ISSN:0003-2654    

DOI:10.1039/AN98207FX009

出版商:RSC    

年代:1982

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN98207BX011

出版商:RSC    

年代:1982

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN98207FP021

出版商:RSC    

年代:1982

数据来源: RSC

ISSN:0003-2654    

DOI:10.1039/AN98207BP027

出版商:RSC    

年代:1982

数据来源: RSC

摘要:

March 1982 The Analyst Vol. 107 No. 1272 High-resolution Carbon-13 Solid-state Nuclear Magnetic Resonance Spectroscopy* G. R. Hays KoniiaklijkelShell-Laboratorium, Shell Research B. V., P.O. Box 3003, 1003AA Amsterdam, The Netherlands The status of high-resolution carbon-13 solid-state nuclear magnetic resonance (NMR) spectroscopy with respect to quantitative organic analysis is discussed. As the techniques for obtaining NMR spectra from solids are somewhat different from those used in liquid-state NMR, the experimental concepts are outlined first. Spectra that illustrate a few of the topics that have been studied in the author’s laboratory are then presented. Finally, the problem of quantitativeness (or indeed non-quantitativeness) of such carbon- 13 solid- state NMR spectra is examined.Keywords : Solid-state carbon- 13 nuclear magnetic resonance spectroscopy Techniques Carbon-13 nuclear magnetic resonance (NMR) spectroscopy of liquids or solutions has long been accepted as an important and quantitative analytical tool. However, a question that arises when one considers the application of carbon-13 NMR spectroscopy to solid samples is why it is not possible to obtain carbon-13 high-resolution spectra of solids using conventional liquid-state techniques, or, to pose this question in another way, why might it be necessary to buy another expensive NMR spectrometer if one wants to examine solids? If one were to run a normal Fourier-transform (FT) NMR spectrum of a solid on a liquid- state spectrometer, all one would see would be a very broad and featureless line.Indeed, the carbon-13 NMR spectrum of a solid could well cover 20 kHz with little or no fine structure. As structure due to chemical shifts may well be of the order of a few hertz, this will obviously be lost. The difference in line widths between solids and liquids is due mainly to the fact that in liquids there is a rapid rotational tumbling and translational diffusion of the nuclear spins, whereas in solids there is, in general, a lack of motion on such a grand scale. Therefore, where- as dipolar coupling between the spins is averaged to zero in a liquid, it is not in a solid, and it is this dipolar coupling that causes the lines to be broad. To put this in another way, each spin “sees” not only the external magnetic field but also local magnetic fields due to neighbouring spins.If this dipolar broadening could be suppressed in some way, it would be possible a t least to begin to see an increase in the information available. In the early 1970s a combination of techniques was developed that made it possible to obtain high-resolution spectra from solids. The underlying idea was that the organic sample contained two spin species, a rare species (by virtue of high chemical or isotopic dilution), which one wished to observe (carbon-13), in the environment of an abundant species (the proton). For such a system, a combination of three techniques, dipolar decoupling, magic- angle spinning1 and cross-polarisation, can often yield liquid-like spectra in a reasonable time. Each of these techniques deals with one of three problems.We will first discuss the one that has already been mentioned, dipolar broadening. As the carbon-13 spins are rare, almost all of the broadening arises from interaction with neighbouring protons. Scalar decoupling in liquid-state NMR is a familiar technique in which the scalar interactions between the protons and carbon atoms are decoupled (by irradiating at the proton resonance frequency) in order to obtain a single peak in the carbon-13 spectrum for each magnetically non-equivalent carbon nucleus. This simplifies the spectrum and aids * Keynote Lecture presented at the Joint NL - UK Symposium on Quantitative Organic Analysis, Noordwijkerhout, The Netherlands, April 22-24, 1981. Five other papers from this Symposium appeared in the October, 1981, issue of The Analyst.241242 HAYS HIGH-RESOLUTION CARBON-13 SOLID-STATE Analyst, VOl. I07 quantitative organic analysis. In solids, dipolar broadening of the carbon atom signals by the protons can be eliminated in basically the same way, Le., by strong decoupling of the abundant proton spins. However, the power needed is far greater for solids than for liquids (perhaps 100 W compared with 10 W or less) because the strength of the interaction is far greater. However, even with dipolar decoupling only a few, overlapping broad lines are usually observed. This is due to a second broadening effect, chemical shift anisotropy. The chemical shift is actually a tensor q ~ a n t i t y . ~ In liquids one usually sees only one value, the isotropic average value, due to motional tumbling.But as it is a tensor, there are in principle three chemical shift components, which in most instances will not be identical. So one obtains chemical shift anisotropic powder patterns for each carbon atom in the spectrum. These general patterns are shown in Fig. 1. In order to visualise how these powder patterns may arise it is perhaps useful to consider a carbonyl bond or an aromatic ring. It is then easy to imagine the carbon atoms to have different chemical shifts dependent upon whether the bond or ring is rigidly aligned parallel to the external field or perpendicular to it. These are two limiting conditions. In a powdered sample one can expect all possible orientations and the corresponding powder patterns are obtained.Thus, the solid-state spectrum of frozen benzene would be very similar to the pattern for an axially symmetric carbon atom, as shown in Fig. 1. This is certainly not a new technique. It was observed4 as long ago as 1958 that rapid rotation of a solid caused narrowing of the lines in its NMR spectrum if the speed of rotation was fast enough. Moreover, it was found that the lines were made narrower by an amount orcos(3cos28-l)-1, where 6’ is the angle between the applied external magnetic field and the axis of sample rotation. If cos2B = 1/3 then this factor vanishes and hence the spectral lines should be narrowed completely with no chemical shift anisotropic broadening. Not surpris- ingly, the angle 8 (54” 44’) is called the “magic angle.” In order to obtain the greatest effect, one has to spin the sample at frequencies equal to or greater than the broadening one is trying to average out.Thus, to remove the full dipolar interaction, speeds of 15 kHz or higher might be required, which are obviously very difficult to obtain. But if magic-angle spinning is performed concurrently with dipolar decoupling, spinning need only eliminate chemical shift anisotropic broadening and speeds of between 3 and 9 kHz would be sufficient. It has been shown that even if spinning cannot be performed fast enough, the lines are still considerably narrower, the only problem being spinning side-bands, which occur at multiples of the spinning speed.5 The experimental arrangement for magic-angle spinning used at present on our Bruker CXP-300 spectrometer (7.05 T magnetic field) is shown in Fig. 2.It should be noted immedi- ately that for spinning speeds of 6 kHz standard glass NMR tubes are not suitable. Instead, mushroom-shaped rotors are These can be either completely solid and made of the sample under investigation (a polymer, for example) or consist of hollow barrels with screw caps, for use with powders. The barrel diameter is 8 mm, the rotor is 14 mm across the “mushroom” and 18 mm from tip to top. These rotors spin in a stator on an air bearing, with dry nitrogen serving as the driving gas. Our rotors were originally fashioned from delrin (polymethylene oxide) [ (-CH,O,-) ] but this has the disadvantage of giving a signal in the middle of the spectrum (and spinning side- bands a t lower spinning speeds).To circumvent this problem, more than a year ago we switched to boron nitride for our spinner material and coated the mushroom end of the boron It is possible to remove chemical shift anisotropy by magic-angle sample spinning. 0 0L (JII 0 1 1 0 2 2 033 isotropic case Axially symmetric case Anisotropic case Fig. 1. Chemical shift anisotropy powder line shapes.March, 1982 NUCLEAR MAGNETIC RESONANCE SPECTROSCOPY 243 nitride rotor with a methyacrylate glue to prevent wear. These spinners give absolutely no background signal. Finally, to get the spinners in and out of the coil, the stator is moved back. Even with dipolar decoupling and magic-angle spinning there is still a limitation to an FT NMR experiment on a solid in that a recycle delay of five times the spin - lattice relaxation time ( T I ) is needed between acquisitions to provide a strong signal by a time-averaging process.This situation is the same as in liquid-state NMR. After sampling the signal it is necessary to wait a time long enough to allow the magnetisation to return to its equilibrium Boltzmann distribution. It is clear that this is important for quantitative liquid-state work. As T , for carbon-13 nuclei in solids can be tens of seconds or even minutes, the whole process could soon become tedious and inefficient. This is avoided by cross-polarisation,2 which allows faster recycling and also enhances the carbon-13 signal. A series of radiofrequency (r.f.) pulses are applied at both the carbon and the proton frequencies, resulting in a transfer of magnetisation from the abundant protons to the rare carbon atoms. This means not only that the signal is enhanced immediately, but also that recycling at five times the T , for protons is possible.As T , for protons can be of the order of seconds, the whole process becomes much quicker. It is only necessary to wait for the proton magnetisation, rather than the carbon magnetisation, to return to its equilibrium population as polarisation is being transferred from the protons to the carbon atoms and the carbon atoms do not have to relax of their own accord. The pulse sequence is discussed in more detail later. The modified magic-angle adjustment was built in our laboratory. Applications It has been stated that liquid-like carbon-13 NMR spectra can now be obtained from solids, a situation similar to that for carbon-13 liquid-state NMR ten years ago, but the question arises as to whether the method can be used to give quantitative information.In order to give an idea of the appearance of solid-state NMR spectra and to illustrate differences from liquid-state spectra, a few representative examples of the application of solid- state NMR will be presented for organic crystalline compounds, chemisorbed surface species and polymers. This should then lead naturally to discussion of the possibility of obtaining quantitative data. Adamantane is one of the most useful solid-state NMR test compounds. The molecule is basically spherical, with one carbon atom at each corner (Fig. 3), and can be examined easily as there is so much motion occurring in the solid even at room temperature.There are two types of carbon atom, CH, and CH, in the ratio 6 : 4 and there is no difficulty in obtaining a quantitative spectrum. For example, it can give information concerning the conformation of molecules in the solid state. Solid-state NMR can sometimes provide more information than liquid-state NMR. 1 / Axis rotation Sample K a g i c angle Flutes chamber Fig. 3. Adamantane structure and spectrum. Stator Driving gas (nitrogen) Fig. 2. Magic angle spinning design.244 HAYS : HIGH-RESOLUTION CARBON-13 SOLID-STATE Analyst, VOZ. IOY The molecular formula of the Hantzsch ester (I) is apparently symmetrical but X-ray analysis8 shows that, in fact, the carbonyl bonds point in different directions (11).The solution spectrum CH3 \ FH2 H3 I i5L00c2H5 H H&OOC CH3 I H I II [Fig. 4(b)] shows that the ring olefinic carbon pairs (a and b, and c and d) are equivalent owing to rotation about the ring-carbonyl bonds. However, the conformation is frozen in the solid state, and Fig. 4(a) shows that the ring olefinic carbon pairs are non-equivalent, becoming doublets in the solid-state spectrum. This spectrum also illustrates the different line widths that occur frequently within the same solid-state spectrum. For example, resonances are generally sharp for non-protonated carbon atoms and broad for carbon atoms next to a nitrogen atom (at a magnetic field of 7.05 T). The spectrum is not obviously quantitative, however. (It should be noted that spectra taker, with a delrin rotor can be recognised by the presence of the corresponding delrin peak.) The spectrum of 2,4- dinitrotoluene (Fig.5 ) displays a doublet for the single methyl group resonance and a similar Solid-state NMR can also yield crystallographic information. Delrin sample rotor CH3(et hyl) CH2(et hy I) CH,(ri ng ) I I I I I I 200 100 0 200 100 0 Chemical shift, p.p.m. Fig. 4. Hantzsch ester: (a) solid-state 13C NMR spectrum; and (b) solution-state 13C NMR spectrum. 160 80 Chemical shift, p.p.m. 0 Fig. 5 . 2,4-Dinitrotoluene : structure and spectrum.March, 1982 NUCLEAR MAGNETIC RESONANCE SPECTROSCOPY 245 doublet for the ring carbon atom to which the methyl group is attached. This has been shown by X-ray analysis to be due to the presence of two non-congruent molecules in the unit ceL9 These occur if the sample is not spun fast enough.For truly quantitative spectra, it is preferable to have all of the intensity concentrated in single peaks but when spinning side-bands are present it is necessary to sum the intensities for all of the side-bands and this becomes impossible if they overlap real peaks or other side-bands. It is therefore better, perhaps, to use a spectro- meter of lower magnetic field strength if one is really interested in quantitative spectra. Ours works at a frequency of 300 MHz for protons and, as the chemical shift anisotropy is propor- tional to the magnetic field, it is necessary to use a far faster spin-rate at a higher field. How- ever, this has to be weighed against the disadvantage of the sensitivity becoming lower at a lower field. Chemisorbed surface species of this type are widely used as stationary phases in chromatographic columns.These were well characterised samples and were used as a basis for the study of surface species in general and catalysis in particular. This is an area in which we have already done a lot of work and in which we see a large market for solid-state NMR. All the samples are of the general structure I11 and Fig. 6 shows the first of the series where One can also begin to see problems arising from spinning side-bands in this spectrum. We have also looked at a large series of chemically modified silica gels. R = CH,, referred to as the C, compound. This spectrum is of truly “high-resolution,” displaying narrow line widths (about 0.5 p.p.m.).All seven carbon resonances can be seen and the intensities appear to be roughly quantitative. Fig. 8 is the spectrum of the C, species. Here the middle two carbon atoms in the chain have coincident chemical shifts. The intensities are again roughly quantitative although perhaps this becomes less so towards the end of the chain. This is probably due to the fact that as the end of the chain becomes more mobile the static dipolar interactions that allow cross-polarisa- tion start to average out; cross-polarisation then takes longer so these carbon atoms need a longer “contact time” (the time that the protons and carbons atoms are in “contact” with each other, i.e., the time during which cross-polarisation occurs or magnetisation is transferred from the protons to the carbon atoms) to polarise fully. The spectrum of the C,-phenyl compound is shown in Fig.9 and appears again to be quantitative. I t is interesting to compare this with the spectrum of the phenyl species in Fig. 10. Again, spinning side-bands are present (and not in the spectrum of the C,-phenyl compound) because here the ring is held firmly to the surface The C, derivative is shown in Fig. 7. I 1 180 80 Chemical shift, p.p.m. Fig. 6. Spectrum of C, surface species. - 20246 0 - SI-CH~-CH~-CH*-CH~-CH~-CH~-CH~-CH~ hH,2 3 4 5 6 7 8 9 I Y HAYS : HIGH-RESOLUTION CARBON-13 SOLID-STATE Analyst, VOZ. 107 a y 3 0 -Si -CH2-CH2-CH2-CH2 - 2 3 4 5 ’ hH3 1 ? i -!- I I 180 80 3+6 1 I I - 20 Chemical shift, p.p.m. Spectrum of C, surface species. Fig.7. 180 80 Chemical shift, p.p.m, Fig. 8. Spectrum of C, surface species. - 20 129 124 5 /!8 6+7 2 3 4 I 1 1 . 2.. 1 180 80 - 20 Chemical shift, p.p.m. Fig. 9. Spectrum of C,-phenyl surface species. and hence the chemical shift anisotropy is larger, whereas motion at the end of the C, chain already averages a part of the chemical shift anisotropy. Fig. 11 shows the spectrum of vulcanised cis-l,4-polybutadiene, displaying similar (but not identical) spectra both in the solid state and in a swollen gel, using “conventional” liquid-state techniques. Three solid- state spectra of cis-l,4-polyisoprene, obtained under different experimental conditions, are presented in Fig. 12. It is obvious that upon variation of the cross-polarisation contact time Finally, in this section some spectra of rubbers are presented.March, 1982 NUCLEAR MAGNETIC RESONANCE SPECTROSCOPY 247 I 3+5 4 I ! 1 Aromatic I spinning , side 1 Aromatic spinning side bands 180 80 -20 Chemical shift, p.p.m.Fig. 10. Spectrum of phenyl surface species. (5, 10 and 60 ms) something strange is happening to the intensities. Lastly, Fig. 13 shows four spectra of a piece of rubber squash ball. Fig. 13(a) is the spectrum obtained on a static sample using “conventional” FT NMR techniques. Broadening of all resonances is due to chemical shift anisotropy. The addition of magic-angle spinning [Fig. 13(b)], although still using “conventional” methods, enables us to see that the squash ball is made of an isoprene- isobutene copolymer. Figs. 13( c) and 13(d) were obtained using both magic-angle spinning and cross-polarisation (contact times 1 and 10 ms, respectively).It can be seen that peaks and quantative information are now being lost and this is because cross-polarisation techniques produce signal enhancements and permit the use of repetition rates faster than those allowed by the carbon-13 spin - lattice relaxation times. Thus, distortion of relative intensities must always be considered a possibility and this will be discussed in the next section. ( a ) 131.5 29.5 I Fig. 11. Spectrum of cis-1,4-polybuta- diene (vulcanised): (a) solid state; and ( b ) solution state. i Fig. 12. Spectra of cis-1,Qpolyisoprene with variable contact times: (a) 5 ms; (b) 10 ms; and (c) 60 ms.HAYS : HIGH-RESOLUTION CARBON-13 SOLID-STATE AnaZySf, VOZ.I07 180 0 180 Chemical shift, p.p.m. 0 Fig. 13. Four spectra of a piece of squash-ball rubber: (u) static sample, resolution lost through broadening by CSA; (b) addition of MAS but with “conventional” methods; (c) both MAS and cross-polarisation (1-ms CP time) ; and (d) both MAS and cross-polarisation (10-ms CP time). Quantitative Application The application of the cross-polarisation technique, resulting in signal enhancements and allowing repetition rates faster than the carbon-13 T,s would allow, means that distortion of relative spectral intensities must always be considered a possibility and that quantitative spectra will not always be obtained. In the cross-polarisation pulse sequence (shown in Fig. 14), the contact time is important. When both the proton and carbon pulses are on, there is a thermal contact between the proton and carbon-13 spin reservoirs and magnetisation flows from the proton spin bath to the carbon spin reservoir.(The dotted line in Fig. 14 represents a small loss in proton magnetisation, as protons are so abundant, and a large gain in carbon atom magnetisation. The shifting about of magnetisation means that a large number of simultaneous relaxation processes must be considered if we are concerned with the production of quantitative spectra. A type of thermodynamic model, a system of reservoirs of water and interacting pipeslO (shown in Fig. 15), can be used to simplify the problem. The water represents magnetisa- tion, the reservoirs represent the carbon and proton spin baths and the pipes represent the relaxation processes.The diameter of a pipe controls the flow-rate of the water; thus wide pipes are the equivalent of short relaxation times and narrow pipes represent long relaxation times . The first point to note is that there is a fundamental difference between the protons and carbon atoms. This is related to the abundance of protons and the low concentration of “rare” carbon-13 nuclei. In most organic solids, the protons are physically close enough together to undergo fast spin exchange. When the first flips, the next one flips and so on throughout the sample. The effect is that the protons can often be thought of as one group, all with the same relaxation behaviour. The carbon atoms, on the other hand, are separated by large distances and spin exchange occurs at very slow rates (if at all).So the proton bath in Fig. 15 represents all of the protons and the carbon bath just one carbon type. (This difference has further implications regarding quantitativeness and is discussed later in more detail.) Returning to Fig. 15, the protons are connected to the “lattice” and the rate of magnetisa- tion decay (whilst “spin-locked” during the contact pulse) can be described by the relaxation (This is sometimes called “spin diffusion.”) Each individual carbon-13 type therefore exhibits its own relaxation behaviour.March, 2982 NUCLEAR MAGNETIC RESONANCE SPECTROSCOPY 249 * . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Irradi$i !I frequency ’ Contact I Observation Recycle; k time 7 time T time I I ! I I Lattice Fig.15. Thermodynamic model for the relaxation processes occurring during the cross-polarisation experiment. Fig. 14. Cross-polarisation sequence. time TI: (the spin-lattice relaxation time in the rotating frame). The pipe T,, represents cross-polarisation (T,, is the cross-polarisation relaxation time). The pipe is connected when the contact pulse is on and remains so for the duration of the contact pulse, then it is disconnected. These TCR pipes go to each separate carbon-13 type following the arguments given above and hence Fig. 15 shows all of the protons and only one type of carbon atom. Eventually the carbon-13 magnetisation will decay, as demonstrated by the final pipe. Using this type of analogy it is immediately possible to see two requirements for quantitative application : (i) T,, has to be open long enough to fill each carbon-13 reservoir; keeping it open for some (ii) There is competition between TCH and TI:; thus it is no good trying to fill up the carbon reservoirs if the magnetisation is disappearing from the proton bath faster than this can be done.Therefore, the conditions necessary if quantitative results are desired are that the contact pulse should be about five times the value of the longest T,, and that TI: should be much greater than the value of the longest TCH. Hence, it is necessary to consider the possible values of Tip" and TcH. Most of the work in the field has been carried out in the area of solid coals, mainly by Dr. D. L. VanderHart (National Bureau of Standards, Washington), and in this section of the paper his arg~rnentsll-~~ will be followed.Coal is, of course, a complicated substance and all of the problems associated with obtaining quantitative spectra, such as conflicting proton TI: values, a whole range of T,, values and maybe even areas without protons at all, might be expected to be present. To return to the competing processes : during the contact time, the carbon atoms are trying to reach thermal equilibrium with a proton polarisation that is itself decreasing with a time constant, TI:. Hence, one of the key questions concerns the relative rates of these two processes. Now, if the fact that the protons undergo spin diffusion and can be treated to- gether whereas the carbon atoms behave individually is recalled, an immediate implication is that when the carbon and proton spin reservoirs are connected we expect, eventually, to achieve a carbon-13 spin polarisation proportional to the proton polarisation.Then the relative intensities will be a true representation of all of the carbon atoms in the sample. (To put this another way, the carbon atoms and protons have their own heat capacities and, before cross-polarisation, all the magnetisation is in the proton system, whereas there is none in the carbon system. During cross-polarisation, the two reservoirs are allowed to come to thermal equilibrium and hence a redistribution of magnetisation will occur, finally dependent upon the ratio of the heat capacities. Eventually, therefore, the carbon-13 spin polarisation will be proportional to the proton spin polarisation.) Fig.1611 illustrates the behaviour of both the proton and carbon-13 polarisations during the contact time. If conditions in which T,, is much smaller than T,,H prevail, it is clear that after a few T,, periods, there will be a carbon-13 polarisation proportional to the proton polarisation and if this situation holds, we will obtain a quantitative cross-polarisation spectrum. This approximation becomes less true as T,, becomes comparable to, or greater than, TI:. So it is clear from Fig. 16 that if one has a sample with a wide range of T,, to half-fill and others to fill fully is not satisfactory.250 HAYS : HIGH-RESOLUTION CARBON-13 SOLID-STATE Analyst, VOZ. I07 A C D Cross-polarisation time Fig.16. Competing rate processes in CP: 1, lH (TlpH); 2, 13C (TCH .< TIPH); 3, 13C (TCH m TloH); and 4, 13C ( T C H > TlpH). TIPH in coals -5-10 ms). (Taken from D. L. VanderHart and H. L. Retcofsky, reference 11.) values, then sampling at time A, B, C or D could selectively enhance or suppress the relative intensities of any one of the components, depending on the selected cross-polarisation time. If the spectra of coals are considered, the TI: values are often between 5 and 10 ms and this is probably because the relaxation in most coals is dominated by paramagnetic impurities. Therefore, a restriction is immediately placed on the TCH values in the sample if quantitative spectra are required, ie., T,, should be 1 ms or less. Experimentally we know that T,, values for protonated carbon atoms are usually of the order of 50 ,us and that non-protonated carbon atoms will cross-polarise in about 1 rns if there are other protonated carbon atoms around.However, we do not know how many carbon atoms in coal will be in similar environments. As the hydrogen to carbon atom ratio is often less than unity, we can safely assume that there will be a reasonable fraction of unprotonated carbon atoms. Assuming that the pipes and reservoirs arguments are followed regarding competition between TCR values and T,:, and knowing representative values of these parameters, we should be able to choose our experimental conditions concerning the length of the contact time (1-2 ms ?) and obtain quantitative spectra for coal. However, there are conditions in which quantitative spectra may still not be obtained and three of these situations have been suggested by VanderHart and co-worker~.~~-~~ Firstly, hydrogen to carbon atom ratios suggest that there must be areas of non-protonated carbon atoms.If the non-protonated carbon atoms lie at some distance from protons then for these, T,, will be greater than TIE and this would result in the unwanted effects upon the spectra of there being an attenuated signal for short contact times and too strong a signal for long contact times. Secondly, there could be large areas of fused aromatic rings. This would lead to extensive regions of non-protonated material. Here cross-polarisation would be infinitely slow (T,, = 00) and result in no signal contribution from these carbon atoms.Thirdly, if the spin-diffusion assumption is wrong and the protons cannot be treated as a single entity and there are areas of distinctly different proton spin diffusion then, T1f will not be uniform across the sample and the carbon-13 magnetisation will display a complicated behaviour . This brings us back to the rubber spectra with the different contact times (Fig. 1.3) Signal distortions are almost bound to occur in cross-polarisation spectra of multiphase sys- tems, such as rubbers or semi-crystalline polymers, especially when one of the phases has a high liquid-like mobility. One is that the cross-polarisation process is slowed down (as for the long alkyl chain surface species when compared with the shorter chains), necessitating longer and longer cross-polarisation times.The second is that it accentuates the problem associated with treating all of the protons as one group, i.e., there might be different areas of protons. To return to coals, how can the ideas regarding the original choice of experimental conditions to obtain quantitative coal spectra be tested? One way is to measure the apparent aro- maticity values (fa, the fraction of aromatic carbon atoms) as a function of contact time. Table I shows the experimental results for two coals, BS 400 (a bituminous coal) and BS 100 (a There are two reasons for this.March, 1982 NUCLEAR MAGNETIC RESONANCE SPECTROSCOPY TABLE I MEASUREMENT OF APPARENT AROMATICITY (fa) IN TWO COAL SAMPLES AS A FUNCTION OF CONTACT TIME f a 251 Contact timelms 0.2 0.4 0.8 1.0 2.0 3.0 4.0 BS 400 61.1 62.8 66.8 70.3 78.7 76.3 73.3 BS 106 81.0 82.0 82.0 88.3 90.2 90.4 84.0 Dry, ash-free analysis, %- C .. . . .. . . 86.6 93.3 H . . . . . . * . 5.5 3.3 H/C ratio . . .. .. 0.76 0.42 high-rank anthracite). obtained. longer TCH values but drops off as it becomes longer than T$ would allow. It can be seen that the predicted behaviour for a short TI: is The apparent aromatic content builds up as the contact time is long enough for the Other tests for quantitativeness that have been used are : (a) The comparison of a tedious “conventional” FT NMR spectrum using much longer waiting times (900 s) to allow for long carbon-13 T , values with a cross-polarisation spectrum of the same ~ o a l . l l - ~ ~ J ~ (b) The addition of a known amount of adamantane (aliphatic carbon) and observation of how the apparent aromaticity changes.16 This should change linearly with added aliphatic carbon if all of the aromatic carbon atoms of the coal are indeed being seen.( c ) Studies of model ~ o m p o u n d s . ~ ~ - ~ ~ J ~ ( d ) Comparison of frozen coal-derived liquids with the same components in the liquid st ate. 11-14 All of these tests seem to suggest that fairly quantitative solid-state spectra of coals can be obtained. Conclusions It is now possible to obtain high-resolution carbon-13 NMR spectra of solids fairly routinely. This was not the situation ten years ago. However, quantitativeness is not a subject that has been generally tackled yet. Work has been done with respect to coal and related areas (oil shales) and some work has been started on polymers.For quantitativeness, however, it is necessary to consider the various relaxation processes. Carbon-13 solid-state NMR is a t very much the same stage as carbon-13 liquid-state NMR was ten years ago. I t can be confidently expected that as solid-state NMR becomes easier and more routine, quantitative aspects will begin to be studied in earnest, especially as more solid- state spectrometers are installed in industrial environments. All the solid-state spectra presented in this paper were obtained in collaboration with Dr. A. D. H. Clague and R. Huis at the Koninklijke/Shell-Laboratorium, Amsterdam. The conformational work on the Hantzsch’ester was part of a project carried out in conjunc- tion with Dr.J. Verhoeven and F. Rob (University of Amsterdam) and Dr. B. Coleman (Koninklij ke/Shell-Laboratorium, Amsterdam). The spectrum of 2,4-dinitrotoluene was recorded at the request of Dr. K. J. Packer and Professor R. K. Harris (University of East Anglia) to complement work carried out at a lower magnetic field by S. F. Tanner, who provided the sample. The study of chemisorbed surface species was performed together with Dr. G. van der Velden (D.S.M., Geleen) and the samples were prepared by Ir. K. Pikaart in the group of Professor L. de Galan at the Technical University of Delft.252 1. 2. 3. 4. 6. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. HAYS References Schaefer, J., Stejskal, E. O., and Buchdahl, R., Macromolecules, 1977, 10, 384. Pines, A., Gibby, M. G., and Waugh, J. S., J . Chem. Phys., 1973, 59, 569. Andrew, E. R., Prog. Nucl. Magn. Reson. Spectrosc., 1971, 8, 1. Andrew, E. R., Bradbury, A., and Eades, R. G., Nature (London), 1958, 18, 1659. Maricq, M. M., and Waugh, J. S., J . Chem. Phys., 1979, 70, 3300. Andrew, E. R., Farnell, J. F., Firth, M., Gledhill, T. D., and Roberts, I., J . Magn. Reson., 1969, 1, Beams, J. W., Rev. Sci. Instrum., 1930, 1, 667. Lenstra, A. T. H., Petit, G. H., Domisse, R. A., and Alderweireldt, F. C., Bull. SOC. Clzim. Belg., Balimann, G. E., Groombridge, C . J., Harris, R. K., Packer, K. J., Say, B. J., and Tanner, S. F., Garroway, A. N., Moniz, W. B., and Resing, H. A., Faraday Symp. Chem. Soc., 1978, 63. VanderHart, D. L., and Retcofsky, H. L., Preprints of the 1976 Coal Chemistry Workshop, August VanderHart, D. L., and Retcofsky, H. L., Fuel, 1976, 55, 202. Earl, W. L., and VanderHart, D. L., Macromolecules, 1979, 12, 762. VanderHart, D. L., “13C NMR in Solid Polymers,” A Review Lecture Including Quantitative Aspects, NATO AS1 on “High Resolution NMR in Solids,” Villasimius (Cagliari) , Sardinia, September 1980, to be published. Maciel, G. E., Bartuska, V. J., and Miknis, F. P., Fuel, 1979, 58, 391. Wemmer, D. E., Pines, A., and Whitehurst, D. D., Philos. Trans. R. SOC. London, Ser. A , 1981, 300, 27. 1979, 88, 133. Philos. Trans. R . SOC. London, Ser. A , 1981, 299, 643. 26-27, 1976, Stanford Research Institute, p. 202. 15.

ISSN:0003-2654    

DOI:10.1039/AN9820700241

出版商:RSC    

年代:1982

数据来源: RSC

摘要:

Analyst, March, 1982, Vol. 107, PP. 253-259 253 Determination of Lead and Antimony in Urine by Atomic-absorption Spectroscopy with Electrothermal Atomisation B. M. Smith and M. B. Griffiths Morganite Electrical Carbon Ltd., Clase Road, Morriston, Swansea, SA 6 8PP Two methods have been developed for determining antimony and lead in urine. In the first method the urine is wet oxidised and the metals are extracted as chelates into an organic phase. This is then analysed by atomic- absorption spectroscopy after carbon-rod atomisation. Nitric acid must be eliminated from the wet-oxidation mixture because i t causes incomplete recovery of the antimony. In the second method the lead and antimony are extracted directly from the urine and are again determined by atomic-absorption spectroscopy with carbon-rod atomisation.This is the preferred method for the routine analysis of normal urine samples. The wet-oxidation method is used for samples that contain abnormal amounts of coproporphyrins, chelating agents or other organic materials. Both methods gave comparable results in the range 0-200 pg 1-1 of lead and antimony. Keywords Antimon-y determination ; lead determination ; urine analysis ; atomic-absorption spectroscopy ; electrothermal atomisation Workers engaged in the processing of lead and antimony or their alloys at high temperatures can be exposed to significant amounts of metal vapour; this may result in increased urinary lead and antimony levels. It is important, for the purpose of biological monitoring, to have a rapid and accurate method for determining these elements in urine, The “Control of Lead at Work Regulations 1980”l state that a routine measurement of urinary lead is the appropriate biological control for a worker exposed to lead alkyls and the Act lays down a limit of 15Opg1-1 of lead, above which workers should be suspended from working with lead.No such limit exists for antimony; nevertheless it is useful to have a means of measuring urinary antimony that can be equated with an individual’s exposure and that will supplement antimony in air monitoring and clinical assessments. This paper describes the development of a simple and rapid method for determining lead and antimony in urine, which is used for the routine screening of workers engaged in the processing of these metals.There are several well established atomic-absorption procedures for the determination of lead in ~ r i n e ~ - ~ but none appear to have been published for deter- mining antimony. Antimony is usually determined, after wet oxidation of the urine, by the formation of a coloured complex between antimony(V) and a xanthene or triphenyl- methane dye je.g., Rhodamine B, Brilliant Green and Crystal Violet). Here, the urine is digested with sulphuric, nitric and perchloric acids to destroy any organic matter and to convert all of the antimony into its pentavalent state. The antimony(V) is then reacted with Rhodamine B to produce a red- violet complex that is extracted with benzene. An alternative pro- cedure, which reduces interference from iron, extracts the pentavalent antimony directly into diisopropyl ether from 1.5 N hydrochloric acid.The antimony - Rhodamine B complex is then developed in the diisopropyl ether. The importance of adding the reagents in the correct order, to eliminate the hydrolysis of antimony(V), is highlighted in a paper by Van Aman et a1.6 We found that the Maren method5 gave very erratic results in the range 0-150 pg 1-1 of antimony in urine. Generally, results were lower than would be expected and in some analyses the antimony appeared to have been lost in the digestion stage. The danger of losing the antimony as its volatile chloride is well known and great care needs to be taken with the evaporation of the nitric - sulphuric - perchloric acid mixtures from samples of urine that contain chlorides.Maren also identified another source of low results as being the formation of the comparatively unreactive “antimony(1V)” in the digestion stage. The We have tried the method that was developed by Marem5254 SMITH AND GRIFFITHS : DETERMINATION OF PB AND SB IN Analyst, Vol. 107 addition of perchloric acid is supposed to prevent this by quantitatively oxidising all of the antimony to the pentavalent state. However, the perchloric acid itself may be reduced by residual organic matter at high temperatures and produce enough hydrogen chloride to volatilise the antimony as the trichloride. The Analytical Methods Committee, who investi- gated methods for determining small amounts of antimony in organic matter,’ suggested that some of their low results might be attributed to this effect.We found that if we removed the perchloric acid from the digestion mixture, reduced any “antimony( IV)” by boiling with sulphite and subsequently oxidised all the antimony to antimony(V) with cerium(1V) ions then the results were more reproducible, although this method was still not completely reliable and occasionally gave very low results. Direct analysis of urine for lead and antimony is not possible by atomic-absorption spectro- scopy with electrothermal atomisation because the high inorganic solids content of the urine interferes. There is a marked suppression in the lead and antimony peak signals with the atomisation of successive samples. Willis4 examined the possibility of extracting lead and other heavy metals from urine with a complexing agent in an organic solvent. This was followed by vaporisation of the organic phase in the flame of an atomic-absorption spectro- meter.Kamada and Yamamoto* developed a method for the selective determination of antimony(II1) and antimony(V) in water by solvent extraction followed by atomic-absorption spectroscopy with electrothermal atomisation. We have developed a method for determining lead and antimony in urine by a solvent extraction procedure followed by electrothermal atomisation. This can be applied directly to urine without the need for prior wet oxidation. For purposes of comparison a method has been developed for determining antimony and lead in urine samples after wet oxidation. Some interesting observations have been made on the adverse effect of nitric acid on the determination of antimony. These may also explain many of the difficulties experienced with Maren’s method.Experimental Apparatus The atomic-absorption spectrometer is a Varian AA175 incorporating a hydrogen continuum lamp for background correction. The samples are atomised in a Varian CR90 carbon-rod atomiser, fitted with pyrolytically coated carbon cups. These are protected from oxidation by sheathing in a gas mixture of nitrogen and hydrogen. The injections into the atomiser are made with an Oxford micropipette, which had disposable plastic tips. A Perkin-Elmer, Model 56, flat-bed recorder is used to trace the absorbance readings. The samples are extracted in 50-ml glass centrifuge tubes that are sealed with leakproof caps. The centrifuge is capable of a speed of 3500 rev min-l.The tubes are shaken with a mechanical shaker. The wet oxidations are carried out on a hot-plate in 250-mI tall-form borosilicate glass beakers with cover glasses. These are cleaned by fuming with sulphuric acid followed by washing with de-mineralised water. All other glassware and plastics are cleaned in dilute nitric acid and de-mineralised water. Reagents All solutions are prepared from analytical-reagent grade chemicals and de-mineralised water and stored in polythene bottles. Concentrated nitric acid, sp. gr. 1.42. Concentrated sulphuric acid, sp. gr. 1.84. Concentrated hydrochloric acid, sp. gr. 1.18. Concentrated ammonia solution, sp. gr. 0.91. Hydrogen peroxide, 30% mlm. 4-Methylpentan-2-one (isobutyl methyl ketone, IBMK) .Anhydrous sodium sulphite. Ammonium tetramethylenedithiocarbamatP (antmonium pyrrolidinedithiocarbamate APDC). This is used as a 4% m/V solution in de-mineralised water and should be prepared fresh daily. After preparation the solution is extracted with three successive 10-ml portions of IBMK, which are then discarded.March, 1982 URINE BY AAS WITH ELECTROTHERMAL ATOMISATION 255 Standards Dissolve 1.334 g of antimony potassium tartrate in water and make up to 11. This should be prepared fresh daily. Dilute 10 ml of the 500 pg ml-I antimony(II1) standard solution to 1 1 with water. Dissolve 0.500 g of pure antimony powder in 10 ml of concentrated nitric acid. Heat to boiling and evaporate off the excess of nitric acid. Cool and then dissolve the hydrated antimony(V) oxide in 50 ml of cold, concentrated hydrochloric acid.This should be prepared fresh daily and should be checked for cloudiness, which is evidence that the antimony(V) chloride has hydrolysed. Dilute 10 ml of the 500 pg ml-I antimony (V) solution to 1 1 with 5% V/V hydrochloric acid. Make up to 1 1 with water. Dilute 10 ml of this solution to 1 1 with water and adjust the pH to 2.5 with nitric acid. Antimony(II1) standard solution, 500 pg ml-1. Dilute antimony(1II) standard solution, 5.0 pg ml-1. Antimony( V ) standard solution, 500 pg ml-l. The solution is stable for up to 3 months. Make up to 1 1 with water. Dilute antimony(V) standard solution, 5.0 pg ml-l. Lead standard solution, 5.0 pg ml-l. Dissolve 0.500 g of lead in 1 + 1 nitric acid.Procedure Collection and preliminary treatment of urine To obtain a true indication of the amount of lead and antimony being excreted it is preferable to collect a 24-h urine specimen. In practice, this is done by issuing each worker with 2 x 11, acid washed, polypropylene bottles. Sampling commences at the start of the morning shift. The first urine sample is discarded and every subsequent sample is collected for the next 24 h. At the end of this time, the contents of the two bottles are bulked together and acetic acid is added to the urine to give a final acid concentration of 1-3% V/V. Unless this is carried out there is a danger that the urine will slowly decompose to precipitate calcium phosphate, which may carry down some of the heavy metals. The stabilised specimen is then stored in a refrigerator.Wet digestion of urine samples Initially the samples were digested with mixtures of sulphuric - nitric and sulphuric - nitric - perchloric acids but these digestion mixtures frequently gave low and suspect results for antimony. Our investigations have shown that nitric acid in the oxidising mixture causes serious interference in the determination of antimony. Subsequent digestions were made with sulphuric acid and hydrogen peroxide and this mixture gave significantly higher and more reproducible results. Residual nitrite ions in the solution after wet oxidation are known to interfere in the deter- mination of antimony by the hydride generation method.7 In the hydride method the interference by nitrate can be removed by the addition of iodide but nitrite cannot be tolerated at any concentration in the final solution.I t is not possible to use iodide to eliminate interference from nitrate when determining antimony by solvent extraction followed by atomic-absorption spectroscopy with electro- thermal atomisation. This is because the iodide itself interferes, probably through the formation of volatile antimony(II1) iodide. That is why sulphuric acid and hydrogen peroxide have been used for all our wet oxidations of urine. Recommended wet-digestion procedure Add 5 ml of concentrated sulphuric acid and 2 ml of hydrogen peroxide (30% m/m). Cover the beaker and place it on a hot- plate, heated at about 150 "C. The sample will gradually char. Oxidise the sample by the drop-wise addition of hydrogen peroxide and gradually increase the hot-plate temperature to about 320 "C.Cool, wash down any trace amounts of charred sample from the walls of the beaker with hydrogen peroxide. Return the beaker to the hot-plate, continue the oxidation until the solution is completely Pipette 20 ml of urine into a cleaned 250-ml beaker. Continue to add hydrogen peroxide until the solution is clear.256 clear and heat to fumes of sulphuric acid. Dilute the solution with 10 ml of water and transfer into a 50-ml glass centrifuge tube. SMITH AND GRIFFITHS : DETERMINATION OF PB AND SB IN Analyst, VoZ. 107 Cool the beaker, remove and wash the cover glass. Extraction of lead and antimony after wet oxidatiort and make up to about 20 ml with water. 5 ml of IBMK.Centrifuge until the emulsion clears and the two phases separate. pipette transfer 5 pl of the upper (ketone) phase into the carbon-cup atomiser. atomise the sample under the conditions given for lead and antimony in Table I. Adjust the pH of the oxidised solution in the 50-ml centrifuge tube to between 4 and 5 Add 2 ml of freshly prepared APDC solution and Stopper the tubes and shake vigorously for 1 min in the mechanical shaker. With an Oxford micro- Dry, ash and Direct extraction of lead and antimony from wine add 2ml of freshly prepared APDC solution and 5ml of IBMK. tube. cup atomiser. Add 5ml of concentrated hydrochloric acid to the extracted urine. acid strength of greater than 2 N. of the ketone layer to the carbon-cup atomiser. given for antimony in Table I.Pipette 20ml of urine into a 50-ml centrifuge tube. Adjust the pH to between 4 and 5, Stopper and shake the Transfer 5 pl of the upper layer to the carbon- Dry, ash and atomise the sample under the conditions for lead in Table I. This will give a final Transfer 5 pl Dry, ash and atomise under the conditions Centrifuge until the emulsion clears. Stopper, shake and centrifuge as before. Preparation of standards Standards are prepared by a modified standard-addition method. A sample of urine with a low antimony content is analysed by the multiple-addition method. This sample and the solutions prepared by adding measured amounts of antimony are used as secondary standards. To 20-ml portions of urine collected from someone who is not regularly exposed to antimony, add 0.1, 0.2, 0.3 and 0.4ml of the 5 pgml-l antimony(II1) solution and the 5 pgml-l lead solution.Plot the absorb- ance of each standard against the concentrations of added lead and antimony (in the range 0-100 pg 1-1). Extrapolate to zero absorbance to give the concentration in the sample plus any blank. Subtract the blank reading from all the standards and re-draw the graph so that it passes through the origin. This method provides standards that closely match the physical and chemical properties of the samples. Treat these standards in exactly the same way as the samples. Instrument operating conditions The operating conditions are given in Table I. The atomic-absorption spectrometer is set up according to the manufacturer's instructions. TABLE I INSTRUMENTAL PARAMETERS FOR DETERMINING LEAD AND ANTIMONY Time/s Temperaturelac Stage -Antimon; GAntimon; Drying ... . 20 20 110 110 Ashing. . .. . . 20 20 700 700 Atomisation . . . . 2 2 2 000 2 200 Parameter Ramp rate/OC s-l . . Atomiser . . . . Wavelength/nm . . Slit . . .. . . Lamp current/mA . . Hydrogen lamp for correction . . I . Sample volume/pl . . Sheath gas . . .. Lead . . . . 800 . . . . Carbon cup . . . . 283.3 . . . . Carbon furnace mode .. ,. 5 .. .. Yes . . . . 5 . . . . Hydrogen - nitrogen background Antimony 800 Carbon cup 217.6 Carbon furnace mode 6 Yes 5 Hydrogen - nitrogenMarch, 1982 URINE BY AAS WITH ELECTROTHERMAL ATOMISATION 257 The carbon-rod atomiser settings are given as a guide and should be optimised before each run of samples. This is done by switching the spectrometer to the absorption mode, con- necting the output to a flat-bed recorder and then repeatedly running the highest standard until optimum conditions are obtained.This is when the drying stage removes the solvent without spluttering. The ashing should be carried out at the maximum temperature possible without giving an absorption signal for the metal being determined. The atomisation temperature and ramp rate are set to give the best sensitivity and reproducibility possible. The most sensitive absorption line for lead, at 217.0 nm, lies close to the 217.6 nm antimony line. With the slit set in the carbon furnace mode, which on this spectrometer corresponds to between 0.2 and 0.5 nm, no interference was found in the antimony signal with up to a 2000-fold excess of lead.Results and Discussion Extraction of Lead, Antimony( 111), Antimony(V) and Excreted Antimony from Urine The results of our investigation confirm the findings of Willis4 that the lead- APDC complex is completely extracted at a pH of greater than 4, at lower pH the recovery is incomplete. This has been shown by Yeager et aL9 to be due to iron preventing the APDC chelation of lead. The extraction behaviour of antimony(III), antimony(V) and excreted antimony in urine is shown in Fig. 1. To two separate samples of urine, collected from someone not exposed to antimony, were added 1OOpg1-1 of antimony(II1) and lOOp~gl-~ of antimony(V). The extraction efficiency at different acidities was investigated for these samples and for a urine sample, which, after wet oxidation, was found to contain 44pg1-1 of antimony.Both antimony(II1) and antimony(V) are completely extracted at a pH of below 5 but the extrac- tion of excreted antimony is not complete until an acidity of 2 N is reached. We also found that if the urine was first made strongly acidic (3 N) and then buffered to pH 4 there is still low recovery of the antimony. This might be explained if the antimony were to form a strong complex with some species in the urine. Even where the antimony was added as a simple ion, the extraction efficiency of its APDC complex at various pH values was different for water and for urine. That is why it is important to make up standards in urine. 0.25 r I Fig. 1 . Effect of acidity on the extraction of 100 pg 1-1 of antimony(II1) and antimony(\;) from urine.Also included is the extraction curve for an actual urine sample containing 44 pg 1-l. A, Antimony(II1) solution; B, antimony(V) solution; and C, actual urine sample. Investigation of the Reproducibility of the Wet-oxidation and Direct-extraction Methods for the Determination of Antimony This bulk sample was analysed successively over a period of several weeks by the wet-oxidation method (ten times) and the direct-extraction method (five times). The results are given in Table 11. Several samples of urine from persons exposed to antimony were bulked together.258 SMITH AND GRIFFITHS : DETERMINATION OF PB AND SB IN Analyst, VoZ. I07 TABLE I1 REPRODUCIBILITY OF WET-OXIDATION AND DIRECT-EXTRACTION METHODS FOR DETERMINATION OF ANTIMONY I N A BULK SAMPLE OF URINE Antimony concentration/pg 1-l A f 7 Wet-oxidation method (%SO, - H203 Direct-extraction method 47 45 44 43 44 44 40 44 46 43 46 42 47 40 46 45 Mean ..44.2 Standard deviation . . 2.6 43.8 0.74 The narrower spread of results from the direct-extraction method probably reflects the greater likelihood of errors being introduced by the extra wet-oxidation stage (contamination from glassware, loss of sample, etc.). The fact that the measurements were made over a period of several weeks shows that the repeatability of the method is good and that the urine samples are stable. A further 16 urine samples were analysed for antimony by the wet-oxidation and direct-extraction methods. The results are given in Table 111. TABLE I11 DETERMINATION OF ANTIMONY BY THE WET-OXIDATION AND DIRECT-EXTRACTION METHODS Sample A .... B .. .. c .. .. D .. .. E .. .. F .. .. G .. .. H . . .. I .. .. K .. L .. .. M . . .. N .. .. 0 .. .. P .. .. J .. Wet oxidation .. 11 .. 60 .. 24 ,. 61 .. 75 .. 22 .. 43 .. 224 .. 40 .. 16 .. 101 .. 52 .. 96 .. 27 .. 58 .. 49 Direct extraction from 2 N HC1 10 67 23 58 73 20 45 220 39 14 105 53 104 34 56 52 The results given in Tables I1 and 111 show that the two methods give comparable results. Taking the results in Table I11 the correlation coefficient of the two methods was 0.9970. To see how well the methods work for low levels of antimony, 18 samples of urine were collected from persons who had no previous exposure to antimony. All of the samples were analysed by the direct-extraction method and the five samples that appeared to contain the least amounts of antimony were analysed by the wet-oxidation method.The results are given in Table IV. These results show that levels of urinary antimony in the unexposed population are generally below 1 pg 1-1 and that exposure to antimony results in a significant increase inMarch, 1982 URINE BY AAS WITH ELECTROTHERMAL ATOMISATION TABLE IV ANTIMONY IN URINE SAMPLES COLLECTED FROM PERSONS NOT EXPOSED TO ANTIMONY Concentration of antimony/ A \ ELg 1-1 Direct-extraction method Wet-oxidation method 1 or less 11 5 2 2 3 3 4 1 5 1 No. of results - - - - 259 the value. This is unlike lead, which is more widespread in the environment than antimony and where normal excretion levels in the population can range between 20 and 60 pg 1-1 of lead in urine.Four samples of urine were also analysed for lead following wet oxidation and direct extraction. The results are given in Table V. TABLE V DETERMINATION OF LEAD BY THE WET-OXIDATION AND DIRECT-EXTRACTION METHODS Lead concentration/pg 1-1 h < \ Sample Wet oxidation Direct extraction a t pH 4 1 71 2 61 3 50 4 60 76 68 48 57 The greater divergence between the two sets of lead results, from the wet-oxidation and direct-extraction methods, may be due to slight contamination. With a sample size of 20 ml we are only measuring between 1 and 2 pg of lead. Conclusions The direct extraction method has been regularly used in our laboratories over the past two years. With this simple and rapid method, which gives results comparable to the wet- oxidation method, we find that we can easily analyse 30 samples a day. Both methods include an extraction stage, which means that they are relatively free from interference. The use of nitric acid in the wet oxidation of organic material should be avoided when determining antimony. This could well be a major source of error in the colorimetric Rhodamine B method. The wet-oxidation method might be extended to the determination of trace amounts of antimony in other organic materials. The authors thank the Directors of Morganite Electrical Carbon for permission to publish this paper. 1. 2. 3. 4. 5. 6. 7. 8. 9. References Control of Lead a t Work Regulations, H.M. Stationery Office, London, 1980. Selander, S., and Cramer, K., BY. J . Ind. Med., 1968, 25, 139. Stevens, B. J., Sanders, J. B., and Stux, R., “Lead Determination in Blood and Urine by Atomic- Willis, J. B., Anal. Chem., 1962, 34, 614. Maren, T. H., Anal. Chem., 1947, 19, 487. Van Aman, R. E., Hollibaugh, F. D., and Kanzelmeyer, J . H., Anal. Chem., 1959, 31, 1785. Analytical Methods Committee, Analyst, 1980, 105, 66. Kamada, T., and Yamamoto, Y., Talanta, 1977, 24, 330. Yeager, D. W., Cholak, J., Henderson, E. W., Environ. Scz. Technol., 1971, 5, 1021. absorption Spectroscopy,” Varian Techtron, 1972. Received June 3rd. 1981 Accepted September 23rd, 1981

ISSN:0003-2654    

DOI:10.1039/AN9820700253

出版商:RSC    

年代:1982

数据来源: RSC

摘要:

260 Analyst, March, 1982, Vol. 107, pp. 260-268 Micro-determination of Tin in Organotin Compounds by Flame-emission and Atomic-absorption Spectrop hotomet ry lain L. Marr and Jamil Anwar Chemistry Department, The University of Aberdeen, Meston Walk, Old Aberdeen, A B 9 2UE A general niethod is described for the micro-determination of tin in organotin compounds, using either flame-emission or atomic-absorption spectrophoto- metry. Air - acetylene, dinitrogen oxide - acetylene and air - hydrogen flames have been compared: either of the air flames may be used, but the last mentioned is to be preferred as it is less noisy. The effect of different instrumental and chemical parameters on the emission and absorption of tin has been investigated. A number of organotin compounds have been analysed by both methods as well as by an alternative extraction - spectrophotometric method.The proposed method, in which samples are dissolved in a mixed solvent of water - hydrochloric acid - ethanol - butan-2-one and are aspirated into a near stoicheiometric air - hydrogen flame in an atomic-absorption spectrophotometer, is simple, accurate, rapid and equally useful for organic and for inorganic compounds. Metallic tin is used as the standard. Keywoyds : Tin determination ; organotin cornpounds ; atornic-absorption spectrophotometry ; flame-emission spectrophotornetry ; micro-analysis The determination of tin in organotin compounds poses a number of problems, whichever approach is tried. Gravimetric methods based on ignition to the oxide, sometimes in the course of a carbon - hydrogen combustion analysis, run into problems with volatile tin compounds.Wet-acid decomposition may also give rise to the loss of some tin-containing species, and the subsequent titrimetric determination steps are either slow (e.g., when pre- reduction of the tin is followed by oxidimetric titration) or may suffer from unwelcome interferences (particularly from the sulphuric acid, as in the EDTA titration). Spectrophotometry is a more commonly used technique for the determination of tin, and a number of reagents that form coloured complexes with the element have been investigated. Although the decomposition step is still unavoidable and special precautions have to be taken to avoid losses owing to volatility of the species containing the tin, at least one method has been found to be precise, sensitive and reliable, and has been applied to a large number of samples containing tin, including many organotin compounds.l As far as flame-spectroscopic methods are concerned (both emission and absorption), tin has been considered as one of the more difficult elements, and this is possibly, therefore, why it has received considerable attention from a number of investigators.The emission has been studied in relation to the excitation mechanism and the role of chemilumine~cence.~-~ Although flame-emission spectrophotometry has been used for rather few analytical applica- tions to tin, atomic absorption has been used far more extensively for many different kinds of samples.6-8 The effects of some organic solvents and of some other cations on the atomic absorption of tin have also been studied.9-15 A few attempts have been made to determine individual organotin compounds by using the same organotin compounds for the standards.l6-lg Petree and Smith20 tried to find a relationship between atomic-absorption sensitivity and structure of the compound for a number of organo- silicon and organotin compounds dissolved in hexan-2-one.They concluded that for practical purposes all compounds would have to be converted into the same form before determination by atomic absorption. On the other hand, Dean and FuesZ1 successfully determined arsenic in organoarsenic compounds by simply dissolving the compounds in an organic solvent and nebulising the solution into the flame, then measuring the arsenic emission.Because organic solvents play an important role in our present studies, it is worth briefly reviewing the literature concerned with their effect on the flame spectroscopy of tin. Several The determination of tin in organotin compounds has received scant attention.MARR AND ANWAR 261 workers agree that organic solvents do enhance the emission by tin, and that the excitation mechanism can best be explained by including chemiluminescence in the discu~sion.2-~ But statements concerning the effect of organic solvents on the atomic absorption of tin are not in agreement. Harrison and Julianog found that absorption by tin in an air - hydrogen flame was highly depressed by the presence of solvents, while Gibson et al. reported enhance- ment of absorption by organic solvent^.^ While many procedures for the determination of tin completely avoid the use of solvents, others use a solvent primarily in order to be able to separate the tin by a liquid -liquid extraction step,l0,l7 but do nevertheless aspirate the organic solution into the flame.There are advantages and disadvantages each way : water cools the flame, usually resulting in lower sensitivity, whereas organic solvents on their own tend to make the flame smoky and very reducing. It was felt that for this study a mixed aqueous - organic solvent would be most suitable and that this choice would make it possible to work with either inorganic or organic compounds. Such a mixed solvent would have four components : an organic solvent (5-10y0) to dissolve organotin compounds (solvents tried included butan-2-one, acetone, tetrahydrofuran and 1,2-dimethoxyethane) ; water (5-10%) to accommodate inorganic tin compounds; ethanol, the bulk solvent, to blend the other two components and to control the flame conditions; and hydrochloric acid (1-5y0) to help keep inorganic tin in solution, particularly in the standards prepared from metallic tin.This investigation has shown that the proposed method is not only simple, accurate and rapid, but is also tolerant of reasonable variations in solvent composition and has the advantage of using metallic tin as the standard. Experimental Spectrophotometric Method This has been described in full elsewherel; it was used in this study for comparison purposes, although as the sealed-tube decomposition step was not used, it cannot be con- sidered as a reference method for all compounds analysed in this study.The samples (5-10 mg) were decomposed in 1 ml of concentrated sulphuric acid with 1 ml of 30% hydro- gen peroxide. Flame-emission Method Calibration Dissolve 0.500 g of metallic tin in 50 ml of concentrated hydrochloric acid and dilute to 500 ml with ethanol to give a 1000 p.p.m. solution of tin. Transfer aliquots of this solution containing 1-5 mg of tin into 50-ml calibrated flasks. Add 5 ml of the organic solvent (the choice will depend on the solubility of the compounds to be analysed, but butan-2-one or 1,2-dimethoxyethane are to be preferred, although acetone and tetrahydrofuran may also be used) followed by 0.9-0.5 ml of concentrated hydrochloric acid, respectively, for the 1-5 mg samples, so as to equalise the concentration of acid in the different flasks.Dilute to the mark with ethanol and mix well. Prepare a blank solution from 5 ml of solvent and 1 ml of acid in 50 ml of ethanolic solution. Nebulise the solutions into an air - acetylene or an air - hydrogen flame under the optimum working conditions (the ratio of fuel to air is found by experiment, but certainly an oxidising flame has to be used: bright blue for acetylene and colourless for hydrogen). Measure the emission at 284 nm, about 5 mm above the burner for the acetylene flame and as close to the burner as possible for the hydrogen flame. Determination calibrated flask. ethanol. tion stage and read off the amount of tin from the calibration graph.Weigh out 5-10 mg of sample and dissolve it in 5 ml of the selected solvent in a 50-ml Add 1 ml of concentrated hydrochloric acid and dilute to the mark with Nebulise the solution into the flame under the same conditions as for the calibra- Equipment A Pye-Unicam SP 900 flame spectrophotometer was used with a circular burner for the development work, but a Perkin-Elmer 305 atomic-absorption instrument was used in the emission mode for some of the analyses. Read-out in both instances was on a chart recorder.262 MARR AND ANWAR: S N IN ORGANOTIN COMPOUNDS BY FLAME- Analyst, VOZ. 107 Atomic-absorption Method Calibration Prepare a 1000 p.p.m. solution of tin by dissolving 0.5 g of metallic tin in 50 ml of con- centrated hydrochloric acid and diluting to 500 ml with water.Transfer aliquots containing 1-5 mg of tin into 50-ml calibrated flasks followed by 5 ml of organic solvent (see comment above on choice of solvent) and the amount of hydrochloric acid required to bring its con- centration to 5% V/V. Prepare a blank by taking the same volumes of reagents, but no tin. Nebulise the solutions into an air - acetylene or an air - hydrogen flame under the optimum working conditions (the acetylene flame is blue coloured, turning slightly yellow when the solutions are aspirated, while the hydrogen flame is colourless, turning red - blue when a tin solution is aspirated). Measure the absorbance at 286.5 nm in the acetylene flame, or a t 224 nm in the hydrogen flame, in both instances about 2-3 mm above the burner, i.e., looking through the blue cone of the flame. Add 30 ml of ethanol and dilute to the mark with water.Determination Weigh out 5-10 mg of sample and dissolve it in 5 ml of the selected solvent in a 50-ml calibrated flask. Add 2.5 ml of concentrated hydrochloric acid and 30 ml of ethanol, then make the solution up to the mark with water. Nebulise the solution into the flame under the same conditions as for the calibration stage and read off the amount of tin from the calibration graph. Equipment A Perkin-Elmer, Model 305, atomic-absorption spectrophotometer was used with a 10-cm single-slot burner and an Activion hollow-cathode lamp operated at 20 mA. For trials with a dinitrogen oxide - acetylene flame a 5-cm burner was used, according to the maker’s recommendations.Results and Discussion Flame Emission Because the sensitivity for the determination of tin in aqueous solutions by flame-emission spectrophotometry is very poor, all the solutions used in this study were prepared in solvent mixtures consisting largely of ethanol. For the atomic-absorption studies described later the ethanol concentrations were lower, usually around 60% V/V. There are two emission lines that are useful for the determination of tin, at 286.3 and 284.0 nm, the latter being about twice as intense as the former. The intensities are about 60% higher for the acetylene flame (Fig. 1) but the noise is lower for the hydrogen flame as there is much less background emission from this flame in the relevant part of the spectrum.A B Fig. 1. Emission of tin at 284nm in two flames: A, air - hydrogen and B, air -acetylene. Tin at 100p.p.m. in 95% V / V ethanol with 5% V / V concentrated hydro- chloric acid. Each measured at optimum height above burner and optimum fuel - air ratio. Measured on a Perkin- Elmer 305 spectrophoto- meter, slit 4, damping 2. Optimum conditions for jame emission The effects of varying the fuel flow-rate and the burner height were studied using standard ethanolic solutions prepared from metallic tin. It was not possible to measure absolute fuel - oxidant ratios on the equipment used (and they would not be meaningful on account of entrained air entering the flame), but it is clear from Fig. 2 that there is an optimum fuelMarch, 1982 EMISSION AND ATOMIC-ABSORPTION SPECTROPHOTOMETRY 263 flow-rate for each flame where emission intensity is at a maximum. The maximum intensity is observed almost immediately above the burner, and lies within the blue cone seen in both flames when solvents are aspirated.This possibly reflects the distribution of the tin atoms in the flame, but it certainly relates to the frequency of collision processes giving rise to excited-state tin atoms in the flame. 12 c. .- S 3 2 2 e .I- .- .- 5 8 - rn S .I- .- S rn .- .- E w 4 r B I I I 1 1 I Fig. 2. Effect of fuel flow-rate on the emission of tin at 284 nm, measured 2 mm above the burner; air pressure, 25 p.s.i.g. A, Air - acetylene; 100 p.p.m. uf tin in 99% V / V ethanol; gain, x32. B, -4ir - hydrogen; 1000 p.p.m. of tin in 99% V / V ethanol; gain, x 2.E f e c t of hydrochloric acid The presence of hydrochloric acid is thought to be necessary to prevent hydrolysis of inorganic tin species and hence also any loss of tin from solution. Fig. 3 shows the effect for both flames when the solvent is largely ethanol. The marked depressive effect of the acid is according to expectations, because mixing more acid with less ethanol increases the viscosity of the solution and reduces the rate at which the tin in solution is conveyed to the flame. The difference in the shape of the two curves in Fig. 3 suggests that the presence of the acid also has some effect on the flame processes giving rise to emission from the tin. 0 5 10 15 20 25 Concentration of 12 N hydrochloric acid, % V/V Variation of emission intensity with concentration of hydrochloric acid, compared with solution uptake rate.Conditions and A and B as in Fig. 2 ; C, time taken for uptake of 2 ml of solution. Fig. 3. Efect of additional organic solvent include a second organic solvent to dissolve the sample. As many organotin compounds have only a low solubility in ethanol, it is necessary to The effect of adding varying con-264 MARR AND ANWAR: SN IN ORGANOTIN COMPOUNDS BY FLAME- Analyst, VoZ. I07 cent rat ions of other solvents (acetone, but an-2-one, 1,2-dime t hox yet hane, t e trah ydro f uran and toluene) to the ethanolic solutions of tin (prepared from the metal) was investigated for both flames. Each of the solvents slightly depressed the emission in the air - acetylene flame whereas the effect in the air - hydrogen flame was one of depression at higher solvent concentration but enhancement at lower concentrations. If the concentration range of the added solvent is l0-15% V/V (this amount replacing ethanol), this is not critical as the effect is negligible.This means that a small and variable loss of solvent occurring during the dissolution step can be tolerated if these optimum working conditions are adopted, i.e., additional solvent at a concentration of 10% V/V and hydrochloric acid at 2% V/V. Q 4 ' 9 Atomic Absorption Tin can be determined, with moderate sensitivity, by atomic-absorption spectrophoto- metry (typical working range 20-100 p.p.m.) using one of the lines at 286.3, 235.5 or 224.6 nm. Contrary to what one is led to believe by some statements in the literature9S2O the use of organic solvents does not necessarily lead to severe depression of the signal.For this investi- gation ethanol was again used as the blending solvent, but at somewhat lower levels than was used for the flame-emission work. Optimum conditions for atomic absorption The effect of added organic solvents on the atomic-absorption sensitivity was found to be dependent on the fuel flow-rate, as it was for flame emission: added solvents increase the absorption signal in oxidising flames and decrease it in reducing fuel-rich flames. Similar trends were observed for both the air - acetylene and the air - hydrogen flames. The experiments in this section of the work involved aqueous solutions of tin with varying concentrations of ethanol, and then 40% V/V ethanolic solutions containing varying amounts of additional solvents.Fig. 4 indicates the magnitude of the solvent effect for different fuel flow-rates in the air - acetylene flame. Only line A, for a fuel-rich flame, is in agreement with the finding of Harrison and Julianog that increasing the proportion of organic solvent causes further depression of the absorption signal. It could be that these workers studied the effect only in reducing flames. The data shown in Fig. 4 suggest that there is some measure of compensatory effect between the through-puts of fuel gas and of solvent to the flame, and that the optimum sensitivity will be obtained with a moderate proportion of solvent and a near stoicheiometric flame mixture. Choice of $ame ethanol sprayed into three different flames.Comparable absorption traces are shown in Fig. 5 for a 50 p.p.m. solution of tin in aqueous The acetylene flame gave a high sensitivity I 0 10 20 30 40 50 60 70 Concentration of ethanol, Yo V/V Fig. 4. Variation of tin atomic absorption, in different air - acetylene flames, as ethanol con- centration is increased ; 100 p.p.m. of tin in 1 yo V / I.' hydrochloride acid ; wavelength 286.3 nm; height above burner, 4 mm; air pressure, 32.5 p.s.i.g. A, Rich flame, acetylene pressure 28 (manometer reading) ; B, stoicheiometric flame, acetylene pressure 22 (manometer reading) ; and C, lean flame, acetylene pressure 18 (manometer reading).March, 1982 EMISSION AND ATOMIC-ABSORPTION SPECTROPHOTOMETRY 265 A B rc P C D E F Fig. 5 .Comparison of sensitivity and noise level for atomic-absorption determination of tin in different flames. The flow-rates were optimum for running solutions. A, Air - hydrogen, aqueous solution of 100 p.p.m., 100-mm single-slot burner; (trace reduced by half) ; B, air - hydrogen, 100 p.p.m. in 60% V / V ethanol, 100-mm single-slot burner; C, air - acetylene, aqueous solution of 100 p.p.m. ; 100-mm single-slot burner; D, air - acetylene, 100 p.p.m. in 60% ethanol, 100-mm single-slot burner; E, dinitrogen oxide - acetylene, aqueous solution of 100 p.p.m., 50-mrn single-slot burner; F, dinitrogen oxide - acetylene, 100 p.p.m. in ethanol, 50-mm single-slot burner. when it was very reducing, but it also gave much more noise. On the other hand, the hydrogen flame gave about twice the sensitivity and very much less noise: as it is easy to change between these two fuel gases, this latter choice can be strongly recommended. The dinitrogen oxide - acetylene flame gave a sensitivity comparable to that for the a i r - hydrogen flame, allowing for the former being used with a 5-cm burner, but the noise was much worse.Added to that is the problem of having to stop after every 4-5 samples to clean the burner from deposited carbon when organic solvents are aspirated : this dinitrogen oxide - acetylene flame therefore has little to recommend it for this determination. E$ect of hydrochloric acid Up to 15% V/V of concentrated hydrochloric acid can be used without introducing any interference in either flame, and for practical reasons this concentration is best kept as low as possible (5% in the procedure given).The absence of interference by hydrochloric acid has been reported by a number of other worker^.^ At first, the difference between this behaviour and that shown in Fig. 3 for emission work might seem surprising. However, the solutions for the atomic-absorption measurements contained only around 50% V/V ethanol, and the uptake rate did not change with increasing concentration of hydrochloric acid, as it did with the higher ethanol concentration for the emission studies. E$ect of additional organic solvents The effect of varying the solvent composition and also the fuel flow-rate (while keeping the air pressure constant) was investigated for the atomic-absorption determination of tin with both flames. The results displayed in Fig.6 indicate that the optimum fuel - oxidant ratio differs for solutions of different composition: all the solutions showed the same general kind of dependence, but the maximum absorption signal was recorded at different fuel flow- rates for different solvents. Three additional solvents were investigated in admixture with ethanol and water : acetone, butan-2-one and 1,2-dimethoxyethane. The more detailed picture that emerges when Fig. 7 is considered illustrates again the apparent compensatory relationship between fuel flow-rate and the proportion of organic solvent. Similar relationships were found for all266 MARR AND ANWAR: SN IN ORGANOTIN COMPOUNDS BY FLAME- Analyst, VoZ. 107 A B ' C I I Flow-rate of fuel, manometer units 15 20 25 30 35 40 Fig.6. Effect of fuel flow-rate on absorption of 50 p.p.m. of tin a t 286.3 nm; air pressure, 35 p.s.i.g. A, Hydrogen - air, 1 yo hydrochloric acid; B, acetylene - air, 1% hydrochloric acid; C, acetylene - air, 50% V/V ethanol; D, acetylene - air, 50% V / V ethanol plus 10% butan-2-one. 5 10 15 20 25 Concentration of butan-2-one in aqueous ethanol, % VIV Variation in absorption of 50 p.p.m. of tin with butan-2-one concentration in aqueous ethanol in an air - hydrogen flame. Air pressure: 32.5 p.s.i.g.. Hydrogen pressure: A, 80; B, 60; and C, 40 (manometer readings). Solution: 70% V / V ethanol, 625% V / V butan-2-one and 30-5y0 V / V water. Fig. 7. TABLE I COMPARISON OF RESULTS FOR ANALYSIS OF SOME ORGANOTIN COMPOUNDS Tin content, yo Compound (C,H,),SnCI .. .. . . Triphenylvinyltin . Hexaphenylditin . . . . (C,H,),SnCH,I . . .. . . C,H,),Sn . . . . . . Flame emission Theoreti- Spectrophoto- cal metric method 30.8 30.6 23.5 23.0 31.5 31.5 24.3 23.9 33.9 32.5 24.2 24.6 30.0 29.65 22.6 21.9t 24.7 24.5 41.2 38.5 27.8 27.5 Air - acetylene 30.7 23.3 31.8 24.1 34.2 24.5 29.7 22.2 24.4 40.2 27.3 Air - hydrogen 30.5 23.2 31.7 24.5 34.1 24.7 30.2 22.1 24.1 40.1 26.8 Atomic absorption -- Air - acetylene 30.5 23.7 31.3 24.4 34.1 24.4 30.5 22.2 24.5 39.9 27.4 Air - hydrogen 30.4 23.6 31.2 24.5 33.7 24.4 29.8 22.00 24.6 40.00 27.3 Solvent used for emission and absorption Butan-2-one Butan-2-one Butan-2-one Bu tan -2-one Butan-2-one Tetrahydrofuran Butan-2-one Tetrahydrofuran Butan-2-one Tetrahydrofuran Tetrahydrofuran * Sample decomposed by sealed-tube method' gave 30.0% tin.t Sample decomposed by sealed-tube method' gave 22.6% tin.March, 1982 EMISSION AND ATOMIC-ABSORPTION SPECTROPHOTOMETRY 267 three solvents. From a practical point of view, the behaviour in near stoicheiometric flames with moderate amounts of additional solvent (ie., around 10% V/V of acetone, 5-10% V/V of butan-2-one or 10-15% V/V of dimethoxyethane) is very attractive, because under these conditions the measured absorption signals show a generous tolerance to variation in solvent composition coupled with satisfactory sensitivity. The choice of such working conditions has formed the basis of a useful and reliable method for the determination of tin in organo- metallic compounds and also in inorganic salts, by atomic-absorption spectrophotometry.Analysis of Organotin Compounds The results obtained for the analyses of a number of organotin compounds are summarised in Table I. Results obtained by the spectrophotometric method tended, .in general, to be slightly low, particularly for the methyltin compounds. This could well be due to some losses occurring during the open-tube wet-acid decomposition step, which could be more serious for the more volatile methyltin compounds. The advantage of the flame method is seen clearly here. Both the flame-emission and the atomic-absorption methods gave generally acceptable results, correct within the standard deviation ranges. In some instances, notably with (CH,) ,SnS.C,H,.NH,, discrepancies in the results were significant and can probably be attributed to impurities in the compounds and to their instability.Table I1 compares the performance of the three methods in terms of relative standard deviations for the deter- mination of tin in some selected compounds. It is of interest that the only method that gave significantly different standard deviations for two compounds was the spectrophoto- metric one; one of the selected compounds contained methyl groups bonded to the tin atom. TABLE I1 COMPARISON OF THE PERFORMANCE OF METHODS FOR DETERMINATION OF TIN Method Compound Spectrophotometry . . . . (C6H,),SnC1 Flame emassion- Air - acetylene . . . . (C,H,),SnCH,I (C6H5)3SnC1 Air - hydrogen . . * (C6H5)4Sn Atomic absorption- Air - acetylene . . . . (C6H,),SnC1 W Air - hydrogen .. * * (C6H5)4Sn No. of samples 6 6 8 6 5 6 6 6 6 7 Noise R.S.D.,* % level,? % 0.82 - 1.01 1.45 1.51 1.53 1.55 0.72 0.68 0.62 0.60 * R.S.D. = relative standard deviation. t Noise level expressed as a percentage of the signal from the most concentrated working standard used to construct the calibration graph. Considering the flame methods, atomic absorption is seen to be markedly better than flame emission in terms of reproducibility, although the accuracy of both methods is268 MARR AND ANWAR acceptable. The flame-emission method has proved useful in analysing extracts of soft- wood treated with an organotin fungicide, bis(tributyltin)oxide, as the recommended extrac- tant is ethanol containing a little concentrated hydrochloric acid. The poorer performance of the emission method may be due, in part, to the age of the instrument used, but may also be accounted for to some extent by the noise level from the flame background emission.The significance of the noise level in these flame methods is well illustrated by the improve- ment in performance achieved by changing from acetylene to hydrogen for the atomic- absorption measurements (Table 11). Some workers have preferred to use the dinitrogen oxide - acetylene flame when determining tin in organic solvents, claiming that it gives a higher sensitivity than when air is used as the oxidant, and also reasonable freedom from interferences from other metal^.^^^^^ For this study the sensitivity offered by the air - hydrogen flame is adequate, and it does not have the problem of carbon being deposited on the burner.Conclusions This investigation has clearly shown that tin can be determined with satisfactory results and with very little manipulative effort in a wide range of compounds, both inorganic and organometallic, by simply dissolving the sample in a suitable mixed solvent and aspirating it into an air - hydrogen flame in an atomic-absorption spectrophotometer. Other procedures described in the literature for the determination of organotin compounds have relied on the availability of a sample of the pure organotin compound for use as a standardl6s18,l9: this procedure offers the advantage of needing only metallic tin for preparation of the calibration solutions. The authors thank the Ministry of Education, Government of Pakistan, for providing a scholarship (for J.A.), Dr.R. A. Chalmers for his interest in the work and Dr. J. L. Wardell for providing the organotin compounds. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. References Marr, I. L., Talanta, 1975, 22, 387. Buell, €3. E., Anal. Chem., 1963, 35, 372. Gibson, J. H., Grossman, W. E. L., and Cooke, W. D., Anal. Chem., 1963, 35, 266. Gilbert, P. T., in Lippincott, E. R., and Margoshes, M., Editors, “Proceedings of the Tenth Colloquium Spectroscopicurn Internationale,” Spartan Books, Washington, D.C., 1963, pp. 17 1- 215. Dagnall, R. M., Thompson, K. C., and West, T. S., Analyst, 1968, 93, 518. Capacho-Delgado, L., and Manning, D. C., Spectrochim. Acta, 1966, 22, 1505. Bowman, J. A., Anal. Chim. Acta, 1968, 42, 285. Agazzi, E. J., Anal. Chem., 1965, 37, 364. Harrison, W. W., and Juliano, P. O., Anal. Chem., 1969, 41, 1016. Headridge, J. B., and Sowerbutts, A., Analyst, 1972, 97, 442. Welsch, E. P., and Chao, T. T., Anal. Chim. Acta, 1976, 82, 337. Vickers, T. J., Cottrell, C. R., and Breakey, D. W., Spectrochim. Acta, 1970, 258, 437. Schallis, J. E., and Kahn, H. L., A t . Absorpt. Nemsl., 1968, 7, 84. Levine, J. R., Moore, S. G., and Levine, S. L., Anal. Chem., 1970, 42, 412. RubeBka, I., and MikSovskg, M., A t . -4bsorpt. Nemsl., 1972, 11, 57. Freeland, G. N., and Hoskinson, R. M., Analyst, 1970, 95, 579. Burke, K. E., Analyst, 1972, 97, 19. Williams, A. I., Analyst, 1973, 98, 233. George, G. M., Albrecht, M. -4., Frahm, L. J . , and McDonnell, J . P., J . Assoc. Off. Anal. Chem., Peetre, I. B., and Smith, B. E. F., Mikrochim. Acta, 1974, 301. Dean, J. A., and Fues, R. E., Anal. Lett., 1969, 2, 105. Pickett, E. E., and Koirtyohann, S. R., Spectrochim. Acta, 1969, 24B, 325. 1973, 56, 1480. Received July 9th, 1981 Accepted October Sth, 1981

ISSN:0003-2654    

DOI:10.1039/AN9820700260

出版商:RSC    

年代:1982

数据来源: RSC

摘要:

Analyst, March, 1982, Vol. 107, p$, 269-275 269 Determination of Mercury in Coal by Non-oxidative Pyrolysis and Cold Vapour Atomic-fluorescence Spectrometry L. Ebdon and J. R. Wilkinson and K. W. Jackson Department of Environmental Sciences, Plymouth Polytechnic, Drake Circus, Plymouth, Devon, PL4 8AA Department of Chemistry, Shejield City Polytechnic, Pond Street, Shefield, S1 1 WB Mercury was liberated from coal by non-oxidative pyrolysis at 800 "C with a nitrogen purge and collected in acidified potassium permanganate solution. Subsequent determination was by cold vapour atomic-fluorescence spectro- metry using an argon-sheathed windowless cell. The mercury vapour was generated continuously by pumping tin(I1) chloride and sample to a novel reduction cell, This continuous flow system offered excellent precision, with a 27; relative standard deviation at 0.5 ng ml-l and a detection limit ( 2 ~ ) of 0.043 ng ml-l.The accuracy and precision were assessed by performing repli- cate analyses on a number of coals, including NBS SRM 1632a bituminous coal, for which a mean value of 134 ng g-l and a standard deviation of 3.1 ng g-l were obtained. The excellent sensitivity, precision and extended linear working range of the method (5 ng g-l to 25 p g g-l) are noted. Keywords : Mercury determination ; coal ; cold vapour atomic-fluorescence spectrometry ; non-oxidative pyrolysis ; continuous flow system Although coal has been a major source of energy in industrialised countries for many years, the rising cost of other forms of energy, and the increased recognition that they will be consumed before coal supplies are exhausted, has refocused attention on coal.Natural gas has, for example, been widely welcomed as a relatively clean and simply delivered energy source. Eventually supplies of such gas will end and then it seems likely that we shall return to gas supplies produced from coal. Coal utilisation is, however, a potential source of a range of environmental pollutants including volatile trace metals. Many elements have been identified in coal,lP2 and it is a subject of intensive study as to which become airborne after combustion or gasification. Hence there is increasing interest in determining toxic metals in coal, for ex- ample so that toxic metal balances at coal combustion, gasification and liquefaction plants may be determined.Mercury causes particular concern because of its severe toxicological effects and because the large tonnages of coal to be consumed could lead to relatively large amounts of mercury reaching the en~ironment.~ Determinations of mercury in coal are fraught with a number of problems because of the highly intractable nature of the coal matrix, the volatility of mercury and the very low levels thought to be present (typically significantly less than 1 pg g-l). It has been proposed that spark-source mass spectrometry offers a technique whereby mercury can be determined in coal without sample d i s s ~ l u t i o n , ~ ~ ~ but this has not been widely confirmed. In any event, it is unlikely that good precision will be obtained in this way.Although whole coal may be irradiated in order to determine mercury by neutron-activation analysis, it is reportedly necessary to c o r n b ~ s t , ~ or digest,6 the coal later in order to concentrate the mercury prior to counting in order to obtain the necessary sensitivity. Re- cently, low-temperature ashing (150 "C) using radiofrequency excited oxygen at low pressure has been used to prepare coal samples for analysis. Although most metals are retained using this method, some of the more volatile may still be l o ~ t . ~ ? ~ Low-temperature ash also requires further treatment, e.g.. fusion or dissolution with hydrofluoric acid to bring it into a form suitable for analysis. A variety of oxidising acid mixtures and conditions have been used, typically under reflux, e.g., sulphuric acid - hydrogen peroxide,' sulphuric acid - nitric acid,8 sulphuric acid - nitric acid - perchloric Most alternative analytical methods for coal entail prior destruction of the matrix.A few reports have appeared on methods of coal dissolution without prior ashing.270 EBDON et d. HG IN-COAL BY NON-OXIDATIVE PYROLYSIS Analyst, VOZ. I07 acid,9 perchloric acid - periodic acid,1° aqua regia - potassium permanganatell and sulphuric acid - potassium permanganate.12 Some of these methods may result in incomplete digestion or losses of volatile elements, especially mercury owing to the vigorous digestion conditions. The use of PTFE-lined digestion bombs is another alternative reported, and again a variety of acid mixtures have been used, e.g., aqua regia - hydrofluoric acidla and fuming nitric acid - hydrofluoric acid.14 These bombs present some handling difficulties6,11 and acid attack on metallic components may be experienced. In our laboratory, none of these methods of wet digestion was found to be very satisfactory, yielding low recoveries of mercury or incomplete digestion. Possibly- this is because some of the mercury present is in the sulphide form and even the use of reflux conditions could not retain the mercury formed upon oxidative heating conditions.For mercury, however, a specific approach is possible. Oxidative combustion and non- oxidative pyrolysis can be used to generate mercury vapour at temperatures reported between 600 O C 1 5 and 850 O C . 1 6 The liberated mercury may subsequently be trapped and isolated on a series of gold frits6s16-l8 or collected in a suitable absorbing solution, commonly acidified potassium permanganate6,15,19.20 or in a liquid nitrogen-cooled trap.21 Some workers have experienced problems of over-rapid combustion when using oxygen as the purge gas,15 and others have reported low recoveries if auxiliary oxidising agents or catalysts, e.g., copper oxide,21 heated quartz-woo116p20 or hot platinum wires,21 are not included in the combustion train.In view of these conflicting opinions, especially concerning the possibility of explosive ignition, it seemed that non-oxidative pyrolysis under nitrogen, as a method of extracting mercury from the coal matrix for subsequent collection in acidified potassium permanganate solution, was worthy of investigation.Atomic-absorption and atomic-fluorescence spectrometry are commonly used to determine mercury at trace levels in both flame and electrothermal atom cells. Using flames, typical detection limits are 0.5 pg ml-l for atomic-absorption spectrometry22 and 0.1 pg ml-1 for atomic-fluorescence spe~trometry.~3 When the cold vapour reduction - aeration technique is used, the reported detection limits are much lower, typically 0.6 ng for atomic-absorption spectrometry using 5-ml samples24 and 0.02 ng for atomic-fluorescence spectrometry using 1-ml ~amples.~5 The cold vapour atomic-absorption method, however, suffers from several disadvantages that can be overcome by using atomic-fluorescence spectrometry. With cold vapour atomic-absorption spectrometry fogging of the cell windows by moisture necessitates, at low mercury levels, drying the vapours prior to their entry to the atom cell and the desiccant can be both a source of contamination and mercury loss26; when a windowless cell is used in atomic-fluorescence spectrometry such problems are avoided.Whereas in cold vapour atomic-absorption spectrometry non-specific and molecular absorption are recorded as mercury and background correction is highly desirable, in atomic-fluorescence spectrometry molecular absorption poses a lesser problem. Greater sensitivity and an extended linear working range are also available with atomic-fluorescence spectrometry. We have recently reported a simple, improved, argon-sheathed cell for the cold vapour atomic-fluorescence method that offers excellent ~ensitivity.~7 This was used as the basis of the detection system in this work.Our previous report used the conventional manual injection technique with discrete samples. This method is relatively time consuming and a fully automated system using the atomic- absorption, reduction - aeration, procedure has been described.28 Although relatively expens- ive and complex, this report indicated the feasibility of developing a continuous flow procedure. The advantages of such a procedure may be summarised as follows : steady-state signals lead to greater convenience and improved accuracy; precision is increased through the elimination of injection and other discrete manipulations ; control over experimental variables is greater; the base line establishes the background level unambiguously and this remains essentially constant.The system described in this paper uses two elementary peristaltic pumps to deliver the sample and reducing agent to the reduction cell. This was purged with argon and acted as a constant- head gas - liquid separator. The evolved mercury vapour was swept to the windowless, argon- sheathed, atomic-fluorescence cell previously reported. 27 Apparatus Experimental The apparatus was the same as that used previously to determine mercury in orchard leavesMarch, 1982 AND COLD VAPOUR ATOMIC-FLUORESCENCE SPECTROMETRY 27 1 and barley seedsz7 except for the addition of the continuous flow system and the equipment for the pyrolysis of the coal. Instrumental operating conditions for the excitation and measure- ment of mercury atomic fluorescence were as described previ0usly.~7 Continuous Flow System Two small peristaltic pumps (Schuco Scientific, London) were used to deliver tin(I1) chloride solu- tion and standard or sample to a reduction cell.This cell consisted of a pear-shaped flask of capacity 25 ml, a constant-head drainage tube and two inlet ports (4 mm i.d.) for the pumped inputs. A constant flow of argon was bubbled through the solution to flush the evolved mercury vapour into the atom cell, The continuous flow system for the generation of mercury vapour is shown in Fig. 1. Fig. 1. Continuous flow system for the generation of mercury vapour. A, Peristaltic pump, tin(I1) chloride solution, 2.5 ml min-l; B, peristaltic pump, sample/standard solution, 7 ml min-l ; C, reduction cell; D, drain, to waste; E, carrier gas inlet, 0.4 1 min-l of argon; F, sheathing gas inlet, 5.0 1 min-' of argon; G, laminar flow device; and H, atom cell.Pyrolysis System This is described in the procedure below and shown in Fig. 2. Reagents All chemicals used were of analytical-reagent grade, except where stated. Acidified potassium permanganate collecting soluihws. Add potassium permanganate (50 g) to 1 1 of a mixture containing concentrated nitric acid (100 ml) , concentrated sulphuric acid (200 ml) and distilled, de-ionised water (706 ml). Dilzlent (used for all standards as a presefivative). Nitric acid (loyo, 1.6 M) containing 0.03% m/V of potassium dichromate. Hydroxytammonium chloride solution, 2by0 m/ V in distilled, de-ionised water.Purify by a double extraction with dithizone (0.05% @/V) in chloroform.272 EBDON et al. : HG IN COAL BY NON-OXIDATIVE PYROLYSIS Analyst, VOZ. I07 E H Fig. 2. Schematic diagram of coal pyrolysis system. A, Nitrogen inlet, 300 ml min-l; B, scrubbing solution, 50 ml; C, fused silica tube; D, fused silica boat containing powdered coal; E, furnace; F, heating tape and power supply; G, collecting solution, 20 ml; and H, collecting solution, 5 ml in a 50-ml calibrated flask. Tin(l1) chloride solution, 10% m/V. Dissolve tin(I1) chloride dihydrate in hydrochloric acid (1 M). Purge the solution with argon for about 5 min each day before use. Procedure for Determining Mercury in Coal Weigh powdered coal (ca. 2.0 g, <250 pm) into a fused silica boat, previously cleaned by heating at 1000 "C in a stream of nitrogen. Insert the boat into a fused silica tube (25 mm i.d., 500 mm long) located in the cavity of a tube furnace (Griffin and George Electric Furnace, Carbolite, Sheffield).Fill a Drechsel bottle (to act as a scrubber) and the two collecting bubblers (see Fig. 2) with 50, 20 and 5 ml, respectively, of acidified potassium permanganate collecting solution. Connect the scrubber to the front of the train and the collectors to the end, the latter by means of a heated tube maintained at 150 "C by a heating tape to prevent condensation of water or organic vapours on the glassware. Switch on the nitrogen supply (300 ml min-l) and then the furnace. On reaching 800 "C (after about 40 min at 20 "C min-l) switch off the furnace and allow it to cool to 400 "C (about 1 h).Transfer the contents of the bubbler into the 50-ml flask, rinsing once with 10 ml of concentrated nitric acid and then twice with small volumes of the acidified potassium dichromate diluent. Transfer the washings into the 50-ml flask and destroy the excess of potassium permanganate by adding hydroxyl- ammonium chloride solution (20% m/V) until the purple colour is discharged. Dilute to 50 ml with the acidified potassium dichromate solution. Prepare a blank by "pyrolysing" an empty boat every fourth or fifth sample. Using the continuous flow system, generate mercury vapour and determine the concentration of mercury using atomic-fluorescence spectrometry with the gas-shielded atom cell.After subtraction of the blank signal from the sample signal, deter- mine the mercury content by comparison against aqueous calibration standards. Using the acid potassium dichromate solution, standard solutions remained stable for several days at the 1 ng ml-l level and below, and for several weeks at levels of 10 ng ml-1 and greater. The coal digests remained stable for at least one working day. Results and Discussion Optimisation of the Continuous Flow System The signal versus time graph obtained from the continuous flow system was typical of an automated flow system, as shown in Fig. 3. The signal size was independent of both the flow- rate and the concentration of the tin(I1) chloride solution in the ranges 2.5-25 ml min-l and 540% m/ V , respectively.The signal increased linearly with increasing flow-rate of sample (or standard) up to a maximum of 30 ml min-l, above which it became difficult to drain the cell sufficiently rapidly. Flow-rates of 2.5 ml min-l of tin(II) chloride solution (10% m/V as the dihydrate) and 7.0ml min-l of sample/standard were chosen to minimise the volume of reagent and sample required, consistent with acceptable signal size. Using these flow-rates the rise time, as shown in Fig. 3, was about 60 s and the sample delivery tube contained suffici-March, 1982 AND COLD VAPOUR ATOMIC-FLUORESCENCE SPECTROMETRY 273 0 1 2 3 Ti me/m i n Fig. 3. Typical signal ven'sus time plot for the continuous flow system. A, t = 0, tube is placed in sample solution; B, t = 60 s, steady-state signal is obtained and delivery tube is removed from solution; B-C, 15-s integration period; C, t = 75 s, solution ceases to enter cell; and D, t = 160 s, base line is re-established.ent residual solution to allow a 15-s integration time on the spectrometer to be used with removal of the tube from the sample solution upon commencement of the integration. Conse- quently, the amount of sample consumed was about 7 ml. Residual mercury was purged from the cell and the base line re-established about 2 min after solution ceased to enter the cell. The continuous flow system gave improved precision, particularly at low levels of mercury, with typically a 2y0 relative standard deviation at 0.5 ng ml-l compared with 7% at 0.5 ng rn-1 for the manual injection technique. If 10% is taken as the maximum tolerable relative stand- ard deviation, the lower end of the useful working range of the method was 0.2 ng ml-1 and calibration was linear over nearly 4 orders of magnitude, from 0.2 to 1000 ng ml-l.The limit of detection, expressed as the mass or concentration of mercury that produces a signal equal to the mean of the blank plus twice the standard deviation of the blank mean, was 0.043 ng ml-1. The total analysis time was 2-3 min per solution. Preparation of Coal Samples Following the decision to investigate non-oxidative pyrolysis under nitrogen as a method for liberating mercury from coal, a furnace temperature of 800 "C was selected as being high enough for complete mercury distillation. Initially acidified potassium dichromate was used as the collecting solution because of its proved utility for storing dilute mercury solutions.Various concentrations of potassium dichromate and nitric acid were investigated, but apparently high recoveries (200-1 Z O O ~ o ) were obtained. On the other hand, initial experiments using potass- ium permanganate - sulphuric acid collecting solutions yielded low recoveries. This was apparently due to the reducing effect of some of the organic pyrolysis products on the perman- ganate, as evidenced by the formation of manganese dioxide and an orange coloration of man- ganese(1V) ions. When air was used as the purge gas, in order to oxidise these organic products, low recoveries were again observed and there was some evidence of over-rapid reaction and sputtering of the sample from the silica boat.The problem of low recoveries was overcome by returning to nitrogen as the purge gas, increasing the concentration of the potassium perman- ganate and adding concentrated nitric acid to the collecting solution, making it both more oxidising and acidic. The collecting solution described above gave quantitative recoveries for the only coal with a certified mercury value and would be expected to be sufficiently oxidising for a range of coal types. Determination of Mercury in Coal The method thus developed was applied to the analysis of several coal and ash samples. Results obtained for some samples with independently determined mercury levels, including the National Bureau of Standards (NBS) standard reference material SRM 1632a bituminous coal, are shown in Table I.The mercury content of the ash sample is unusually high and, although it is therefore atypical, some demonstration of the wide working range of the technique is given.274 EBDON et aZ. : HG IN COAL BY NON-OXIDATIVE PYROLYSIS Analyst, VoZ. 107 TABLE I RESULTS OF MERCURY DETERMINATIONS ON VARIOUS SAMPLES Independent analysis r 3 Mercury found/ Mercury quoted/ Sample ng g-’ ng g-l Method* NBS SRM 1632a coal . . 134.17 130 f 30: AAS and NAA NCB A coal . . .. 140 <400 NAA NCB B coal .. .. 175 < 500 NAA BCRA coal ash . . . . 202 pgg-1 195 p g 8-l CVAFS CVAFS = cold vapour atomic-fluorescence spectrometry. Standard deviation, 3.1 ng g-l; relative standard deviation, 2.7%. * AAS = atomic-absorption spectrometry ; NAA = neutron-activation analysis ; t Mean for nine replicate coal samples with each of these solutions analysed nine times.Certificate value. Conclusions It was shown previously27 that the use of an argon-sheathed atom cell improved the precision of cold vapour atomic-fluorescence spectrometry. This, together with the reproducible pyrolysis conditions and the continuous flow system, provided more than adequate precision for coal analysis (2.7% relative) ; this is especially so when the inherent sampling problems of coal are considered. The large linear working range for mercury in coal (5 ng g-l to 25 pg g-l), with a detection limit of 1.13 ng g1 using a 2-g sample, could be even further improved by varying the sample size taken or the amount of collecting solution. This confirms the high sensitivity and extended working ranges of the cold vapour atomic-fluorescence technique.The analysis time is at present 2-3 min per solution, but the time-consuming step involves the slow heating and cooling rates of the furnace available. A smaller, more versatile furnace with more rapid heating and cooling rates should markedly reduce the over-all analysis time of approximately 1.75 h per sample. This would make the method even more suited to the routine determination of mercury in coal and fly ash, with its probable extension to coal tar and related materials. The method shows excellent agreement with certified and independent coal analyses. We are grateful to the Science Research Council and the London Research Station, British Gas Corporation, for the award of an SRC CASE studentship to one of us (J.R.W.), and to the East Midlands Regional Laboratory, National Coal Board, and the British Carbonization Research Association for providing analysed coal samples for comparative purposes. 1.2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. References Abernethy, R. F., Peterson, M. J., and Gibson, F. H., “Spectrochemical Analysis of Coal Ash for Trace Elements,” Rep. Invest. US. Bur. Mines, 1969, No. 7281. Kessler, T., Sharkey, A. G., Zr., and Friedel, R. A., “Analysis of Trace Elements in Coal by Spark Source Mass Spectrometry, Rep. Invest. U.S. Bur. Mines, 1973, No. 7714. Piperno, E., i n Babu, S . P., Editor, “Trace Elements in Fuel,” Advances in Chemistry Series, No. 141, American Chemical Society, Washington, D.C., 1975, pp.192-209. Slates, R. V., “Methods for Analysis of Trace Elements in Coal, Coal Fly Ash, Soil and Plant Samples,” Report DP 1421, National Technical Information Service, 1976. Ruch, R. R., Gluskoter, H. J., and Shimp, N. F., “Occurrence and Distribution of Potentially Volatile Trace Elements in Coal : An Interim Report,” Illinois State Geological Survey, Environ- mental Geology Notes, April, 1973, No. 61. O’Gorman, J. V., Suhr, N. H., and Walker, P. L., Jr., APpl. Spectrosc., 1972, 26, 44. Murphy, J., A t . Absorpt. Newsl., 1975, 14, 151. BS 1016: Part 10: 1960, “Methods for Analysis and Testing of Coal and Coke. Rains, T. C., and Menis, O., J . Assoc. Off. Anal. Chem., 1972, 55, 1339. Spielholtz, G. I., and Diehl, H., Talanta, 1966, 13, 991. Nadkarni, R. A,, Anal. Chem., 1980, 52, 929. Huffman, C., Jr., Rahill, R. L., Shaw, V. E., and Norton, D. R., U.S. Geol Sum. Pvof. Pap., 1972, Ward, A. F., and Marciello, L., Jawell-Ash Plasma Newsl., 1978, l,, 10. Part 10: Arsenic in Coal and Coke,” British Standards Institution, London, 1977. No. €4004, p. C203.March, 1982 AND COLD VAPOUR ATOMIC-FLUORESCENCE SPECTROMETRY 275 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. Hartstein, A. M., Freedman, R. W., and Platter, D. W., Anal. Chem., 1973, 45, 611. Lo, F. C., and Bush, B., J . Assoc. 08. Anal. Chem., 1973, 56, 1509. Anderson, D. H., Evans, J . H., Murphy, J . J., and White, W. W., Anal. Chem., 1971, 43, 1511. Joensuu, 0. I., Appl. Spectrosc., 1971, 25, 526. Heinrichs, H., 2. Anal. Chem., 1975, 273, 197. Bailey, B. W., and Lo, F. C., J . Assoc. 08. Anal. Chem., 1971, 54, 1447. Jones, P., and Nickless, G., Proc. SOC. Anal. Chem., 1973, 10, 269. Cavalli, P., and Rossi, G., Analyst. 1976, 101, 272. Price, W. J., “Spectrochemical Analysis by Atomic Absorption,” Heyden. London, 1979, p. 321. Zacha, K. E., Bratzel, M. P., Jr., Winefordner, J. D., and Mansfield, J . M., Jr., Anal. Chem., 1968, Gilbert, T. R., and Hume, D. N., Anal. China. Ada, 1973, 65, 461. Thompson, K. C., and Godden, R. G., Analyst, 1975, 100, 544. Gardner, D., Anal. Chim. Acta, 1980, 119, 167. Ebdon, L., Wilkinson, J. R., and Jackson, K. W., Anal. Chim. Ada, 1981, 128, 48. Bailey, B. W., and Lo, F. C., Anal. Chem., 1971, 43, 1525. 40, 1733. Received July 27th, 1981 Accepted August 27th, 1981

ISSN:0003-2654    

DOI:10.1039/AN9820700269

出版商:RSC    

年代:1982

数据来源: RSC

摘要:

276 Analyst, March, 1982, Vol. 107, $9. 276-281 Optical Emission Spectrometry with an Inductively Coupled Radiofrequency Argon Plasma Source and Direct Sample Introduction from a Graphite Rod G. F. Kirkbright and S. J. Walton Department of Instrumentation and Analytical Science, University of Manchester Institute of Science and Technology. P.O. Box 88, Manchester, M60 IQD Preliminary studies of analytical performance are described for an instru- mental assembly in which microlitre volumes of liquid samples are applied to a graphite rod, desolvated and the rod is inserted axially directly into a continuously operating low-power inductively coupled argon plasma. Simplex optimisation of operating parameters has been undertaken for manganese to determine the most favourable signal to background intensity ratios and detection limits are reported for manganese, cadmium, cobalt, copper, iron, nickel and lead.Keywords : Optical emission spectrometry ; inductively coupled plasma ; direct sample introduction ; graphite rod Apart from conventional pneumatic nebulisation, sample introduction into inductively coupled plasmas employed for optical emission spectrometry (ICP - OES) has been concerned with external vaporisation using electrothermal heating,lP2 spark excitation3 or laser ablation4; in each instance the generated aerosol has been transported to the plasma in a stream of carrier gas. Recently, Salin and Horkick5 described a method in which samples are inserted directly into a 2.5-kW argon ICP using a graphite electrode inserted axially via the injector of a Fassel-type torch and using the electrode to ignite the plasma on each insertion.Continuous operation has been achieved by Sommer and Ohls,6 who inserted samples into a Greenfield-type torch sup- porting an argon - nitrogen ICP at a continuous power of 3 kW. The study described here was an attempt to develop a technique for the direct insertion of sample solutions into a low-power (<1.5 kW), continuously running argon ICP using an axial graphite rod on to which the sample is applied with a micropipette. Apparatus The source consisted of a commercial 27.12-MHzJ 1.5-kW generator (Model HFD-EOOD, Plasmatherm Inc.) fitted with a demountable torch of the type described by Bombelka.7 The torch was modified only in that the 3.7 mm i d .injector tube had no constriction at the tip, and was lengthened so that the lower end extended through an aperture cut into the base of the torch box. The insertion device consisted simply of a standard 30.5-cm graphite rod, diameter 3.05 mm (Ringsdorff RW-0), attached to an insulating handle. An adjustable stop controlled the position of the rod in the plasma. The detection system consisted of a 1-m monochromator (Monospek 1000, Rank Hilger), which was employed with a slit height of 3 mm and a slit width of 35 pm (reciprocal linear dispersion 0.8 nm mm-l). The signal output was to a Servoscribe chart recorder (Model RE 511.20; Smiths Industries Ltd.) with a response time of 1 s for full-scale deflection, a Series 3000 fast-response chart recorder (Oxford Instrument Co.Ltd.) (0.25 s for full-scale deflection), or a digital storage oscilloscope (Trio MS.1650). All standard solutions were prepared in 1% nitric acid and dispensed with a micropipette. Experimental Initial technique development was carried out without the usual central argon injector flow. Whereas it may prove desirable to include an injector flow for optimum sample containment and decreased background emission, the insertion apparatus was thereby simplified as thereKIRKBRIGHT AND WALTON 277 was no chance of extinguishing the plasma by air entrainment in the injector argon. Plasma initiation was best effected by initially closing the injector tube; thereafter the plasma could be operated normally with the injector tube open. In operation the graphite rod was simply raised manually until the stop fixed its final position in the plasma.The rod was then seen to glow within a few seconds. Initial attempts to desolvate aqueous solutions using direct insertion of the rod into the plasma were unsuccessful. The evaporation was difficult to control under these conditions; it is important that evaporation is complete, otherwise the plasma may be disturbed or even extinguished at the insertion stage. Aqueous samples were therefore applied to the rod and dried by strongly heating the rod below the sample with a heat gun (R. S. Components 545-137). As the rod was inserted, there was little re-tuning of the automatic matching network and the reflected power remained less than 3 W. There was an immediate visible diminution of the plasma intensity as power was coupled into the rod, accompanied by a fall in the recorder base line.Typical peak shapes are shown in Figs. 1 and 2 , demonstrating that the background - - * - 10 s 0.5 nn Ni J, 1 7 s L los’ - 10 s 0.5 na Cd 5.0 nn Fe 0.5 ng Pb 20 mVf.s.d. 50 m i f.s.d. 50 mV-f.s.d. 20 mVf.s.d. Fig. 1. Typical signal versus time relationships obtained for nickel, cadmium, iron and lead with direct sample introdution on a graphite rod. Fig. 2. Typical peak utrsus time relationship for cobalt, copper and manganese with direct sample intro- duction.278 KIRKBRIGHT AND WALTON: OES - ICP WITH DIRECT Analyst, VOZ. 107 change is sufficiently well resolved from the atomic peak (with the possible exception of cad- mium) and, as expected, the background change was less dramatic at lower wavelengths.In several instances it is apparent that the background after the atomic signal from the analyte does not return to its original level; prolonged heating does not change this and the effect may be attributable to change in rod position or physical condition after heating. The effect of rod shape was investigated by progressively increasing the depth of a 2 mm diameter hole in the end of the rod and measuring the peak heights from 5 pl of 50 ng ml-1 manganese solution at constant rod height and other plasma parameters. As shown in Fig. 3, the best response occurred with a flat-ended rod and subsequent measurements were made with this. This limited the acceptable solution volume to 5 pl, which results in the necessity for careful addition and drying.Y I I I a : 3; 1 2 3 Recess depth/mrn Fig. 3. Variation of peak signal intensity at 257.6nm for 0.25ng of manganese with depth of recess in rod. Optimisation of conditions was studied for the determination of manganese by simplex optimisation as described by Ebdon et aZ.,8 using five parameters: power (defined as the forward power less the reflected power), observation height, rod height, plasma flow and auxiliary flow; these parameters were varied to optimise the signal to background ratio (SBR). The simplex was terminated after 18 moves and univariate searches were carried out on all five para- meters (Figs. 4-8). The SBR was taken as the gross signal to background ratio obtained with the rod inserted. The manganese concentration employed was 100 ng ml-1 (Le., 0.5 ng).20 16 a m * 12 4 ' ' I t I I I I 600 700 800 900 1 OCO 1 100 1 200 1 300 PowerMl Fig. 4. Univariate search for optimum power of operation (observation height, 19 mm; rod height, 4 mm; plasma gas flow- rate, 16 1 min-l; auxiliary gas flow-rate, 0.9 1 min-l) for manganese at 257.6 nm. 20 15 a m * 10 ~~~ 5 10 15 20 25 Observation height/mm Univariate search for optimum observa- tion height above top of coil (power 1000 W; rod height, 4 mm; plasma gas flow-rate, 16 1 min-l; auxiliary gas flow-rate, 0.9 1 min-I) for manganese at 257.6 nm. Fig. 5.March, 1982 SAMPLE INTRODUCTION FROM A GRAPHITE ROD 279 20 a15 m v) 10 1 1 I -5 0 +5 Rod position relative to top of coilimm Fig. 6. Univariate search for optimum rod position above coil (power, 1000 W; observation height, 19 mm above coil; plasma gas flow-rate, 16 1 min-l; auxili- ary gas flow-rate, 0.9 1 min-l) for man- ganese at 257.6 nm.*O 1 10' ' I I I 12 14 16 18 20 Plasma gas flow-ratell min-' Fig. 7. Univariate search or optimum plasma gas flow-rate (power, 1 000 W; observation height, 19 mm; rod height, 4 mm; auxiliary gas flow-rate, 0.9 1 min-') for manganese a t 257.6 nm. A lateral scan was also carried out to ensure observation of the area of plasma which pro- duced best SBR (Fig. 9). It appears that the region of optimum SBR is fairly closely defined even in the absence of an injector flow, although there is some detectable emission at least 2-3 mm from the axis. The slightly skewed distribution is presumably due to torch asym- metry.The effect of an injector flow was also investigated at the other previously determined conditions, by fitting a side-arm to the injector tube to allow introduction of argon, and a rubber seal to prevent the injector gas flowing downwards with the rod inserted. Fig. 10 shows that the optimum SBR occurred at a low flow-rate (about 0.2 1 min-l). This flow-rate 22 I I , I 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Auxiliary gas flow-rate/l min-' 1 6 ' ' Fig. 8. Univariate search for optimum auxiliary gas flow-rate (power, 1000 W; observation height, 19 mm; rod height, 4 mm; plasma gas flow-rate, 16 1 min-l) for manganese at 257.6 nm. Lateral position/mm Fig. 9. Variation of signal to background ratio with lateral viewing position in the plasma tail flame (viewing height 19 mm) for manganese at 257.6 nm.280 5O KIRKBRIGHT AND WALTON: OES - ICP WITH DIRECT Analyst, VOZ.I07 0:l 0'.2 0:3 0.; 0:s 0:6 0:7 0:8 0.9 25 20, CT v, 15 m 10 0------- -.- --0----- '. --- -- --- , ~~ _ _ _ _ _ _ ~ _ ~ _ _ _ ~ _ ~ ~ _ ~ .~~~---~-- - _ _ __________ _ _ _ observed for molybdenum ; presumably insufficient power was available. Results Detection limits were determined for cadmium, cobalt, copper, iron, nickel, lead and mangan- ese under the conditions used for manganese (power 1000 W, observation height 19 mm, rod height 14 mm, plasma gas flow-rate 16 1 min-l, auxiliary gas flow-rate.0.9 1 min-l). These are y_^u ,, *_ -*_ - I--- -_ ---- . ----- -- ------- .. --- ------------- *--*-* vu*vu*u-*.,~* .,* **-- * " I U C . A . v U C U A A dard deviation obtained for ten samples of each of these elements in 5-pl aliquots of sample in 1% nitric acid.The detection limit was then calculated as that concentration which was equivalent to twice this relative standard deviation. The precision attainable by this technique is dependent on the physical condition of the graphite rod employed. A tendency for peak broadening and consequently a lower peak height was observed as the rod aged in use. Even with peak-area measurement, the precision degraded significantly for manganese at the 0.5-pg level as the number of determinations increased (as shown in Table 11). TABLE I DETECTION LIMITS OBTAINED FOR DIRECT GRAPHITE ROD SAMPLE INSERTION DEVICE Element Cadmium . . Cobalt .. Copper . . Iron . . .. Manganese .. Nickel .. Lead . . .. Wavelength/nm .. 228.80 .. 345.35 .. 324.75 . . 371.99 .. 257.61 . . 341.48 .. 283.31 Detection limit (20 blank)/ ng ml-l (5 pl) 3.9 4.1 4.0 7.8 1.2 12.7 7.0 Detection limit/ Pg 19.5 20.5 20 39 6 63.5 35March, 1982 SAMPLE INTRODUCTION FROM A GRAPHITE ROD TABLE I1 28 1 VARIATION OF RELATIVE STANDARD DEVIATION OBTAINED FOR 5-pI SAMPLES OF MANGANESE (0.5 ng) AT 257.6 nm WITH INCREASING NUMBER OF DETERMINATIONS USING THE SAME ROD No. of determinations Relative standard deviation, % 10 0.05 15 0.05 20 0.067 25 0.10 30 0.11 Rod wear was, however, not obvious until more than 50 insertions had been effected. Eventually erosion was seen to occur at the junction of the rod with the base of the plasma. Some preliminary investigations have been carried out using a single loop of 0.25-mm tungsten wire.This had a much faster rate of heating, e.g., a 0.5-ng manganese insertion gave a peak with a rise time of 25 ms and it was found essential to use a storage oscilloscope for accurate recording. Under the conditions determined previously (which may not be optimum for a tungsten filament) manganese gave a detection limit of 7.7 ng ml-l. Considerable blank variability was experienced and contributed to the poorer detection limit. Additionally, embrittlement shortened the life of the wire. Conclusion Direct insertion of samples into a continuously running low-power plasma shows promise as a technique for the determination of trace elements in small liquid samples. The sensitivity, particularly for volatile elements, compares favourably with electrothermal atomisation with similar equipment2 ; optimisation of individual elements, electronic integration and, in some instances, the use of more sensitive analytical lines would be expected to improve it.In addition, interferences due to variations in sample transport efficiency, which are known to occur in electrothermal atomi~ation,~ should be absent. Future developments are expected to follow electrothermal atomisation improvements. It may be possible to utilise the plasma to carry out a programmed sample ashing stage prior to analyte vaporisation. The use of pyrolitic graphite and a higher operating power should assist in the determination of refractory elements. The system should be amenable to automa- tion of both sample addition to the rod and rod insertion into the plasma; this should lead to an improvement in precision. Direct solid analysis is to be investigated and will require only minor modification to the insertion device. We thank Dr. M. J. Cope for his assistance in the early stages of this work. 1. 2. 3. 4. 5. 6. 7. 8. 9. References Nixon, D. E., Fassel, V. A., and Kniseley, R. N., Anal. Chem., 1974, 46, 210. Gunn, A. M., Millard, D. L., and Kirkbright, G. F., Analyst, 1978, 103, 1066. Human, H. G. C., Scott, R. H., Oakes, A. R., and West, C. D., Analyst, 1976, 101, 265. Thompson, M., Goulter, J . E., and Sieper, F., Analyst, 1981, 106, 32. Salin, E. D., and Horlick, G., Anal. Chem., 1979, 51, 2284. Sommer, D., and Ohls, K., Fresenius 2. Anal. Chem., 1980, 304, 97. Bombelka, R.. PhD Thesis, University of London, 1978. Ebdon, L., Cave, M. R., andMowthorpe, D. J., Anal. Chim. Acta, 1980, 115, 179. Millard, D. L., Shan, H. C., and Kirkbright, G. F., Analyst, 1980, 105, 502. Received August loth, 1981 Accepted October 21st, 1981

ISSN:0003-2654    

DOI:10.1039/AN9820700276

出版商:RSC    

年代:1982

数据来源: RSC

摘要:

282 Analyst, March, 1982, Vol. 107, pp. 282-287 2-Thiobarbituric Acid as a Reagent for the Determination of Bismuth( 111) by Normal and Derivative Spectrop hotometry Basilio Morelli Istituto di Chimica Analitica, Universitd di Bari, Via A mendola, 173, 70126-Bari, Italy A spectrophotometric study of the reaction of 2-thiobarbituric acid (TBA ; 4,6-dihydroxy-2-mercaptopynmidine) with bismuth( 111) is presented. Bismuth(II1) forms a coloured complex with an absorbance maximum a t 416 nm with TBA in acid solution (pH 1.2-1.7). The maximum absorbance value is achieved within about 25 min. Beer’s law is obeyed for a bismuth concentration of up to 10.2 p g ml-l. The molar absorptivity of the complex is 2.4 x lo* 1 mol-1 cm-l a t 416 nm and the sensitivity of the reaction is, according to Sandell’s calculation, 8.7 x pg cm-2 per 0.001 absorbance unit. Cobalt (11), nickel(II), magnesium(I1) , tin (I I), lead (11) , zinc(II), barium(II), manganese(I1) and aluminium(II1) do not interfere in the deter- mination. The same procedure can also be used for the spectrophotometric determination of copper (11) , which seriously interferes in the bismuth deter- mination. A satisfactory method for the determination of bismuth in the presence of copper by derivative spectrophotometry is, however, reported.Keywords : Bismuth(III) determination ; 2-thiobarbituric acid ; derivative spectrop hotometry Some derivatives of barbituric acid and thiobarbituric acid (TBA ; 4,6-dihydroxy-2- mercaptopyrimidine) have been used for analytical purposes (other than for clinical and pharmacological applications), e.g., barbituric acid with pyridinel or y-picoline,2 and sodium barbiturate with sodium isonicotinate3 have been used in the spectrophotometric deter- mination of cyanide.2-Thiobarbituric acid, because of its reactions with aldehydes to form coloured compounds, has been utilised in the detection of aldehydes in photographic gelatine~,~ in chromatography5 and in determining the degree of oxidation of natural fats (e.g., in milks,’). Furthermore, TBA forms a coloured complex with rutheniumS (which has a maximum absorbance at 570 nm) and with iron9 (maximum absorbance at 380 nm) and reacts with bismuth10 and copper11 in strongly acidic solution to form complexes (3: 1, maximum absorbance 390 nm, and 4: 1, maximum absorbance 400 nm, respectively), which can be utilised in the spectrophotometric determination of these elements.More recently, the possibility of reactions between some divalent ions and unsubstituted TBA has been studied by polarographic techniques.12 In particular, the formation of two copper(I1) - TBA complexes (1: 1 and 1:2) in alkaline medium was suggested by ampero- metric titrations. The existence of a 1 : 2 copper(I1) - TBA coloured complex in 1 M sodium acetate solution at pH 8 was confirmed spectrophotometrically by a Job’s plotl2: the complex exhibited a very broad absorbance maximum at 690 nm but it was recommended that the absorbance was measured within 1 min of mixing the reagents owing to the rapid appearance of a blue precipitate in the solution, which is a disadvantage in the spectrophotometric determination of the copper(I1).This paper reports a method for the spectrophotometric determination of bismuth(II1) in the concentration range 1 .O-10.2 pg ml-l using TBA in acid solution (pH 1.2-1.7). The same method is also suitable for determining copper(I1) (0.3-12.4 pg d-l) at pH 1.5-6.5, i.e., in a larger pH range than other methods. In addition, a method is presented for determining bismuth(II1) in the presence of copper(1I) by first and second derivative spectrophotometry. The composition of the complex, the influence of the experimental conditions and the effect of foreign ions are discussed. Experimental Reagents Bismutlz(ll1) standard solutions. An approximately 0.1 M solution was prepared andMORELLI 283 standardised with EDTA in the usual way.A working standard solution, 2.7 x M , was obtained by diluting with water and the pH was adjusted to the desired value with nitric acid. CoPper(I1) standard solution. A working standard solution, 2.7 x M, was prepared as for bismuth. The pH was adjusted to the desired value. 2-Thiobarbituric acid solution. A 1.35 x 10-2 M solution of TBA was prepared by dissolving, with stirring, a suitable amount of the solid in 100 ml of distilled water; as the TBA is difficult to dissolve in water, about 0.3 ml of 4 M sodium hydroxide solution was added to speed up the process. The pH was adjusted to the required value with hydrochloric acid. Apparatus 1-cm silica cells. All absorbance measurements were made on a Perkin-Elmer 555 spectrophotometer using The pH of solutions were measured on an Orion Research Ionalyzer 901.Procedure for Determination of Bismuth( 111) A 4.5 ml aliquot of the 1.35 x 10-2 M TBA solution was placed in a 5-ml calibrated flask; a few mkrolitres of the 2.7 x 10-3 M bismuth(II1) standard solution were added and the resulting solution was made up to volume with distilled water. After standing for about 30 min, the absorbance was measured at 416 nm against a reagent blank. Blanks were prepared in a similar manner by transferring 4.5 ml of the 1.35 x M TBA solution into a 5-ml calibrated flask and diluting to volume with water : the resulting solutions were 1.21 x M in TBA. A bsorption spectra The absorption spectrum of the bismuth - TBA complex, obtained by following the described procedure, shows a broad maximum at 416 nm.The absorption spectrum of TBA shows two maxima at 343 and 357 nm; because the reagent absorbs quite significantly at the wavelength at which the absorbance of the complex is meas- ured, it seems advisable to use the same concentration of TBA in the sample under test and in the reagent blank. By operating under conditions of pseudo first-order reaction (large excess of TBA), the results were not affected by the fact that bismuth consumes some of the reagent. Efect of pH The effect of pH on the absorbance of the resulting solutions after mixing the reagents was examined between 550 and 370 nm. From a series of measurements in the pH range 1.2-1.7, the absorbance appeared constant. At higher pH, bismuth( 111) solutions slowly hydrolysed.Efect of time on colour development At room temperature (approximately 18-25 "C) the yellow colour of the bismuth - TBA complex begins to develop immediately after mixing and the maximum absorbance is achieved in 25 min; the absorbance then remained constant for at least 20 min. The reaction time at 18-25 "C was thus established as being about 25 min. These results were obtained from graphs of absorbance of the bismuth - TBA complex measured as function of time after mixing of the reagents. With the more concentrated solutions, the appearance of a finely divided precipitate may occur. Stability of the reagent with graphs obtained using the fresh reagent. Calibration graphs recorded with TBA stored at room temperature for 2448 h agreed well Calibration graphs By following the described procedure, the calibration graphs for the determination of bis- muth were obtained under optimum conditions.The system obeys Beer's law in the range of concentrations tested, i.e., up to a concentration of 10.2 pg ml-l of bismuth. In more con- centrated solutions a finely divided precipitate may appear a few minutes after sample prepara- t ion. The molar absorptivity of the bismuth-TBA complex at 416 nm was calculated to be 2.4 x lo4 1 mol-l cm-l and, according to Sandell's calculation, the sensitivity of the reaction for bismuth was estimated as being 8.7 x pg cm-2 per 0.001 absorbance unit.284 MORELLI : 2-THIOBARBITURIC ACID FOR DETERMINATION OF Analyst, VOl. 107 To test the reproducibility of the method, repeated experiments were carried out on five solutions containing 6.3 pg of bismuth per 5 ml: from the results obtained a standard deviation of 0.0038 in the absorbance was measured.Composition of the complex The reaction of the formation of the bismuth - TBA complex was studied at various pH values by the method of continuous variation. From the spectrophotometric data, the molar ratio of bismuth(II1) to TBA, at pH 1.2-1.7, was found to be 1:2. A typical Job's plot at pH 1.5 is shown in Fig. 1 : the total concentration of bismuth and TBA was 1.2 x 10-3 moll-1 and the measurements were made against water. 0 0.1 0.3 0.5 [Bi3+] lei3+] + [TBAI Fig. 1. Molar ratio of Bi - TBA complex by the method of con- tinuous variations. [Bi(III)] + [TBA] = 1.2 x 10-3M; pH = 1.5; 416 nm; reference, water.Efect of foreign ions The influence of foreign ions on the determination of bismuth was studied under optimum conditions using samples of bismuth that also contained certain amounts of the diverse ions. The analytical results for a series of determinations are given in Table I. Cobalt(II), nickel(II), magnesium(II), tin(II), lead(II), zinc(II), barium(II), manganese(I1) and alumi- nium(II1) did not interfere. With chromium(II1) and cadmium(I1) the formation of a finely divided precipitate was observed ; centrifuging followed by filtration to remove the precipitate was inefficacious because it resulted in a noticeable diminution in the colour intensity of the TABLE I THE PRESENCE OF FOREIGN IONS AT pH 1.5 DETERMINATION OF BISMUTH(II1) USING 2-THIOBARBITURIC ACID I N Foreign ion CO/II) .. . . Ni(II\ . . .. Mg(JI\ . . .. SnUI) .. .. P b W . . .. Zn(1I) . . .. Ba(I1) . . .. Mn(1I) .. .. Al(II1) . . .. Hg(II\ . . .. FeUII) . . .. Amount added 34 35 35 15 17 33 34 15 16 20 16 per 6 ml/M Form added C1- s0,a- sop C1- NO,- c1- Cl- SO,S- so,2- Cl- c1- Bi taken Per 5 ml//# 39.5 40.1 38.8 39.5 38.1 36 2 42.8 32 9 40.1 26 3 46.0 Bi found/pg 39.3 40.3 38.9 39.4 38 3 36.4 42.5 33.1 40.2 31.5 64 4 Relative error, yo -0.5 +0.5 t 0 . 2 - 0.2 t 0 . 5 + 0.5 -0.7 + O 6 + 0.2 + 19.8 4-40 0March, 1982 BI (111) BY NORMAL AND DERIVATIVE SPECTROPHOTOMETRY 285 solution, presumably owing to adsorption of the complexes on the precipitate, which intro- duced large negative errors. For this reason, a preliminary stage to remove the chromium(II1) and cadmium(I1) ions by physico-chemical methods, before the TBA is added, is required.Mercury(I1) and iron(II1) interfered ; some attempts were made to overcome these interferences (e.g., by using EDTA, acetate, tartrate or oxalate as masking agents) but none was success- ful. Determination of Bismuth( 111) in the Presence of Copper(I1) Significant difficulties arise in the spectrophotometric determination of bismuth( 111) using TBA in the presence of copper(I1) because of a large overlap of the spectra of the bismuth - TBA and copper - TBA complexes : by following the same procedure as described previously for bismuth, TBA has also shown favourable colorimetric properties as a reagent for the deter- mination of copper(I1).At pH 1.5-6.5, copper(I1) forms a 1 : 2 coloured complex with an absorp- tion maximum at 381 nm. The composition of the copper - TBA complex was investigated, at various pHs, by the molar ratio method with samples containing 1.08 x M TBA and increasing amounts of copper( 11) as required ; absorbance measurements were made against water. At 18-25 "C the maximum absorbance value of the complex is achieved within about 12 min and then remains constant for at least 90 min. Beer's law is obeyed over the whole range of concentrations tested, i.e., 0.3-12.4 pg ml-l of copper; from repeated experiments on five solutions containing 6.9 pg of copper per 5 ml, a standard deviation of 0.003 1 in the absorb- ance of the copper complex was measured. The molar absorptivity of the copper - TBA com- plex at 381 nm was calculated as being 6.5 x lo3 1 mol-l cm-1 and the sensitivity of reaction for copper is, according to Sandell's calculation, 9.8 x pug cm-2 per 0.001 absorbance unit.Copper( 11) seriously interferes in the determination of bismuth(III), so preliminary separa- tion procedures are required. For reasons of speed and simplicity, a more direct approach would be preferable : derivative spectr~photometryl~~~~ allows the direct determination of bismuth using TBA in the presence of copper (in the same concentration range as reported for determining bismuth alone). In derivative spectrophotometry the first or higher derivative of an absorption spectrum is recorded versw wavelength to enhance the detectability and the measurement of minor spectral features, such as weak shoulders, and it is therefore possible to follow subtle changes in a spectrum.In general, this technique can be used profitably in a situation where a background spectrum overlaps the band of interest. A typical absorption spectrum (zero derivative) of a mixture of complexes (16.9 pg of bismuth per 5 ml plus 17.2 pg of copper per 5 ml) is shown in Fig. 2. The spectrum presents a weak shoulder corresponding to the absorption maximum of the bismuth - TBA complex. The sample was prepared using the same procedure as described for determining bismuth alone. The methods most commonly used to measure the analyte concentration from a normal spectrum (zero derivative), i.e., measurement of the total absorbance at the analyte maximum or tangential base-line approximation to correct partially for overlap of the analyte spectral band by the interfering band, are not suitable for the determination of bismuth because the shoulder is very flat and, consequently, the measurements to be taken are not uniquely defined.A noticeable enhancement of this bismuth spectral feature is obtained by means of a derivative technique. Fig. 3(a) and (b) show the respective first and second derivatives of the spectrum in Fig. 2, from which it is possible to take measurements, proportional to the bismuth concentration, to prepare analytical working graphs. Obviously, the same measurements can be taken for the standards as for the analytical samples. The type of graphical measurements taken in actual experiments are illustrated in Fig.3, ie., tangential base-line measurements (hJ and peak-to- peak measurements (h2). A typical example of experimental working graphs obtained from h, and h, measurements, relevant to standards containing 17.2 pg of copper per 5 ml and increas- ing amounts of bismuth, are shown in Fig. 4. The method gives linear graphs of height measurements versus bismuth concentration with an intercept of zero. The results from a typical series of determinations on art&ial mixtures are reported in Table 11. Both first and second derivative methods give results with a high accuracy and precision.286 540 380 460 -0.06 0.6 al & 0.4 e s 2 0.2 -0.002 MORELLI : 2-THIOBARBITURIC ACID FOR DETERMINATION OF A rtalyst, YOZ. I07 i-0.06 a .- .I- > '0 .- b o z ii .I- 0 380 460 54 Fig.2. Undifferentiated sum spectrum of a mixture of Bi - TBA and Cu - TBA complexes. Bismuth(II1) concentration, 16.9 pg per 5 ml; copper(I1) concentration. 17.2 LLR per 6 ml; Wavelengthlnm - t o.ob2 a) 1 b = o 8 .I- > I- v) 420 500 reference, reagent biaili (1.21 x lo-% M TBA). Fig. 3. (a) First and (b) second derivative of the spectrum in Fig. 2. The only observed failing of the second derivative method compared with the first one is, as is expected, a small decrease in the signal to noise ratio; a large decrease in this ratio with the process of differentiation is a common failing of the higher derivative spectra compared with the undifferentiated spectra: generally, a factor of two for every order of differentiation is observed and one would not expect derivative spectrophotometry to be very helpful if, in the undifferentiated spectrum, a poor signal to noise ratio is already apparent.To confirm that a spectral feature observed in a differentiated spectrum is not a random noise event, it is useful to record the derivative spectrum several timed* In practice, in the determination of bismuth, this inconvenience would become necessary only for very dilute solutions. 1 I 0 1 3 5 [Bi3+l/yg ml-' Fig. 4. Analytical working graphs for the determination of bismuth(II1) in the presence of copper(I1) by first (h,) and second (k,) derivative spectrometry.March, 1982 BI(III) BY NORMAL AND DERIVATIVE SPECTROPHOTOMETRY TABLE I1 DETERMINATION OF BISMUTH( 111) BY FIRST AND SECOND DERIVATIVE SPECTROPHOTOMETRY I N THE PRESENCE OF COPPER 287 The amount of copper present is 17.2 pg per 5 ml and each value is the mean of three determinations. Bi found by Bi found by Bi taken 1st derivative Relative 2nd derivative Relative per 5 ml/pg method/pg error, 76 methodp/g error, % 18.0 18.0 0.0 18.2 + 1.1 22.6 22.5 - 0.4 22.6 0.0 28.2 28.1 -0.3 28.1 - 0.3 30.0 30.0 0.0 30.1 + 0.3 Conclusions The method described gives results of high precision and accuracy for the entire range of concentrations.tested and, with the added advantage of speed and simplicity, represents a good alternative to other methods. In particular, the determination of bismuth in the presence of copper( 11) by derivative spectrophotometry, without prior separation procedures, could be useful in routine analytical work. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. References Asmus, E., and Garschagen, H., Fresenius 2. Anal. Chem., 1953, 138, 414. Nagashima, S., Anal. Chim. Acla, 1977, 91, 303. Nagashima, S., Anal. Chim. Acta, 1978, 99, 197. Pomadier, J. M., and Venet, A. M., Sci. Ind. Photogr., 1952, 23, 303. Saslaw, L. D., and Waravdekar, V. S., Arch. Biochem. Biophys., 1960, 90, 245. Wertheim, H. J., and Proctor, B. E., J. Dairy Sci., 1956, 39, 391. King, R. L., J. Dairy Sci., 1962, 45, 1165. Knight, S. B., Parks, R. L., Leidt, S. C., and Parks, K. L., Anal. Chem., 1957, 29, 571. Sikorska-Tomicka, H., Fresenkus 2. Anal. Chem., 1968, 234, 414. Sikorska-Tomicka, H., Mikrochim. A d a , 1969, 715. Sikorska-Tomicka, H., MiRrochim. Acta, 1969, 718. Murphy, R. J., and Svehla. G., Anal. Chim. Ada, 1978, 99, 115. O'Haver, T. C.. and Green, G. L., Anal. Chem., 1976, 48, 312. O'Haver, T. C., Anal. Chem., 1979, 51, 91A. Received July 24th, 1981 Accepted September 21st, 1981

ISSN:0003-2654    

DOI:10.1039/AN9820700282

出版商:RSC    

年代:1982

数据来源: RSC