' H3P04. Flow rates: H20? 11.1 ml min-l; luminol, 2.5 ml min-I; H202, 2.1 ml min-I. Injection volume: 0.405 ml 4 8 12 Flow rate/ml min-' Fig. 5 Effect of A, water; B, luminol; and C, H202 flow rates on chemiluminescence intensity. Conditions: [H202], 0.14 moll-' in 0.2 moll-' C032--HC03- buffer (pH 10.87) and 1 x moll-' EDTA; [luminol], 1 X moll-' in 0.2 rnol I-' C032--HC03- buffer (pH 10.87); 1 ppb Crrl' in 3 x rnol I-' H3P04.Injection volume: 0.405 ml Injection volume The influence of the injection volume on the chemilumin- escence intensity is shown in Fig. 6. The maximum emission intensity was attained with an injection volume of 0.4 ml. Hence this value was used for sample analysis. EDTA concentration The luminol-H202 chemiluminescence reaction is well known. This reaction is catalysed by Cr"' and other metals, and the addition of EDTA greatly reduces the luminescence of the reaction because of the formation of metal-EDTA complexes. The Cr"'-EDTA complex is thermodynamically stable but kinetically slow to form at room temperature, whereas all the other metal ions that catalyse the luminol- H202 reaction rapidly form complexes with EDTA.33 There- fore, the addition of EDTA to mask foreign ions improved the selectivity of the determination of CrI" with the proposed method.Seitz et aZ.34 proposed the use of 1 x mol 1-1 EDTA for masking potentially interfering ions, whereas Hoyt and Ingle35 considered that 1 x mol I-' EDTA was adequate. The effect of EDTA was investigated in two different ways: (i) addition of this compound to a Cr'lI solution as a masking reagent; and (ii) addition of EDTA to H202 solution for stabilization of the latter. No noticeable decrease in light emission was found in the presence of 1 x mol 1-1 EDTA in Cr"' solution. However, when the EDTA con- centration was higher than 1 x mol 1-l, a significant decrease in light emission was observed.Hence, 1 x moll-' EDTA was chosen for masking metal ions. 10 0 0.2 0.4 0.6 0.8 Injection volume/ml Fig. 6 Influence of injection volume on chemiluminescence inten- sity. Conditions: [HzO,], 0.14 rnol I-' in 0.2 mol 1-* C03*--HC03- buffer (pH 10.87) and 1 X 10-5 mol 1-1 EDTA; [luminol], 1 x moll-' in 0.2 rnol I-' C032--HC03- buffer (pH 10.87); 1 ppb Cr"l in 3 x rnol 1 - I EDTA. Flow rates: H 2 0 , 11.1 ml min-'; luminol, 2.5 ml min-'; H202, 2.1 ml min-' rnol I-' H3P04 and 1 x + I I I-- I Log ([EDTAl/mol I - ' ) - 6 - 5 - 4 - 3 Fig. 7 Effect of EDTA in A, H202; and B, in sample solutions on chemiluminescencc intensity. Conditions: [H202], 0.14 mol I-' in 0.2 mol 1-1 C032--HC03- buffer (pH 10.87); [luminol], 1 X mol 1-1 in 0.2 rnol 1-1 C032--HC03- buffer (pH 10.87); 1 ppb Cr"' in 3 x rnol I-' H3P04.Flow rates: H 2 0 , 11.1 ml min-I; luminol, 2.5 ml min-l; H202, 2.1 ml min-'. Injection volume: 0.405 ml It was also found that when EDTA was added to the H202 solution at a concentration of 1 x mol I - l , the baseline noise of the chemiluminescence reaction decreased slightly and the H202 solution became stable and generated fewer bubbles. Hence, 1 X rnol I-' EDTA in H202 solution was used in the analysis. The results are shown in Fig. 7. H3P04 concentration The effect of the H3P04 concentration on chemiluminescence intensity is shown in Fig. 8. The maximum intensity was reached for a concentration of H3P04 of 3 x mol I-' and the sample contained 1 ppb of Cr"' and 1 ppm of Fe"'.When the sample only contained 1 ppb of Cr"' without iron, the maximum intensity was reached with 2 x moll-' H3P04. Interferences Many metal ions catalyse the luminol-H202 reaction. In order to remove the interferences, 1 x rnol I-' EDTA was added to the Cr"' solution before analysis. Table 1 lists the ions that were tested. Of the solutions tested, Co", Fell1 and CeIV interfered even with the addition of EDTA. However, in many samples, the amount of Co" is lower or not much higher than that of Cr"'. Cerium(1v) reduces the chemiluminescence intensity, presumably by oxidizing Crl'l to Crvl. This is not true interference as this oxidation would have already taken place in naturally occurring samples. Iron(n1) could be masked by careful adjustment of the pH of the sample solution.As shown646 ANALYST, JUNE 1993, VOL. 118 I Table 2 Results of the determination of Cr"' in different samples of water* Amount of CrllL added Recovery (PPb) ("/I 0.1 98 0.05 92 Dilution of sample 1 + 15.6 - Cr"' found (PPb) 0 0.024 Cr"' in sample <0.01 0.4 (PPb) Sample Distilled water Mineral water Drinking water:? Sample 1 Sample 1 Sample 2 Sample 3 Sample 4 Sample 1 Sample 2 Polluted Water:$ 1 + 9 1 + 99 1 + 99 1 + 49 1 + 49 2.79 0.26 0.128 0.184 0.257 27.9 26.0 12.8 9.2 12.9 - - 0.2 99 0.2 91 0.2 103.5 0.2 94 0 2 4 6 8 [H3PO41/1O-4 rnol 1 - l Fig. 8 Effect of H3P04 in the Sam le solution on chemiluminescence Cr"' and 1 ppm Fell' solution. Conditions: [Hz02], 0.14 mol 1- in 0.2 mol I-' C03*--HC03- buffer (pH 10.87) and 1 X lo-' mol 1-1 EDTA; [luminol], 1 x mol I-' in 0.2 mol I-' C032--HC03- buffcr (pH 10.87); 1 ppb Cr"' in 3 x rnol 1-I H3P04.Flow rates: H20, 11.1 ml min-I; luminol, 2.5 ml min-I; HzOz, 2.1 ml min-l. Injection volume: 0.405 ml P intensity. A, H3P04 in 1 ppb Cr" P solution; and B, H3P04 in 1 pb 1 + 99 1 + 99 0.361 0.31 1 0.2 108.5 0.2 95 36.1 31.1 * Results are averages from at least six replicates. t The samples of drinking water were from the mains of the city of * The samples of polluted water were from different sites of the Seville taken at different sites. Guadalquivir river. Table 1 Tolerance levels of interferents in the determination of Cr"'. The concentration of Cr"' in all instances was 1 ppb Table 3 Results of the determination of Cr"' in food samples Cr"' Cr"' Recovery Sample found/yg g-l added (ppb) (Yo ) Shrimp 0.1 0.2 102 Brewer's yeast 0.3 0.2 110 Brown bread 0.48 0.2 96 Bovine muscle 0.14 0.2 S8 Ions added Tolerance levcl Ca", F - , C1-, Br-, I- >10 ppm Mn", Hg", Ni", Cd", Th'", Zn", Uvl, MeV' Mg" , , , , , , CrV' Al"' Pb" Ba" Vv In"', Fe", Fe"' '1 PPm 1 PPm Ce'V 0.2 ppm CO" 10 ppb I plot of log peak height as a function of log Crlll concentration.The results illustrate the high sensitivity and wide dynamic range of this technique. o m c Y a c Applications The proposed method was applied to the analysis of several samples of water and food. Table 2 shows the results of the analysis of several samples of water. Distilled, drinking, mineral and waste water were analysed with good results. The data obtained for the analysis of different types of food sample are presented in Table 3.These data are coincident with the values obtained for CrlI1 in food reference materials of this type. I I I 1 Log ([Crll']/g ml-1) Fig. 9 Calibration graph for the determination of Cr"'. Conditions: [H202], 0.14 rnol 1-l in 0.2 mol 1 - I C032--HC03- buffer (pH 10.87) and 1 x 10-smol 1-' EDTA; [luminol], 1 X 10W3 mol I-' in 0.2 rnol 1-1 C03*--HC03- buffer (pH 10.87); Cr"' in 3 X lop4 rnol I-' H3P04 and 1 x 10 -4 mol 1-1 EDTA. Flow rates: H20, 11.1 ml min-l; luminol, 2.5 ml min-l; H202, 2.1 ml min-l. Injection volume: 0.405 ml -8 0' -11 -10 -9 Conclusions The proposed method combines the sensitivity of chemilu- minescence with the rapidity of FI, and, at the same time, affords good selectivity by the addition of EDTA.Under the optimum conditions, commonly encountered ions do not interfere with the determination. The method can be used to determine from 0.01 to 6 ppb of C P , and its high sensitivity permits the determination of Cr in drinking water and foods. Moreover, because the method is very sensitive, many samples must be diluted and interferences are, therefore, reduced. The relative standard deviation found (15 replicates) was 2.5%. The method is rapid and up to 70 samples per hour can be analysed. in Fig. 6, when the H3P04 concentration was below 3 x moll-' (pH 3.7), Fell1 reduced the chemiluminescence intensity, presumably because of the adsorption of Cr"' by Fe(OH)3 particles. At approximately pH 3.7, Fell1 did not interfere with the determination, which could be due to the formation of the very stable complex, [Fe2(0H)#+.On increasing the concentration of H3P04 further, Ferrl increased the chemiluminescence intensity, owing to the decomposition of this complex to generate free Fe1Ii ions. In addition, H3P04 was the best acid of those tested, viz., HC1, HN03, and other acids. Therefore, 3 x rnol I-' H3P04 was used to adjust the pH of the sample solution. This work was supported by Plan Andaluz de Investigacibn. Q. L. is grateful for financial help from Consejeria de Educacion of Junta de Andalucia. References 1 2 Mertz, W., in Chromium in Nutrition and Metabolism, eds. Shapcott, D., and Hubert, J., Elsevier, Amsterdam, 1979, p. 1. Mertz, W., Physiol. Rev., 1969, 49, 163. Calibration Graph The calibration graph for CrI", obtained under the optimum conditions, was linear from 0.01 to 6 ppb of Cr.Fig. 9 shows aANALYST, JUNE 1993, VOL. 118 647 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Haylock, S. J . , Thornton, R. J . , Buckley, P. D., and Bachwell, L. F., Exp. Mycol., 1982, 6, 335. Toepfer, E. W., Mertz, W., Polansky, M. M., Roginsky, E. E., and Wolf, W. R., J. Agric. Food Chem., 1977, 25, 162. Prasad, A. S . , and Oberleas, D., Trace Elements in Human Health and Disease, Academic Press, New York, 1976, p. 79. Chromium-Health and Safety Precautions, Guidance Note EH2, Health and Safety Executive, London, 1977. Raymond, T. L., and Hercules, D. M., Anal. Chem., 1974,46, 917. Steiner, J. W., Moy, D. C.. and Kramer, H. L., Analyst, 1987, 112, 1113. Chang, C.A., Patterson, H. H., Mayer, L. M., and Bause, D. E., Anal. Chem., 1980, 52, 1264. de Andrade, J . C., Rocha. J. C., and Baccan, N., Analyst, 1985, 110, 197. R%iEka, J . , and Hansen, E. H., Anal. Chim. Acta, 1975, 78, 145. Valcarcel, M . , and Luque de Castro, M. D., Flow Injection Analysis. Principles and Applications, Ellis Horwood, Chiches- ter, 1987. RSiiEka, J . , and Hansen, E. H., Flow Injection Analysis, Wiley, New York, 2nd edn., 1988. RSiiEka, J . , and Hansen, E. H., Anal. Chim. Acta, 1980, 114, 19. Vanderlisce, J. T., Stewart, K. K., Rosenfeld, A. G., and Higgs, D. J . , Talanta, 1981, 28, 11. Worsfold, P. J., Anal. Chim. Acta, 1983, 145, 117. Riley, C., Rocks, B. F., and Sherwood, R. A., Talanta, 1984, 31, 879. 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 RGiiEka, J . , and Hansen, E. H., Anal. Chim. Acta, 1988. 214, 1. R6iiEka, J . , Fresenius' Z . Anal. Chem. , 1988, 329, 653. Reijn, J. M., Poppe, H., and van der Linden, W. E., Anal. Chim. Acta, 1983, 145. 59. Osborne, B. G., and Tyson, J . F . , J . Food Sci. Technol., 1988, 23, 541. Isacsson, U., and Wettermark, G., Anal. Chim. Acta, 1974,68, 339. Paul, D. B., Talanta, 1978, 25, 377. Miller, J . N., Analysr, 1984, 109, 191. Kricka, L. J., and Thorpe, G. H. G., Analyst, 1983, 108. 1274. Grayeski, M. L., Anal. Chem. , 1987, 59, 1243A. Townshend, A., Analyst, 1990, 115,495. Burguera, J . L., and Townshend, A., Proc. Anal. Div. Chem. Soc., 1979, 16, 263. Burguera, J . L., Townshend, A., and Greenfield. S., Anal. Chim. Acta, 1980, 114, 209. Townshend, A., Anal. Proc., 1985, 22, 370. Alwarthan, A. A., and Townshend, A., Anal. Chim. Actu, 1987, 196, 135. Hool, K., and Nieman, T. A., Anal. Chem.. 1988, 60, 834. Hamm, R. E., J. Am. Chem. Soc., 1953,75, 5670. Seitz, W. R., Suydam, W. W., and Hercules, D. M., Anal. Chem., 1972,44, 957. Hoyt. S. D., and Ingle, J . D., Anal. Chim. Acta, 1976, 87,163. Paper 2l0515.51 Received September 25, 1992 Accepted January 4, I993

ISSN:0003-2654    

DOI:10.1039/AN9931800643

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ANALYST, JUNE 1993, VOL. 118 649 Phase-selective Alternating Current Adsorptive Stripping Voltammetry of Aminopterin on a Mercury Thin Film Carbon Fibre Ultra m icroelect rode* Michael A. Malone,+ Agustin Costa Garcia and Paulino Tunon Blanco* Department of Physical and Analytical Chemistry, University of Oviedo, Asturias, Spain Malcolm R. Smyth School of Chemical Sciences, Dublin City University, Dublin 9, Ireland The electrodeposition of mercury thin films onto carbon fibres for the determination of aminopterin and its analogues has been optimized following an investigation of the electrochemical reduction processes of aminopterin obtained at a static mercury drop electrode. The advantageous characteristics of ultramicroelec- trodes combined with adsorptive preconcentration and phase-selective a.c.stripping voltammetry were found to yield a very sensitive and reproducible method. By using this electrode, accumulation was performed at five different concentrations of aminopterin ranging from 5 x IO-lO to 5 x rnol dm-3. The electrode yielded a calibration graph from 2 x IO-IO to 8 x rnol dm-3 ( r = 0.994) with a limit of detection [signal-to-noise ratio (S/N) = 31 of 1 x IO-lO rnol dm-3 aminopterin in aqueous solutions. The reproducibility of the signal was evaluated at three different concentrations of aminopterin producing relative standard deviations ranging from 3.57% at the 5 x 10-10 rnol dm-3 level to 2.49% at the 1 x lo-* rnol dm-3 level ( n = 10). The electrode was applied to the determination of aminopterin in urine resulting in a limit of detection (S/ N = 3) of 2.5 x Keywords: Mercury thin film ultramicroelectrode; adsorptive stripping voltammetry; aminopterin; urine rnol dm-3 without the employment of any pre-treatment of the urine.Much of the work reported to date relating to mercury thin film electrodes has been carried out using conventional sized graphite or glassy carbon electrodesl-3 due to their inertness and simplicity of use. It was found that the deposition of mercury occurs at sites of varying activity and that the deposition of larger amounts leads to formation of mercury droplets, the size and distribution of which depends on the deposition potential. More recently, the popularity of microelectrodes has grown rapidly due to the recognition that many of the undesirable aspects of electrochemical and electroanalytical techniques can be reduced or eliminated by virtue of their use.Microelectrodes possess several advantages over conventional sized electrodes ,4-6 including reduced capacitative charging currents and increased mass transport rates due to the radial component of mass transport; these consequently lead to excellent signal-to-noise (S/N) characteristics. Thus, much work has been directed towards the formation and characteri- zation of mercury films of microelectrodes using various substrates. Various workers studied the possibility of using iridium as a substrate7-11 and found that it was suitable for application to the adsorptive stripping analysis of several metals without problems of intermetallic compound formation.Platinum has also been widely studied as a substrate12-14 and applied to the flow injection anodic stripping voltammetry of various heavy metal ions.15 Silver has also been used to support mercury depositsl"17 and was reported to yield a coherent surface of the entire mercury deposit and to cover the base disc completely . Numerous studies and applications of mercury thin film carbon fibre electrodes to the anodic stripping voltammetricl8- 22 and potentiometric stripping analyses23 of heavy metals have appeared in the literature. Carbon fibre has been * Presented at SAC '92, an International Conference on Analytical + Permanent address: School of Chemical Sciences, Dublin City * To whom correspondence should be addressed. Chemistry, Reading, UK, September 20-26, 1992.University, Dublin 9, Ireland. reported to be a suitable inert substrate and yields stable and reproducible films in the form of mercury micro-droplets. Despite the numerous studies of mercury thin film carbon fibre microelectrodes using inorganic substances no reports have been located relating to their use in cathodic stripping voltammetry of organic compounds. Aminopterin was the first antifolate compound to show proven success in the treatment of cancer24 and in the past has been used widely for the successful treatment of acute leukemia. Today, it has limited use in cell fusion experiments to select for hybrid cells by killing the unfused cells, which are deficient in enzymes for nucleotide salvage pathways.25 However, more importantly the analogues of aminopterin are under constant study26-31 in an effort to develop more active and less toxic drugs for the treatment of various cancers, including leukemia, lung cancer and head and neck cancer.Thus aminopterin analogues continue to hold investigational interest and many are currently undergoing phase 1 and phase 2 clinical trials. The majority of the analogues maintain the pteridine ring of aminopterin intact as substitution at the N-10 position is one of the main routes to increasing the antileukemic effectiveness.32 Thus the first electrochemical reduction process, which is the process of analytical importance (described later) will be common to all. Few methods have been reported in the literature for their determination in biological fluids. A promising approach has been reported however by Tellingen et aZ.33 using high- performance liquid chromatography with fluorimetric detec- tion.In this paper the parent compound, aminopterin, was studied as a model for the aminopterin analogues bearing in mind that the parent compound and the analogues of major investigational interest have the analytically important revers- ible reduction process in common. The cyclic voltammetric behaviour of aminopterin at a static mercury drop electrode (SMDE) is presented. The conditions for optimum mercury deposition on carbon fibre electrodes, optimum phase-selec- tive a.c. voltammetric conditions and optimum methodology for the determination of aminopterin in aqueous solutions and urine samples are subsequently presented and discussed.650 ANALYST, JUNE 1993, VOL. 118 Experimental Reagents and Materials Aminopterin was purchased from Sigma and used without further purification.Stock solutions (1 x rnol dm-3) were prepared daily in 1 x 10-* rnol dm-3 sodium carbonate and stored at 4 "C in the dark. A 0.1 rnol dm-3 ammonium acetate buffer (pH 5 ) was prepared by adjusting 0.1 rnol dm-3 acetic acid to pH 5 using ammonia solution and used as the background electrolyte throughout. All other reagents were of analytical-reagent grade, including: Hg(N03), (Merck); hydrochloric acid (Panreac); and acetic acid (Panreac). All solutions were prepared using de-ionized water obtained by passing distilled water through a Milli-Q water purification system (Millipore). All de-aerations were carried out using purified (N-48) nitrogen (<1 ppm 0,) (Sociedad Espaiiola del Oxigeno).The urine analysed consisted of pooled human urine samples obtained from healthy individuals spiked with various amounts of aminopterin stock solutions to achieve the desired concentration of the drug in urine. Instrumentation All cyclic voltammograms were obtained using a Metrohm VA scanner (E-612) linked to a Metrohm VA detector (E- 61 l ) and a Graphtex x-y recorder (wx-4421). A Metrohm EA- 290 (Kemula) SMDE was used as the working electrode. All potentials are referred to an Ag-AgC1-KCI (3 rnol dm-3) reference electrode. The microelectrode studies were carried out using a Metrohm (Herisau) Model E-506 Polarecord. A 20 cm3 electrochemical cell was used, which allowed the working electrode, reference electrode, counter electrode and nitrogen delivery tube to be fixed in position through a Plexiglas cover. The carbon fibres used were supplied by Donnay and had a nominal diameter of 7.5 pm.All pH measurements were made using a Crison micropH Model 2001 pH meter. Microelectrode Preparation Initially all fibres were washed sequentially in 10% v/v nitric acid, water and acetone, respectively, and then dried at 70 "C. The electrodes were then prepared by microscopically insert- ing a single fibre in the eye of a 100 mm3 plastic micropipette. The eye was sealed either by heat sealing using a soldering iron or by using low viscosity resin, kit TK4 (A. R. Spurr). This resin was polymerized by placing in an oven at 70°C overnight. Following polymerization the electrode was back- filled with mercury and the electrical contact was made using a copper rod that had been filed to remove surface oxides.The electrode was then sealed using the low viscosity resin. Various lengths of fibre were studied with respect to stability of the mercury film and peak height and shape of the aminopterin cathodic stripping peak, and a length of 3 mm was found to be optimum. By using shorter fibres very low currents were produced. Longer fibres yielded higher currents but lower stability of the mercury film, particularly during medium exchange as a result of physical vibration. Determination of Aminopterin Aminopterin was found to precipitate in the presence of mercury(I1) because of the formation of insoluble mercury salts; therefore, two separate cells were employed.After prior activation of the carbon fibre, by initially dipping in chromic acid for 2 min followed by 2 min in concentrated nitric acid, the mercury film was generated under the optimum conditions (described in Results and Discussion). Once formed, the mercury thin film microelectrode was rapidly transferred to the analytical cell containing the ammonium acetate (pH 5 ) electrolyte. This transfer, inevitably, caused a decrease in the amount of mercury present in the film due to both physical detachment of the mercury droplets and air oxidation of the mercury. The transfer procedure was studied in order to minimize these losses. The optimized procedure involved the rapid but careful removal of the first cell in a downward direction, followed by careful cleaning of the reference and counter electrodes and placement of the analytical cell in an upward direction making sure that vibrations of the fibre, which is the main potential cause of loss of mercury film, were minimized.This transfer lasted approximately 10 s. Through- out the procedure a closed circuit was maintained with the potential set at the film deposition potential of -800 mV. Once the electrodes had been placed in the analytical cell the potential was moved to -1400 mV for 30 s to ensure that a clean film was present for the first analysis. Both solutions were de-aerated using nitrogen for 15 min prior to film formation and analysis, respectively, and a nitrogen blanket was employed over the electrolyte throughout the analysis.Results and Discussion Cyclic Voltammetry In this work voltammograms were recorded in a 0.1 rnol dm-3 ammonium acetate (pH 5 ) medium containing 1 x mol dm-3 aminopterin. At this low concentration, the responses seen were adsorption controlled. The cyclic vol tam- metric behaviour of aminopterin in this medium is shown in Fig. l(a), which shows three reduction processes, the first process being reversible, followed by two irreversible pro- E N versus Ag-AgCI- -0.4 -0.8 EN Fig. 1 (a) Cyclic voltammogram of 1 x 10Ph rnol dm-3 aminopterin in 0.1 mol dm-3 ammonium acetatc (pH 5 ) electrolyte using an SMDE of drop area = 2.2 mm2; and scan rate = 100 mV S Y ' . ( b ) Direct current adsorptive stripping voltammctry of a uiesccnt solution of 1 x rnol dm-3 aminopterin in 0.1 mol dmq ammonium acetate electrolyte on an SMDE of drop area 2.2 mm2; scan rate = 100 mV s-l; E,,, = 0 V; accumulation times: A.0; B , 60; and C, 240 sANALYST, JUNE 1993, VOL. 118 65 1 cesses. It has been proposed34 that the first process (1c at -620 mV) is a 2e-/2H+ reduction of the pteridine ring to yield the 5,8-dihydro derivative. The smaller current of the anodic response ( l a at -530 mV) is due to the subsequent tautomeri- zation of the 5,8-dihydro derivative to yield the 7,8-dihydro derivative, which cannot be re-oxidized back to aminopterin. The second (2c at - 1030 mV) and third (3c at - 1210 mV) reduction processes are under kinetic control because they are dependent on the chemical rearrangement outlined above. The second process is due to the 2e-/2H+ reductive cleavage of the dihydro derivative, produced during the first reduction process, between the C-9 and N-10 positions to yield R-NHI and the 7,8-dihydro derivative of the pteridine moiety.The third reduction process is then due to the subsequent 2e-/2H+ reduction of this 7,8-dihydro derivative to the 5,6,7,8- tetrahydro derivative. If the direction of the potential scan is switched immediately after the first process, the anodic response increases because the majority of the 5 &dihydro derivative has not yet been tautomerized to the 7,8-dihydro derivative. The effect of accumulation at 0 V using both open circuit and applied electrolysis was then studied with respect to the first (reversible) process. Similar accumulation rates were observed with and without applied electrolysis.Fig. l(h) demonstrates accumulation under the influence of applied electrolysis. Accumulation in open circuit would obviously present practical advantages in terms of selectivity. However, with a view to working with a mercury thin film microelec- trode, accumulation in open circuit is not possible because electrolysis must be applied at all times to ensure the stability of the mercury film. Mercury Thin Film Optimization The formation of the optimum mercury thin film is a critical factor in the development of a sensitive and reproducible electrode. Mercury deposition on carbon fibres can be in the form of a thin film or can be increased to an almost spherical shape. The deposition of larger amounts of mercury for adsorptive stripping voltammetric applications might seem to be advantageous, facilitating the adsorption of more analyte; however, in this work it was seen that larger amounts of mercury produced unstable films and it seemed that the physical structure of the film was more important than the actual amount of the mercury present. Following a study of the literature and previous studies (unpublished) carried out in these laboratories, mercury nitrate in hydrochloric acid was selected as the mercury salt solution to be studied for the film formation.Therefore, a series of experiments were carried out to find the optimum concentrations of both the mercury salt and the hydrochloric acid. The mercury salt concentration was varied between 1 x 10-I and 1 x rnol dm-3 and a film was formed at each concentration by applying a deposition potential of -1200 mV €or 30 s.After formation the film was anodically stripped and the stripping peak was studied as an indication of the morphology of the film. At high conccntra- tions of the mercury salt (1 X 10-I rnol dm-3) the anodic stripping peak was broad and short. As the concentration was decreased to 1 x mol dm-3 the stripping peak became sharper and the peak height increased dramatically. At this concentration the mercury is thought to exist as numerous micro-droplets of high surface area. Below 1 x rnol dm-3 the peak height decreases as there is a decrease in the number of droplets on the carbon fibre surface. By using the same experimental criteria the molarity of the HC1 was studied in the range 1-6 rnol dm-3.The peak height of the mercury stripping peak increased with molarity up to 5 rnol dm-3 HCI and then began to decrease. Thus, it is thought that at a concentration of 1 x mol dm-3 Hg(N03)I in 5 rnol dm-3 HCI the mercury exists in the form of numerous micro- droplets on the fibre surface. At different concentrations the quality of the film deteriorates in terms of a decrease in the number of droplets or a growth in the size of the droplets with an overall effect of a decrease in the surface area of mercury. Following this, two important parameters, namely, the electrodeposition potential (Efilm) and deposition time (tfilm) were studied. By employing a solution of 1 x rnol dm-3 Hg(N03)2 in 5 rnol dm-3 HCI a series of films were formed sequentially at various deposition potentials using a depo- sition time of 60 s.Each film was subsequently stripped and studied as before. The electrodeposition potential was varied between -100 and -1200 mV in 100 mV intervals. As the applied potential became increasingly more negative the peak height of the mercury stripping increased down to a potential of -900 mV at which point it began to decrease again. It would seem that as the potential becomes more negative the number of surface active sites on the carbon fibre becomes increasingly larger and hence the number of mercury droplets increase. At potentials more negative than -800 mV hydrogen gas production, which potentially causes droplet detachment, became evident. Fig. 2(a) represents the results obtained.In order to ensure a stable film a deposition potential of -800 mV was chosen for further studies. The deposition time (tfilm) was then studied using the same criteria as in the previous study and was followed by a study of the aminopterin phase-selective a.c. cathodic stripping res- ponse with respect to tfilm. As tfilm was increased the height of the mercury stripping peak increased up to a ttilm of 90 s after which time it began to decrease again. This is represented in Fig. 2(b). At each deposition time the electrode was 2t 0 1 ’ ’ ’ I ’ 1 4 - d 1200 1000 800 600 400 200 0 0 50 100 150 200 250 300 50 100 150 200 250 & i d s t .. (+) EImV - Fig. 2 ( a ) Optimization of the mercury thin film formation on carbon fibre in terms of the applied deposition potential (Eel,) using direct current adsorptive stripping voltammetry.Mercury salt solution = 1 X rnol dmP3 Hg(N03)2 in 5 rnol dm-3 HCI; film deposition time (ftilm) = 60 s; film was stripped anodically at 100 mV s-l. (b) Optimization of the mercury thin film formation on carbon fibre in terms of deposition time (tfilm) using direct current adsorptive stripping voltammetry (series 1). Mercury salt solution = 1 X lo-’ rnol dm-3 Hg(N03)2 in 5 rnol dm-3 HCI; Etilm = -800 mV; film was stripped anodically at 100 mV s-l. (c) Optimization of the mercury thin film formation on carbon fibre in terms of tfilm versus aminopterin cathodic stripping response using phase-selective alternating current adsorptive stripping voltammetry. Mercury salt solution = 1 X lo-’ rnol dm-3 Hg(N03)2 in 5 rnol dm-3 HCI; Etilm = -800 mV.5 X mol dm-3 aminopterin solution in 0.1 rnol dm-3 ammonium acetate elcctrolyte (pH 5 ) ; aminopterin accumulation time = 60 s at an applied potential of 0 V. (For procedures and a.c. voltammetric conditions see text). (d) Two typical anodic stripping peaks of the mercury film from the carbon fibre electrode using direct current adsorptive stripping voltammetry. Mercury salt solution = 1 X loP3 rnol dm-’ Hg(N03)2 in 5 rnol dm-3 HCI; ttilm = 90 s; Etit, = -800 mV; scan speed = 100 mV S-I652 ANALYST, JUNE 1993, VOL. 118 transferred to a 5 x mol dm-3 aminopterin solution in pH 5 ammonium acetate medium and aminopterin was accumulated for 60 s at an applied potential of 0 V and then cathodically stripped. Similarly a mercury deposition time of 90 s produced the best aminopterin stripping peak.As outlined in Fig. 2(c) the aminopterin response began to decrease when higher mercury deposition times were employed. It seems that at deposition times of greater than 90 s a growth in droplet size rather than droplet number occurs, thus causing an overall decrease in the surface area of mercury available for analyte adsorption, and also producing a less stable film. Therefore, an optimum deposition time of 90 s was chosen. Between each film formation the fibre was regenerated producing a clean surface free of mercury; this contributed to the reproducibility of the film. The optimum conditions for this regeneration were found to be the application of a potential of +790 mV for 40 s, which oxidized all the Hg(0) on the fibre surface.Fig. 2(d) shows two typical anodic stripping peaks of the mercury film after formation under optimum conditions. The reproducibility of film formation under these optimum conditions was studied by forming and subsequently anodically stripping six consecutive films. Measurement of peak current (ip) of the stripping peaks yielded a relative standard deviation (RSD) of 1.11% (n = 6). To study the reproducibility of medium exchange of the film, the film- coated electrode was rapidly transferred to a blank 5 mol dm-3 HCI solution and stripped. The transfer reproducibility yielded an RSD of 13.3% (n = 6). Therefore, this dictated the necessity to use the same film throughout the whole analysis run using regeneration of the film surface between each measurement.Optimization of Phase-selective a.c. Conditions Once the mercury thin film microelectrode had been trans- ferred to the analytical cell, phase-selective a.c. voltammetry (AC1-Tast) was employed to exploit the reversibility of the reduction process of interest. Thus the conditions were optimized to yield the best aminopterin response possible. A study of the effect of the phase angle (0) was carried out by measuring the peak characteristics of a 5 x mol dm-3 aminopterin solution over the complete range of angles. It was found that at angles close to 90" the current was essentially the charging current component, whereas angles closer to 0" produced the best analytical signals. An optimum angle of 18" was chosen as under the experimental conditions employed it was seen to produce the best discrimination of the faradaic current against the capacative current, allowing easier measurement of the analytical signal.As the phase angle was increased to angles approaching 90" a dramatic increase in the background current was observed resulting in an increasingly poor analytical signal. The applied a.c. voltage amplitude (AE) was studied under the above conditions and a linear relationship between the signal (aminopterin cathodic strip) and AE was observed up to 20 mV, according to the following equation: ilnA = 1.27 x 10-1 AElmV -0.015, (Y = 0.9998). The instrument used worked at a fixed frequency of 75 Hz, and did not permit any frequency variation. A scan speed of 10 mV s-l was chosen as in a.c.voltammetry the scan speed does not significantly affect the signal. Aminopterin Accumulation Studies Accumulation studies were carried out in quiescent solutions for five different concentrations of aminopterin, i.e., 5 X l O - l O , 1 x 5 x 1 x 10Ws and 5 x rnol dm-3. An accumulation potential of 0.0 V was applied in all cases as it produced the best response. More negative accumulation potentials produced slower rates of accumulation and more positive potentials led to oxidation of the film, which is stripped at +260 mV in the acetate medium. Activation of the mercury film between each measurement was carried out by holding the potential at -1.4 V for 30 s. This removed the adsorbed reduction products of aminopterin producing a fresh film for the next measurement.At all concentrations studied the electrode responded in the same way. Initially the response increased linearly with time, then at a certain point the slope decreased and the response began to increase again in a linear fashion at this lower slope. For example, this slope change occurred at an accumulation time (tact) of 15 s for 5 x 10Wg rnol dm-3 aminopterin and at a tacc of 360 s for 5 x lo-'* rnol dm-3 aminopterin. After this second linear portion the slope again decreased to a point where the response became virtually independent of accumulation time (lace). Fig. 3 represents the first two linear regions of the curves. The third region of the curves can be interpreted as saturation of the electrode surface. However, the response still increases with increasing concentration. The changes in slope of the accumulation curves can be explained in terms of the surface of the mercury film becoming modified at higher concentra- tions of the drug, because at higher concentrations the potential of the aminopterin cathodic stripping peak shifts to a few millivolts more negative.A comparison of the curves indicates that at higher concentrations of aminopterin, the slope growth was faster. This phenomenon was previously explained35 as being due to molecular interactions between adsorbed molecules on the electrode surface. Careful exam- ination of the accumulation curves facilitated the selection of suitable accumulation times (tact) for further studies. Calibration Graphs By carefully selecting accumulation times, different concen- tration ranges of aminopterin could be studied.Obviously, at higher accumulation times lower concentrations can be detected, but the linear dynamic range was smaller than with shorter accumulation times. Thus, a compromise between limit of detection and linear range must be made to suit the analysis of interest. By using an accumulation time of 180 s and an applied potential of 0.0 V, a linear calibration plot was obtained between 2 x 1O-IO and 8 x 10-9 rnol dm-3 aminopterin in aqueous solutions according to the following equation: i h A = 2.88 x lo-' x caminoplmol dm-3 + 0.360 (Y = 0.9994). By employing a shorter accumulation time higher concen- trations of the compound can be studied. For example, by employing an accumulation time of 10 s, the electrode could be used in the 1 x 10-8-1 x rnol dm-3 region yielding the 0 50 100 150 200 250 300 350 400 450 t X C l S Fig.3 Aminopterin accumulation curves on the mercury thin film carbon fibre electrode using phase selective a x . adsorptive stripping voltammetry; E,,, = 0 V; electrolyte = 0.1 mol dm-3 ammonium acetate (pH 5 ) . (For procedure and a.c. voltammetric conditions see text). A, Series 1, 5 x rnol dm-3 aminopterin; B, series 2, 1 X mol dm-3 aminopterin; C , series 3, 5 x mol dm-3 aminopterin; D, series 4, 1 x rnol dm-3 aminopterin; and E, series 5 , 5 x lo-'" rnol dm-3 aminopterinANALYST, JUNE 1993, VOL. 118 653 following equation: ilnA = 5.74 x X caminodmol dm-3 + 0.525 ( I - = 0.9991). By using an accumulation time of 180 s the detection limit based on the concentration that yielded a stripping peak three times greater than the background level was 1 x 10-lO mol dm-3 of aminopterin.Fig. 4(a) represents the increase in peak current with the increase in aminopterin concentration at low aminopterin concentrations. This figure also outlines how the baselines were constructed at these low concentrations. It can be seen from Fig. 4(b) that baseline construction becomes increasingly easier due to the improved peak shape at higher aminopterin concentrations. Reproducibility and Stability The reproducibility of the aminopterin cathodic stripping response was studied at low (5 X rnol dm-3), medium (1 x rnol dm-3) and high (1 X rnol dm-3) concentrations. This study involved recording ten consecutive voltammograms at each concentration and calculating the RSD of the response.At a concentration of 5 x 10-lO rnol dm-3 an accumulation time of 120 s was employed and an I I t .- -(-I E -(-I E Fig. 4 (a) Phase-selective a.c. adsorptive stripping voltammograms of aminopterin in aqueous solution using a mercury thin film carbon fibre electrode; E,,, = 0 V, t,,, = 180 s. A, Blank pH 5 acetate medium; B, 2 x 10-lO rnol dm-3; C, 4 X mol dm-3; D, 6 X 1O-IO rnol dm-3; E, 8 x 10-'0 mol dm-3; and F, 1 X rnol dm-3 aminopterin. (For a.c. voltammetric conditions see text). (b) Phase- selective a.c. adsorptive stripping voltammogram of 5 X rnol dm-3 aminopterin in aqueous solution using a mercury thin film carbon fibre electrode. Accumulation potential = 0 V; accumulation time = 10 s.(For a.c. voltammetric conditions see text) Fig. 5 Standard additions analysis of urine spiked with 5 x rnol dm-3 aminopterin using phase-selective a.c. adsorptive stripping voltammetry. 0.5 cm3 of a 1 + 5 dilution of the spiked urine was injected into the 20 cm3 analytical cell followed by standard additions of a 1 x mol dm-3 stock aminopterin solution. E,,, = 0 V; tact = 40 s. A, Blank urine; B, sample; C, 2 mm3 of standard; D, 4 mm3 of standard; E, 6 mm3 of standard; F, 10 mm3 of standard; and G, 14 mm3 of standard. (For a.c. voltammetric conditions see text)654 ANALYST, JUNE 1993, VOL. 118 RSD of 3.57% (n = 10) was obtained. At an aminopterin concentration of 1 X mol dm-3 an RSD of 2.79% ( n = 10) was obtained using an accumulation time of 60 s.At a concentration of 1 X mol dm-3 the RSD was 2.49% (n = 10) using an accumulation time of 20 s. This reproducibility is also an indication of the stability of the mercury film using the optimized deposition and regeneration conditions. Thus, the electrode proved to be very sensitive and reproducible in aqueous solutions of aminopterin. The same electrode could be used for a period of at least 4-6 weeks. The shape of the anodic stripping peak of the mercury film was a good indicator of the state of the fibre surface and when this stripping peak was seen to broaden and decrease in peak height the surface of the fibre was regenerated by dipping in chromic acid for 30 s. In this way the chromic acid re-oxidized any reduction products of aminopterin that might have adsorbed on the surface of the carbon fibre, thus providing a cleaned surface. Aminopterin Determination in Urine Pooled urine samples were spiked with appropriate amounts of aminopterin stock solution to achieve final concentrations of 5 x 1 x and 5 x rnol dm-3 aminopterin in urine.Various dilutions of these urine samples were made using the ammonium acetate electrolyte so as to reduce the effect of interfering compounds naturally present in urine. Dilutions of 1 + 4 , l + 9 and 1 + 19 were made and 0.5 cm3 of the resulting solutions was injected into the 20 cm3 analytical cell. Blank urine samples were injected to ensure that no interference was seen at the same potential as that of aminopterin. In each case after injection of the spiked urine to the cell, appropriate standard additions were made and these standard additions plots were used to calculate the concentra- tion of aminopterin in the original sample, by extrapolation. Two separate analyses were carried out at each spiked concentration involving two separate dilutions.At a spiked concentration of 5 x 10-6 mol dm-3 aminopterin an RSD of 1.75% ( n = 2) was achieved by employing a 1 + 10 dilution using a 90 s accumulation time and a 1 + 19 dilution using 120 s accumulation. When a spiked concentration of 1 x mol dm-3 was studied, dilutions of 1 + 4 (tact = 60 s) and 1 + 9 (tact = 120 s) were executed and an RSD of 2.45% (n = 2) was obtained. At the lower concentration of 5 x mol dm-3 aminopterin, a 1 + 4 dilution was always used with an accumulation time of 120 s as at lower dilutions the signal was difficult to measure.The RSD of the signal at this concentra- tion was 16.5% (n = 2). A seemingly obvious way to improve this signal is to increase the accumulation time. However, at higher accumulation times adsorption of naturally occurring compounds in urine becomes more evident causing increased electrode passivation. Throughout the analysis of urine the gradual build up of these compounds on the electrode was evident. This phenomenon can clearly be seen in Fig. 5 where initially two extra peaks can be seen in addition to the aminopterin peak. These peaks from urine gradually increase and eventually merge. However, these responses occur at a different potential to that of the analyte. Fig. 5 is a standard additions analysis of a 1 + 4 dilution of a 5 x rnol dm-3 spiked urine sample.Baseline construction was carried out in the same way as in Fig. 4. Adsorption of these compounds from urine was less evident at the lower dilutions of 1 + 9 and 1 + 19. As with the analysis of aqueous solutions fibre regeneration was attempted using chromic acid. Although this regeneration resulted in some improvement, the fibre never returned to its original activity as the other compounds in urine caused gradual passivation of the fibre surface. Therefore, the electrode usually had to be replaced after approximately 20 measurements. The use of a pre-analysis extraction of the analyte from the urine using CI8 solid-phase extractisn cartridges would undoubtedly lower the limit of detection and minimize the passivation of the electrode surface.However, the objective of this work was to develop a system that could be applied to the determination of aminopterin analogues without the employment of pre-extraction procedures. Conclusion In this paper the cyclic voltammetric behaviour of aminopterin was presented and discussed. A mercury thin film carbon fibre ultramicroelectrode was developed so that the advantageous features of ultramicroelectrodes, adsorptive preconcentration and phase-selective a.c. voltammetry were combined. This system was applied to the determination of aminopterin as a model for its analogues that retain the pteridine ring within their structure. It yielded a sensitive and reproducible method, facilitating the possibility of both pre-administration and post-administration analysis of these compounds.This ultramicroelectrode system could potentially be applied to other organic compounds that exhibit accumulative behaviour on mercury. The incorporation of this type of system into a flow-injection system format offers a potentially useful area of investigation. Related work is currently underway in these laboratories. The authors thank the ERASMUS programme and the Spanish Ministry of Education and Science (DGICYT) project PB87-1041 for their support of this work. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 References Stulikova, M.. J. Electroanal. Chem. Interfacial Electrochem., 1973, 48, 33. Lieberman, S. H., and Zirino, A., Anal. Chem., 1974, 46, 20.Loung, L., and Vydra, F., J. Electroanal. Chem. Interfacial Electrochem., 1974, 50, 379. Pons, S . , and Fleischmann, M., Anal. Chem., 1987,59. 1391 A. Wightman, R. M., Anal. Chem., 1981, 53, 1125A. Johnson, D. C., Ryan, D. M., and Wilson, S. G., Anal. Chem.. 1988, 60, 147R. Golas, J., Galus, Z., and Osteryoung, J., Anal. Chem., 1987, 59, 389. Golas, J., and Kowalski, Z., Anal. Chim. Acta, 1989,221,305. Kounaves, S. P., and Deng, W., J. Electroanal. Chem., 1991, 301, 77. Kounaves, S. P., and Buffle, J . , J. Electroanal. Chem., 1987, 216, 53. Kounaves, S. P., and Buffle, J., J. Electroanal. Chem., 1988, 339, 113. Gunawardcna, G., Hills, G., and Scharifker, B . , J. Electroanal. Chem., 1981, 130, 99. Wehmeyer, R. K., and Wightman, R. M., Anal. Chem., 1985, 57, 1989. Stojekand, Z., and Osteryoung, J., Anal. Chem., 1988,60,131. Tay, E. B.-T., Khoo, S.-B., and Loh, S.-W., Analyst, 1989,114, 1039. Stojek, Z., and Osteryoung, J., Anal. Chem., 1989, 61, 1305. Ciszkowska, M., Penczek, M., and Stojek, Z., Electroanalysis, 1990, 2, 203. Schulze, G., and Frenzel, W., Anal. Chim. Acta, 1984,159,95. Golas, J., and Osteryoung, J . , Anal. Chim. Acta, 1986, 181, 211. Jennings, V. J., and Morgan, J. E., Analyst, 1985, 110, 121. Sottery, J. P., and Anderson, C. W., Anal. Chem., 1987, 59. 140. Baranski, A. S., Anal. Chem., 1987, 59, 662. Hua, C., Jagner, D., and Renman, L., Talanta, 1988, 35, 597. Farber, S., Diamond, L. K., and Mercer, R. D., New Engf. J . Med., 1948, 238. 787. Fehlner, P. F., Bencsath, A . , Lam, T., and King, T. P., Immunol. Methods, 1987, 101, 141. Souhami, R. L., Rudd, R. M., Spiro, S. G., Lamond, P., and Harper, P. G., Cancer Chemother. Pharmacol., 1992, 30,465. Green, M. D., Sherman, P . , and Zalcberg, J., Invest. New Drugs, 1992, 10, 31.ANALYST, JUNE 1993, VOL. 118 655 28 29 30 31 32 Sirotnak. F. M., Degraw, J. I., Chello, P. L., Moccio, D. M., and Dorick, D. M., Cancer Treat. Rep., 1982, 66, 351. Rosowsky, A., Forch, A. R., Bader, H., and Freisheim, J . H., J. Med. Chem., 1991, 34, 1447. Rosowsky, A., Forsch, A. R., Moran, G. R., and Freisheim, J. H . , J . Med. Chem., 1991, 34, 227. Rosowsky, A.. Bader, H., and Freisheim, H., J. Med. Chem., 1991, 34, 203. Holland, J . F., and Frei, E., in Cancer Medicine, LEA Febiger, Philadelphia, 1982, pp. 77.5-789. 33 Tellingen, 0. V., Beijnen, J. H., van de Woude, H. R., Bruning, P. F., and Nooyen, W. J., J. Chromatogr. Biomed. Appl., 1990, 529, 135. 34 Dryhurst, G., in Electrochemistry of Biological Molecules, Academic Press, New York, 1977, pp. 324-357. 3.5 Laviron, E. J., J . Electroanal. Chem., 1979, 100, 263. Paper 2105157E Received September 25, 1992 Accepted November 23, 1992

ISSN:0003-2654    

DOI:10.1039/AN9931800649

出版商:RSC    

年代:1993

数据来源: RSC