ISSN:0003-2654    

DOI:10.1039/AN98914FX017

出版商:RSC    

年代:1989

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN98914BX019

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, MAY 1989, VOL. 114 541 Dual-beam Thermal Lens Spectrophotometry in Flowing Samples With Chopped Continuous Wave Laser Excitation Joseph Georges and Jean-Michel Mermet Laboratoire des Sciences Analytiques, Batiment 308, Universite Claude Bernard-1 yon I , 69622 Ville urban n e Cedex, France The thermal lens response for flowing samples following chopped continuous wave laser excitation has been studied with respect to the chopping frequency and flow-rate. The use of two types of flow cell and distinct configurations of flow direction with respect to beam propagation enabled the main ways by which the flow may disturb the thermal lens signal to be considered. With a cell in which the flow is perpendicular to the beam axis, the signal was mainly degraded by the bulk flow the effect of which was related to the chopping period and the residence time of the sample in the probe region as defined by the beam spot size.When the flow was parallel to the beam propagation and the optical path length not sufficiently long the decrease in signal amplitude was found to no longer depend on the bulk flow, but rather to originate from turbulence and mixing in the channel cell. With a chopping frequency of 10 Hz and flow-rate of 1 ml min-1, the signal was degraded by a factor of 2 by the bulk flow and by less than a factor of 1.2 by mixing compared with a static sample signal. At 80 Hz, the signal for continuously flowing samples was nearly independent of flow for flow-rates as high as 1.2 and 2 ml min-1 in the transverse and axial configurations, respectively.However, for flow-injected samples the signal was found to depend on the time response of the detection system. Keywords: Thermal lensing; chopped continuous wave excitation; flowing samples; detection configuration Various thermo-optical spectrophotometric methods are currently being developed for the analysis of weakly absorbing samples.l-3 Basically, excitation by a pump laser followed by thermal relaxation produces a temperature rise within the sample. Because the refractive index of a material changes with temperature, the heated part of the sample behaves as an optical element, lens, prism or grating and as such is probed by a second laser. Among the different techniques used to probe the optical element, thermal lens spectrophotometry is the most devel- oped for liquid samples.This method involves co-linear or crossed-beam thermal lensing according to the experimental configuration. In the former,4-7 the probe and pump beams are made co-linear or nearly co-linear before crossing the sample and the optical element behaves as a spherical lens. In crossed-beam thermal lensing,*lg also referred to as photo- thermal refraction, the pump and probe beams are co-planar but cross the sample at right angles, and the heated sample acts as a cylindrical lens. In both instances, the excitation beam may be a modulated continuous wave (CW) laser or a pulsed laser that produces a time-dependent thermal lens which defocuses the probe beam. The resulting modulated probe-beam intensity is measured with a power-sensitive detector and demodulated with a lock-in amplifier or a boxcar averager.In thermal lensing, the signal may be integrated over the entire optical path of the cell allowing low-concentration samples and relatively large sample volumes. In contrast, with crossed-beam geometry, the probe volume is defined as the interaction region of the two beams; the signal is integrated over the pump beam diameter and is independent of the cell length. This configuration, which allows very small probe volumes (-10-l2 I), is well suited to small-volume analysis and especially to detection in microbore liquid chromatography.10 When thermal lensing is intended for liquid chromato- graphy, flow injection or automated analysis the effect of the flow on the thermal lens behaviour must be considered. Indeed, in chopped CW laser excitation, the thermal lens signal increases with time following the onset of illumination in a chopper cycle and a steady state is achieved only when the rate of heat input from the pump laser is just balanced by the rate of heat conduction out of the illuminated region.The response depends on a characteristic signal rise time constant tc (typically in the order of milliseconds) and the steady state is approached after a time t > lot,. If the solution is flowing at a linear velocity sufficiently high to remove the heated volume from the probe beam region on the time scale of t,, the thermal lens signal will be substantially decreased.11-14 On the other hand, with pulsed excitation, the signal rise time should be much faster than the heat-transfer effects caused by conduc- tion; this time is limited by the rate of thermalisation and lies approximately in the microsecond range.15316 As a result, the signal is nearly independent of the flow because the maximum temperature rise induced by absorbance is attained before the heated sample can be removed from the monitored region of the detection cell by the flowing liquid. According to Dovichi and H a r r i ~ , ' ~ the effects of flow may be divided into two components: bulk flow and mixing.Bulk flow includes axial flow, which is parallel to the cell axis, and transverse flow for which both sample introduction and removal are perpendicular to the cell axis. Bulk flow results in a movement of heated material in the flow direction, from the entrance to the exit of the cell.Mixing, due to turbulence within the cell, has the effect of mixing the heated material with the surrounding material in a section of the flow, independently of the bulk flow.15 A kinetic model, based on time-resolved thermal lens spectrophotometry has been con- structed to improve signal to noise ratios and detection lirnits in flowing samples.13 The aim of this work was to study the thermal lens response of flowing samples using chopped CW laser excitation. In order to evaluate the contributions to the signal amplitude of bulk flow and mixing, the thermal lens behaviour was studied with two types of flow cell: a square-duct cell with flow perpendicular to the laser propagation and a cylindrical channel cell with flow parallel to the beam axis.In both instances, the thermal lens response was studied with respect to the chopping frequency and for flow-rates ranging from tens of microlitres to several millilitres per minute. The use of these cells under the flow injection conditions is examined in terms of the peak amplitude with respect to the flow-rate and injected sample volume. Experimental The single-laser - dual- beam thermal lens experimental set-up used was the same as that described previously.17 An He - Ne542 ANALYST. MAY 1989, VOL. 114 laser that produces a 7.5-mW linearly polarised TEMoo beam at 632.8nm was used. The laser beam was separated by a polarising beam splitter into two linearly polarised beams that emerged from the cube at exactly 90” to each other.The transmitted beam was used as the probe beam and the reflected one as the pump beam. The pump beam was then modulated before being re-combined with the probe beam via a polarisation-sensitive beam splitter in order to minimise energy losses. A lens focused both beams through the flow cell, which was located at about fl times the confocal distance beyond the beam waist and tilted slightly with respect to the normal incidence beam to avoid interference effects due to back-reflection. The pump beam was then stopped with a Glan prism. The intensity of the centre portion of the probe beam was sampled by a 0.4mm i.d. aperture and detected using a reversed-biased PIN photodiode. The signal from the photodiode was observed on an oscilloscope and sent to a lock-in amplifier or a boxcar averager.The entire optical system was mounted on a 0.8 x 1.6 m granite optical table. The choice of focusing lens (f = 100 mm) and the distance from the lens to the detector (0.5m) were optimised in previous work.” Indeed, the use of a lens having a short focal length presented several advantages. As t, varies as the square of the spot size (beam radius) in the cell, the steady state is approached more rapidly and the sensitivity of the thermal response to the sample flow should be reduced. From another point of view, the probe beam diverges more rapidly when focused more tightly, hence the detector can be placed closer to the sample cell. The detection stage is, therefore, less sensitive to laser stability and alignment drift and it allows the experimental set-up to be more compact.Two flow cells were used successively. The first was a 1 .5 X 1.5 mm square-channel cell (Hellma, 176.352-QS) with a quartz window 11 mm high. The cell was positioned with the sample flowing up vertically through the laser beams. The laser propagated at about 5 mm from the cell entrance so the effective dead volume was about 11 1.11. The second cell was a 12-yl, 1.4mm i.d. tubular cell with a path length of 8mm (Kratos, 2900-0146). In the latter, the cell axis and therefore the sample stream were parallel to the laser beam. In order to reduce the noise originating from optical interferences between the probe and pump beams, the two beams were slightly crossed in the sample cell (crossing angle, cu. 0.9’). In this condition, the pump - probe interaction length, I,, could be approximated by the equation18 1, = 2w,/sin@ where w, ic the spot size in the cell and @ the angle at which the beams cross.The spot size could be evaluated from the equation4 7 4 2 = w [) 2 [ I + (z/z,)q where Z is the distance of the cell centre from the beam waist (12 mm), 2, the confocal distance (7.9 mm) and wo the spot size at the waist (40 pm).17 In this instance w, = 72 pm, which corresponds to an interaction length of about 9mm. As this value is not much greater than the path length of the tubular cell, the effective interaction length was calculated by compar- ing the signals given by the same sample in the two cells and assuming that, in the square-channel cell, the effective interaction length equals the 1.5-m path length; an effective interaction length of 6.5 mm was calculated for the tubular cell.The samples or the solvent alone were pumped through the flow cell using a precision micropump fitted with a pulse- damping system and connected to an injection valve equipped with an interchangeable sample loop. The injection valve was connected to the flow cell with 0.25 mm i.d. Teflon tubing. Samples of copper(I1) - EDTA in water were prepared from stock solutions of known absorbance. These were then either pumped continuously through the flow cell or injected with the injection valve into a water stream, as required by the experiment. Results and Discussion Thermal Lens Behaviour For a Continuously Flowing Sample Initially it was of interest to study the shape of the periodic signal seen on the oscilloscope.The time-resolved periodic signals for a solution of copper(I1) - EDTA in water, corresponding to the “on” portion of the chopper and monitored with a boxcar averager, are shown in Figs. 1 and 2. For a static non-flowing sample, the probe beam intensity decreases as the thermal lens forms and the signal, expressed as Zo - I ( [ ) , reaches a steady-state value when t > t,. The temporal response of the thermal lens signal for CW laser excitation is given by the equation19 S ( t ) = S(t = a ) ( l + f&-’ Hence the graph may be used to calculate t,, the time at which the signal is half its steady-state value. The values obtained, 11.6 and 11.9 ms, are in good agreement with the theoretical value of 9 ms calculated from the following equation: t, = w,2pCp/4k where w, is the beam radius in the sample cell (72 lkm), p the density (1 g cm-3), C, the heat capacity (4.19 J g-1 K-1) and k the thermal conductivity (6 X 10-3 J s-1 cm-1 K-1).The effects of flow-rate on the signal intensity using each flow cell are also shown in Figs. 1 and 2. For both cells, the signal amplitude was dependent on the flow-rate; the higher the flow-rate, the faster was the rise time and the smaller the signal amplitude. Nevertheless, the tlow disturbance was much greater when the flow was perpendicular to the laser propagation. Moreover, as the flow-rate was increased, the shape of the time-dependent signal changed; the flow-rate had no effect on the initial intensity because the refractive index of the medium was constant, but with increasing time and during the “on“ phase of the chopper, the signal reached a maximum value and then decreased.According to Dovichi and Harris,l4 the thermal lens signal would change as a result of two effects. Firstly, mixing would result in a decrease i n the effective 1 .o - (II C 01 v) > m a, [r .- a, 0.5 .- .w - 0 Fig. 1. . . 12.5 25.0 37.5 5 Timeims .o Time-resolved thermal lens signal obtained with the square- duct cell (Hellma) and transverse configuration during the “on” phase of the chopper cycle. Chopper frequency = 10Hz (data from 104 chopper cycles were averaged using a boxcar). Flow-rate: A , 0; B, 0.34; C, 0.75: and D, 1.5 ml rnin 1ANALYST, MAY 1989, VOL. 114 543 characteristic time constant t, and a more rapid approach towards the steady state.Secondly, the bulk flow or cooling would lead to a decrease in the temperature gradient and hence reduce the thermal lens effect. In fact, the two effects cannot be distinguished by comparison of the graphs obtained with the different cells (Figs. 1 and 2) if it is assumed that the steady-state signal is influenced more when the flow direction is perpendicular to the beam axis (Fig. 1). With the second configuration, for which mixing is the most important factor, the rising part of the curves is not affected by the flow (Fig. 2). With pulsed excitation,I5 the influence of the flow-rate on signal decay provides further evidence that the signal evolves as a result of competition between the signal rise due to heat input from laser excitation and heat removal originating from mass transfer.At times that are long compared with the thermal time constant t,, the signal decreases because the probe laser no longer interacts with the centre of the refractive index gradient and the steady state is not achieved. The frequency dependence of the lock-in amplifier signal for both static and flowing samples is shown in Fig. 3. For a static sample, the signal increases as the chopping frequency decreases; however, there is some discrepancy between these graphs and those representing the periodic signal. In fact, when measurements were performed at 10 Hz, the periodic signal reached a steady-state value well before the end of the “on” phase of the chopper. In contrast, the output of the lock-in amplifier increased continuously even below 10 Hz.The frequency dependence of the demodulated signal is more consistent with t, because for t, = 10 ms, the signal obtained at 10Hz is theoretically 83% of the maximum signal. With sample flowing through the cell, the maximum signal is approached more quickly when the chopping period is increased, as for the periodic time-dependent signal; however, the amplitude of the former decreases as the flow-rate increases. As already stated, the bulk flow acts as a thermal transport mechanism contributing to thermal diffusion. Nevertheless, the observed frequency dependence of the signal is different for each of the two cells; at 10 Hz, a steady-state signal is obtained at flow-rates >0.5 ml min-1 with the square-duct cell [transverse configuration, Fig.3(n)] and >3 ml min-1 with the tubular cell [axial configuration. Fig. 3(b)]. At higher frequencies (280 Hz), the chopping period is comparable to or lower than t, and the signal is nearly independent of the flow-rate. Figs. 4 and 5 show the variation of the thermal lens signal with flow-rate when operating at a constant chopping fre- quency. At low flow-rates, and up to a critical value that depends on the chopping frequency, the signal remains constant. The maximum flow-rate, d,, giving a constant signal is given in Table 1. In order to compare the two types of flow cell, these values must be expressed in terms of the linear velocity, v I , which is used to determine the residence time of the sample in the laser beam.According to Dovichi and Harris.14 the time scale necessary for axial bulk flow (tubular cell, flow parallel to the beam axis) to influence the thermal lens behaviour is given by the residence time of the sample in the cell. In the present context, we prefer to define the residence time as that time for which the sample is in the active part of the beam, i.e., the effective path length divided by the linear velocity. In contrast, for the square-duct cell, the bulk flow is perpendicular to the beam axis and the residence time of the sample is equal to the beam diameter divided by the linear velocity. According to Nickolaisen and Bialkowski ,Is the signal would be substantially degraded if the solution were flowing at a sufficiently high linear rate to remove the heated sample from the laser beam on a time scale of t, and hence the limiting flow-rate could be calculated by dividing the beam radius by t,.In fact, the limiting residence time, t,, calculated from the graphs shown in Fig. 4(a), depends on the chopping frequency and is in good agreement with the chopping period (Table 1); as long as t, is greater than the chopping period, the Table 1. Dependence of the limiting flow-rate and corresponding residence time on the chopping frequency and beam spot size in the square-duct cell (Hellma) with flow perpendicular to the beam axis. Sectional area of the flux = 0.0225 cm2; d, = maximum flow-rate giving an unchanged signal (from Fig. 4); and tr = 2w,/v1 Frequency/ Chopping dm/ Hz period/ms ml min-1 vJcm s-1 trims 10 100 0.16* 0.12 120 0.32.1- 0.23 I00 20 50 0.34* 0.25 56 0.70t 0.52 46 80 12.5 1.20’ 0.89 16 * w, = 72 pm [data from Fig.4(a)]. i- w, = 119 pm [data from Fig. 4(b)] t - m C m VJ .- A -- _ _ - - - B (;..“ . . . a ..... C ... . ... I I I I 1 0 12.5 25.0 37.5 50.0 Timeims Fig. 2. Time-resolved thermal lens signal obtained with the tubular cell (Kratos) and axial configuration during the “on” phase of the chopper cycle. Chopper frequency = 10 Hz (data from 104 choppcr cycles were averaged with a boxcar). Flow-rate: A, 0; B. 0.34; and C, 1.5 ml min-1 thermal lens will have the same amount of time to evolve and the signal will not be affected by the flow. In order to corroborate this result, the same measurements were carried out with a beam, twice as large, obtained using a focusing lens with a focal lengthf = 200 mm.With this lens the spot size in the sample was calculated to be about 120pm, and the calculated values of t, corresponding to d, [Fig. 4(b) and Table 11 were also found to be in the same range as the chopping period. As t, varies with 14, a dependence of t, on wit, would give a value of d,,, twice as low as that obtained with the smaller cpot size. It is interesting to note that t, corresponds to the chopping period and not to half the time of formation of the thermal lens. Indeed, with a continuously chopped heating source, if the durations of the “on” and “off” phases of the chopper are not sufficiently long with respect to t,, the thermal lens will not reach equilibrium during the “on” phase and will not relax completely during the “off” phase.Hence, the temperature rise in the sample is caused not only by the action of the excitation beam during the current chopper cycle but also by a contribution from the previous chopper cycles. Because in our experiment the “on” and “off” phases of a chopper cycle are equal, the cumulative effect would depend only on the period of the chopper cycle, which determines the equilibrium between repetitive formation and relaxation of the thermal lens. The agreement of t, with the chopping period would indicate that, for a flow perpendicular to the beam axis. bulk flow represents the major contribution to signal disturbance. In contrast, with the tubular cell where the flow is parallel to the beam axis, t, is much greater than the chopping period (Fig.5 and Table 2) and the axial bulk flow is not the limiting544 ANALYST, MAY 1989, VOL. 114 Table 2. Dependence of the limiting flow-rate and corresponding residence time on the chopping frequency for the tubular cell (Kratos) with flow parallel to the beam axis. Sectional area of the flux = 0.0154cm2; d, = maximum flow-rate giving an unchanged signal (from Fig. 5 ) ; and t, = effective path length (0.65 cm) divided by v1 FrequencyiHz d,/ml min--' vllcm s-' t,ls 10 0.37 0.4 1.62 5.5" 5.95 0.11 20 0.77 0.83 0.78 80 2 2.16 0.30 * Value corresponding to X on curve A, Fig. 5 . 0 1 2 F I ow-rateim I mi n - Fig. 4. Dependence of the thermal lens signal (lock-in output) on the flow-rate at constant frequency (Hellma cell, transverse configuration).Focal length of the focusing lens: (a) 100 and ( b ) 200 mm. Chopping frequency: A, 10; B, 20; and C, 80 Hz 0 50 100 Chopping periodims Fig. 3. Influence of flow-rate on the frequency dependence of the lock-in amplifier output for (a) the transverse and ( b ) the axial configuration. (a) Flow-rate: A, 0; B, 0.5; C , 1.4; and D, 2.3 ml min-1. (6) Flow-rate: A, 0; B, 1.5; C, 3; and D , 4 ml min-I factor. Previously, a dependence on the transverse bulk flow was ruled out on the basis that no beam deflection was observed.14 The decrease in signal intensity was thought to be caused by turbulence within the cylindrical channel, which has the effect of mixing the heated material with the surrounding material in a section of the bulk flow. Also, as shown in Figs.1-3, the effect of mixing in the tubular cell is less important than the effect of bulk flow in the square-duct cell. At a flow-rate of 1 ml min-1, and operating at 10 Hz, the signal was about 85% and less than 50% of that for a static sample with the tubular cell (axial configuration) and the square-duct cell (transverse configuration), respectively. At higher flow-rates, the plot of signal intensity versus flow-rate, with the tubular cell, shows a small inflection centred at about 5.5 ml min-1 (Fig. 5 ) ; this corresponds to v1 = 6 crn s-1 and t, = 108 ms. This latter value is close to the chopping period (100 ms) and characterises the limiting axial bulk flow for such a configura- tion. Thermal Lens Response for Flow-injected Samples In order to minimise the effect of flow and to reduce degradation of the thermal lens signal, the chopper was operated at 80Hz.The signals obtained (expressed as peak height) with various flow-rates and injected sample volumes in the range 7-200 p1 are shown in Fig. 6. As expected, the peak 0 2.5 5.0 7.5 Flow-rate/ml min-' Fig. 5. Dependence of the thermal lens signal on the flow-rate at constant frequency (Kratos cell, axial configuration). Chopping frequency: A, 10; B, 20; and C, 80Hz. Focal length of the focusing lens, 100 mm. X corresponds to the limiting axial bulk flow height increased with the injected sample volume and, at moderate flow-rates, reached a steady-state value correspond- ing to the signal obtained for an undiluted sample (Ho). The upper curves in Fig. 6(a) and ( b ) , corresponding to a flow-rate of 0.12 ml min-1, may be used to characterise the dispersion of the system.In flow injection techniques, the dispersion, D , is defined as the ratio of the concentrations before and after the dispersion process has taken place in the segment of sample that yields the signal, i.e., the ratio of signal height Ho obtained when the undiluted sample flows continuously through the cell to peak height, H , in the flow injection mode2O ( D = HO/H). The minimum dispersion is unity, and a dispersion of two was obtained for both cells with an injected volume of about 10-12 pl. By injecting a sample volume equal to the dead volume of the cell, a signal readout of about half the steady-state signal was obtained giving a good compromise between sample volume and peak intensity.ANALYST, MAY 1989, VOL.114 545 100 c .- C 3 2 2 e c .- 2 50 OY a, I: Y m a .- 0 200 v) C 3 4- .- 2 E k 100 + .- a + I: a Q, I: Y m a, a .- 0 100 200 Injected vo I u me/pl Fig. 6. Dependence of peak height on the injected sample volume and flow-rate. (a) Square-duct cell and transverse configuration; and ( b ) tubular cell and axial configuration. Chopping frequency, 80 Hz. Flow-rate: A, 0.12; B, 0.75; and C, 1.5 mI min-1 (lock-in operating with a 3-s time constant); and D, 0.75 ml min-1 (lock-in operating with a 1-s time constant). Hn = steady-state signal for an undiluted sample t c .- v) C a, c Y m Q, 4- .- a 1 2 3 4 5 6 Timeimin Fig. 7. Distortion of the rising part of the peak of a flow-injected sample (tubular cell, axial configuration).Chopping frequency = 80 Hz; flow-rate = 50 PI min-1; and injected volume = 60 pl When the flow-rate was increased to a range where the thermal lens response for a continuously flowing sample was independent of flow, the result was a loss of sensitivity. This may have originated partly from an increase in the dispersion and mainly from the response time of the detection system. As the photodiode had a rise time of 1 ns, the large decrease in the peak intensity arose from the response of the lock-in amplifier, which operated at a 3-s time constant. As is also shown in Fig. 6, operation at a 1-s response time cancels the Time Fig. 8. Typical recording of flow-injected sample peaks (square-duct cell, transverse configuration). Chopping frequency, 80 Hz; flow-rate, 0.34mlmin-1; and solvent, water.(a) Base-line noise with a 1-mV lock-in scale. ( b ) Flow-injected sample peaks in triplicate with a 5-mV lock-in scale. Copper(I1) - EDTA in water; A = 0.024; cell path length, 1.5 mm; and injection loop volume, 7 PI sensitivity loss almost completely, at least at moderate flow-rates. Nevertheless, if the 1-s time response is convenient at low flow-rates, allowing satisfactory spatial resolution of the peaks, it leads to a greater band pass of the fluctuations and spikes on the recording and, consequently, to less reprodu- cible peak heights at higher flow-rates. Considering that, for a continuously flowing sample, the effect of flow is less with the tubular cell (Kratos), the sensitivity loss with this configura- tion is, by comparison, more important in the flow injection mode. This result can be discussed with respect either to the volume or the length of the probed element, which is different for each configuration.In the transverse configuration (Hellma), the probed length is equal to the beam diameter (about 0.14 mm), whereas in the axial configuration (Kratos), it equals the effective path length (about 6.5mm), a path length ratio of 45 : 1. This may be a serious drawback at low flow-rates and for small injection volumes, as can be seen in Fig. 7 in which the rising portion of the peak obtained at a low flow-rate in the axial configuration has a shoulder the width of which is exactly equal to the time necessary for the front of the sample zone to cross the path length of the cell (6.5 mm).In this part of the curve, which actually corresponds to the dead volume of the cell (10 pl), the intensity increases with both the sample concentration and the effective path length. Although both cells have the same dead volume, the transverse configuration enabled smaller volumes to be probed and this should result in a better spatial resolution of the peaks for small sample volumes and low flow-rates. On the other hand the axial configuration provided greater sensitivity. Systematic comparison of cell performance with respect to flow-rate and injected sample volume would have been tedious and was not the aim of this work. As an example of such a comparison, the flow injection response of the thermal lens at a moderate flow-rate and with an injected sample volume approximately equal to the dead volume of the cell is shown in Fig.8. The blank response was fairly constant and was not significantly affected by the flow for flow-rates <1 ml min-1. The peak to peak base-line noise corresponded to an absorbance of 7.2 x 10-5, which, with a peak signal to peak to peak blank noise ratio of 2 (i.e., loo), gave a minimum measurable absorbance of 1.5 x 10-4. In conclusion, the effects of axial bulk flow and mixing, which arise from the flow cell geometry and beam propagation with respect to the flow direction, may significantly affect the thermal lens signal. Nevertheless, with an appropriate confi- guration and at flow-rates compatible with most applications of flowing sample detection, the signal is reduced only to a small extent.The small volume probed with the transverse configuration is more convenient for low flow-rate and small546 ANALYST, MAY 1989, VOL. 113 injection volume applications such as liquid chromatography; the axial configuration is well suited to high flow-rate and large injection volume applications, such as high-speed titrations by flow injection or automated analysis. The performance of the transverse configuration (Hellma) can be improved by increasing the optical path length and reducing the channel width to obtain a higher sensitivity without increasing the dead volume of the cell. However, the combination of high sensitivity and very low detection volume is not compatible with co-linear thermal lensing because of the dependence of Beer’s law on the path length and because of optical constraints.Finally, the development of a very small volume detector for microbore liquid chromatography requires the use of the crossed-beam geometry implemented in photothermal refraction.21.22 1. 2. 3. 3 . 5. 6. References Morris, M. D . , and Peck, K., Anal. Chem., 1986, 58, 811A. Dovichi, N. J . , Crit. Rev. Anal. Chem., 1987, 17, 357. Harris, T. D . , Anal. Chem., 1982, 54, 741A. Harris, J . M., and Dovichi, N. J . , Anal. Chem., 1980,52,695A. Fang. H. L.. and Swofford, R. L., in Kliger, D . S . , Editor. “Ultrasensitive Laser Spectroscopy,” Academic Press, New York, 1983. p. 176. Harris, J . M., in Piepmeier, E. H., Editor, “Analytical Applications of Lasers,” Volume 87, Wiley-Interscience, New York, 1986, p. 451. 7. 8. 9. 10. 11. 12. 13. 13. 1s. 16. 17. 18. 19. 20. 21. 22. Georges, J., and Mermet, J.-M., Analusis, 1988, 16, 203. Dovichi. N. J . , Prog. Anal. Spectrosc., 1988, 11, 179. Dovichi, N . J . , Nolan, T. G., and Weimer. W. A,. Anal. Chem.. 1984, 56, 1700. Nolan, T. G., and Dovichi, N. J . , Anal. Chem., 1987.59.2803. Weimer, W. A . , and Dovichi, N. J . , Appl. Opt.. 1985,24,2981. Weimer, W. A , , and Dovichi. N . J.. Appl. S‘pectrosc.. 1985.39, 1009. Weimer, W. A , , and Dovichi. N. J.. Anal. Chem., 1985, 57, 2436. Dovichi, N. J . . and Harris, J . M., Anal. Chem.. 1981, 53, 689. Nickolaisen, S. L., and Bialkowski. S. E.. And. Chtw7., 1986, 58. 215. Nickolaiscn, S. L.. and Bialkowski, S. E.. Anal. Clzein.. 1985, Georges, J . , and Mermet, J.-M.. Analysr, 1988, 113, 11 13. Jackson, W. R . , Amer, N. M . , Boccara, A. C., and Fournier, D . , Appl. Opt., 1981, 20, 1333. Carter. C. A . , and Harris, J . M., Appl. Opt.. 1983, 23, 376. Rfiiieka, J . , and Hansen, E. H . , “Flow Injection Analysis,” Wiley, New York, 1981 Kettler, C. N.. and Sepaniak, M . J . , A i ~ u l . Chein., 1987, 59, 1733. Nolan, T. (3.. Hart. B. K., and Dovichi, N. J . , Anal. Clzem., 1085, 57, 2703. 57,758. Paper 8103383H Received August 22nd, 1988 Acc~pted Junuar? 4th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891400541

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST. MAY 1989. VOL. 114 547 Study of Dynamic Processes by Impulse Response Photoacoustic Spectroscopy Richard M. Miller Unilever Research, Port Sunlight Laboratory, Quarry Road East, Bebington, Wirral, Merse yside L63 3J W, UK Graham R. Surtees and Christopher T. Tye Department of Instrumentation and Analytical Science, University of Manchester Institute of Science and Technology, P.O. Box 88, Manchester M60 KID, UK Impulse response photoacoustic spectroscopy has been used to study variations in chromophore distribution with time in a dynamic system. A system in which a dye diffuses through a polymer film has been examined, and it is shown that it is possible to monitor the diffusion process in situand non-destructively. Comparison of the experimental results with a theoretical model demonstrates that the results are consistent with classical diffusion processes.Keywords: Photoacoustic spectroscopy; impulse response measurements; diffusion; polymers Alexander Graham Bell discovered that when an absorbing solid mounted in a closed container was illuminated by an amplitude modulated light source, an acoustic signal was produced at the same frcquency as the modulation of the light.1 This phenomenon became known as the optoacoustic or photoacoustic effect, and has subsequently found applica- tion in the analysis of gases,2 the measurement of vibrational relaxation times of gases,'.A the spectroscopy of opaque and light-scattering solids5 -7 and the imaging of sub-surface details in thin fi1ms.g." The mechanisms behind the photoacoustic effect are well documented,7.1() and of its many applications the spectro- scopic characterisation of solids is the most widespread.Besides its obvious advantages in the analysis of difficult samples, photoacoustic spectroscopy (PAS) offers a unique, no n - de s t ruc t i ve method of o I-, t ai n in g depth re 1 ate d i n form a- tion in solids.7.1' When a sample is illuminated by a modulated radiation source of a wavelength absorbed by the sample, a modulated heat source will be produced by internal conver- sion of excited molecular states. The spatial distribution of the heat source will be a function of the optical absorption of the sample. A sample of high absorbance will produce a locaiised heat source close to the illuminated face of the sample.A weaker absorber will produce a more distributed heat source in the sample. A sample with strong sub-surface absorption will produce a localised sub-surface heat source, and so on. As the heat released by radiation absorption must diffuse through the sample to the surface before detection, different initial heat source distributions produce variations in the evolution of the photoacoustic signal with time. Analysis of the photoacoustic signal in the time domain provides informa- tion on the absorption coefficient, distribution of the chromo- phore and thermal properties of the sample. Attention has focused on the measurement of the photo- acoustic impulse response of a sample as a means of gaining the necessary time domain information. Studies have included the measurement of film thickness,l2 dye distribution in coloured films'3-ls and the examination of chromophore distributions in intact leaves and petals.'3.16 It has been shown that the shape of the photoacoustic impulse response depends on the chromophore distribution and that by measuring the impulse response at a series of wavelengths, a response surface can be constructed which gives information about variations in the photoacoustic signal with both wavelength and depth.This technique is usually called impulse response photoacoustic spectroscopy (IMPAS) or correlation photoacoustic spectro- scopy (CPAS). Impulse response photoacoustic spectroscopic measure- ments can be performed by illuminating a sample in a photoacoustic cell with a brief pulse of light and recording the resulting photoacoustic transient.However, the duty cycle of such a measurement system is very low and time-consuming signal averaging is required to produce usable results.17 A more convenient method of making impulse response measurements is to use multi-frequency modulation of the incident radiation and to recover the system impulse response by cross-correlation of the modulation with the detected photoacoustic signal. This technique is used widely in engin- eering and telecommunications1~.19 and has proved very successful in recovering impulse responses in noisy environ- ments. It can be shown that the cross-correlation between the input and output of a system is the convolution of the auto-correlation of the input to the system with the system weighting function.*y If the input signal has the same auto-correlation function as an impulse, then the cross-corre- lation function will be a good estimate of the system impulse response. A suitable multi-frequency test signal for this application is the pseudo-random binary sequence (PRBS).This has the same auto-correlation function as a single pulse, but has a 50% duty cycle. As the strength of the photoacoustic signal is a function of the total amount of energy released in the sample per unit time, the use of PRBS modulation with cross-correlation signal recovery provides a dramatic improve- ment in signal to noise ratio over single pulse methods.12 It has been the basis of most of the IMPAS studies.12-'6 An alternative approach has been demonstrated successfully by Mandelis and co-workers,20J1 who used a swept frequency sine wave modulation coupled with a correlation signal recovery process to obtain the system transfer function and impulse response. As the photoacoustic impulse response can be measured in a short time using the cross-correlation technique, it is logical to apply the method to systems where the chromophore distribu- tion is a function of time.Hence dynamic physico-chemical systems can be studied. In a preliminary paper,22 we have reported the observation of dynamic changes in the impulse response as a result of varying chromophore distribution. In this paper we extend the work to a study of dye diffusion in a polymer matrix. Experimental Apparatus A block diagram of the experimental apparatus used is shown in Fig.1. The radiation source was a krypton ion laser (Innova548 - Beam expander Laser - Modulator - ANALYST, MAY 1989, VOL. 114 Sample cel I Film I Clock clenerator U I - 1 Fig. 1. Block diagram of the experimental apparatus Dye reservoir S u bstr a t e Fig. 2. Schematic diagram of the dye diffusion test sample 90-K, Coherent UK, UK). Measurements were made using the 647-nm laser line, at an output power of 40mW. Amplitude modulation was carried out using an acousto-optic modulator (Model 304, Coherent UK). The PRBS modula- tion signal was generated using a PSI A108 correlator (Prosser Scientific Instruments, UK). A 127-bit PRBS was used for all experiments. The clock frequency of the PRBS was deter- mined by a variable frequency square wave generator (Model TG 310, Levell, UK).A clock frequency of 333 Hz was used for all experiments, giving a PRBS equivalent to a 3 x 10-3 s impulse excitation. The modulated source was directed via a beam expander on to samples contained in a gas-microphone photoacoustic cell (Model OAS 401, EDT Research, UK). The sample chamber of this cell had a usable area of 15 x 5 mm and was ca. 1 mm deep. The photoacoustic signal from the cell was amplified by a low noise pre-amplifier (Model 450S, E G & G Brookdeal, UK) and passed into a digital signal analyser (Solatron 1200, Solatron, UK), where it was cross-correlated with a reference PRBS from the generator. Data acquisition and the storage of impulse responses were controlled by an HP 9816 microcom- puter [Hewlett-Packard (UK), UK] connected to the Solatron 1200 analyser through an IEEE-488 interface bus.Software which allowed the remote control of the signal analyser, the acquisition and storage of data and its subse- quent manipulation and display was written in HP BASIC 2.0. Impulse responses were plotted on a digital plotter [Model HP7470A, Hewlett-Packard (UK)] connected to the IEEE- 488 interface bus. Sample Preparation Small amounts of a blue, solvent-based ink were placed on to the underside of pieces of a waterproof sealing film (Parafilm, Gallenkamp, UK). The Parafilm was then mounted on to a piece of double-sided adhesive foam strip, providing a thermally thick substrate. The ink drop was sandwiched between the film and substrate (Fig. 2). As soon as the ink made contact with the film, it began to diffuse through the film to the upper surface.The samples were placed in the photoacoustic cell, film uppermost, and illuminated through the film. Light would be absorbed by the dye and heat released. Method At specified time intervals, the microcomputer initiated the impulse response measurement by the signal analyser. Each impulse response was the result of averaging 50 measurement cycles, typically taking 35s in total. At the end of each measurement the impulse response was transferred to the microcomputer. Each experiment consisted of 20 individual impulse responses. The time between impulse response measurements could be varied at the beginning of each experiment to take account of varying diffusion rates. At the end of each experiment the 20 impulse responses were stored on disk for later analysis. Digital Simulation In order to be able to compare the results with theoretical predictions, a digital simulation model was used.A numerical method of solution was chosen rather than an analytical solution, because of the difficulties in handling complex distributed heat sources in the currently available closed form solutions. The numerical method of solution permits flexible definition of the problem conditions and rapid modification of the underlying assumptions. It is therefore a more appropriate approach for the type of problem being considered here. The model used was a one-dimensional simulation of the optically induced heat flux in the sample under pulse excitation, using the finite difference method of solution.The finite difference simulation involves iteratively calculating the temperature distribution throughout the model system for successive small increments in time. The changes in tempera- ture distribution are calculated by allowing for heat transfer between adjacent discrete elements in the model, taking into account any known heat sources or sinks within the model. To simulate a system where the chromophore distribution through the thickness of the film was varying with time due to diffusion, a separate diffusion model was constructed in which the rear surface of the sample was assumed to be in contact with an infinite reservoir of diffusing species. Using similar methods of calculation, the fractional concentration of the diffusing species in each of the thickness elements of the sample was calculated for a series of time intervals after the start of the diffusion experiment.The results of these calculations were then fed to the main finite difference mode!ANALYST, MAY 1989, VOL. 114 549 where the concentration distribution at a particular instant in time was used, together with the specified optical absorption coefficient of the chromophore, to generate the initial heat source distribution which would give rise to the photoacoustic impulse response. The modified impulse response was calcu- lated for each time increment and the resulting series of impulse responses were compared with the experimental results. By varying the parameters and assumptions used in the dye diffusion calculations, various possible interpretations of the experimental data could be assessed.The simulation model was written in FORTRAN 77 on a VAX 11/785B (Digital Equipment, USA). Double precision arith- metic was used to avoid the accumulation of truncation rounding errors in the iterative calculation of heat diffusion. Graphical output was obtained using the FREELANCE + graphics package (Lotus Development, USA). Full details of the model and the simulations carried out have been published elsewhere .23.*4 Results and Discussion Fig. 3 shows the sample impulse response at the beginning of a diffusion experiment and Fig. 4 shows the response after diffusion for 30 min. Significant differences between the impulse responses are evident. In Fig. 3 there is a small sharp peak at a short delay, followed by a much broader, flatter peak with a much greater peak delay. The first feature is probably due to residual surface absorption.This phenomenon has also been observed in other systems. 14.15 Most samples have a residual broad band optical absorption at the sample/gas interface. This residual absorp- tion produces a small sharp peak in the response at a very small delay. Improved sample handling can reduce the size of 0 50 100 Timeil O-3 s Fig. 3. at time zero Photoacoustic impulse response of the dye diffusion sample 0 50 100 TimellO-3 s Fig. 4. after diffusion for 30 min Photoacoustic impulse response of the dye diffusion sample this surface feature, but cannot eliminate it. However, this surface feature can be useful in providing a reference time marker in the system, which is relatively constant between impulse responses.The second feature is typical of the impulse response obtained from a sub-surface chromophore. 14 It is broadened by diffusion and of relatively low amplitude due to energy dissipation within the sample as the heat propagates. In Fig. 4, the diffusion of the dye has brought the chromophore much closer to the surface of the sample. As a result, the peak of the impulse response has moved to much shorter delays and the peak height is much greater. The width of the impulse response peak is also significantly reduced. All these changes are consistent with a shift in the chromophore position from deep within the sample to much closer to the surface.14 The surface feature is still present, but has been obscured by the much bigger response from the chromophore.This interpretation is supported by the results shown in Fig. 5. Three impulse response curves are shown for static systems. Curve A shows the impulse response obtained from a sample of film applied directly to the standard adhesive foam strip. For this sample there is no strong sub-surface absorption and the impulse response shows only the residual surface absorp- tion. The total signal intensity is very low for this sample. Curve B shows the impulse response obtained for a sample in which carbon black was trapped between the foam substrate and the film. As a non-diffusing material, the carbon black gives an indication of the type of impulse response which would be observed at diffusion time zero if it were possible to perform this experiment.The expected broad sub-surface signal is seen. Curve C shows the result of applying a high concentration of a strongly absorbing dye directly to the surface of sample A. A very strong surface signal is obtained which grows in during the 3 X 10-3 s period equivalent to the pulse width of the excitation and which then decays exponen- tially. This result should be equivalent in form to the result obtained from the diffusion experiment after infinite diffusion time. The surface features obtained from these static samples are sharper than those seen in the diffusion experiments. This is attributed to a different batch of film being used for the diffusion studies that had a small residual absorption through- out the film thickness giving a more distributed surface feature.Fig. 6 shows curves A and C from Fig. 5 normalised with respect to each other and overplotted. Despite the large difference in signal amplitude, the signal form is very similar for the two curves. Both are surface signals and exhibit virtually identical time courses. This is further evidence that the system is linear over a large range of signal amplitudes.12 The response from the dye-coated sample shows a slight I t I 0 25 50 75 Time/10-3 s Fig. 5 . Photoacoustic impulse response for static systems. A, Untreated film; B, untreated film backed with carbon black; and C, film surface coated with a strongly absorbing dye550 ANALYST, MAY 1989, VOL. 114 0 25 50 75 TimeilO-3 s Fig.6. normalised t o the same amplitude Photoacoustic impulse responses A and C from Fig. 5 I A I 0 50 100 Timeil 0-3 s Fig. 7. recorded at 3-min intervals over a period of 30 min Photoacoustic impulse responses for the dye diffusion sample positive deviation from the typical exponential decay in the region just after the peak of the response. This may be due to penetration of the dye into the matrix or an absorption coefficient for the dye at the laser wavelength which allows some penetration of the incident radiation below the sample surface. The response is consistent with a resulting surface- weighted distributed heat source.23 Because of the higher noise level on the uncoated sample, the two responses are not significantly different. Fig. 7 summarises results from a complete experiment.It consists of ten overlaid impulse response measurements taken at 3-min intervals over a period of 30 min. At the start of the run the impulse response appears as in Fig. 3, with two distinct peaks, one arising from a surface feature and the other from a sub-surface chromophore. During the course of the experi- ment there is a gradual movement of the sub-surface peak to shorter time delays. This is due to dye diffusing towards the surface of the film. The changes in peak delay and peak height with time are illustrated in Fig. 8. The peak delay shows a decaying trend with increasing diffusion time as the chromophore diffuses towards the surface of the sample. The peak height shows an approximately linear increase with time. A linear regression line has been plotted for the changes in peak height.These results are typical of a number of experiments which were carried out and which show influences due to the experimental conditions. For example, the apparent linear increase in peak height with time was observed for all experiments, although the slope varied. This variation was associated with the amount of dye solution used and may indicate variations in the rate of diffusion. - 6 a, -0 Y m Q 0 Is, a L Y m a L 2 .- n 0 n o o o 1 I I I 1 10 20 30 40 50 Ti rne/min Fig. 8. Experimental variation of peah height (0) and peak delay (0) with time for thc dye diffusion sample. The linear regression line for the peak height data is indicated 0 0 I I 10 20 30 40 50 Timeimi n Fig. 9. Calculated variation of peak height (0) and peak delay (0) with time assuming a moving boundary transport mechanism.The linear regression fine for the pcak height data is indicated Although there is a clear empirical relationship between the changes in the impulse response and changes in the cbromo- phore distribution with time, we needed to determine if the experimental results were consistent with a reasonable physical model of the processes involved. To attempt to resolve this question, the experimental results were compared with the results obtained from the digital simulation model. Initially, we wished to determine whether the experimental results were consistent with the assumption that the dye was being transported through the membrane in accordance with Fick’s law.25 Two alternative hypotheses were evaluated; that the chromophore was diffusing according to Fick’s law or that the chromophore formed a sharply defined boundary which progressed through the thickness of the polymer membrane at a constant velocity.The second hypothesis is not physically realistic, but represented a test of whether we could distin- guish between two very different assumptions about the transport mechanisms using the experimental data available. Fig. 9 summarises the predicted responses assuming the moving boundary hypothesis. The peak delay shows a positive deviation from a straight line rather than the negative deviations seen experimentally and the peak height shows a strong negative deviation from the approximately linear dependence observed in the experimental data. It is clear that the form of the curves are completely different to those shown in Fig.8. Variation of the absorption coefficient for the chromophore and the boundary velocity did not change the shape of the curves substantially and it is therefore clear that the experimental data cannot be accounted for by postulating a moving boundary transport mechanism.ANALYST, MAY 1989, VOL. 114 55 1 Ti meim in Fig. 10. Calculated variation of peak height (0) and peak delay (13) with time assuming a diffusion according to Fick’s law. The linear regression line for the peak height data is indicated Fig. 10 summarises the theoretical predictions based on the assumption of classical diffusion in accordance with Fick’s law. The linear relationship between peak height and diffusion time observed in the experimental data is also found in the simulated data and the curve representing the variation of peak delay with diffusion time is of the same general form in both experimental and simulated data.The linear relationship between peak height and diffusion time is preserved over a range of diffusion coefficients and the slope varies in a manner analogous to that observed in the experimental data. Fig. 11 illustrates the variation in peak delay with diffusion time for the experimental data and a simulation experiment where the variation in peak height with time agrees closely with the experimental data. It is important to note that the simulation results are expressed in arbitrary units. It is therefore easy to scale the simulation results so that the end points of the curve agree with the experimental data.Tn assessing the fit between the simulation and experimental data, the criterion of agreement must be the deviations of the simulation and experiment between these two points. It is clear that the simulated curve is not inconsistent with the experimental one, although the deviations are still greater than can be accounted for by the scatter in the experimental points. In addition, because of the way in which the simulation has been carried out it is not possible to specify the diffusion coefficient in physically meaningful units. Further work will be required on a range of samples to quantify the transfer function of the cell and to complete the mapping between the model and sample.At present, the technique can be used on a comparative basis, but not on an absolute basis. Conclusions Impulse response photoacoustic spectroscopy has been used successfully to monitor chromophore distribution in a dynamic physico-chemical system. The results have been shown to be consistent with simulations based on physically realistic assumptions. The method is non-destructive and offers considerable advantages over alternative approaches to characterising dynamic systems. It is anticipated that the technique will find application in a wide range of systems, including the uptake of dye by films and fibres. the dynamics of controlled release drug formulations and the study of transport mechanisms in biological and clinical systems. Future developments of the technique should include a better theoretical understanding of the relationship between the chromophore distribution and the resulting impulse response, its use with tunable light sources for time dependent IMPAS and the application of non-contacting detection systems to allow a wider range of sample types to be studied.Y a $ 1 n 6, Time Fig. 11. for the variation o f peak delay with diffusion time Comparison of experimental (0) and simulated ( A ) data We thank Simon Johnson and Ed Staples of Unilever Research, Port Sunlight Laboratory, for their assistance in the studies carried out on the static systems. References 1. 2. 3. 4. 5 . 6. 7. 8. 9. 10. 11. 12. 13. 13. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. Bell, A. G., Philos. Mag., 1881, 11, 510. Pfund.A. H., Science, 1939, 90, 326. Read, A. W., Adv. Mol. Relaxation Processes, 1967, 1 , 257. Cottrell, T. L.. Macfarlane. I. M., Read, A. W., and Young. A . H., Trans. Faraday Soc., 1966, 62. Adams, M. J., King. A. A , , and Kirkbright, G. F., Analyst, 1976. 101, 119. Rosencwaig, A , . in Pao. Y.-H., Editor, “Optoacoustic Spec- troscopy and Detection,” Academic Press. New York, Lon- don, 1977, Chapter 8. Rosencwaig, A.. “Photoacoustics and Photoacoustic Spectro- scopy,“ Wiley-Interscience, New York, 1980. Kirkbright, G. F., and Miller, R. M., Analyst, 1982. 107. 798. Kirkbright, G. F., Liezers. M., Miller. R. M., and Sugitani. Y., Analyst, 1984, 109, 465. McDonald. F. A., Am. J . Phys.. 1980, 48, 31. Adams, M. J . , Beadle, B. C., King, A. A . , and Kirkbright, G. F., Analyst, 1976. 101. 553. Kirkbright, G. F., and Miller, R . M., Anal. Chem., 1983, 55, 502. Sugitani. Y., Uejima, A.. and Kato, K., J . Phommwrics, 1982, 1, 217. Kirkbright, G. F., Miller, R. M., Spillane, D. E. M., and Sugitani, Y . , Anal. C‘hem., 1984. 56, 2043. Uejima, A. , Sugitani, Y., and Nagashima, K., Anal. Sci., 1985, Kirkbright, G. F.. Miller, R. M., Spillane, D. E. M., and Vickery, I. P., Analyst, 1984, 109, 1443. Cox, M. F., and Coleman, G. N., Anal. Chem.. 1981,53,2034. Davies, W. D. T., “System Identification for Self Adaptive Control,” Wiley, London, 1970. Lynn, P. A . , “An Introduction to the Analysis & Processing of Signals,” Second Edition, Macmillan, London, 1984. Mandelis, A . , Rev. Sci. Instrum., 1986, 57, 622. Mandelis, A., and Sin, E. K. M., Phys. Rev. B , 1986,34,7209. Miller, R. M., Surtees, G. R., Tye, C. T., and Vickery, I. P . , Can. J . Phys., 1986, 64, 1146. Miller, R. M., Can. J . Phys., 1986, 64, 1049. Miller, R. M., Spectrochim. Acta, Part B , 1988, 43, 687. Crank, J., “The Mathematics of Diffusion,” Second Edition, Oxford University Press, London ~ 1975. 1, 5 . Paper 81039980 Received November 15th, 1985 Accepted Junuary 6th, I989

ISSN:0003-2654    

DOI:10.1039/AN9891400547

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST. MAY 1989, VOL. 114 553 Production and Certification of Ten High-purity Polychlorinated Biphenyls as Reference Materials Alan S. Lindsey and Peter J. Wagstaffe Community Bureau of Reference, BCR, Commission of the European Communities, Rue de la Loi 200, B- 1049 Brussels, Belgium The development by the Community Bureau of Reference of ten polychlorinated biphenyl reference materials of certified purity is reported. The identity of the specially synthesised compounds was mostly confirmed by either nuclear magnetic resonance spectroscopy or X-ray crystallography. Measurements to establish the purity of the ten compounds were carried out in a collaborative certification campaign by 11 European Laboratories. The purity was based on the determination of individual impurities in each compound by applying various well-established measurement techniques. A statistical analysis of the results obtained provided the basis of the certified purity values, which also take account of the presence of trace inorganic i m pu rities.Keywords : Polychlorinated biphenyls; reference materials; high -purity; certification The contamination of the biosphere by polychlorinated biphenyls (PCBs) has been long established,l-4 and their potential health hazards have been well documented.”-” The European Economic Community and many other countries have promulgated various regulatory controls on the manufac- ture of PCBs and acceptable PCB levels.9-11 The regulation of acceptable levels of these compounds in environmental samples and foodstuffs, where the levels can be very low, rests on the accuracy and reliability of their identification and quantitative analytical measurement.In turn the analytical techniques which are applied for this purpose, based mainly on gas - liquid chromatography (GC) and high-performance liquid chromatography (HPLC) with various detection systems, are essentially dependent on the availability of very pure reference samples of prominent PCBs which can serve as calibration materials for the measurement of specific PCB compounds. A comprehensive review of the analytical chemistry of PCBs has been published recently.’* This paper describes the production and certification of ten PCBs which possess purities of between 0.986 and 0.999 g 8-1. The choice of this series of compounds was based on the following considerations: key representatives of PCBs found in the environment; desirability of the availability of at least one high-purity isomer representing each of the six isomeric groups of dichloro- to heptachloro-biphenyls; toxicological aspects; non-availability from commercial sources; and requirements of legislative control within the European Community.In the experimental work associated with the production of the series of PCB reference materials (RMs) used here, the following quality criteria were adopted: the materials should possess a purity of preferably at least 0.990 g g-1; the purity analysis should be carried out by at least three different analytical methods in each participating laboratory; and impurities detected at mass fractions of 0.001 g g-1 or higher should be identified, if possible, and determined quantita- tively. The general aim of the method was to certify the purity of the reference material with a total uncertainty of less than 0.01 g g-1, by subtracting the impurities from unity.The certified PCB reference materials are intended mainly for the qualita- tive and quantitative calibration of analytical apparatus and methods (e.g., determination of retention times, response factors and reference spectra in chromatographic and spectro- scopic analyses) and for the study of biological activity. As the individual PCB compounds have been systematically numbered based on the IUPAC rules of substitution for biphenyl,l3,14 for convenience of reference in this paper, these systematic numbers are indicated in addition to the chemical nomenclature and are given in the form PCB-n where n represents the position number in the sequential series.Experimental Syntheses of Polychlorinated Biphenyls One PCB isomer (PCB-28) was prepared by Sandmeyer’s method in which the diazonium group was replaced with chlorine using copper(1) chloride. Two of the symmetrically substituted PCBs (PCB-52 and PCB-153) were prepared from an appropriate, specially synthesised, iodo-derivative by an Ullmann-type condensation reaction. The other seven com- pounds were prepared from suitable chloroanilines, which were diazotised and reacted with an appropriate chloroben- zene to provide the required compound. In some instances the product was admixed with one or more isomeric compounds.The methods of synthesis and the identification techniques used are given in Table 1 (see also reference 22). Purification of Polychlorinated Biphenyls After removal of the reaction solvent by steam distillation and an initial crystallisation of the crude product (for PCB-20 and PCB-35 after separation by vacuum distillation), the materials were purified by column chromatography (on eithcr silica gel or alumina) and then recrystallised. A contaminating by-product which is formed during the synthesis of PCB-153 is 2,3,7,8-tetrachlorodibenzofuran (TCDF). After purification of PCB-153 TCDF was found to be present at a concentration of about 400 vg g-1, which, in view of its potential toxicity, was considered to be too high. Therefore, an additional chromatographic purification was carried out, and a GC and mass spectrometric (MS) examina- tion of the final purified sample indicated that TCDF, if present, was at concentrations below 1 pg g-1. Characterisation of the Polychlorinated Biphenyls Proton nuclear magnetic resonance (NMR) spectra were recorded using a Varian Model XL-100 FT instrument; tetramethylsilane was used as the internal reference and CDC13 as solvent.The X-ray analyses were carried out using an Enraf-Nonius CAD-4 diffractometer (w,O scan, Mo K or Cu K radiation monochromated by pyrolytic graphite) .23,25 Mass spectra recording the relative isotopic molecular ion intensities, shown in Fig. 1, were measured using a Finnigan Model 4510 quadrupole mass spectrometer equipped with an554 ANALYST.MAY 1989, VOL. 114 Table 1. Synthesis and identification of polychlorinated biphenyl RMs Synthesis RM Biphenyl compound reference 289 PCB-8 2,4’-Dichloro- 15-17 Identity M.p.l“C confirmed by Reference 22 23 15,24 23 19,24 19 19 24 25 19 40.5 M.p., IR, X-ray crystallography 290 PCB-20 2,3,3 ’-Trichloro- 17 42.0 X-ray crystallography 291 PCB-28 2,4,4’-Trichloro- 15 56.0 M.p., lR, mass spectrometry 292 PCB-35 3,3 ’ ,4-Trichloro- 17 62.0 X-ray crystallography 293 PCB-52 2,2’ ,5 ,5 ‘-Tetrachloro- 15,18 88.0 M.p., IR, relative molecular mass, NMR 294 PCB-101 2,2’ .4,5,5’-Pentachloro- 19 76.5 NMR,IR 295 PCB-118 2,3‘,4,4’,5Pentachloro- 20 111.5 NMR.IR 296 PCB-138 2,2‘ ,3,4.4‘ ,5 ‘-Hexachloro- cf. 19 81 .O NMR. IR, mass spectromc try 297 PCB-153 2,2‘ ,4,4’ ,5 .5 ’-Hexachloro- 21 102.0 X-ray crystallography 298 PCB-180 2,2’,3,4,4’ ,5,5‘-Heptachloro- 19 114.5 NMR,IR 100 60 20 - 36C 222 222 L 258 256 ’[;62 J 256 262 I 52 360 j2 ’ 360 364 I I , , ?70 I BPClp PCB-8 BPC13 PCB-20 PCB-28 PCB-35 BPCls PCB-138 PCB-153 100 60 20 20 BPC14 PCB-52 BPCIS PCB-101 PCB-118 BPC17 PCB-180 mlz Fig.1. intensities for biphenyls substituted with n chlorine atoms ( n = 2-7). Corresponding measured intensities shown against each PCB number.] Mass spectra: comparison of theoretical and measured relative intensities of isotopic parent ions. [BPCl, = theoretical relative ion lncos data system. Electron impact mass spectra were OV-101, CP-Sil-SCB, SE-54, BP-5, DB-5 and other similar recorded at an electron energy of 70 eV. types, in addition to Apiezon L, were utilised as stationary phases.Both isothermal and temperature programmed (in the Methods of Quantification of Impurities range 60-300 “C) conditions were applied. A solution of the PCB in a suitable solvent such as hexane, isooctane, di- Gas - liquid chromatography was carried out using both packed and capillary columns under various conditions. Commercially available products such as the silicone-based chloromethane, toluene and cyclohexane at mass concentra- tions of between 0.5 and 20 mg g-1, was applied either as a cold on-column injection or as a split or splitless injection. TheANALYST, MAY 1989, VOL. 114 555 eluted peaks were detected by tlame ionisation or by coupling to a mass spectrometer. For HPLC, carried out at room temperature, columns of 15 to 25 cm in length and 4-5 mm internal diameter were used. A variety of both normal and reversed-phase type packings were employed which included (normal type) : LiChrosorb Si-60 (5 pm).Silica A (8 pm), CP Spher Si (10 pm) and Zorbax S91 ( 5 pm); and (reversed-phase type): LiChrosorb RP18 (5 pm), Hypersil-ODS ( 5 ctm), Supelcosil LC-PAH (10 pm) and Zorbax ODS ( 5 um). The peaks were measured by ultraviolet detection. Prior measurements of the UV spectra of isooctane solutions of PCBs were made at the Biochemisches Institut fur Umweltcarcinogene, through the courtesy of Professor J . Jacob. Organic Impurities Detected by GC - MS The m/z values of trace organic impurities present in each of the reference materials were recorded by GC - MS and showed that the impurities consisted mainly of isomeric and higher PCBs, e.g..for RM 293 hexachloroterphenyl was found to be present. A list of these trace impurities and the amounts detected have been given elsewhere.2‘ Homogeneity Study of the Candidate Reference Itlaterials All materials were homogenised in solution followed by single batch recrystallisation, or evaporation to dryne5s and subse- quent thorough mixing, to promote uniform distribution of possible impurities. The materials were dispensed as 25-mg units into glass vials, and the between-bottle homogeneity of each series was then investigated in a laboratory employing a method which had been shown in preliminary studies to give reliable results. Analysis by HPLC with reversed-phase adsorbent columns (HPLC-RP) was used for RMs 289, 291, 293, 296, 297 and 298, and with normal phase adsorbent columns (HPLC-Ad), for RMs 290,294 and 295.Capillary GC (GC-Cap) was used to analyse RM 292. The between-bottle homogeneity was assessed using six vials taken at random from each series. Typically, 0.1-mg test portions were dissolved in a suitable solvent (acetonitrile - water or acetonitrile for HPLC-RP; hexane for HPLC-Ad and GC-Cap) and two sub-ramples from each of the six vials were analysed by the method indicated above. The purity of each test portion was obtained by quantifying the impurities. In each instance the standard deviation of the resultc obtained for six different vials was not significantly different from that of the replicate analyses of a single solution. Detailed results have been reported elsewhere.22 Stability of the Candidate Reference Materials The analyses carried out over a period of 1 year, in conjunction with the initial purity and homogeneity controls carried out in two laboratories, together with the certification analyses showed no evidence of instability of the ten com- pounds in the solid form.Certification Procedure The certification procedure adopted for the PCB compounds was fimilar to that utilised for the polycyclic aromatic compound (PAC) series (see reference 26) which involved dividing the reference materials and the 11 participating laboratories into two groups. Each group consisted of five or six laboratories of which two were common to both, each laboratory in each group analysing at least five materials.All laboratories reported triplicate results obtained by three or more methods. This procedure permitted the number of results available for technical and statistical evaluation to be of adequate size and comparable to similar PAC certification rounds. Results and Discussion Confirmation of Identity of the PCB Compounds The synthetic routes utilised generally gave one main product, but in some instances an appreciable amount of a secondary product was formed. Hence, although PCB-20 and PCB-35 were synthesised together they were purified separately. For all compounds a comparison of their melting-points and infrared spectra with those recorded in the literature (where these data were available) gave little reason to doubt the identity of the isolated compound, although the data cannot be taken as providing an absolute confirmation.The struc- tures of PCB-8 (RM 289), PCB-28 (RM 291) and PCB-52 (RM 293) are well established; however, as many of the PCB isomers are similar, the other synthesised products were subjected to additional physical measurements to provide further confirmation. Examination of the individual mass spectra of the isomers confirmed that the miz value of the parent molecular ions of the all T 1 compounds corresponded to the expected relative molecular mass of the isomer. If the natural abundance ratio T 1 : 37Cl, which is close to 3 : 1. is assumed to hold true for the PCB compounds then the binomial distribution of the isotopic molecular ions, and hence their theoretical relative intenities can be calculated by expansion of the binomial (a + b)il where a = 3, b = 1 and Y Z = number of chlorine atoms present in the biphenyl.Although in practice the relative intensities may suffer distortion through the presence of adventitious ions, or the asymmetry of the molecule, a good agreement was observed when the theoretical and measured relative intensi- ties were compared (Fig. 1). The structures of PCB-101, PCB-118, PCB-138 and PCB- 180 (RMs 294,295,296 and 298, respectively) were confirmed by examination of their proton NMR spectra. The proton NMR spectra of these compounds revealed chemical shifts relative to tetramethylsilane (TMS) ( b p.p.m.) ac followc (interpretation in brackets): PCB-101 (RM 294): 7.24 (H-6’), 7.35 (H-4‘), 7.36 (H-6), 7.42 (H-3’) and 7.60 (H-3); PCB-118 and 7.59 (H-3); PCB-138 (RM 296): 7.09 (H-b), 7.34 (H-6’), 7.46 (H-5) and 7.60 (H-3’); PCB-180 (RM 298): 7.28 (N-6), 7.33 (H-6‘) and 7.62 (H-3’).These data provide good evidence for the accepted structures (cf., references 19 and 27). X-ray diffraction data were obtained for the prepared PCB-8 (RM 289), PCB-20 (RM 290), PCB-35 (RM 292) and PCB-153 (RM 297) compounds. These were identified unequivocally as 2,4’-dichlorobiphenyI, 2,3,3’-trichloro- biphenyl, 3,3’,4-trichlorobiphenyl and 2,2’,4,4’,5,5’-hexa- chlorobiphenyl , respectively.23,25 All the measured data, together, provide a sufficient basis for the conclusion that the synthesised products have the expected structures. (RM 295): 7.24 (H-6’). 7.41 (H-6), 7.49 (H-2’), 7.52 (H-5’) Methods of Determining Purity An indirect method for determining the purity of the PCBs for certification was used.The mass fractions of the observed impurities were determined and the purity of the material was then established by subtraction from unity of the total mass fraction of all detected impurities. The implications of this procedure on the accuracy of the purity determination are discussed below. Each laboratory received a sample, taken at random, of 25 mg of each material to be analysed. Three analytical methods, in which the laboratory had particular experience selected from the following techniques, were then applied. Gas - liquid chromatography with capillary or packed columns and quan- tification of resolved impurities by means of a flame ionisation detector (FID) or a mass spectrometer.High-performance liquid chromatography on normal- or reversed-phase adsor- bents with ultraviolet (UV) detection at a suitable wavelength and quantification of each resolved impurity by measurement556 ANALYST, MAY 1989, VOL. 114 of its peak parameters. At least three replicate measurements by each method applied were carried cut to demonstrate that the instrumental technique provided repeatable results. Only two laboratories provided GC measurements using an elec- tron capture (EC) detector. These measurements were not accepted for inclusion in the final certification due to their doubtful quantitative reliability in view of the unknown nature of the impurities present and the large and unpredictable differences in detector response which may occur with such impurities.Large differences in detector response are known to occur between PCB congeners of similar chemical composi- tion and structure.27 Laboratories were asked to guard against the possibility of cross-contamination of samples by using a new syringe for each PCB isomer and to obviate errors due to “memory effects” from the chromatographic column by running solvent blanks between samples. Where HPLC was utilised for quantification, the UV wavelength used to determine the absorbance of the solution was selected, where possible, to provide the maximum sensitivity from the instrument used. Laboratories were also asked to run each compound at least once on each analytical system for an extended period of time to ensure that errors did not arise through non-detection of late-eluting peaks. Where flame ionisation detection (FID) was utilised to obtain measurements, the response factors of the impurities and of the principal component were assumed to be the same; subsequent to the present work Zoller et aZ.28 have examined the effect of hydrogen flow-rate on the relative molar response factors of the PCBs described in this paper.Where a mass spectrometer coupled to a gas chromatograph was employed, quantification of impurities was based on the total ion current chromatogram. In some instances selected ion recording (SIR) allowed the detection of impurities present at very low levels. Sources of Error Although a systematic examination of the analytical errors which can occur in the purity determination of PCB materials is not feasible because of the differences in equipment (instruments, chromatographic columns) and experimental conditions used by the participating laboratories, some of the potential errors can be discussed.A fundamental source of error arises from the method used for the determination of purity by the subtraction of mass fractions of impurities from unity. It is clear that those impurities that remain undetected because they are not separated from the main component or are not detected in its presence, or are retained on the column, will lead directly to an over-estimation of the purity. The certification procedure was therefore designed to minimise the risk of non-detection of impurities by the use of a wide range of separation systems.Hence, the HPLC methods covered the possibility of detect- ing organic non-volatile impurities which might not otherwise be detected by the GC measurements. A second, general, source of error in the evaluation of analytical results obtained by the chromatographic techniques is linked with the various methods of peak-area quantification and with the difficulty in assigning correct response factors to minute impurity components. A third source of error lies in the possible presence of trace amounts of inorganic impurities in the candidate reference materials. The determination of inorganic impurities after ashing samples of the PCB isomers at a moderate temperature led to the detection of their presence only in PCB-8 (RM 289) at a mass fraction of 0.0003 g 8-1, in PCB-35 (RM 292) at 0.002 g g-1 and in PCB-52 (RM 293) at 0.0004 g g-1. In the other compounds inorganic impurities were present at levels of less than 0.0002 g g-1 which was the detection limit of the method used.Results and Their Evaluation All the results received were grouped according to the method used. The statistical means of the individual measurements supplied by each laboratory for each analytical method (hereafter referred to as the set means) were used for the evaluation. The results obtained were subjected to an evaluation at a meeting of the participants and, as far as possible, technical reasons were sought before any result or set of results was eliminated by following the general principles adopted for the certification of Community Bureau of Reference (BCR) (Brussels, Belgium) reference materials.29 Results rejected during the evaluation were in accordance with the following criteria: results showing a substantial divergence from the means of results by similar methods applied in other labora- tories and technically judged to be incorrect values.Reasons for these discrepancies could not be established in every instance however; those results where no impurities were detected by the method and the given value was the laboratory’s lower estimate of purity based on the detection limits for impurities of their methods and equipment. Results obtained with similar methods applied in another laboratory indicated the presence of impurities, such results were eliminated on the grounds of a failure of the application of the technique by the laboratory rather than because of an inherent weakness in the method itself; in several instances, it was apparent that measurements made by certain methods system- atically gave the wrong results, all results by these methods were rejected.Only the accepted results ( P ) were used for the purity calculation. The organic impurity values, the values of P , the estimated inorganic impurities present and the certified purity values of the ten reference materials are given in Table 2. Table 2. Certified purity values PCB Organic RM No. Biphenyl compound impuritiedg g ~ 1 289 290 29 1 292 293 294 295 296 297 298 8 20 28 35 52 101 118 138 153 180 2,4‘-DichIoro- 2,3,3‘-Trichloro- 2,4,4’-Trichloro- 3,3’ ,4-Trichloro- 2,2 ’ ,5,5 ’ -Tetrachloro- 2,2‘ ,4,5,5‘-Pentachloro- 2,3‘ ,4,4‘ ,5-Pentachloro- 2,2‘ ,3,4,4’,5’-Hexachloro- 2,2’ ,4,4’ ,5,5’-Hexachloro- 2,2’ ,3,4,4‘,5,5’-Heptachloro- 0.0034 + 0.0024 - 0.0009 0.0014 f.0.0006 0.0020 +_ 0.0006 0.0119 f. 0.0012 0.0037 k 0.0012 0.0059 -t 0.0008 0.0038 -t 0.0012 0.0007 f. 0.0003 0.0005 + 0.0004 - 0.0002 0.0042 f. 0.0007 P* 23 18 22 9 22 19 19 12 12 19 Inorganic impuritieslg g- 0.0003 k 0.0003 0.0002 k 0.0002 0.0002 -t 0.0002 0.0020 * 0.0020 0.0004 k 0.0004 0.0002 k 0.0002 0.0002 k 0.0002 0.0002 -t 0.0002 0.0002 f. 0.0002 0.0002 * 0.0002 Certified purity/g g- I 0.9963 + 0.0012 0.9984 k 0.0008 0.9978 f 0.0008 0.9861 L 0.0032 0.9959 k 0.0016 0.9939 -t 0.0010 0.9960 f 0.0014 0.9991 f 0.0005 0.9993 + 0.0004 - 0.0006 0.9956 -t 0.0009 - 0.0027 * P = number of accepted sets of results for organic impurities.ANALYST, MAY 1989, VOL.114 557 0.9800 1 1 I L 2 Q 1 .oooo 289 290 291 292 293 294 295 296 297 298 Reference mate ri a I Fig. 2. Range bar graphs of certified purity of the reference materials. (For each reference material the first bar represents the uncertainty limits for the organic purity when inorganic impurities are disregarded and the second bar represents the uncertainty limits when these are included. The mean purity value is indicated by a cross-stroke. ) Statistical Treatment of the Results for Organic Impurities The statistical evaluation of the accepted results of the purity measurements made on the PCB reference materials was essentially the same as that adopted in the earlier PAC series.26 The results were expressed as follows: the certified organic impurity was derived from the mean of the individual set means, each set mean representing a given method in a given laboratory; the uncertainty was expressed in terms of the 95% confidence limits of the mean organic purity; in addition to the calculated mean organic purity, a certified over-all purity was calculated which took account of both the organic and inorganic impurities present.In the initial phase of the statistical evaluation, the distribution of the accepted set means was tested for normality for each of the materials. Inspection of the data suggested that, especially for those compounds of very high purity, the skewed distribution would be more consistent with a logarith- mic-normal distribution.Hence, the set means, expressed as mass fractions of the impurities, and, in a separate calculation, their logarithms, were tested for conformity to a normal distribution using the Kolmogorov - Lilliefors test. It was found generally that the accepted results for the majority of the compounds conformed to a normal distribution. The results for two of the compounds (RMs 289 and 297) however were found to conform more satisfactorily to a logarithmic- normal distribution. The 9.5% confidence limits were estab- lished by standard procedures. Table 2 gives a summary of these final values together with results for the determination of inorganic impurities. The mean values of the purities of the ten PCB isomers with their associated uncertainties with and without taking account of the presence of trace inorganic impurities are shown in Fig.2. The members of the following laboratories are thanked for their participation: J. Jacob, Biochemisches Institut fur Umweltcarcinogene, Ahrensburg, FRG*,§; Tin Win, Bunde- sanstalt fur Materialprufung, Berlin, FRGO; P. Payen, Centre d’Etudes et Recherches des Charbon.nages de France, Ver- neuil, Frances; A. Liberti, CNR Istituto Inquinamento Atmosferico, Rome, Italy§; J. Gielen, Instituut voor Toege- paste Chemie (Hoofdgroep Maatschappelij ke Technologie) , TNO, Delft, The Netherlands§; E. Gevers, Instituut voor Toegepaste Chemie (Hoofdgroep Maatschappelij ke Technol- ogie) , TNO, Zeist , The Netherlands* ,t ,§ ; M. Henderson, Inveresk Research International, Musselburgh, Scotland, UK*; L.Boniforti, Istituto Superiore di Sanita, Rome, Italy§; * Laboratories where compounds were synthesised. t Laboratories where homogeneity was evaluated. :i: Laboratories where structural studies were carried out. 3 Laboratories that contributed measurement data for use in the certification programme. A. Head, National Physical Laboratory, Teddington, UKS; H. Kienhuis, Prins Maurits Laboratory, TNO, Rijswijk, The Netherlands* ,$; T. Rymen, Studiecentrum voor Kernenergiel Centre d’Etudes Nucleaires, Mol, Belgium$; K. Ballschmiter, University of Ulm, Ulm/Donau, FRG”; and M. Metzler, University of Wurzburg, Wurzburg, FRG.5 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. 28. 29. References Jensen, S ., PCB Conference, National Swedish Environmental Protection Board, Research Secretariat, Solna, Sweden, 1970. Anon., New Sci., 1966, 32, 612. Jensen, S., Johnels, A . G., Olsen, M., and Olterlind, G., Nature (London), 1969, 224, 247. Fishbein, L., J. Chrornatogr., 1972, 68, 345. Haley, T. J . , “Dangerous Properties of Industrial Materials Report,” 1984, 4, No. 6, p. 2. IARC Monographs, “Some Antithyroid and Related Sub- stances, Nitrofurans and Industrial Chemicals,” International Agency for Research on Cancer, Lyon, 1974, Volume 7, p. 261. ZA RC Monographs, “Polychlorinated and Polybrominated Biphenyls,” International Agency for Research on Cancer, Lyon, 1978, Volume 18. IARC Supplement 4 to Monographs Volumes 1-29, “Chemi- cals, Industrial Processes and Industries Associated With Cancer in Humans,” International Agency for Research on Cancer, Lyon, 1982, p.217. Official Journal of the European Communities: 1976, L108,41, Directive 76/403; 1976, L262, 201, Directive 76/769; 1985, L269. 56, Directive 85/467. Regulation No. 81735 of Permitted Maximum Content of PCB-components in Food Products (September, 1985). Pub- lished in the Netherlands Staatscourant 200, 1985. Official Journal of the European Communities: 1980, L229,11, Directive 80/778. Erikson, M. D., “Analytical Chemistry of PCBs,” Butter- worth, London, 1986. Hutzinger, O., Safe, S . , and Zitko, V., “The Chemistry of PCBs,” CRC Press, Cleveland, OH, 1974. Ballschmiter, K . , and Zell, M., Fresenius Z. Anal. Chem., 1980, 302, 20. Hutzinger, O., Safe, S., and Zitko, V., Bull. Environ. Contarn. Toxicol., 1971, 6, 209. Willis, D. E . , and Addison, R. V., J. Fish. Res. Board Can., 1972, 29, 592. Cadogan, J . I . G . , J. Chem. SOC., 1962, 4257. Hutzinger, O., and Safe, S . , Bull. Environ. Contam. Toxicol., 1972, 7, 374. Erb, F., Pommery , J . , van Aerde, C., and Vermeersch, J . , Bull. SOC. Chim. Fr., 1976, 964. Sundstrom, G., Acta Chem. Scand., 1973, 27, 600. Moron, M., Sundstrom, G . , and Wachtmeister, C . A . , Acta Chern. Scand., 1973, 27, 3121. Jacob, J . , Lindsey, A. S . , and Wagstaffe, P. J . , “The Certification of the Purity of Polychlorinated Biphenyl Isomers. BCR Reference Materials Nos. 289-298. EUR 10998 En-1987 ,” Commission of the European Communities, Luxem- bourg, 1987. Moes, G. W. H., and Lenstra, A. T. H . , Toxicol. Environ. Chern., 1986, 12, 255. Safe, S . , and Hutzinger, O., J. Chem. SOC., Perkin Trans. 1, 1972, 686. Geise, H. J . , Lenstra, A. T. H., de Borst, C., and Moes, G. W. H., Acta Crystallogr., Sect. C, 1986, 42, 1176. Jacob, J . , Belliardo, J . J . , and Wagstaffe, P . J . , “The Certification of Polycyclic Aromatic Compounds. Part VI: CRM Nos. 152, 265-272. EUR 10295 En-1985,’’ Commission of the European Communities, Luxembourg, 1985. Mullin, M. D . , Pochini, C. M., McCrindle, S . , Romkes, M., Safe, S . H., and Safe, L. M., Environ. Sci. Technol., 1984, 18, 468. Zoller, W., Schafer, W., Class, T., and Ballschmiter, K., Fresenius Z. Anal. Chem., 1985, 321, 247. Marchandise, H., and Colinet, E., Fresenius 2. Anal. Chem., 1983, 316, 669. Paper 8103539C Received September 19th, 1988 Accepted January 4th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891400553

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, MAY 1989, VOL. 114 559 Direct Determination of Zinc, Lead, Iron and Total Sulphur in Zinc Ore Concentrates by X-ray Fluorescence Spectrometry Josefina de Gyves Departamento de Quimica Analitica, Facultad de Quimica, Universidad Nacional Autonoma de Mexico, Mexico Montserrat Baucells" Servei d'Espectroscopia, Universitat de Barcelona, 080028 Barcelona, Spain E. Cardellach and J. L. Brianso Depa rtam en to de Geo log ia, Un ive rsida d A u ton om a de Barcelona, Barcelo na, Spa in A direct X-ray fluorescence (XRF) method has been developed for the rapid determination of zinc, lead, iron and total sulphur present in zinc ore concentrates. The concentrations of the elements present, mainly as sphalerite, galena, pyrite and pyrrhotite, vary widely: Zn, 40-65; Pb, 0.1-10; Fe, 0.5-1 1; and S, 2533% m/m.Using a wavelength-dispersive X-ray spectrometer and samples prepared as briquettes, Zn and Pb may be determined directly using the KP and Ly, lines, respectively. For Zn, a mathematical model (concentration- based correction equation) was applied to correct for inter-element effects. For the determination of Fe the ratio of the fluorescence intensities of the Fe Ka and Zn KP lines was used and, for the determination of total S , the intensity of the S Ka line was corrected with the Rayleigh scatter tube line, Rh La. For both iron and total sulphur, subsequent application of the mathematical model helped to correct for the residual inter-element effects. Good agreement was obtained between experimental and certified values for two international reference materials.Keywords: Zinc; lead; iron; zinc sulphide matrix; X-ray fluorescence spectrometry The accurate analysis of geological samples with complex matrices by X-ray fluorescence (XRF) spectrometry requires the use of a correction method. Such samples include zinc ore concentrates, for which the large variation in the composition of several major elements (Zn, 4&65; Pb, 0.1-10; Fe, 0.5-1 1; and total S, 25-33% mim) gives rise to severe absorption and enhancement effects. The many approaches proposed for the solution of inter-element effects may be divided into two main groups': (a) mathematical methods, in which inter-element effects are calculated (e.g. , fundamental parameters and influence coefficients) and (b) comparative methods (internal standard, standard additions or spiking, simple or double dilution, scatter intensity, etc.) in which such effects are compensated for.Mathematical correction methods based on the calculation of influence coefficients can be classified further into two categories: intensity-based (e.g., Lucas Tooth - Pyne model') and concentration-based (e.g. . Lachance - Trail1 model3) correction methods. The latter offers the advantages of requiring relatively small computer facilities and of being applicable over relatively wide concentration ranges. The X-ray spectrometer used in this work has built-in software for the concentration-based correction equation proposed by de Jongh ,A namely c, = (0, + E,R,)(l + a,,c,) . . . . . . where c, is the concentration to be determined of element i, D, and E, are elementhnstrument dependent parameters, R, is the intensity ratio, a , is the influence coefficient for the effect of element j on i and C, is the concentration of j .Alpha constants are calculated by regression analysis of data obtained from a relatively large number of calibration standards or from fundamental constants. To apply equation (1) the concentrations of all elements in the mid to high concentration range that could give rise to inter-element effects must be known. These concentrations can be calcu- lated from intensity data by regression analysis. The concen- * To uhom correspondence should be addrased. tration of the unknown is then estimated by an iterative calculation. For the work described here, as insufficient, well charac- terised standards for zinc ore concentrates were available, synthetic pelleted standards were prepared.Because the chosen calibration procedure involves matching the matrix composition of standards and samples, standards were pre- pared to contain all major elements which could significantly modify the matrix mass absorption coefficient ( prndrrlx) at the analytical working wavelengths. Account was also taken of the elements that could give rise to enhancement effects. Matrix mass absorption coefficients of all the specimens used as standards calculated at the wavelengths of the XRF emission lines of Pb Lyl (A, 0.84 A) and Zn K(3 (A, 1.295 A) show that this parameter does not vary significantly: pmatrlx. Pb L~~~ - 61.65 k 1.59 and pmatrlx, Zn Kp = 111.63 k 3.21, in spite of the wide concentration variations (Table 1).Further, no enhance- ment effects might be expected for these lines from any of the elements present in the standards and samples. Consequently, Zn and Pb could be determined directly with good precision and accuracy. Alpha coefficients for Fe can be used in the determination of Zn to improve the accuracy even more, as they correct the weak inter-element effects in the analysed specimens. The analogous calculations of pmatrlx at the emission lines of Fe K a (A, 1.936 A) and S K a ( h , 5.375 A) showed a more important variation in this parameter: jimatl,x, enhancement effect of Zn over Fe might be expected because of the lower atomic number of Fe and the very high Zn concentration; however, a direct analysis for these two elements did not give satisfactory results.Accuracy was improved with empirical application of the alpha correction method together with alternative approaches such as the use of the Zn K(3 line to correct the Fe K a line and the Rh L a Rayleigh scatter line to correct the S K a line; such procedures compensate for inter-element effects to a great extent and decrease the divergence from the calibration line. Hetero- geneity effects due to particle size were overcome by carefully grinding the specimens to obtain a particle size of less than 10 pm. A study using optical microscopy was carried out in order - Fe Kor = 124.1 k 4.9 and pmatrlx, s Koc = 1544 t 113. An560 ANALYST, MAY 1989, VOL. 114 Table 1. Analysis of synthetic standards (specimens 1-13) and reference materials Specimen 1 2 3 4 5 6 7 8 9 10 11 12 13 BCR108 ., CZN-1.. . . Fe, % m/m A" B? 12.50 12.42 10.00 9.98 5.00 4.84 2.50 2.39 1.25 1.22 0.60 0.60 0.30 0.32 11.00 11.18 7.50 7.45 4.00 4.12 4.00 4.07 4.00 3.97 4.00 3.90 7.21 7.41 10.93 10.85 * Calculated or certified concentrations. t Concentrations found. Total S, % m/m Zn, % m/m Pb, YO rnlm A* Bt 20.3 20.0 24.1 24.2 28.4 28.4 30.6 30.3 32.5 32.6 32.7 32.5 32.6 32.3 25.0 25.4 27.6 27.6 25.6 25.8 28.9 29.3 30.2 30.5 30.7 30.6 31.0 30.9 30.2 30.0 A* Bt 39.0 39.0 48.7 48.0 57.4 56.7 62.2 61.7 64.6 64.7 65.8 66.2 66.6 66.8 49.1 49.4 55.3 55.2 56.2 56.9 58.4 58.0 61.2 61.0 63.0 62.4 54.0 54.5 44.7 44.7 A* Bt 7.50 7.47 4.95 5.04 2.50 2.42 1.25 1.18 0.59 0.56 0.30 0.28 0.098 0.085 6.25 6.03 3.75 3.58 0.48 0.53 0.48 0.52 0.48 0.49 0.48 0.50 0.90 0.89 7.45 7.50 Table 2.Measuring conditions for Zn, Pb, Fe and total S Element (line) Collimator Detector Zn(K6) . . . . . . F" FS t Pb(Ly1) . . . . . . F S$ Background Pb (Lyl) . . F S BackgroundPb(LyJ . . F S Fe(Kp) . . . . . . F FS BackgroundFe(K(3) . . F FS Fe(Ka) . . . . . . F FS S(Ka) . . . . . . C§ Flow BackgroundS(Ka) . . C Flow Rh(La) . . . . . . C Flow BackgroundRh(La) . . C Flow Crystal LiF 200 LiF 200 LiF 200 LiF 200 LiF 200 LiF 200 LiF 200 Ge 111 Ge 111 Ge 111 Ge 111 Voltage/ kV 50 50 50 50 55 55 55 40 40 40 40 Current/ mA 50 50 50 50 45 45 45 75 75 75 75 28/" 37.53 24.07 25.27 22.87 51.73 49.73 57.52 110.86 113.86 89.605 87.605 Timels 20 40 40 40 40 40 20 20 20 40 40 * F = Fine.f FS = Flow - scintillation. $ S = Scintillation. § C = Coarse. to define the grinding process and to control strictly the average particle size. Experimental Apparatus X-ray fluorescence intensities were measured with a Philips PW 1400 computer-controlled wavelength-dispersive X-ray spectrometer equipped with an Rh source. The operating conditions used are specified in Table 2. Calibration Standards Analytical-reagent and spectroscopic grade chemicals were used to prepare synthetic pelleted standards of "infinite" thickness for calibration. The ZnS used was 99.99% pure and PbS was obtained by precipitation with H2S from Pb(N03)2. For the pellet formation, commercially available Albacite 2044 (butyl methacrylate resin), prepared as a 20% rnlV acetone solution, was used as a binder.In order to simulate the composition of the unknown as closely as possible, 13 standards were prepared to contain the major elements found in such minerals: Zn", FeIII, Pb", S and 0. Also, minor gangue elements such as AlIIl, SiIV, CaII and Ba" were added in amounts usually found in zinc ore concentrates (A1203, 0.025-0.25%; S O 2 , 0.1-3.0; BaO, 0.005-0.5; and CaO, 0.005-0.3% rnlm). Gangue elements were added as oxides (A1 and Si) and carbonates (Ba and Ca), Zn" and Pb" as sulphides and Fell1 as the oxide. Calibration standards (5 .O g) were prepared from accurately weighed portions of the metal compounds which were mixed together for 20 min in a Spex Mixer-Mill (Model 8000). The particle size of the specimens was determined to be <10 pm.The 5.0-g specimens were mixed with 2 ml of Albacite - acetone solution in an agate mortar. Pellets of 40-mm diameter were prepared at pressures of 200 kN applied for 60 s. The precision of pellet formation (RSD), obtained by measuring the Pb Ly,, Zn K(3, Fe Ka and S Ka lines for eight pellets of the same specimen, is given in Table 3. Procedure Zinc and Pb are determined directly using the operating conditions specified in Table 2. The alpha coefficients for Fe to be used in the determination of Zn are calculated by regression analysis from the concentration data of standards. For the determination of Fe, the intensity of the Ka line is corrected using the intensity of the Zn K(3 line. In addition, the alpha coefficients of Zn are calculated by regression analysis of data obtained from synthetic standards.In the determination of total S, the intensity of the S Ka line is corrected using the Rayleigh dispersion Rh La line. Both Zn and Pb alpha coefficients are calculated as described pre- viously.ANALYST, MAY 1989, VOL. 114 561 ~~~ Table 3. Precision of pellet formation ( n = 5 ) Mean concentration, SD/ Element '/o m/m counts 5-1 RSD, Yo Zn . . . . 54.10 0.21 0.39 S . . . . 26.40 0.34 1.28 Fe . . . . 3.77 0.037 0.98 Pb . . . . 0.324 0.0074 2.28 + I I 0 30 60 90 120 150 180 210 240 270 I n t e nsi t y, a r b i t ra ry u nits Fig. 1. Calibration graph for Fe obtained using the Kcu line ,+ /+ t , I I 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Relative intensity Fig. 2. with the Zn K(3 line (and using the alpha coefficients of Zn) Calibration graph for Fe obtained after correcting the Kcu line Results and Discussion Iron Determination The presence of high concentrations of heavy elements such as Zn and Pb gives rise to important absorption effects at the XRF emission wavelength of Fe.Also, as Zn has a higher atomic number than Fe, its XRF radiation is sufficiently energetic to excite Fe atoms and a specific enhancement effect might be observed on the XRF emission lines (Ka and KP) of Fe (Fig. 1). When the mass absorption coefficient of a sample differs from the average mass absorption coefficient of the standards by about +5% use of the de Jongh mathematical model,4 which involves calculating the alpha coefficients for Zn and Pb to correct for inter-element effects, gives results of acceptable accuracy. However, for the application of this correction procedure, the use of a considerable number of standards is recommended.The presence of an inter-element effect means that an extra degree of freedom must be determined in the analysis. As a guide, each degree of freedom determined requires at least three standards in order for it to be statistically significant.5 The combination of a compensation method, using the intensity of the Zn K(3 line to correct the intensity of the Fe Ka line, and the de Jongh mathematical model, to calculate the alpha coefficients of Zn, reduces the dispersion of the calibration points and improves the accuracy of the Fe determination (Fig. 2). As only one alpha coefficient is calculated in this procedure, the number of standards required is reduced by three.20 ; i E 30 Relative intensity 40 Fig. 3. Calibration graph for total S obtained after correcting the K a line with the coherently dispersed Rh La line (and using the alpha coefficients of Zn and Pb) Similar results are achieved in the determination of Fe using the Fe K(3 line. The Zn K(3 line is also used as the internal standard in this instance. Total Sulphur Determination Sulphur is particularly well suited to determination by XRF spectrometry because of the magnitude of its atomic num- ber.h.7 However, the presence of elements that will change the mass absorption coefficient of the sample by more than t5% from the average mass absorption coefficients of the standards will introduce errors into the determination of S owing to changes in the absorption behaviour of the S K a line.Enhancement effects due to the presence mainly of Zn, Fe and Pb and of the tube element, Rh, might also be expected. The use of only one correction procedure, either the de Jongh mathematical model (to correct for inter-element effects by calculating the alpha coefficients for Zn and Pb) or the compensation method (using the Rayleigh scatter dispersion tube line, Rh La. to correct the measured intensity of the S K a line), gives values of moderate accuracy. By combining the two procedures dispersion about the calibration graph is reduced and the accuracy is significantly improved (Fig. 3). The moderately accurate concentrations obtained in the determination of total S by the single-correction method might be explained as follows.Use of the Rayleigh scatter dispersion procedure does not correct the enhancement effect due to the very high concentrations of Zn present (the use of the scatter radiation method is applicable, in principle, only to elements with an atomic number higher than those of all major constituents of the matrix, as the method does not compensate for enhancement effects).' The mathematical procedure corrects enhancement effects due to the major elements present in the matrix; however, it does not correct for enhancement effects due to the tube element. By combining both procedures all inter-element effects can be corrected as indicated by the good agreement between the results obtained by the proposed method and the certified values for two international reference materials (Table 1).The presence of an absorption edge of S between the two selected radiations apparently has no important consequences due to the fact that, although pmatrlx, s K~ > pmatnx, Rh for all the standards used, pmatrlx, s Ka/p,atrlx, Rh L~ does not vary significantly. Zinc Determination In zinc ore concentrates it is possible to determine Zn using either Ka or K(3 radiation. In this work K(3 was used because the count rates measured for the Ka line were so high that a direct procedure could not be used. The calibration graph for Zn was linear for concentrations in the range 39-67% mlm and could be applied to specimens with a mass absorption coefficient no greater than +5% with respect to the average562 ANALYST, MAY 1989, VOL.113 mass absorption coefficient of the standards. The use of Fe alpha coefficients corrected for dispersion about the calibra- tion graph and improved the accuracy. For the determination of @30% rnlrn Zn in Zn - Pb minerals Mahapatrag has suggested the use of the Cu K a line as the internal standard. Lead Determination In principle, for the determination of Pb in zinc sulphide minerals it is possible to use the Lf3 and Lyl lines; the most useful Pb La line suffers from spectral interference due to the close proximity of the As Ka line. For this work the Pb Lyl line was selected because the mass absorption coefficient of the specimens used as standards is less influenced by the Zn concentration at this wavelength than at the Pb L(3 wavelength. It was subsequently observed that the determina- tion of Pb in specimens with a relatively low Zn content and a high content of Pb was more satisfactory when the Lyl line was used.In addition, to improve the accuracy of the Pb determination we recommend that PbS is used for the preparation of standards; when PbO is used differences are obtained in the measured XRF intensities.9 1. 2. 3. 4. 5 . 6. 7 . 8. 9. Accuracy and Precision The accuracy of the recommended procedures was assessed by analysing the standard reference materials BCR 108 (Com- mission of the European Communities) and CZN-1 (Canada Centre for Mineral and Energy Technology). As can be sccn from Table 1, the Zn, Pb, Fe and total S concentrations obtained are in agreement with the certified values. The RSDs of the four elements, based on eight replicate analyses of each sample, were as follows: Zn, 0.136% (52.66 -t- 0.071% mlrn); Pb, 0.72% (0.555 -t 0.004% rnlrn); Fe, 0.304% (3.75 2c 0.011% nzlnz); and total S, 0.313% (25.95 k 0.08 Yo rnirrz) . References Tertian, R.. and Claisse, F.. in “Principles of Quantitative X-ray Fluorescence Analysi\,” Heyden, London, 1082, p. 118. Lucas Tooth, H. J . , and Pyne, C., Adv. X-Ruy Anal., 1964,7. 523. Lachance, G . K., and Traill, R. J., Cun. Spectrosc., 1066, 1 1 , 43. de Jongh, W. K., X-Ray Spectrom., 1973, 2. 151 Jenkins. R., “An Introduction to X-ray Spectrometry,” Hey- den, London, 1974. Chapter 7. Fabbi, B. P . and Moore, W. J.. Appl. Spectrac., 1970. 24, 426. Weber, H T., van Willigcn, J . I I . El. G.. and van der Linden. E., Anal. Chim Actu, 1984, 160, 271. Mahapatra, N . S . , X-RUJ Spectrom., 1987. 16, 171. Baucells, M., Lacort, G., Roura, M., and de Gyves. J . , Analyst, 1988, 113. 1325. Paper- 81042961 Received October 28tl.1, 1988 Accepted Janisary 4tlq 1989

ISSN:0003-2654    

DOI:10.1039/AN9891400559

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, MAY 1989, VOL. 114 563 Determination of Ammonium-, Nitrite- and Nitrate-nitrogen by Molecular Emission Cavity Analysis Using a Cavity Containing an Entire Flame Ali Celik and Emur Henden Department of Chemistry, Faculty of Science, University of Ege, Bornova, Izmir, Turkey Ammonium-, nitrite- and nitrate-nitrogen were determined by molecular emission cavity analysis using a cavity containing an entire flame. Ammonium-nitrogen was converted to ammonia by injection on to solid sodium hydroxide. The calibration graph was linear for 5-100 pg ml-1 of nitrogen when the ammonia generated was swept directly into the cavity and for 0.05-1.0 pg ml-1 of nitrogen when it was collected in a liquid nitrogen cold-trap. Concentrations of 1.5 and 0.01 pg ml-' of nitrogen could be detected using the direct and cold-trap methods, respectively.Nitrite was determined after conversion to nitrogen monoxide by iodide. Nitrate was reduced to nitrite using a copperised cadmium column and then determined as nitrite. The calibration graphs for both anions were linear up to 7 pg mi- of nitrogen and 0.1 pg ml-1 of nitrogen could be detected. The methods were applied successfully to the determination of nitrite and nitrate in meat products, and nitrate-nitrogen in drinking water samples. Keywords: Molecular emission cavity analysis; ammonium-nitrogen; nitrite-nitrogen; nitrate-nitrogen Although most spectrophotometric methods for the determi- nation of ammonium, nitrite and nitrate ions are highly sensitive, special precautions are required when applying them to samples of coloured solutions or suspensions.1 In contrast, the flame emission methods, which produce spectral bands for molecules containing nitrogen,'.-? are fast and are not affected by the colour of the solutions; however, they are generally not highly sensitive. The determination of ammoniacal nitrogen in effluents and fertilisers by molecular emission cavity analysis (MECA) with an oxy-cavity has been described by Belcher et al.4.5 Am- monium ion was converted to ammonia and swept by nitrogen into the cavity. By monitoring the NO-0 continuum, 10-200 pg ml-1 of nitrogen could be determined with a detection limit of 1 pg ml-I. Al-Zamil and Townshendh also applied MECA with an oxy-cavity to the determination of nitrite and nitrate after their reduction to nitrogen monoxide by iodide or zinc; 5-300 pg ml-1 of nitrogen could be determined with a detection limit of 0.5 pg m - 1 of nitrite-nitrogen and 2 pg nil-1 of nitrate-nitrogen.The use of MECA with an oxy-cavity for producing oxide-based emissions of various elements involves the introduction of a small amount of oxygen into the cavity held in a hydrogen - nitrogen diffusion flame.' It has been shown that oxide-based emissions of boron, selenium, arsenic and antimony can also be obtained using another cavity containing an entire flame within the oxy-cavity.8 This cavity eliminates the need for the burner system required by the oxy-cavity and consumes much smaller amounts of gases. In this paper, the application of a cavity containing an entire flame to the determination of ammonium-nitrogen, with and without cold-trap collection, and of nitrite and nitrate after their reduction to nitrogen monoxide using a copperised cadmium column is described.Significant improvements in the detec- tion limits compared with previous MECA studies were obtained. Experimental Apparatus A modified Pye Unicam SP 90A flame spectrometer, with a 1.4-mm slit (band width at 400nm = 45nm) was used to measure the emissions as described previously.9 The emission intensity was recorded on a Varian G-2500 chart recorder (response time, 0.5 s for full-scale deflection). The stainless- steel cavity used (4mm in diameter, 10mm deep) was as described previously,V but was cooled by passing a small flow of water through the copper tube at the rear of the cavity (Fig.1). The volatilisation systemlo was modified as shown in Fig. 2. It consisted of a bent glass reaction vessel (5.5 cm long, 1.7 cm in diameter) with an injection hole 3 cm from its closed end. A silicone-rubber septum in an O-ring was placed over the injection hole and tightened with a clip. The reaction vessel was connected by means of PTFE tubing to a stainless-steel tube (0.8mm i.d.) in one of the rear openings of the cavity through a drying tube and a three-way valve. When the cold-trap was required, a coiled PTFE tube (2 mm i.d., 70 cm long) was connected between the three-way valve and the cavity and immersed in liquid nitrogen in a Dewar flask. The drying tube was packed with calcium chloride powder for the determination of nitrite- and nitrate-nitrogen and with sodium hydroxide pellets for the determination of ammonium- nitrogen.When the cold-trap was not in use, the reaction vessel was heated at 85-90°C by placing it in a small heating mantle constructed in a 25-ml beaker using a heating coil. The tubes between the reaction vessel and the cavity were also wrapped in the heating coil and kept at 85-90 "C. A small plug of glass-wool ( 5 cm long) was placed in the PTFE tube before the cavity. The three-way valve was connected after the drying tube so that the carrier gas could be supplied to the cavity while waiting for the reactions to go to completion. ( a1 (6) I Fig. 1. Schematic diagram of the cavity, which contains the entire flame. (a) Cross-section; and ( b ) front view (all dimensions in mil 1 i metres)564 ANALYST, MAY 1989, VOL.114 Three-way valve Glass wool PTFE \ \ tubinn IMagnetic stirrer] Dewar flask Fig. 2. nitrite- and nitrate-nitrogen (not to scale) Volatilisation system for the determination of ammo ium-. Reagents All chemicals used were of analytical-reagent grade. Stock solutions of ammonium, nitrite and nitrate ions (1000 pg ml-1 of nitrogen) were prepared by dissolving ammonium chloride, sodium nitrite and sodium nitrate, respectively, in distilled water. The nitrite solutions, the low-concentration solutions of ammonium and nitrate and the iodide solution were prepared fresh. Procedure for the Determination of Ammonium-nitrogen Procedure without cold-trup collection Transfer 0.5-0.6 g (about three pellets) of solid sodium hydroxide into the reaction vessel and close the volatilisation system.De-aerate the system for 20 s with nitrogen carrier gas and turn the three-way valve to direct the nitrogen into the cavity without it passing through the reaction vessel. Inject 0.2 ml of the sample solution into the reaction vessel and wait for 2 min. Sweep the ammonia generated into the cavity by turning the three-way valve to the direction of the reaction vessel. Record the emission intensity at 640nm and use the peak height for the determination. Procedure using cold-trap collection Transfer ten pellets of sodium hydroxide into the reaction vessel and close the system. Purge the system for 15-20 s with nitrogen carrier gas and immerse the trap in liquid nitrogen in a Dewar flask. Inject 5 ml of the sample solution and collect the ammonia generated for 4 min.Remove the cold-trap from the liquid nitrogen and immerse it in a water-bath at 90-95 "C. Record the emission at 640 nm and measure the peak height. Procedure for the Determination of Nitrite-nitrogen Introduce 0.5 ml of a reducing solution containing 0 . 5 ~ potassium iodide in 0.5 M hydrochloric acid into the reaction vessel containing a magnetic bar. Purge the system with nitrogen and turn the three-way valve so that the nitrogen carrier gas by-passes the reaction vessel. Turn on the magnetic stirrer and inject 0.5 ml of the sample solution. One minute after injection turn the three-way valve so that the nitrogen monoxide produced is swept with the nitrogen carrier gas into the cavity.Record the emission as above and use the peak height for measurements. Procedure for the Determination of Nitrate-nitrogen Reduce nitrate to nitrite using a copperised cadmium col- umn1.11 and then proceed as described above for nitrite- nitrogen. Results and Discussion The spectrum of the white emission obtained by presenting ammonia vapour or nitrogen monoxide at a constant speed to the flame in the cavity was similar to that reported previously,h but with maximum intensity at 640nm. The emission is probably due to the NO-0 continuum.6 The spectrum appeared to have been shifted to longer wavelengths when compared with the reported spectrum.4-6 This shift may be attributed to an increase in the spectral band width of the prism monochromator used as the wavelength increased and/or to the differences in the composition and temperature of the flame used in this work.Similar shifts have also been observed in the spectrum of arsenic.7.9 The gas flow-rates were optimised in order to obtain the highest signal to noise ratio. Flow-rates of 115, 45 and 75 ml min-1 for hydrogen, oxygen and nitrogen. respectively, were found to be optimum for all the determinations. Determination of Ammonium-nitrogen Injection of an aqueous solution on to solid sodium hydroxide caused a sudden increase in the temperature of the solution, which in turn caused sputtering of the solution into the gas phase in the reaction vessel and carry-over of a sodium- containing mist into the flame. A very noisy flame with a strong and unstable sodium emission was obtained after each injection.This problem was overcome by placing a small glass-wool plug in the PTFE tubing before the nitrogen inlet into the cavity. Because ammonia is highly water soluble, it is difficult to de-gas the solution. Therefore, when the volume of the sample solution was increased the peaks became lower and broader. The optimum sample volume was 0.2m1, as reported by Belcher et al.,4,5 and this volume was used in the determi- nation of ammonium-nitrogen without the cold-trap. When the cold-trap was used to collect ammonia before measurements were made, the volatilisation system could not be heated and so the condensation of water vapour in the tubing could not be prevented. It was, however, found to be necessary to prevent the condensation of water in the tubing, otherwise erratic results were obtained, probably due to the dissolution of ammonia in the condensed water.Various drying agents were used to dry the ammonia vapour; sodium hydroxide pellets did not reduce the emission intensity and were, therefore, used in the following experiments. A linear calibration graph (Fig. 3) was obtained for 5.&100 pg ml-1 of ammonium-nitrogen by sweeping the ammonia directly into the cavity. The relative standard deviations (seven experiments) for the determination of 20 and 40 pg ml-1 of ammonium-nitrogen were 5.9 and 4.9%, respec- tively. The limit of detection (signal equal to twice the background noise) was 1.5 pg ml-1 of nitrogen. The limit of detection could be improved by collecting the ammonia generated; this was achieved by injecting much larger sample volumes into the liquid nitrogen cold-trap and introducing the ammonia collected into the flame for a short period of time.Hence, sharp emission peaks were obtained and 0.01 pg ml-l of nitrogen could be detected. The calibration graph obtained with the cold-trap method was linear for 0.05-1.0 pg ml-1 of nitrogen, whereas for 2 pg ml-1 of nitrogen an apparent self-absorption was observed (Fig. 3). The relative standard deviation for the determination of 0.5 pg ml-1 of ammonium- nitrogen with the cold-trap system (seven experiments) was 4.1%. Determination of Nitrite- and Nitrate-nitrogen Nitrite was reduced by iodide to nitrogen monoxide which was then swept into the flame by the nitrogen carrier gas and the emission was monitored. When the nitrogen monoxide, generated by injecting 0.5 ml of a sample solution into 0.5 mlANALYST, MAY 1989, VOL.114 565 0 20 40 60 80 100 NH3direct(A) 0.4 0.8 1.2 1.6 2.0 NH3 trapped (6) 2 4 6 8 10 NO*-(C) Concentration of nitrogen/pg ml-1 Fig. 3. Calibration graphs: A, 5-100 pg ml-1 ofammonium-nitrogen by direct sweeping; B, 0.05-2.0 pg ml-1 of ammonium-nitrogen by the cold-trap method; and C, 0.5-10 pg ml-' of nitrite-nitrogen of the iodide solution, was swept directly into the flame, a broad peak with pronounced tailing was obtained. This could be due to the slowness of the reaction and/or the slowness of the diffusion of the nitrogen monoxide out of the solution. The reaction was complete in 1 min and a sharp peak with a base of 123 duration was obtained when a 1-min reaction time was allowed before the measurements were made. The volumes of the sample and iodide solutions were chosen so as to obtain the highest sensitivity and reproducibility.A 0.5-ml volume of each solution was used in the following experiments. The calibration graph was linear up to 7 pg ml-1 of nitrite-nitrogen (Fig. 3). The relative standard deviation for the determination of 5 pg ml-1 of nitrite-nitrogen (seven experiments) was 3.9% and 0.1 pg ml-1 of nitrite-nitrogen could be detected. Nitrate was determined by the iodide method after its reduction to nitrite by means of a copperised cadmium column (17 cm long, 8 mm in diameter) as described elsewhere.'." The pH of the nitrate solution was adjusted using boric acid - borax buffer instead of ammonia solution to avoid any spectral interferences from the latter.A change in the pH of the buffer within the range 5.3-9.6 did not affect the reduction of nitrate to nitrite. The efficiency of the conversion of nitrate to nitrite for a sample containing 0.5-10 pg ml-1 of nitrogen was 9.5-100%. The calibration graph and the limit of detection were the same as that obtained for nitrite-nitrogen. The relative standard deviation for the determination of 5 pg ml-1 of nitrate-nitrogen was 4.3%. Interferences Various cations and anions have been reported to have no effect on the generation of nitrogen monoxide from nitrite by iodide6 (apart from oxidising nitrite to nitrate) nor on the reduction of nitrate to nitrite by the copperised cadmium column.11 However, in this work carbon dioxide and an excess of hydrogen sulphide gave emissions within the cavity and produced spectral interferences.These two interferents were eliminated by placing a trap between the reaction vessel and the cavity. The trap was packed in series with cotton-wool moistened with lead acetate solution, potassium hydroxide pellets and barium hydroxide powder. Using this trap, the effect of 0.01 M sodium carbonate solution on the determina- tion of 0.5-10 pg ml-1 of nitrite-nitrogen was eliminated. Table 1 . Determination of nitrite in meat products Amount of nitrite per 10 g of sample/mg Sample SoudjikI* . . SoudjikII* . . SoudjikIIt . . SoudjikIII" , . SoudjikIIIt . . Salami* , . . , Salamit . . . . SalamiII" . . SalamiIII" .. SausageI" . . SausagcII* . . SausageIIT . . Found . . 0.085 . . 0.21 . . 0.21 . . $ . . -1: . . $ . . $ . . -1: . . 0.20 . . 0.12 . . 0.12 . . Added 0.33 4.11 4.11 2.0s 0.82 2.05 2.05 0.82 0.82 0.82 2.05 2.05 Recovered 0.31 4.03 3.90 2.14 0.79 1.95 1.99 0.81 0.81 0.83 1.94 1.88 " Proteins precipitated before the measurements were made. t Measurements made without separating the proteins. j Not detectable. Table 2. Concentration of nitrate in meat products Concentration of Sample nitrateimg kg- 1 SoudjikI . . . . 17 Salami1 . . . . . . 45 Salami11 . . , . 53 SoudjikII . . . . 11 Sausage1 . . . . 62 Table 3. Determination of nitrate-nitrogen in drinking water. All values in mg 1-1 Spectro- Sample technique method 1 1.40 1.30 2 1.52 1.48 3 6.80 6.80 4 5.20 5.20 S 1.72 1.68 MECA photometric Determination of Nitrite and Nitrate in Meat Products The proposed MECA method was applied to the determina- tion of nitrite and nitrate in soudjiks, salami and sausage.The samples were taken randomly from the supermarket. The samples were prepared for analysis according to a standard procedure.12 A 10-g amount of the minced samples was extracted with hot water at 7540°C. The proteins in the extracts were separated by precipitation with potassium hexacyanoferrate( 11) and zinc acetate. Nitrite was determined directly in the extract. For the recovery studies, known amounts of sodium nitrite were added to 10 g of the minced samples. The results obtained (Table 1) indicate that the recovery of added nitrite is satisfactory when the standard sample preparation procedure is coupled with MECA. The nitrate in the extracts was also determined without separating the proteins. The results were similar to those obtained after separation of the proteins (Table 1).The nitrate in the extracts was reduced with the copperised cadmium column12 and the total nitrite and nitrate was then determined. The nitrate concentration was calculated as the difference between the total nitrite and nitrate, and nitrite concentrations (Table 2). The column did not affect the results of the nitrite determination. Proteins in the extract blocked the reductor column and, therefore, had to be separated prior to the determination of nitrate. Determination of Nitrate-nitrogen in Drinking Water The water samples were taken from various sources in Izmir.No detectable levels of nitrite- and ammonium-nitrogen wereANALYST, MAY 1989, VOL. 114 found in the samples. The nitrate in the samples was reduced to nitrite using the reductor column and then determined by both the MECA and a standard spectrophotometric method' [diazotisation of nitrite ion with sulphanilamide and coupling with N-( 1-naphthy1)ethylenediamine dihydrochloride to form an azo dye which is then measured spectrophotometrically]. Table 3 shows that the results given by both methods are comparable. Conclusion The MECA technique using a cavity containing an entire flame and coupled with a volatilisation system constitutes a simple, fast and highly sensitive method for the determination of ammonium-, nitrite- and nitrate-nitrogen; the limits of detection are 100, 5 and 20 times better, respectively, than those reported previously using MECA with an oxy-cavity.*fi The proposed method was applied successfully to the determination of nitrate-nitrogen in drinking water and nitrite and nitrate in meat products.References 1. "Annual Book of ASTM Standard$," American Society for Testing and Materials, Easton, Philadelphia. PA, 1982. Part 31, Nos. D 1426, D 992 and D 3867. 2. 3. 4 5 . 6 7 8. 9. 10. 1 1 . 12. Honma, M., and Smith, C . L.. Anal. Chem.- 1954. 26, 358. Butcher, J . M. S., and Kirkbright, G. F.. Arta/y.rr. 1978, 103, 1104. Belcher, R . , Rogdanski. S . L.. Calokerinos. A. C., and Townshend, A . , Atzulyst, 1977, 102. 220. Belcher. R . , Bogdanski, S . I,.. Calokerinos. A. C., and Townshend, A . , Analyst, 1981. 106, 625. Al-Zamil. I. Z.. and Townshend, A , . Anal. Chirn. Acta. 1982, 142. 151. Belcher, R.. Bogdanski, S . L., Ghonaini, S. I,.. and Town- shend, A., Anal. Chim. Acta, 1974, 72, 183. Bogdanski. S. L.. Henden. E., and 'Iownshend, A , . Aizal. Chim. Acta, 1980. 116. 93. Henden, E.. Arid. C'him. Acta, 1985, 173, 89. Bclcher, R.. Bogdanski. S . L., Henden. E . , and Townshend. A , , And. Chim. Actu, 1977. 92. 33. Henriksen. A . . and Sclmer-Olscn, A. R.. Arzalysr, 1970, 95. 5 14. International Standards Organization, Standard Nos. I S 0 2918 (1975) and I S 0 3091 (1974), International Organization for Standardization, Geneva, Switzerland. Paper K103688H Received September 2/st, 1988 Accepted December I4th, 1988

ISSN:0003-2654    

DOI:10.1039/AN9891400563

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST. MAY 1989, VOL. 114 567 Gas Chromatographic Determination of Ethyl 2-Cyanoacrylate in the Workplace Environment Virindar S. Gaind and Kazik Jedrzejczak Occupational Health Laboratory, Ontario Ministry of Labour, 101 Resources Road, Weston, Ontario M9P 3T1, Canada Low levels of airborne ethyl 2-cyanoacrylate (ECA) were monitored by drawing air through sampling tubes containing Tenax GC. After desorption with acetone, ECA was quantified by gas chromatography using a therrnionic specific detector in the nitrogen mode. Polymerised ECA was shown to yield ECA monomer under the conditions of gas chromatographic analysis. Keywords: Gas chromatography; airborne ethyl 2-cyanoacrylate; glue Ethyl 2-cyanoacrylate (ECA) and other alkyl2-cyanoacrylates such as methyl 2-cyanoacrylate (MCA) are highly reactive monomers that undergo rapid anionic polymerisation when a small amount of the monomer is pressed between two surfaces.The polymerisation is catalysed even by the small amount of water vapour normally present in the atmosphere. Alkyl cyanoacrylates set rapidly resulting in their widespread use as adhesives in a variety of industrial, household and medical applications. Krazy Glue adhesive is reported to contain 99.9% ECA. Both ECA and MCA are moderate eye irritants and can cause transient blurred vision at levels of 2 p.p.m.1 Further, ECA has toxic effects if ingested or absorbed through the skin. The American Conference of Governmental industrial Hygie- nists* has adopted a threshold limit value for an 8-h time- weighted average of 8 mg m--7 (2 p.p.m.) for MCA and the Ontario Ministry of Labour' has recommended a working exposure guideline of 2 p.p.m.of ECA measured over 15 min. The gas chromatographic and infrared analytical procedures used for the determination of ECA and MCA in manufactur- ing processes are not sufficiently sensitive for monitoring the low concentrations of these monomers likely to be present in the workplace atmosphere. Leonard et al.4 developed a more sensitive procedure in which MCA was determined by measuring the amount of formaldehyde released by poly- merised MCA cn its reaction with 4,s-dihydroxynaphthalene- 2,7-disulphonic acid (chromotropic acid). However, this method is not specific. This paper describes a highly sensitive and specific proce- dure for the determination of airborne ECA based on the collection of ECA on a Teflon (polytetrafluoroethylene bonded to polypropylene) filter and a tube containing Tenax adsorbent. The quantification of ECA or MCA is carried out by gas chromatography using a thermionic specific detector in the nitrogen mode.Experimental Reagents Ethyl 2-cyanoacrylate and methyl 2-cyanoacrylate were received as a gift from 3M (London, Ontario, Canada). Acetone (analytical-reagent grade) was obtained from Cale- don Laboratories (Georgetown, Ontario, Canada). Apparatus The portable air sampling pump was a Bendix Model 44. The air sampling tubes (Catalogue No. 226-35-03) and Teflon filters (Catalogue No. 225-17-01) were obtained from SKC (Eighty Four, PA, USA). A Varian Vista Model 44 gas chromatograph equipped with a thermionic specific detector (TSD) and a Model 401 data system, and a Hewlett-Packard Model 5985 gas chromatograph - mass spectrometer with an HP 7920 data system were used.Sample Collection Air was drawn ?t 1 1 min-1. for 30 min, through a Teflon filter and then through a sorbent tube containing two sections of Tenax. The filter and Tenax from both sections of the tube were extracted separately using acetone and the extracts of ECA and MCA were quantified by gas chromatography (GC). Gas Chromatographic Analyses with Tenax GC, 50180 mesh. 230 "C. Under these chromatographic conditions, the peak for ECA was well resolved from both that of the solvent and MCA, as shown in Fig. 1. Further experiments on stability and recovery were conducted on ECA only.Column. Glass, 1.8 m x 6 mm 0.d. and 2 mm i.d., packed Temperatures. Oven, 200; injector, 220; and detector, Carrier gas. Nitrogen, 30 ml min-1. Preparation of Polymeric ECA Approximately 0.5 ml of liquid monomeric ECA was spread thinly on a watch-glass and left overnight in a fume cupboard. t - m C 0, v) ._ 0 5 10 Ti meimin Fig. 1. ECA GC - TSD chromatogram showing separation o f MCA and568 ~~ 80 60 40 ANALYST, MAY 1989, VOL. 114 (a) - - - 100 7 I 1 Table 1. Precision data obtained by measurement of peak areas (arbitrary units) Concentration of ECA/pg ml ~ I ~~~ 5 49 718 49 749 50 266 50 321 48 186 49 214 49 204 47 717 x: 49 297 SD: 935 CV: 0.019 10 108 887 103 339 98 080 96 153 106 595 110 023 99 235 104 575 x: 103361 SD: 5123 c v : 0.049 20 205 586 197 332 201 111 206 840 196 053 204 213 192 983 204 245 x: 201 045 SD: 5045 CV: 0.025 Table 2.Recovery of ECA from spiked Tenax tubes ECA ECA Average addedlyg foundlpg recovery, YO 5.0 4.4 4.8 93 4.8 10.0 10.1 9.6 20.0 19.8 19.3 97 19.2 10.2 100 x: 97% CV: 0.036 SD: 3.5% Table 3. Stability of ECA on Tenax tubes under various storage conditions Day sample was analysed ECA addedlpg Ambient temperature (2CL23 "C)- 0 5 .0 0 10.0 0 20.0 3 5.0 3 10.0 3 20.0 7 5.0 7 10.0 7 20.0 14 5.0 14 10.0 14 20.0 0 5.0 0 10.0 0 20.0 3 5.0 3 10.0 3 20.0 7 5.0 7 10.0 7 20.0 14 5.0 14 10.0 14 20.0 Refrigerated at 4 "C- ECAfoundiyg 4.6 9.5 19.2 4.7 9.4 19.3 4.8 10.2 19.8 4.4 9.1 19.5 4.8 10.2 19.3 4.2 9.8 20.4 4.7 9.6 19.8 4.5 9.2 19.7 Average recovery, YO 94 95 99 92 98 95 96 93 The brittle solid that formed was weighed and dissolved in acetone to prepare standard solutions of polymeric ECA. Results Precision and Linearity The detector response to ECA showed excellent precision and linearity when replicate injections (2 yl) were made of three 80 t 80 98 ' l l ? , 126 M + 1 100 120 140 160 180 200 miz 60 80 100 120 140 160 180 200 mi z Fig.2. mode for both (a) a monomeric ECA and ( b ) a polymeric ECA Identical mass spectra obtained in the chemical ionisation standard solutions ( 5 , 10 and 20ygml-1). The results are given in Table 1. Recovery of ECA Spiked on Tenax The front sections of several tubes packed with Tenax were spiked with 5 , 10 or 20 pg of ECA using a syringe and air (30 I) was drawn through the spiked tubes at 1 1 min- 1.Each section of Tenax was placed in a separate vial, which was then crimp-sealed, and extracted with 1.0 ml of acetone. The acetone extracts were analysed for ECA and the percentage recovery was calculated by comparison with standard solu- tions of ECA. The results given in Table 2 indicate near quantitative recoveries at all levels of spiking. Stability of ECA Spiked on Tenax Replicate Tenax tubes were spiked with known amounts of ECA and then stored at 4°C and at ambient temperature (2@23"C) for various periods of time. The ECA was determined after storing the tubes for 0, 3, 7 and 14d. The amount of ECA recovered is given in Table 3. It was found that samples collected on Tenax remained stable for at least 14 d with recoveries in the range 93-100% at different levels of spiking.Mass Spectra of ECA The electron impact mass spectrum of ECA obtained at 70 eV yielded a negligible molecular ion peak at rnlz 125 (1%) and a base peak at rnlz 98 that was of little value for identification or quantification purposes. However, the chemical ionisation spectrum obtained with methane as the reagent gas yielded the molecular ion peak at rnlz 126 ( M + 1) as the base peak. Other characteristic peaks at mlz 154 ( M + 29) and m/z 166 (A4 + 41) were also discernible (Fig. 2). The presence of a molecular ion peak in the chemical ionisation mode provides an alternative procedure for confirming the presence of and quantifying ECA in samples. Adam$ studied the chemical ionisation pyrolysis mass spectra of a number of polymers other than polycyanoacrylates and found that polystyrene , polybuta- diene, polymethylmethacrylate, etc., yielded characteristicANALYST, MAY 1989, VOL.114 569 Filter Table 4. Relationship between the injector temperature and the depolymerisation level. Sample mass. 10 pg Injector Peak area, Depolymerisation temperature/’C arbitrary units level, YO Polymeric ECA- 90 70 958 8 100 296 571 35 120 429 854 51 140 641 113 77 160 718 878 86 200 830 222 99 220 831 420 99 220 836 460 100 Monomeric ECA- molecular ions, especially ( M + l), corresponding to the respective monomer on pyrolysis. Our work, however, shows the reproducible and quantitative nature of the poly- merisation - depolymerisation process for ECA when it occurs at temperatures above 220°C and in an inert atmosphere in the gas chromatograph.The mass spectra obtained for solutions of monomeric and polymeric ECA, injected through the heated injection port (220 “C), were identical in all respects, confirming that the polymer depolymerises to the monomer under these conditions. Discussion The quantitative recovery of ECA spiked on Tenax tubes, achieved even after passing air through the tubes or keeping the tubes for 2 weeks, was intriguing in view of the known high reactivity of the monomer. One plausible explanation was that monomeric ECA collected on Tenax polymerised at some stage. However, on dissolution in acetone and subsequent injection into the gas chromatograph the polymer depoly- merised completely in the heated injection port. In order to validate the above hypothesis, 500mg of monomeric ECA were spread as a thin film on a watch-glass and allowed to solidify.After 16 h, the brittle solid formed was dissolved in acetone (50ml) to yield a standard solution (10.0 mg ml-I). After further dilution, a standard solution prepared from polymeric ECA (10 LLg ml-1) was compared against a standard solution of the same concentration freshly prepared from monomeric ECA. The peak areas and reten- tion times obtained for both solutions were essentially similar. The chemical ionisation mass spectrometric profiles of both solutions were indistinguishable. Further support for our hypothesis that polymerised ECA depolymerised to monomeric ECA as a result of heating in the gas chromato- graph injection port was obtained when it was found that the peak area obtained by injecting 10 pg of polymerised ECA increased progressively with an increase in the injection temperature up to 200 “C, after which it became constant and almost equal to that obtained on injection of 10pg of monomeric ECA.The peak area obtained by injecting monomeric ECA did not change significantly with injection temperature (Table 4). Rooney6 studied the effect of heating cyanoacrylate oli- gomers using thermogravimetric techniques on a qualitative level and showed using NMR spectra of ECA oligomers that during pyrolysis there were no structural alterations other than a gradual decrease in the relative molecular mass. The work described in this paper shows that the process of depolymerisation, under the controlled conditions of GC analysis, is highly reproducible and quantitative.From the preceding discussion it is apparent that the GC determination of ECA collected on sampling tubes containing Tenax would include both the monomer and polymerised ECA as the analytical procedure is unable to distinguish N2 U Fig. 3. Configuration of apparatus for ECA sampling between the two. However, completely polymerised ECA is unlikely to have sufficient vapour pressure to allow its collection on the Tenax tubes during sampling. Any particu- late ECA (arising from polymeric ECA) can be distinguished from the monomer vapours by placing a Teflon filter between the sampling pump and the Tenax tube. To validate this, the following experiment was conducted to simulate the actual sampling process. A sampling glass bulb with a rubber septum and two Teflon stopcocks was connected to a cassette containing a Teflon filter followed by a Tenax tube (Fig.3). Dry nitrogen was passed through the system at 11 min-1. The ECA monomer (100 pg) in acetone was injected through the rubber septum and the flow of nitrogen maintained for 30 min (30 1). At the end of this period, the Teflon filter and two sections of Tenax from the sampling tube were extracted with acetone and ECA was quantified in each extract. For three replicate determina- tions at ambient temperature (20-23 “C) the recoveries of ECA from the Tenax tubes were 87, 94 and 95% with a standard deviation (SD) of 4.3% and a coefficient of variation (CV) of 0.48. No ECA was found on the Teflon filter and the major portion of ECA added to the sampling bulb was found on the Tenax tube.The experiment was repeated with the injection of 100 pg of polymerised ECA. Analysis of the filter and Tenax tubes indicated no detectable amount of ECA on either of the sampling media. This shows that polymerised ECA does not have sufficient vapour pressure at ambient temperature to be carried even in a stream of inert gas, in contrast to monomeric ECA, most of which is retained by Tenax under similar conditions. Conclusions Airborne ECA and MCA can be monitored by drawing air through tubes containing Tenax followed by desorption with acetone and quantification of cyanoacrylate by GC with thermionic specific detection or selected ion monitoring using chemical ionisation mass spectrometry. If particulate material is suspected to be present in the atmosphere, a Teflon filter placed before the Tenax tube can be used to distinguish between polymeric and monomeric ECA or MCA. This method is capable of monitoring concentrations of ECA well below 0.01 mg m-3. The authors thank M. A. Nazar, Chief Scientist, Occupational Health Laboratory of Ontario Ministry of Labour for helpful suggestions during preparation of the manuscript. References 1. 2. 3. 4. 5 . 6. McGee, W. A., Oglesby, F. L., Raleigh, R. L., and Fassett, D. W., Am. Znd. Hyg. Assoc. J., 1968, November/December, 558. “Threshold Limit Values and Biological Exposure Indices for 1988-1989,” American Conference of Governmental Industrial Hygienists, Cincinnati, OH, p. 26. “Regulation Respecting Control of Exposure to Biological or Chemical Agents,” 0. REG. 654/86, Ontario Ministry of Labour, Ontario, p. 60. Leonard, F., Kulkarni, R. K., Brander, G., Nelson, J., and Cameron, J. J., J . Apyl. Polyrn. Sci., 1966, 10, 259. Adams, R. E., Anal. Chem., 1983, 55, 414. Rooney, J . M., Br. Polym. J . , 1981, December, 160. Paper 81044.53 H Received November 9th) 1988 Accepted Januury 11 th, 1989

ISSN:0003-2654    

DOI:10.1039/AN9891400567

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, MAY 19x9, VOL. 114 57 1 Chemically Bonded Cyclodextrin Stationary Phase for the High-performance Liquid Chromatographic Separation and Determination of Sulphonamides Abd-El Hamid N. Ahmed and Samia M. El-Gizawy College of Pharmacy, Department of Pharmaceutical Chemistry, Assiut University, Assiut, Egypt Mixtures of seven sulphonamides have been separated successfully on a fi-cyclodextrin stationary phase by high-performance liquid chromatography (HPLC) using phosphate buffer (pH 7.0) - methanol (85 + 15). The HPLC determination of trisulphapyrimidines and Polysulpha in dosage forms is described. That the proposed method is accurate was established from the percentage recoveries of standard sulphonamide solutions. Keywords: High-performance liquid chromatography; (3-cyclodextrin-bonded stationary phase; trisulp h ap yrim idines; Pol ysulp ha and sulp h on am ides Methods for the quantification of sulphonamides were per- formed traditionally by sodium nitrite titration or thin-layer chromatography - ultraviolet spectrophotonietry and were generally not specific, slow and tedious.1 Sulphonamides have been separated by ion-exchange chromatography on a cation- exchange resin ,2 ion-pair partition chromatography3 and adsorption chromatography on silica.3 These methods have been applied to the determination of these drugs in phar- maceutical dosage forms, body fluids and tissues.5 Recently, reversed-phase high-performance liquid chro- matography has been developed for the quantification of sulphonamides in various combinations.6.7 There has also been considerable interest in the utilisation of cyclodextrins as a stationary or mobile phase in chromato- graphy because of their ability to form inclusion complexes with a variety of organic molecules or ions both in the solid state and in aqueous solution.8-l4 The interaction of /3-cyclodextrin (P-CyD) with some sulphonamides in aqueous solution has been investigated by circular dichroism (CD), ultraviolet (UV) absorption, solubil- ity techniques and by high-performance liquid chromato- graphy (HPLC).15.16 In this paper, we describe the use of a (3-CyD-bonded stationary phase in HPLC for the successful separation of sulphonamides and for the quantification of certain sulphon- amides in dosage forms. Experimental Apparatus For HPLC a Du Pont 8800 pump module, equipped with a stainless-steel column (100 x 4.6 mm i.d.) packed with V-CyD chemically bonded to a high-purity silica gel (Cyclobond I, Advanced Separation Technologies, USA), was used.Detec- tion was effected spectrophotometrically at 257 nm using a Du Pont variable-wavelength UV spectrophotometer. A Ser- vogor 310 recorder combined with an SP 4100 computing integrator was used to monitor the chromatographic charac- teristics. used to adjust the pH. A PW 9418pH meter (Pye Unicam, Cam Chemicals and Reagents All chemicals and reagents used were either I (USP) - National Formulary or ACS grade. ,ridge, UK) was S Pharmacopeia Dosage Forms Trisulphapyrimidine tablets (USP) (Triple Sulpha). Each tablet (SO 1 mg) contained equal parts of sulphadiazine, sulphadimidine and sulphamerazine (1 67 mg of each).Polysulpha tablets (CID Laboratories, Egypt). Each tablet contained sulphadiazine (0.133 g), sulphathiazole (0.133 g), sulphadimidine (0.134 g) and sulphamerazine (0.100 g). Solutions Sodium orthophosphate (0.05 M) solution was prepared using distilled water and its pH adjusted to 7 with 0.1 M NaOH. Stock solutions (1 mg ml-1) of sulphathiazole, sulphaguan- idine , sulphadiazine , sulphamerazine, sulphadimidine , sul- phaphenazole and sulphacetamide were prepared in methanol and diluted serially with phosphate buffer (0.05 M NaH2P04, pH 7) so as to give final concentrations of 1@-30, 5-20, 10-30. 5-20, 5-20, 10-30 and 10-30 pg ml-1 of each sulphonamide, respectively, for the construction of calibration graphs.Chromatographic Conditions The mobile phase was phosphate buffer (pH 7.0) - methanol prepared in the ratios of SO + 50,S5 + 45.60 + 40,6S + 35,70 + 30,75 + 2S,80 + 20,85 + 15 and 90 + 10. The mobile phase was degassed directly before use. The flow-rate was varied from 1.5 to 0.5 ml min-1 and the temperature adjusted to 35 "C. The chart speed was 0.25 cm min-1 and the attenuation unit for full-scale deflection was set at 1 mV. Preparation of Assay Solutions For trisulphapyrimidine tablets (Triple Sulpha) One tablet was placed in each of ten 100-ml calibrated flasks and SO ml of methanol were added. The sample was sonicated (15 min) and the solution made up to the mark with methanol. A portion of the solution was centrifuged (15 min), then 0.1 ml of the clear supernatant solution was transferred quantita- tively into a 10-ml calibrated flask and made up to the mark with phosphate buffer (pH 7.0) - methanol (85 + 15).The final concentrations were equivalent to 16.7 pg ml-1 of sulphadiazine, 16.7 pg ml-1 of sulphadimidine and 16.7 pg ml-1 of sulphamerazine. For Polysulpha tablets The procedure was identical with that described above for trisulphapyrimidine tablets. The final concentrations were equivalent to 13.3 pg ml-1 of sulphadiazine, 13.3 pg ml-1 of sulphathiazole, 13.4 pg ml-1 of sulphadimidine and 10.0 pg ml- 1 of sulphamerazine.572 ANALYST, MAY 1989, VOL. 114 Assay The standard and prepared solutions (50ml of each) were injected into the chromatograph and the peak areas were determined. The amount of active components was deter- mined by comparing the peak area of the sample with the representative peak area of the standard of known concentra- tion.Results and Discussion The P-CyD-bonded stationary phase (Cyclobond I) retains a wide variety of compounds via the formation of inclusion complexes. Consequently, compounds that were previously thought to be difficult to separate by conventional liquid chromatography can be resolved easily. To form the inclusion complexes, the guest sulphonamide molecules approach and penetrate the hydrophobic cavity of the cyclodextrin from the more open and accessible secondary hydroxyl group of the glucose unit. As a result, some of the physico-chemical properties of the guest sulphonamide mol- ecule can be modified.15 The stability of the inclusion complexes varies with the size of both the guest sulphonamide molecule and the host P-CyD.The basis of the separation and resolution of some of the sulphonamides studied was observed from the selectivity in binding towards the CyD cavity.16.17 Knowing the stability of the P-CyD inclusion complexes formed, an accurate predic- tion could be made concerning the expected elution behaviour on the column of the seven sulphonamides studied. The sulphonamides could be detected easily and monitored using HPLC. Sulphathiazole interacts strongly with P-CyD and is retained longer; therefore it exhibits a relatively longer Time/m in Fig. 1. Separation of sulphonamides using phosphate buffer (pH 7) - methanol (85 + 15) at a flow-rate of 0.5mlmin-1. A, Sul- phacetamide; B, sulphadiazine; C, sulphamerazine; D, sulphaphen- azole; E, sulphaguanidine; F, sulphadimidine; and G, sulphathiazole Table 1.Determination of sulphonamides in known synthetic mixtures Added1 Found k SD*I Recovery, No. Prepared mixture mg mg Yo 1 Sulphadiazine Sulphathiazole Sulphadimidine 2 Sulphadiazine Sulphamerazine Sulphathiazole 3 Sulphadiazine Sulphamerazine Sulphadimidine 133 133 134 185 185 185 167 167 167 133.0 t 0.25 132.7 t 0.61 134.2 t 1.03 185.5 t 0.49 185.0 t 1.27 184.8 k 1.07 166.6 k 0.16 167.2 k 0.35 167.5 t 1.31 * Based on five replicate analyses of known mixtures. 100.00 99.77 100.15 100.27 100.00 99.89 99.76 100.12 100.30 retention time than all the other sulphonamides studied. Sulphacetamide interacts weakly with P-CyD and is eluted rapidly with a consequent short retention time.The elution order of the seven sulphonamides was depen- dent on the amount of methanol in the mobile phase. The composition of the mobile phase [phosphate buffer (pH 7.0) - methanol (55 + 45, 60 + 40, 65 + 35,70 + 30,75 + 25, 80 + 20, 85 + 15 and 90 + 101 was varied to investigate its effect on the resolution and retention times. The presence of the phosphate buffer at the expense of methanol was essential for complete separation. The effect of the flow-rate (1.5, 1 .O and 0.5mlmin-1) of the mobile phase on the resolution of the seven sulphonamides was also studied. Efficient separation of the seven sulphonamides was achieved using phosphate buffer (pH 7.0) - methanol (85 + 15) at a flow-rate of 0.5 ml min-1 (Fig.1). The proposed HPLC method was applied to the analysis of trisulphapyrimidine (Triple Sulpha) and Polysulpha tablets using phosphate buffer (pH7.0) - methanol (85 + 15) as the mobile phase at a flow-rate of 0.5 ml min-1. The results for the synthetic mixtures were quantitative and showed complete recovery, Table 1. A linear regression was obtained by plotting the standard concentrations of sulphamerazine, sulphadiazine, sulphadim- idine and sulphathiazole versus peak area. Table 2 gives the slope, correlation coefficient and intercept for each sulphon- amide. Figs. 2 and 3 illustrate the utility of the Cyclobond I column for the separation of sulphonamides in trisulphapyrimidine and Polysulpha tablets. There was no indication that the tablet excipients interfered with the separated peaks.It is apparent that the effect of the accumulation and competition of seven sulphonamides in the standard mixture (Fig. 1) on the resolution from the P-CyD cavity is different from the competition of three or four sulphonamides (Figs. 2 and 3) so that minor differences in the retention times were obtained. Also the differences in the composition of the mobile phase from day to day should also be taken into consideration. Table 2. Calibration data for standard drug solutions Concen- Correlation tration/ coefficient, Inter- Compound pg ml-1 r(SD)* Slope cept Sulphadiazine . . 10-30 0.998(0.015) 0.893 0.046 Sulphamerazine . . 5-20 0.999 (0.001) 0.151 0.001 Sulphadirnidine . . 5-20 0.998 (0.027) 0.275 0.176 Sulphathiazole . . 10-30 0.997 (0.012) 0.623 0.117 * Average of five determinations.Fig. 2. Chromatogram of 50 PI of a solution of trisulphapyrimidine tablets containing: A, sulphadiazine; B, sulphamerazine; and C, sulphadimidineANALYST, MAY 1989, VOL. 114 l-----l I I I Ti meimin Fig. 3. Chromatogram of 50 p1 of a solution of Polysulpha tablets containing: A, sulphadiazine; B, sulphamerazine; C, sulphadimidine; and D , sulphathiazole Table 3. Analysis of dosage forms Label Assay claim results Recovery, Preparation mg per tablet % SD CV,* % Trisulphapyrimidine tablets- Sulphadiazine . . 167 166.1 99.46 0.51 0.31 Sulphamerazine . . 167 165.7 99.22 1.29 0.77 Sulphadimidine . . 167 167.8 100.48 1.03 0.61 Sulphadiazine . . 133 134.0 100.75 1.04 0.78 Sulphamerazine . . 100 101.1 101.1 1.30 1.03 Sulphadimidine .. 134 132.9 99.18 1.27 0.95 Sulphathiazole . . 133 134.0 100.75 1.09 0.81 * Standard deviation of the mean of five determinations divided by Polysulpha tablets- the mean expressed as a percentage. 573 The authors thank Nabil M. Omar, Faculty of Pharmacy, Assiut University, Assiut, Egypt, for helpful discussions and support of this work. 1. 2. 3. 4. 5 . 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. References “US Pharmacopeia XXI, National Formulary XVI,” US Pharmacopeial Convention, Rockville, MD, 1985, pp. 988- 993. Poet, R . B., and Pu, H. H . , J . Pharm. Sci., 1973, 62, 809. Su, S. C., Hartkopf, A . V., and Karger, B. L.,J. Chromatogr., 1976, 119, 523. Cobb, P. H . , and Hill, G. T., J . Chromatogr., 1976, 123, 444. Pryde, A., and Gilbert, M.T., “Applications of High Per- formance Liquid Chromatography.” Chapman and Hall, London and New York, 1979, p. 73. Farmi, A. A , , Tylor, D., and Mashi, S. Z . , Drug Develop. Ind. Pharm., 1984, 10, 31. Parasrampuria, J., and Das Gupta, V., Drug Develop. Ind. Pharm., 1986, 12, 251. Hinze, W. L., Sep. Purif. Methods, 1981, 10, 195. Armstrong, D. W.. and Demond, W., J. Chromatogr. Sci., 1984, 22, 411. Kujimura, K., Ueda, T., and Ando, T., Anal. Chem.. 1983,55, 446. Kawaguchi, Y . , Tanaka, M., Nakae, M., Funazo, R., and Shono, T., Anal. Chem., 1983, 55, 1852. Armstrong, D. W., Demond, W., Alak, A., Hinze, W. L., Riehl, T. E., and Bui, K. H., Anal. Chem., 1985, 57, 234. Hinze, W. L., Riehl, T. E., Armstrong, D . W., Demond, W., Alak, A., and Ward, T., Anal. Chem., 1985, 57, 237. Armstrong, D . W., Demond, W., and Czech, B. P., Anal. Chem., 1985, 57, 481. Bender, M. L., and Komiyama, M., “Cyclodextrin Chem- istry,” Springer-Verlag, New York, 1978. Uekama, R., Hirayama, F., Otagiri, M., Otagiri, Y., and Ikeda, R . , Chem. Pharm. Bull., 1978, 26, 1162. Uekama, R., Hirayama, F., Nasu, S., Matsuo, N., and Irie, T., Chem. Pharm. Bull., 1978, 26, 3477. The accuracy of the method was established by achieving reproducible results of 99.22-100.48% for the trisulphapyr- imidine tablets and 99.18-101.1°/~ for the Polysulpha tablets, Table 3. Paper 8101367E Received April 6th, 1988 Accepted December 14th, 1988

ISSN:0003-2654    

DOI:10.1039/AN9891400571

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

ANALYST, MAY 1989, VOL. 114 575 Thin-layer Chromatographic Scanner, Spectrophotometric and High-performance Liquid Chromatographic Methods for the Determination of Colchicine Taha M. Sarg, Maher M . El-Domiaty and Mokhtar M. Bishr Faculty of Pharmacy, Zagazig University, Zagazig, Egypt Osama M . Salama and Ahmed R. El-Gindy Islamic Centre for Medical Sciences, Kuwait One high-performance liquid chromatographic (HPLC) and two thin-layer chromatographic (TLC) methods are proposed for the determination of colchicine in crude drugs and pharmaceutical preparations. The TLC scanner method is based on measurement of the absorbance of the separated colchicine spot; alternatively, after scraping the spot from the plate and elution the absorbance can be measured spectrophotometrically. The HPLC assay was carried out isocratically on a reversed-phase column using MeOH - H20 (60 + 40).The recoveries were 99.2 k 1.23,gg.l k 1 .I2 and 99.1 k 2.01 % for the TLC scanner, spectrophotometric and HPLC methods, respectively. The methods were shown to be sensitive and specific and can be used as an alternative to the pharmacopoeia1 methods having been applied to the determination of colchicine in corms ot Merendera persica and in three pharmaceutical preparations. Keywords : Th in-la yer chroma tog rap h ic scanner method; spectrop ho tom e tr y; high -perf0 rmance liquid chromatography; colchicine determination; pharmaceutical preparations The plant alkaloid colchicine i s known to provide effective pain relief in acute gouty arthritis.l.2 It is the predominant alkaloid in plants of the family Liliaceae, especially of the genera Colchirunz and Merendera.si Typically, the official methods for the determination of colchicine in crude drugs and pharmaceutical preparations are gravimetricGY or colorimetric.x.1O7I I Other methods, viz., colorimetric and spectrophotometric,2.12-'" fluorimetric,'7.19 volumetricl" and chromatographic2(&23 assays have also been described.Most of these methods are designed to measure the total alkaloid content and hence are not specific. They involve multiple extraction and purification procedures, often fol- lowed by a derivatisation step or by chemical reaction prior to the actual determination. Reactions with hydroxylamine hydrochloride,24 iron(II1) chloride,25 nitric acid,26 iodine27 and aluminium lithium hydride15 have been exploited.Polaro- graphic investigations28 have not been widely undertaken and have been electrochemically oriented.29 Therefore, the re- ported methods lack either simplicity, precision or specificity and have various other disadvantages.22 More recently, an HPLC method has been described-?() that is concerned more with the separation of colchicine derivatives than with quantification, which requires a complicated gradient elution technique. The work described in this paper was undertaken to demonstrate explicitly the exploitation of both a simple MPLC method and a TLC scanner method for the rapid and precise determination of colchicine. Also discussed is a low-cost method that involves TLC followed by elution and the \pectrophotometric determination of colchicine.The three proposed methods were applied successfully to the determina- tion of colchicine in corms of Merendera persica, in which colchicine is the major alkaloid,i and in a number of pharmaceutical preparations. Experimental and Results Materials In this work, one crude drug and several commercial pharmaceutical dosage forms containing colchicine were used. A pure colchicine sample was used to confirm the accuracy and efficiency of the proposed methods. The following samples were employed: (1) corms of Merendera persica L., family Liliaceae (Hamdard Foundation, Pakistan, Waqf); (2) colchicine tablets (Evans Medical, Greenford, Middlesex, UK) containing 500 pg of colchicine per tablet; (3) ColBene- mid tablets (Merck, Sharp and Dohme, St.Louis, MO, USA) containing 500 pg of colchicine and 500 mg of probenecid per tablet; and (4) Urosolvine effervescent granules (Nile, Cairo, Egypt) each 5 g containing colchicine (300 pg), piperazine hydrate (0.128 g), atropine sulphate (0.128 mg) and sodium citrate (the difference). Apparatus The TLC scanner method employed a Shimadzu C5-930 TLC scanner. Spectrophotometric determinations were carried out using a Shimadzu UV-260 spectrophotometer and silica gel G pre-coated plates (Merck). Chromatographic assays were conducted on a Shimadzu LC-4 HPLC instrument equipped with a UV detector. Determination of Colchicine in Corms of M . persica L. Sample preparation Coarsely powdered corms (5.0 g) were transferred into an apparatus for the continuous extraction of drugs and extracted exhaustively with 70% V/V ethanol.The alcohol was removed by evaporation under reduced pressure and the remaining aqueous residue was extracted with chloroform (4 x 30 ml). The combined extracts were evaporated to dryness and the resulting residue was then either dissolved in 5 ml of methanol (stock solution for the TLC scanner and HPLC determina- tions) or in 1 ml of methanol (stock solution for spectro- photometric assays). TLC scanner assay Procedure. The prepared stock extract of M . yersicu corms (10 pl) was spotted on silica gel G pre-coated plates, which were developed using chloroform - methanol (95 + 5 ) as the solvent system. The air-dried plates were subjected to UV measurement using a TLC scanner operating at 243 nm.This procedure was repeated six times. Quantification. The peak areas for six colchicine determina-576 ANALYST, MAY 1989, VOL. 114 tions were measured and the mean was calculated. The amount of colchicine in the prepared sample (Table 1) was then obtained using a calibration graph of average peak area versus known concentration of pure colchicine. The calibra- tion graph was linear in the range 1.0-6.0 yg (spotted amount) of colchicine and had a zero intercept. Precision and recovery. The precision and recovery were determined by repeated analysis of a single sample ( 5 g) of powdered corms of M . persica L. to which 1 mg of pure colchicine had been added ( i . e . , each 10-pl spotted volume should contain 2 pg of the added colchicine).The results given in Table 2 indicate good repeatability with a coefficient of variation (CV) of 1.24% and a standard deviation (SD) of 1.23%. Spectrophotometric assay Procedure. The stock solution (10 pl) was spotted (together with pure colchicine as a marker) on a pre-coated silica gel G plate. The plate was developed using CHC13 - MeOH (95 + 5) and then air-dried. The colchicine spots (RF, 5.5) were detected under UV light (366 nm) and checked against the colchicine marker. These spots were then scraped from the plate and the colchicine was eluted using methanol. The methanolic extract was filtered quantitatively, the volume adjusted to 5 ml and the UV absorbance measured at 350 nm. Each sample was analysed six times. Table 1. Determination of colchicine in corms of M .persica by three different methods Colchicine */ pg per gram Colchicine, Method of powder % mtrn TLC scanner . . . . . . , . 571.6 0.057 Spectrophotometric . . . . . . 580 0.058 HPLC . . . . . . . . . . 570 0.057 * Mean of six determinations Table 2. Precision achieved by three different methods Recovery, Yo Run No. TLC scanner Spectrophotometric HPLC 1 2 3 4 5 6 7 8 9 10 Average . . SD . . . . CV . . . . 98.0 99.4 99.8 101.9 99.1 98.9 97.9 100.3 98.5 98.2 99.2 1.23 1.24 99.8 97.2 100.5 99.4 98.3 98.5 99.6 97.9 99.1 100.7 99.1 1.12 1.13 96.8 98.2 101.5 99.9 99.5 98.1 100.2 97.6 100.1 99.1 99.1 2.01 2.03 Quantification. The mean absorbance for six determina- tions was calculated and the concentration of colchicine determined with the aid of a calibration graph of known concentration of pure colchicine versus absorbance. The calibration graph was linear in the range 1.76-8.8 pg with a zero intercept.The results are shown in Table 1. Precision and recovery. The precision and recovery of this method were determined by the procedure described for the TLC scanner method but using 0.3 mg of pure colchicine (i.e., each spotted volume should contain 3 pg of added colchicine). The results, with an SD of 1.12% and a CV of 1.13%) are given in Table 2. ' HPLC assay Chromatographic system. A Shimadzu LC-4 chrornato- graph, fitted with a UV detector operating at a wavelength of 243 nm, was employed. A reversed-phase Zorbax-ODS C18 column (15 cm x 4 mm i.d.) was used at 40 "C. The mobile phase was MeOH - H20 (60 + 40) in the isocratic mode with a flow-rate of 0.5 ml min-I.The injection volume was 1 pl. Quantification. The mean peak area for six determinations of colchicine ( R t , 9.5 min) was calculated and the concentra- tion of colchicine in the sample (Table 1) was obtained from a calibration graph, which was constructed by plotting the peak areas against the known concentrations of the pure com- pound. The calibration graph was linear over the range 0.2-1.5 pg of pure colchicine and passed through the origin. Precision and recovery. The efficiency of this method was verified by the good reproducibility obtained for repeated injections of the same sample, prepared after the addition of 2.5 mg of pure colchicine to 5 g of powdered corms (i.e. ~ 0.5 pg of pure colchicine in each injection) (Table 2).The CV was 2.03% and the SD 2.01%. Determination of Colchicine in Pharmaceutical Preparations The three proposed methods were applied to the determina- tion of colchicine in three commercial pharmaceutical prep- arations (see under Materials). Sample preparation Tablets. Twenty tablets (colchicine or ColBenemid) were weighed accurately and pulverised. An amount of finely powdered tablets equivalent to 5 mg of colchicine was extracted with methanol, the combined methanol extracts were filtered and the volume was adjusted to give a stock solution containing 0.5 mg ml-1 of colchicine. Effervescent granules. An accurately weighed amount of finely powdered effervescent granules (5 g) equivalent to 0.3 mg of colchicine was dissolved in water. After the efferves- cence had ceased, the resulting solution was extracted with chloroform. The chloroform was dried using anhydrous sodium sulphate and removed by distillation and the residue was dissolved in methanol to give a stock solution containing 0.5 mg ml-1 of colchicine.Table 3. Determination of colchicine (pg) in three dosage forms using the proposed methods TLC scanner method Spectrophotometric method HPLC method Dosage form Mean* k SDt CV, YO t Recovery, YO Mean" * SDt CV, YO? Recovery. YO Mean* & SD?- CV, YO t Recovery, YO I$ 500.5 k 1.7 0.21 100.1 499.0 * 2.7 0.54 99.8 499.1 f 3.2 0.64 99.8 II§ 499.8 f 3.0 0.60 99.9 498.0 * 3.7 0.74 99.6 498.6 f 3.3 0.66 99.7 1111 298.6 _+ 2.3 0.79 99.5 298.4 * 1.6 0.53 99.4 297.6 _+ 4.6 1.54 99.2 * Results are the mean of six determinations.?- SD = standard deviation; CV = coefficient of variation. $ Colchicine tablets (500 pg per tablet). 0 ColBenemid tablets (500 pg of colchicine and 500 mg of probenecid per tablet). 1 Colchicine effervescent granules. Each 5 g contains 300 pg of colchicine, 0.128 g of piperazine hydrate, 0.128 mg of atropine sulphate and sodium citrate making up the difference.ANALYST, MAY 1989, VOL. 114 577 Procedure The prepared samples were assayed using the TLC scanner, spectrophotometric and HPLC methods as described for the determination of colchicine in corms of M . persica. The results are given in Table 3. Discussion and Conclusion The work described in this paper was prompted by interest in the corms of M . persica, which have been used successfully for the treatment of rheumatoid arthritis and other disea~es31.3~ in some Asian and Arabian countries.The main purpose of the investigation was to develop simple, sensitive and rapid methods having a high degree of specificity for the determina- tion of colchicine in crude drugs and pharmaceutical prepara- tions. In order to determine the efficiency of the proposed methods, pure colchicine was added in known amounts to the samples prior to the assay procedures. The recoveries were quantitative. The results indicated that high recovery was achieved: 99.2, 99.1 and 99.1% for the TLC scanner, spectrophotometric and HPLC methods, respectively. The CVs were 1.24, 1.13 and 2.03%, respectively. The HPLC method provided satisfactory resolution of colchicine (Rt, 9.5 min) from related compounds such as colchicoside (Rt, 6.7 min) and allowed the determination of as little as 0.2 pg of colchicine in less than 10 min employing a simple isocratic technique.Hence this method is useful for rapid, routine work, requiring no reconditioning of the HPLC column. In terms of simplicity and sensitivity, the results showed that this method has advantages over the reported HPLC assay.3O In the TLC scanner method the absorbance was found to be proportional to the concentration of colchicine up to 6 pg and as little as 1 pg could be determined easily. Similar results (Tables 1-3) were obtained with the spectrophotometric assay. However, the scanning method has the advantage of measuring the absorbance due to colchicine directly on the plate, hence avoiding loss of material and giving savings in time and effort.We believe that these results are sufficiently good to merit replacement of the tedious, non-specific and less sensitive pharmacopoeia1 methods with one of the proposed methods. The method chosen would depend on the laboratory facilities available, the nature of the sample and the scale and purpose of the determination. The HPLC method is recommended for both the quantita- tive and qualitative determination of colchicine in crude drugs containing colchicine and its related compounds, owing to the Table 4. Comparison of sensitivity and requirements of the proposed methods Range of determination/ Time*/ pg,of Method Apparatus min colchicine Spectrophotometric . . UV spectrophoto- 60 1.7-8.8 meter TLC plates (silica gel) gel) TLCscanner .. . . TLCscanner 30 1.0-6.0 TLC plates (silica HPLC . . . . . . Simple HPLC instru- 10 0.2-1.5 ment fitted with a UV detector * For the determination of colchicine in the crude drug an extra 40 min are required for extraction. good separation provided by the chromatographic system. In addition the method can be used to determine sub-microgram amounts of colchicine in pharmaceuticals. The TLC scanner method is capable of determining colchicine in both crude drugs and pharmaceuticals with a high degree of accuracy. It also has the advantage of requiring only simple and relatively inexpensive apparatus. The spectrophotometric method requires even less expen- sive apparatus that is probably available in most laboratories.Although this method gives satisfactory results, it involves additional steps for the scraping and elution of colchicine from the adsorbent, leading to a potential loss of colchicine. The proposed methods provide a wide range from which the analyst can select the technique that best suits the laboratory equipment available and the purpose of the work. Compari- sons of the time and equipment required and of the limits of determination of colchicine for each method are given in Table 4. The authors are indebted to Dr. A. A. Al-Awadi, Minister of Public Health and President of the Islamic Organization of Medical Sciences, Kuwait, and to all members of the Islamic Centre for Medical Sciences for their encouragement and assistance throughout 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. 28. References Laurance, D. R., and Benaeth, P. N., “Clinical Pharmacol- ogy.,” Fifth Edition, Churchill Livingstone, London, 1980. Smith, G., Bullivant, J . M., and Cox, P. H., J . Pharm. Pharmacol., 1963, 15, 92T. Reynolds, J. E. F., and Prasad, A. B., Editors. “Martindale, The Extra Pharmacopoeia,” Twenty-eighth Edition, Phar- maceutical Press, London, 1982, p. 416. Potesilova, H., Alcara. C . , and Santavy, F . , Collect. Czech. Chem. Commun., 1969, 34, 2128. Kaul, J. K., Moza, B. K., Santavy, F.. and Vrublovsky, P., Collect. Czech. Chem. Commun., 1964, 29, 1689. “Egyptian Pharmacopoeia 1953 .” Fouad I University Press, Cairo, 1953, p. 698. “The Pharmacopeia of the United States of America, 1970,” Board of Trustees, Bethesda, MD, 1970, p.142. “British Pharmacopoeia 1973,” HM Stationery Office. Lon- don, 1973, p. 121. “Pharmacopek Francaise,” Volume 1, Ordre National des Pharmaciens, Paris, 1974, p. 181. “British Pharmacopoeia 1968,” Pharmaceutical Press, London, 1968, p. 238. “The Pharmaceutical Codex,” Eleventh Edition, Pharma- ceutical Press, London, 1979, p. 218. Wood, D. R., Pharm. J . , 1957, 178, 188. Pesez, M., Ann. Pharm. Fr.. 1957, 15, 630. Schmit, J. M., Bull. Trav. Soc. Pharm. Lyon, 1968, 12, 31. Dusinsky, G., Machovicova, F., and Tyllova, M., Fam. Obz., 1967, 397. Karawya, M. S., and Diab, A. M., J . Assoc. Off. Anal. Chem., 1975,58, 1171. Arai, T . , and Okuyama, T., Anal. Biochem., 1975, 69, 443. Croteau, R . , and Leblanc, R. M., Photochem. Photobiol., 1978, 28, 33. Arai, T., and Okuyama, T., Seikagaku, 1973, 45, 19. Bonati, A . , and Bacchini, M., Fitoterapia, 1966, 37, 24. Berurier, M. H., and Mathis, M. C., Ann. Pharm. Fr., 1973, 31, 457. Hussein, F. T., and Nasra, M. A . , Plant. Med., 1974, 25, 396. Zweig, G., “Handbook of Chromatography.” CRC Press, Cleveland, OH, 1972. Mack, H., and Finn, E . J . , J . Am. Pharm. Assoc., 1950, 39, 532. Pearce, E . M., J. Chromatogr., 1959, 2, 108. Smolenski, S. J., Grane, F. A , , and Voight, R. F., J . Am. Pharm. Assoc., 1958,47, 359. Gardner, J. E., and Dean, S. J . , Drug Stand., 1960, 28, 50. Santavy, F., Pharm. Acta Helv., 1948, 23, 380.578 ANALYST, MAY 1989, VOL. 114 29. 30. 31. Woodson, A. L., and Smith, D . E., Anal. Chem., 1970, 42, 242. Forni, G., and Massarani, G . , 1. Chromafop., 1977, 131, 444. El-Gindy, A . R., Sabir, M., Nazimuddin, S. K., Zahoor, M., and Shehab, J . , ”Islamic Medicine Conference,” Volume IV, Karachi, Pakistan, 1986. 32. Sarg, T. M., El-Domiaty, M.. SAama. 0. M., Bishr, M., and El-Gindy, A. R., “Conference of Pharmaceutical Sciences XX,” Cairo, Egypt, 1988, p. 62. Paper 8100987B Received Mwch Ilth, 1988 Accepted August 23rd, 1988

ISSN:0003-2654    

DOI:10.1039/AN9891400575

出版商:RSC    

年代:1989

数据来源: RSC