摘要:

NOVEMBER, 1969 THE ANALYST Vol. 94, No. I124 The Spectrographic Analysis of Steelmaking Slag by a. Compaction Method of Sample Preparation BY P. S. BRAMHALL AND P. H. SCHOLES (BISRA, The Inter-Group Laboratories of the British Steel Corporation, Hoyle Street, Shefield, S3 7EY) An analytical process for the analysis of graphite-based compacts of powdered slag has been examined. A standardised procedure is proposed for the preparation of compacts and operation of the direct-reading emission spectrometer. The method can easily be applied by staff already familiar with the use of the instrument for steel analysis. Standardised samples of slag have been used to calibrate the instrument and for the purpose of measuring analytical performance. SLAGS originating during the steelmaking process are multi-component oxide mixtures of widely varying compositions. Materials of this type present a challenge to the physical- methods analyst in the application of instrumental techniques such as X-ray fluorescenq and emission spectroscopy.X-ray fluorescence has the advantage of high precision but inter-element effects are severe and can lead to the attenuation or the enhancement of element intensity measure- ments. Such effects obviously detract from the technique when it is applied to slags of variable matrices. There is, however, a wide literature of reported success when dealing with certahi materials, such as ironmaking sinter and blast-furnace slag with oxide concentrations that vary only within restricted ranges. Current developments in direct electron excitation and the use of computing devices to correct interference effects may lead to significant extension in the future application of X-ray fluorescence to the analysis of non-metallic materials.For certain applications, emission spectroscopy may be considered to be more suitable than X-ray fluorescence, particularly in laboratories that handle slag samples, which are markedly different in composition. The analysis of slags by emission spectroscopy is, however, complicated by the heterogeneous nature of the material and the extremely high concentra- tions of certain components, such as calcium oxide, that may be present in amounts up to 60 per cent. Careful sample preparation, together with some form of dilution, is essential. Techniques used for the analysis of non-metallic materials by emission spectroscopy are well known and can be summarised as follows.(a) Impregnated electrode-Alkali fusion of the sample followed by extraction of the me€$ into dilute acid to provide a solution that is absorbed on to a static or rotating graphite electrode. (b) Tape machine-Alkali fusion followed by crushing and introduction into the spectro- graphic discharge on a moving band of cellulose tape. (c) Fused briqz,jette-Alkali fusion followed by crushing and mixing with graphite to produce an electrically conducting powder for compaction into a flat disc. (a) Direct com$actiolz--Mixing of the powdered sample with graphite followed by corn- pacting to form either a flat disc or a cylindrical briquette. In each case an internal reference element is added during sample preparation.0 SAC and the authors. 946946 3RAMHALL AND SCHOLES: SPECTROGRAPHIC ANALYSIS OF STEELMAKING [A%a&st, vol. 94 COLLABORATIVE WORK- During the period 1962 to 1965, each of the sample preparation procedures outlined above was examined by members of the BISRA Non-Metallic Materials Analysis Sub- C0mmittee.l It was concluded that the impregnated electrode method (a), used initially by Mason,2 was the most flexible and particularly suitable for post-mortem analysis. The two fusion methods ( b ) and (c), used with direct-reading instruments, were shown to be suitable for production control with analysis times of 15 to 20 minutes. Reproducibility of the fusion methods was high, but this was offset by systematic errors in the day-to-day calibration (“setting up”) procedures.Calibration difficulties were eventually overcome, thus enabling most members to analyse five standard slags with results that compared satis- factorily with those obtained by classical and other modern method^.^ The direct compaction method ( d ) was designed for production control analysis of slags and ironmaking sinters496 of restricted compositional ranges and, in consequence, the member operating this technique was unable to participate in the comparative assessment programme. In further collaborative work,6 a careful investigation was made into interference effects in direct-reading spectrometric analysis with preparation procedures (a), (c) and (d), and synthetic slag samples prepared by Johnson.’ It was found that direct compaction was unsatisfactory and pre-fusion of the slag with an alkali flux improved precision and produced more consistent analytical results for the particular operating conditions used by Group members.The degree of interference decreased with increasing dilution of the sample and was virtually eliminated by the use of a flux-to-sample ratio of 26 : 1. Following the collaborative trials, several steelworks laboratories confirmed their choice of one or other of the fusion methods in preference to direct compaction for sample prepara- tion. Successful operation on a production control basis of the fused-briquette and tape- machine methods with Applied Research Laboratories Ltd. spectrometers was first reported (in unpublished work) by Croall of the Guest, Keen and Nettlefold Group, and for about 4 to 5 years his colleagues, Pugh, Straker and Wright, now at three British Steel Corporation pfants, have between them regularly analysed several hundred slag samples per week obtaining results similar in quality to those described in collaborative w0rk.l ,3 Rotrode techniques are also used with a high degree of success for post-mortem analysis in several other steelworks and research laboratories in the British steel industry. DIRECT COMPACTION METHOD- In the selection of a sample preparation method for use with the Polyvac spectrometer, we have re-examined the possible application of the direct-compaction method.There was little published work available on optimising conditions for the compaction of powdered slag and graphite, and it seemed probable that the unfavourable experiences reported by earlier workers could be improved by careful operation and the use of alternative electrical discharge conditions.If successful, the direct approach would be simple to operate and preparation would be complete in two major operations (mixing by grinding and compaction) instead of the three (fusion, mixing by grinding and Compaction) required for the fused- briquette method. It would also be more suitable for an eventual automatic scheme that would reduce handling and analysis time. The evolution of the direct-compaction method extends back about 10 years. In 1958, the Jarrell-Ash Company published a papers indicating that a simple direct method for slags, with their Atomcounter direct-reading spectrometer, produced results that were accurate and precise, but when Croall and Brown visited the Company in 1960 the results obtained were disappointing.This partly explains why, until fairly recently, the direct compaction method has been largely neglected in the U.K. In 1961, Ramsden and Fowler of Applied Research Laboratories Ltd. and Gale of the Steel Company of Wales (in unpublished works) proposed an “Argon Briquette’’ method for use with the Quantovac, but neither organisation pursued the method because of difficulties in preparation. Mason was more successful and in 1965 he described two production control method^^,^ for use in the Spencer Works of Richard, Thomas and Baldwin Ltd. A high voltage a.c. spark discharge was used, and sodium tetraborate was added as a buffer to the graphite-based mixture to control spectrographic emission and improve analytical reproducibility.Mason experienced difficulties arising from reported “variation in pellet emission,” which made mathematical corrections necessary. It seems. probable that these analytical variations may have been caused by variable emissionNovember, 19691 SLAG BY A COMPACTION METHOD OF SAMPLE PREPARATION 947 from the internal reference element (cobalt) because of incomplete mixing of the various components before compaction. A direct-compaction method was published in 1964 by Dickens, Koenig and Dippelg designed for the Polyvac spectrometer with nickel as reference element and with a triggered low voltage d.c.capacitor discharge. In this method the addition of a buffer was omitted and there were no reported analytical variations arising from mixing difficulties. Automatic weighing and pressing equipment was used so that the analysis of open-hearth slag for pro- duction control purposes could be completed within a maximum time of 10 minutes. A similar process has also been used by SchmidtlO for the control analysis of LD converter- furnace slags, blast-furnace slags and ironmaking sinter, except that in this case cobalt was substituted for nickel as reference element. In more recent unpublished work Schmidt has proposed that the sample preparation procedure can be simplified for routine production control by omitting the addition of a reference element andby making constant time exposures.Published work on the use of the direct-compaction method for the spectrographic analysis of slag is conflicting. Clearly, the procedure adopted for preparation of the graphite- based mixture before compaction is of decisive importance. Other factors of major im- portance are electrical discharge conditions, choice of reference element, the pressure of compaction and the selection of emission line wavelengths to minimise spectral interference effects. In this report certain of these factors have been critically examined and, from a greater understanding of their influence, it has been possible to formulate a direct-compaction method that gives a high level of analytical performance. Support is given to the work of Dickens, Koenig and Dippels and Schmidt,lo and the unpublished work of Borrowdale and Kidman of the B.S.C.Midland Group, all of whom use similar methods and equipment to that described in this report. EXPERIMENTAL GEK'ERAL REQUIREMENTS- It would be an advantage in the spectrographic analysis of slag if the analytical procedure were suitable for operation with a direct-reading spectrometer already engaged on the analysis of steel. Any adjustments to the instrument for change-over to slag analysis and vice versa must be rapid and simple to perform. Precautions must obviously be taken to avoid con- tamination from the graphite-based compact particularly in the subsequent determination of carbon in steel. With these considerations in mind, experiments were carried out to establish optimum conditions for sample preparation and the operation of the instrument.ELECTRICAL DISCHARGE COXDITIONS- Previous workersg *lo have shown that discharge parameters similar to those given below are satisfactory for the analysis of compacted non-metallic materials with the Polyvac direct- reading spectrometer. These conditions, based on a thyratron-triggered capacitor discharge, are the same as those employed by most operators using this instrument for the analysis of steel so that change-over procedure is simplified. The conditions are peak capacitor voltage, 550 V; inductance, 0.06 mH; capacity, 20 pF; resistance, 3 s1; sample polarity, negative; analytical gap, 5 mm; and counter electrode, silver, a 5 mm diameter rod with a 90" cone tip. Argon flow : quiescent state, 0.5 litre minute-l; pre-excitation flush, 8 litres minute-l for 10 seconds; and excitation, 2 litres minute-l.Tests were made to determine the optimum pre-integration time to attain stable instru- ment responses for the measurement of individual elements. Examination of chart recordings monitoring emission intensity for each element against time showed that a period of 30 seconds was the minimum time required for element emission to reach a reasonably constant value, suitable for integration. A typical recording for magnesium is shown in Fig. 1, and recordings for calcium, silicon, phosphorus, manganese and iron followed the same pattern. The emission phenomena of aluminium proved to be an exception (see Fig. 1) with intensities increasing over the 30-second period.A 30-second pre-integration period was adopted for the recommended procedure in combination with an integration period of about 20 seconds. Similar times were recommended by Dickens, Koenig and D i ~ p e l , ~ and Schmidt.lo948 BRAMHALL AND SCHOLES : SPECTROGRAPHIC ANALYSIS OF STEELMAKING [A?ta&St, VOl. 94 Time, seconds Fig. 1. Typical emission response of aluminium and magnesium CHOICE OF INTERNAL REFERENCE ELEMENT- There are three main requirements governing the choice of internal reference element to be added during preparation of the sample compact. (i) The element should not introduce spectral interference at the line wavelengths used in the analysis procedure. (ii) The element should not be present in the test samples at a resolvable concentration level.(iii) The element, whether as a metal or in the form of a compound, must be of high purity and sufficiently stable to enable standardised additions to be made. Previous workers have used oxides of nickel9 and cobalt4J0 as reference elements. Schmidt10 compared the performance of nickel and cobalt and found that nickel was subject to interference by iron and silicon. He also observed that when nickel was used the exposure times were less constant than with cobalt. We selected cobalt oxide for preliminary trials and, in addition, tests were made with copper(1) oxide for comparative purposes. The determination of cobalt and, in certain cases, copper in steelmaking control is rarely required and thus the need to adjust photo-multiplier sensitivity controls on change-over from steel to slag analysis can be avoided.TABLE I COMPARISON OF PERFORMANCE WITH COBALT AND COPPER AS INTERNAL REFERENCE ELEMENTS Values are standard deviations Sample Component Fe MnO MgO p*o, CaO NzO, SiO, 7 Approxi- mate content, per cent. 11 7 9 8 46 3 9 MGS 421 - Reference element - Cobalt Copper 0.31 0.40 0.20 0-26 0-40 0.37 0.18 0.18 0.31 1-04 0.18 0.13 0.20 0.33 - Approxi- mate content , per cent. 12 3 2 16 61 0.6 7 MGS 426 ---A-___-7 Reference element & Cobalt Copper 0-33 0.46 0.11 0.12 0.03 0.05 0.66 0.30 1-08 1.66 0.03 0.06 0.26 0.24 7 Approxi- mate content, per cent. 0.3 1 7 0-1 34 20 33 MGS 427 - Reference element & Cobalt Copper 0.10 0.05 0.03 0.04 0.20 0.12 0 0.03 0.67 0.63 0.87 0.86 0.01 0.04 In the comparative tests the mixtures used consisted of 1 g of powdered slag and 5 g of graphite, together with either 0.2 g of cobalt oxide or 0.5 g of copper(1) oxide.These pro- portions of reference elements were chosen as the maximum permissible weights to minimiseNovember, 19691 SLAG BY A COMPACTION METHOD OF SAMPLE PREPARATION 949 weighing errors and to give adequate instrument integration response on the measuring channels available. The components were mixed and ground for 3 minutes in a vibratory mill, and pressed for 20 seconds at a 30-ton load into a 32 mm diameter disc with a laboratory hand-press. Three slag samples were used, and in each case four tests (separate spectrographic excitations) were made on the top and bottom faces of the compacted mixture. The standard deviation of the eight instrument response values was calculated, converted into nominal percentage concentration and expressed as a 95 per cent.confidence limit, as shown in Table I. Attenuator settings were adjusted so that instrument responses on both groups of compacts gave similar values. Inspection of Table I shows eleven cases out of twenty-one in which the precision of the cobalt-reference compacts was superior to that of the copper-reference compacts and eight cases in which the latter showed superiority. It was noted that the copper-reference compacts were more prone to cracking and breakage during handling and this feature, together with the marginal superiority in precision of the cobalt-reference compacts, led to the selection of cobalt oxide as internal reference element.Polished surfaces h 70 Fig. 2. Details of the compact die assembly. All dimensions in mm; all components fully hardened die steel Recess for compact 33 mm diameter I mm deep I Dl 38 mm Fig. 3. Perspex compact holder COMPACTION OF SAMPLE MIXTURE- The die used for compacting the mixture of slag, graphite and cobalt oxide is shown in Fig. 2. After pressing, the surface of the compact was lightly hand-ground on No. 600 grinding paper by using a Perspex holder as shown in Fig. 3 to facilitate handling. Because of variation in compacting pressures recommended in the literature, an investigation was made into the effect of different loads. Six compacts were made from a slag-graphite mixture by pressing at loads of 20, 25, 30, 35, 40 and 45 tons for 20 seconds.In these tests an automatic hydraulic press at the Special Steels Division of B.S.C. was used to provide loads in excess of 30 tons. Eight tests were made on each compact and the standard deviation of the seven major elements present calculated in terms of nominal percentage concentration. The average values of the standard deviations of the seven elements were plotted against compacting load, as shown in Fig. 4. Almost identical integration times were noted, indicating that cobalt emission was constant with varying pressure. There was, however, a decrease in instrument response equivalent to 5 to 10 per cent. of content for phosphorus, calcium, silicon, magnesium and manganese with an increase in load of from 20 to 25 tons, but at loads in excess of 25 tons the difference in instrument response was not significant.950 BRAMHALL AND SCHOLES : SPECTROGRAPHIC ANALYSIS OF STEELMAKISG [Analyst, Vol.94 Load on 32 mm diameter compact, tons Fig. 4. Effect of compacting load on average precision values In terms of handling strength, the mixture compacted at 20 tons showed inferior bonding and the compact powdered freely when handled. From the evidence obtained, a load of 30 tons appeared to be the minimum necessary to achieve optimum stability of instrument response. As this loading could be readily attained with a laboratory hand-press it was adopted for future work. MIXING TECHNIQUES Dickens, Koenig and Dippelg investigated various methods of mixing and grinding the three components (graphite, reference element and slag) to form a homogeneous mixture.Vibratory mills were found to be the most suitable and this type of mill was used in our work. By using samples of slag ground to less than 200 mesh as a starting material, successful results were obtained in preliminary work by shaking the components together in a weighing bottle by hand, followed by mixing and grinding them in a Siebtechnik vibratory mill for 2 to 3 minutes. Some comparative experiments were made with a Spex laboratory mixer mill and also combining both Siebtechnik and Spex mills. In each case the total grinding time was fixed at 3Q minutes and the homogeneity of the resultant compact examined by observing the degree of variation in integration time. If the components are adequately mixed the integration times should be constant, as these values are directly related to the distribution of cobalt in the compact.Typical precision values for integration times (from repeated tests on a compacted sample) by using different mixing techniques are as follows. Integration Standard time, deviation, seconds seconds Mixed in a Spex mill for 36 minutes . . . . .. . . .. . . 26.9 1.07 Mixed in a Siebtechnik mill for 38 minutes . . . . .. . . . . 20.8 0.37 Mixed in a Spex mill for 0-32 0.20 minute then in a Siebtechnik mill for 3 minutes 20.5 20.4 Mixed in a Siebtechnik for 3 minutes then in a Spex mill for + minute . . Direct mixing in a Spex laboratory mill was shown to be unsatisfactory, and best results were achieved by combined mixing and grinding in a Siebtechnik vibratory mill followed by a secondary short mixing period in the laboratory mill.Secondary mixing has the advantage of ensuring that particles, which may be isolated because of compaction on the walls of the vibratory mill, can be thoroughly dispersed in the mixture. When speed in sample preparation is important the treatment in the laboratory mill can be omitted, provided that some slight loss in mixing efficiency is accepted. Some measure of grinding efficiency was obtained by comparing the particle-size distri- bution of the components before mixing and after the mixing by grinding treatment in the vibratory rnill. In a typical experiment, before grinding and mixing all of the material passed a 200-mesh sieve, 14 per cent. was retained on a 300-mesh sieve, 16 per cent.was retained on a 400-mesh sieve and 70 per cent. passed a 400-mesh sieve. After grinding all material passed a 200-mesh sieve, 2 per cent. was retained on a 300-mesh sieve, 4 per cent. was retained on a 400-mesh sieve and 94 per cent. passed a 400-mesh sieve.November, 19691 SLAG BY A COMPACTION METHOD OF SAMPLE PREPARATION 951 RE-PREPARATION OF COMPACT SURFACE- It is normal practice to conduct instrument “setting-up” procedures and calibration checks with standard samples during the work programme. To economise in the preparation of standards it is desirable that the compacts should be re-used several times after removal of previous spectrographic discharge craters. The effect of re-preparation on compacts made from standard slags was examined by hand-grinding on fine grinding paper.Four tests were made on each re-ground surface and typical average response values for a sample of standard slag are shown in Table 11, together with nominal analysis and confidence limits in terms of percentage concentration. Surface TABLE I1 EFFECT OF COMPACT SURFACE RE-PREPARATION Preparation No. 1 2 3 4 5 6 Analysis. per cent. . . .. P,O, 649 641 635 637 634 632 8.10 Instrument response-Sample MGS 422 CaO Fe MgO SO, MnO 986 615 136 454 777 1003 638 144 470 7 80 995 620 134 453 772 994 620 135 454 7 82 990 635 137 450 777 994 647 139 454 782 50.2 7-55 2.10 17.45 6.55 A 7 A1203 203 214 209 210 210 200 1.10 95 per cent. confidence limits for one preparation . . . . f0.28 &Om68 f0.36 f0.09 jc0.27 Az0.08 The tests on six consecutively prepared surfaces showed no indication of trend in response values.The compacted mixture appeared to be homogeneous and, in consequence, each compact can be used several times for the purpose of setting the instrument or for calibration checks. In practice, the surface may be re-ground from six to ten times, the limitation being the eventual cracking or breakage of the compact. METHOD APPARATUS AND MATERIALS- A Hilger and Watts E600 Polyvac spectrometer was used, with discharge conditions as given under Experimental. The pre-integration time prior to exposure was 30 seconds, and the integration time, 20 seconds. Spectrum line wavelengths, nm- co Si Mn Ca P Fe A1 228.6 181-7 192.1 183.8 279.1 Mg 178.3 271-4 186.3 Compacting materials-Ringsdorff RWC mouldable graphite powder” was used, and general-purpose reagent cobalt oxide, low in nickel and iron.Sample preparation-A Siebtechnik vibratory mill; a Spex laboratory mixer mill ; a Research and Industrial Instruments Ltd. 30-ton bench hand-press; a die for pressing sample compacts (see Fig. 2 ) ; a holder for the preparation of the surfaces of compacts (see Fig. 3); and silicon carbide grinding paper No. 600 were used. SAMPLING- a No. 200 B.S. sieve. The slag sample prior to mixing is crushed and then ground to pass completely through PROCEDURE- Preparation of compact-Weigh 5 g of graphite, 0.2 g of cobalt oxide and 1 g of powdered slag and transfer them to a 5 cm diameter x 4 cm stoppered glass weighing bottle. Mix by shaking the bottle vigorously for 30 seconds and transfer the contents to a clean Sieb- technik mill.Mill the mixture for a timed period of 3 minutes with the ring and “stone” * The authors have also used flake graphite, which has improved handling properties compared with the powdered graphite specified.952 BRAMHALL AND SCHOLES : SPECTROGRAPHIC ANALYSIS OF STEELMAKING [Analyst, Vol. 94 in position. Transfer the mixture to a stoppered polythene container (4 cm diameter x 3 cm), clamp the container in a Spex laboratory mixer mill and allow it to vibrate for 30 seconds (Notes 1 and 2). Transfer the mixture to a 32 mm diameter die and compact with a total load of 30 tons for 20 seconds. Remove the compact, place it in the holder and prepare both surfaces by lightly hand-grinding on a No. 600 silicon carbide paper to remove the polished finish from the surface.Operation of spectrometer-Select chemically analysed slag samples to provide high and low response values for each calibration graph. Prepare compacts and make three electrical discharges on a prepared surface of each compact (Note 3). When positioning the compact on the excitation stand, place a flat steel disc (3 cm diameter x 1 cm) on the back face to act as an electrical conductor and as protection against breakage when clamping. Average the response values and adjust the instrument attenuator settings to give pre-determined “high” and “low” readings. For test samples make two discharges on each face of the compact, average the response values obtained on each surface and convert the two mean values into percentage concentra- tion (Note 4).For values above 10 per cent. round off the results to the nearest 0.1 per cent. and for values less than 10 per cent. round off to the nearest 0.05 per cent. During batch analysis make a periodic check on any possible drift of instrument response by using the two “setting-up” samples and an additional sample, preferably similar in composition to the test samples (Notes 5 and 6). CALIBRATION- Make duplicate compacts from independent mixtures of each standard sample with graphite and cobalt oxide. Make two discharges on a prepared face of each compact and plot the mean response values against chemical analysis. The surfaces of the calibration compacts may be re-ground to remove the discharge craters. For short-term storage of calibration compacts, polythene containers fitted with tightly fitting stoppers are adequate, but a desiccator must be used for long-term storage.NOTES- Clean the components of the Siebtechnik mill by brushing thoroughly before use, followed by wiping with a cloth. Place 2 to 3 g of graphite in the mill and operate for about 30 seconds to remove any possible contaminants. Remove the graphite and brush the components once again. This operation should be repeated after milling each slag - graphite mixture. The mixing treatment in the Spex mill can be omitted when speed is more important than the small gain in precision obtained by secondary mixing. The tip of the silver electrode should be cleaned by lightly brushing with a stencil brush after each discharge, and the rubber mat on the excitation stand should be wiped after each compact has been analysed.As a safeguard against possible errors in sample preparation, it is advisable to prepare two compacts from each sample; in this case, make two discharges on one surface of each compact and convert the mean response value of each compact into percentage concentration. The frequency of calibration checks will depend on the degree of instrument drift experienced in the individual laboratory. In changing to steel analysis, precautions must be taken to avoid contamination of the discharge chamber, viz., unscrew the electrode assembly, wipe with a clean cloth, replace in position and change the sample mat on the excitation table. Use separate stencil brushes for cleaning the electrode cone during the analysis of steel and slag.Prepare calibration graphs by using chemically analysed samples of slag. 1. 2. 3. 4. 5. 6. RESULTS CALIBRATION GRAPHS- Calibration graphs were prepared by using twelve MGS standard samples, two BISRA slags selected for their high iron content and one B.C.S. slag. The MGS samples had been analysed by the Non-Metallic Materials Analysis Sub-committee, who used both classical and various modern analytical techniques3 Certain of the standard values of the calibration samples (see Table 111) and the test samples referred to subsequently are based on pre- liminary collaborative trials of new procedures, and these values are marked with an asterisk in Tables I11 and IV.November, 19691 SLAG BY A COMPACTION METHOD OF SAMPLE PREPARATION TABLE I11 CHEMICAL ANALYSIS OF SLAG SAMPLES USED FOR CALIBRATION 953 Sample MGS 401 MGS 402 MGS 406 MGS 410 MGS 412 MGS 421 MGS 422 MGS 423 MGS 424 MGS 426 MGS 427 MGS 428 BCS 174/1 J 707 J 708 SiO,, per cent. 5.95 7.75 10.7 21.5 3.25 9.30 17.45 62-20 11.35 7.05 32.9 36-15 14.7 - - -%OS* per cent.0-55 1.90 5-60* 1.60* 0*90* 3.40 1.10 2-45 3.70 0.55 19.95 10.0 1.70 - - MnO, per cent. 3-45 4.95 9-20 14.35 2-00 6.95 6.55 13.75 6.85 2-55 1.30 0.90 5-10 - - G O , per cent. 48.0 31.9 38.45 29-1* 40.5* 44.65 50-2 6.65 44-35 50.7 34.3 41.7 44-8 - - P 2 0 E i P per cent. 17-4 4.45 0.95 9.85 4.70 8-10 8.10 0.04 1.55 16.25 0-04 0-02 12-3 - - Mgo J per cent. 3-20 9.05 6.10 2.55* 9-35* 8.65 2.10 0.45 7.20 1-50 6.55 7.75 7-15 - - Fe, per cent.11.6 24.3 20.56 14.5 29.6 10.85 7.55 10.2 16.7 12.4 0.30 0-20 8-50 39.5 34.5 TABLE IV ANALYSIS OF TEST SLAG SAMPLES BY DIRECT-READING SPECTROSCOPY Component A p a 0 5 r Sample per cent. MGS 403 Mean . . . . 5.70 Standard value 5-75 Difference . . -0.05 MGS 408 Mean .. . . 0-30 Standard value 0.25 Difference . . +0.05 MGS 409 Mean . . . . 0.10 Standard value 0.10 Difference . . Nil MGS 411 Mean . . . . 6.95 Standard value 6-75 Difference . , f0.20 MGS 413 Mean . . 7-75 Standard value 7-35 Difference . . +0.40 MGS 425 Mean . . 1.20 Standard value 1-15 Difference . . +Om05 BCS 174/2 Mean . . . . 12-60 Standard value 11-95 Difference . . +Om65 CaO, Fe, MgO, SiO,, per cent. per cent. per cent. per cent. 47.50 10-95 9.10 13.00 47.10* 11-30 &96* 13.00 +0.40 -0.35 +Om15 Nil 43.50 0.35 4.30 30.70 43.45* 0.30 4-60* 30-95 +0.05 -0.05 -0.30 +0.25 41-70 0.35 4.45 32-65 42.00 0.35 4.40 32.95 - 0.30 Nil $0.05 -0-05 32-75 12.75 3-10 25.35 32*35* 12-60 3-00* 25.75 +0*40 +0-15 + O s l o -0.45 53.50 4.30 7-10 15-55 53.30* 4-45 6*90* 15-45 +0.20 -0.15 +0*20 +0*10 49.30 8-60 1.85 16.90 48-90 8-70 2.00 17-35 +0*40 -0.10 -0.15 -0.45 (Sample contains 6 per cent.of Cr,O,) 43-20 15-95 4-85 11.75 43.20 15-90 4.65 11.20 Nil $0.05 +0.20 +0-55 ~~ ~~~~ MnO, per cent. 5.40 5.40 Nil 1-30 1.35 - 0.05 0.90 0.95 - 0.30 10.50 10.50 Nil 4.80 4.70 +om10 6-70 6.70 Nil 4-10 3.80 + 0-30 Mean instrument responses were plotted against chemical composition as Figs. 5, 6, 7, 8 and 9. A constant scale relationship was maintained between instrument response and percentage composition in plotting the calibration graphs.It will be noted that the slope of the calcium oxide calibration (see Fig. 8) is less than that for comparable concentration ranges of other components. Despite this there is no obvious impairment in the analytical results obtained for calcium oxide. 7 4 0 S P per cent. 2.85 2-95* - 0.10 15.45 15*05* + 0.40 15-35 15.60 - 0.25 3.35 3-40 - 0.05 1.90 1.85* +045 3-40 3-05 + 0-35 0.70 0.75 - 0.05 shown in954 BRAMHALL AND SCHOLES : SPECTROGRAPHIC ANALYSIS OF STEELMAKING [ArtahYSt, VOl. 94 i 500 /o 0 5 10 15 20 I I I I I S 20 25 .30 35 Total iron, per cent. Fig. 5. Calibration graphs for: A, 0.2 to 17.0 per cent. of iron; and B, 15 to 40 per cent. of iron .Aluminium oxide and manganese oxide, per cent.Fig. 6. Calibration graphs for: A, 0.1 to 15 per cent. of manganese oxide; and B, 0.5 to 20 per cent. of aluminium oxide Phosphorus pentoxide and magnesium oxide, per cent. Fig. 7. Calibration graphs for: A, 0.05 to 12 per cent. of phosphorus pentoxide; and B, 0.5 to 10 per cent. of magnesium oxideNovember, 19691 SLAG BY A COMPACTION METHOD OF SAMPLE PREPARATION 955 30 35 40 4s so B Silicon dioxide (upper scale) and calcium oxide (lower scale), per cent. Fig. 8. Calibration graphs for: A, 16 to 37 per cent. of silicon dioxide (high range) ; and B, 29 to 51 per cent. of calcium oxide Silicon dioxide, per cent. Fig. 9. Calibration graph for 6 to 17 per cent. of silicon dioxide (low range) INTERFERENCE EFFECTS- The wavelengths used for measurement were chosen to avoid obvious interference effects arising from spectral overlap, etc.Examination of the calibration graphs (Figs. 5 to 9) show that most of the calibration points lie within close limits of the “best fit line,” which suggests that interference effects are not excessive. Certain standard samples, however, gave instru- ment responses that diverged from the “best fit line.” Calibration points that deviate by - +0*4 per cent. or more are listed below. MGS 406, -0.6 per cent. of manganese oxide at 9.2 per cent. MGS 410, -0-45 per cent. of magnesium oxide at 2.6 per cent. MGS 412, -0.5 per cent. of calcium oxide at 40.5 per cent. MGS 424, +O-6 per cent. of aluminium oxide at 3.7 per cent. MGS 426, +0*4 per cent. of silicon dioxide at 7-1 per cent. MGS 428, +0-45 per cent.of magnesium oxide at 7.8 per cent. These deviations may possibly be explained by spectrum interference or inter-element effects, and further work is required to establish the causes. ANALYTICAL PERFORMANCE- To test the recommended procedure, six MGS slags, one of which contained 6 per cent, of chromium trioxide, and a B.C.S. slag were analysed by using the prepared calibration graphs. The operator prepared two compacts from each sample; two tests were made on a prepared956 BRAMHALL AND SCHOLES : SPECTROGRAPHIC ANALYSIS OF STEELMAKING [Analyst, Vol. 94 face and the mean instrument response of each compact was converted into percentage concentration. This process was repeated by a second operator and a mean of the four analytical results calculated.During the trials, two of the calibration slags were examined from time to time as a check on instrument drift. Mean results are given in Table IV, together with standard deviations and the differences between the mean and accepted value. With two exceptions, the maximum difference from accepted values did not exceed *0*45 per cent.; this was considered to be satisfactory and there was no significant evidence of bias. The average differences are given below, together with a summary of analytical performance. Confidence limits were calculated by pooling the estimates of standard deviation from samples that contained more than 2 per cent. of a particular component. To pool the estimates, the sum of squares of the deviation of each result from each mean were summed, and the total divided by the total number of degrees of freedom. The square root of this value is the pooled standard deviation.Range, per cent. Pa06 .. .. 6 to 13 CaO .. . . 33 to 54 Fe .. .. .. 4 to 16 MgO .. . . 2 to 9 SiO, .. . . 12to 33 MnO .. .. 2 to 10 Al,O, .. .. 2 to 15 95 per cent. confidence limits of one result f 0.44 f0-88 f0.41 f0.38 f0.73 f0-30 f 0.40 Average difference from standard value, per cent. +0.19 +0-16 - 0.07 - 0.04 - 0.04 + 0.04 + 0.05 Analytical performance compared favourably with that obtained by other workers using the direct compaction method. For example, in the determination of calcium oxide at the 40 per cent. level, Dickens, Koenig and Dippelg and SchmidtlO report typical confidence limits of 1.0 and 0.64 per cent., respectively, and for total iron at the 11 per cent.level of 0.30 and 0-34 per cent., respectively. VARIATION IN TOTAL SPECTROGRAPHIC EMISSION- Mason4 reported variation in the total spectrographic emission from one compact to another. Emission was directly related to particle size and Mason recommended strict adherence to specified conditions for the preparation of the compact. Despite careful pre- paration of the compact, variation was still observed and a correction was made by summation of the apparent oxide contents and comparison with an assumed total to derive a correction factor. We examined the possible variation in total emission during the analysis of the test samples. For this purpose the analytical results for each compact were summed (after conversion of iron into iron(II1) oxide) and the totals were compared with the average total of each sample.Differences from the average totals were then pooled to calculate results of an individual compact. The value obtained, & 1-7 per cent., was only slightly greater than the value of 1.5 per cent. based on the total variation introduced by the determination of individual constituents. Clearly, total spectrographic emission is constant within reasonable limits and corrective action is unnecessary when using the recommended procedure. CONCLUSIONS Direct-reading spectroscopy has been used successfully to analyse steelmaking slag by using a sample preparation technique based on direct compaction of the finely powdered slag with graphite. By comparison with X-ray fluorescence, interference effects are relatively minor; the precision of measurement may not be as high but performance can be improved by making repetitive tests on €he compacted sample. Preparation of the sample has a major influence on performance and care is necessary to ensure the production of a homogeneous compact from the basic materials. The proposed method is simple to operate and is capable of producing satisfactory analytical results that compare favourably with those reported by other workers using alternative forms of sample preparation. It is suitable for post-mortem analysis of batches of slag samples when using an instrument for steel analysis.November, 19691 SLAG BY A COMPACTION METHOD OF SAMPLE PREPARATION 957 Preparation and analysis time for a single compact is about 20 minutes Plzcs 4 minutes if a duplicate compact is made and analysed. Contamination from graphite in the deter- mination of carbon in steel can be avoided provided that simple precautions are taken. We have not attempted to automate the sample preparation process, but such automatic systems suitable for production control purposes have been described elsewhere. We wish to thank Mr. W. Dalby, Mr. M. Baxter and Mr. J. S. Oakland for experimental assistance, and the Members of the BISRA Spectrographic Analysis of Non-Metallic Materials Study 1. 2. 3. 4. 6. 6. 7. 8. 9. 10. - - Group for helpful comments. REFERENCES BISRA Sub-committee Report MG/DB/303/66. Croall, G., Spectrographic Analysis of Non-Metallic Materials Study Group-Progress Report 1966, Mason, L., BISRA Sub-committee Report MG/DD/241/60. Scholes, P. H., BISRA Sub-committee Report MG/DB/686/68. Mason, L., BISRA Sub-committee Report MG/DB/236/66. -, BISRA Sub-committee Report MG/DB/236/66. Johnson, W., BISRA Open Report MG/D/613/68. -, United Steel Companies R. & D.D. Report No. A.5067/2/65. Jarrell-Ash Newsletter, No. 1, March, 1968. Dickens, P., Koenig, P., and Dippel, T., Arch. EisenheittWes., 1964, 35, 87 (abridged translation, Schmidt, K., Berg.+. hiittenm. Mh., 1967, 112, 78 (translation, BISITS 6577). Hilger J., 1965, 9, 13). Received February 3rd, 1969 Accepted April 29th, 1969

ISSN:0003-2654    

DOI:10.1039/AN9699400945

出版商:RSC    

年代:1969

数据来源: RSC

摘要:

958 AIzalyst, November, 1969, Vol. 94, $$. 958-967 The Use of a High Temperature Hollow-cathode Lamp for the Spectrographic Analysis of Steels, High Temperature Alloys and Related Materials for Trace Elements BY K. THORNTON (Henry Wiggin 6 Co. Ltd., Hereford) A study has been made of factors affecting the operation of high tem- perature hollow-cathode lamp discharges for the emission-spectrographic determination of trace elements in steels, high temperature alloys and related materials. The factors studied include electrode geometry, carrier-gas pressure and exposure times. The effect of these parameters on element sensitivity and the precision of analysis is discussed. The analysis of dissimilar alloy types has been shown to present problems associated with inter-element effects, and the magnitude of these effects is illustrated.A simple method for overcoming inter-element effects is described, involving the addition of silicon as a buffer to both samples and standards. Results by this method for the analysis of a wide variety of materials from calibration graphs based on a single set of synthetic powder standards are presented. I t is possible by this technique to obtain quantitative trace- analysis results without the necessity of providing chemically analysed standards. THE hollow-cathode discharge source originally described by Paschen' 9~ has since found wide acceptance for applications normally considered difficult with conventional spectro- graphic sources. The technique has successfully been applied to the determination of sulphur and the halogen~,4s~s~,~,* and for the extensive trace analysis of refractory oxides and The early work of Rosen" and subsequent studies by Webb and Webb16,1s have shown the potentialities of the method for the determination of gases in metals.More fundamental studies, correlating ionisation effects and physical properties of the discharge with, irtter alia, carrier-gas pressure and cathode geometry have been published by Mitchell1? and van Voorhis and Shenstone.l* The latter investigators made no attempt to relate the variables studied to factors such as element sensitivity or analytical precision. Few reports have appeared in the literature concerning the use of this type of source for the analysis of non-refractory metals such as steels and nickel-base alloys. However, a paper by Mitchell and Harris19 referred to the analysis of copper and nickel for a limited number of elements.The manufacture of nickel-base alloys and steels for high temperature applications requires close control of certain impurity elements. Conventional d.c. arc spectroscopy is often suitable, but with the many different alloy types produced the effective standardisation for such procedures is generally inconvenient and often impracticable. The intention in the present studies of the hollow-cathode source was to provide a method for the simultaneous detennination of a large number of trace elements in a wide variety of alloy types and their associated raw materials. APPARATUS- the inert gas supply and circulation unit; and the electrical source unit. HOLLOW-CATHODE LAMP- The lamp is basically a water-cooled chamber machined, in this instance, from stainless steel (Fig.1). The inner cylindrical chamber is about 4 inches long, with a diameter of 24 inches, and can be filled with an inert atmosphere through the gas-circulation tubes, The equipment may conveniently be considered in three parts : the hollow-cathode lamp ; 0 SAC and the author.THORNTON 959 A=Tungsten rod E=Gas inlet B= Pyrex glass joint C= Water inlet D = Water outlet F= Gas outlet G =Quartz window H =Observation window Fig. 1. Hollow-cathode lamp E and F. A front port allows radiation to pass from the discharge region through a quartz window, G, which is mounted between O-ring seals. The discharge can be observed visually through the glass window, H, attached to the viewing port.Graphite hollow-cathode electrodes are mounted by push-fit on a 2 mm diameter tungsten rod, A, which is sealed into the end of a B24 type Pyrex ground-glass joint, B. This joint fits into a corresponding taper machined in the rear of the lamp and is protected by an O-ring seal. The assembly is water-cooled through tubes C and D. Operation of the lamp in the vertical position is made possible by the use of a front- surfaced mirror positioned above the quartz window, G, or, as in these experiments, in a horizontal position, by pointing the discharge port directly at the slit of the spectrograph. The separation between cathode and anode (lamp body) is not critical for this design of lamp. The basic lamp design is similar to that used by Webb and Webb for their studies of gas determination in metal~.~~J6 6 A=Pressure gauge, 0 to 40 B=Pressure gauge, 0 to 760 C = Needle valve D=lnlet valve (gas from E=Outlet valve (to lamp) G and H=Check positions for torr torr F=By-pass flush valve c y I i n d e r) leak detection Fig.2. Gas-circulation system960 THORNTON : USE OF A HIGH TEMPERATURE HOLLOW-CATHODE [ArtaZySt, VOl. 94 INERT GAS SUPPLY AND CIRCULATION SYSTEM- As this type of discharge is severely affected by the presence of atmospheric pollutants (oxygen and water vapour in particular), it is essential that all joints should be vacuum-tight. In some instances, purification of the inert carrier gas may be needed, but this has not been found necessary with the supply of mineral helium used in this laboratory. A schematic representation of the gas-circulation system is shown in Fig. 2.The helium, at 2 to 3 lb inch-2 above atmospheric pressure, is restricted by a fine-control needle valve, C, and the pressure on the output side of this valve is monitored with a capsule-type gauge, A. Gas is pumped through the hollow-cathode lamp by a rotary pump, with a displacement of about 50 1 minute-I, and continuously exhausted to atmosphere. The working pressure is adjusted by means of valve C. To facilitate rapid sample changing, a by-pass valve, F, is provided which, when opened, will bring the lamp to slightly above atmospheric pressure. This eliminates the necessity for frequent air-releasing of the rotary pump and enables the operator rapidly to flush the lamp with helium gas after sample changing.ELECTRICAL SOURCE UNIT- Samples are placed inside cylindrical electrodes and vaporised by the sputtering action of charged particles of the inert gas. Ionic collisions of the second kind between vaporised sample and charged-gas particles result in the excitation of the characteristic- spectrum of the sample. Useful emission spectra have been obtained with currents above 200 mA sustained by potentials between 300 and 1200 V. The rectifying system shown in Fig. 3, which produces currents up to 1 i A at 400 to 500 V, has been found satisfactory. Lamp current is controlled by the auto-transformer supplying the main transformer. 2.70 kn each Fig. 3. Electrical source unit OPERATING PROCEDURE- Metallic samples are analysed in the form of turnings, millings and chippings, but powders require briquetting to avoid instability of the discharge.Weighed amounts (10 to 100 mg) of the sample are loaded into fresh graphite hollow cathodes and inserted into the lamp. The chamber is evacuated, flushed with helium, re- evacuated, and the operating pressure adjusted by a needle valve. The discharge is initiated, and a current of about 200mA obtained by adjustment of the auto-transformer. The spectrum is photographed, the cunent increased and the spectrum again photographed, the spectra being superimposed. This sequential power sweep is continued until a current has been attained that is sufficiently high to effect vaporisation of all of the elements to be determined.In practice, it is convenient to use currents up to about 1 A. This permits the analysis for a large number of elements without undue over-exposure of the matrix elements. The spectra are dispersed by means of a Hilger and Watts Large Quartz Spectrograph and photographed on Ilford Ordinary N30 plates, which are developed in Agfa Rodinal (diluted 1 + 40 with water) for 3 minutes at 68" F and fixed with Ilford Hypam. New plate batches are calibrated by the two-step iron-arc method.November, 19691 LAMP FOR THE SPECTROGRAPHIC ANALYSIS OF STEELS 961 The percentage transmission values are recorded for the element lines and the internal standard line (Table I) and converted into relative logarithmic intensities by using the emulsion calibration curve. These values are plotted against logarithmic concentrations of the impurity elements. No background correction is necessary.The stability of the discharge can be improved by prolonged out-gassing of the graphite crucible (up to 40 minutes is required for this operation), whereupon the sample can be exposed with a single, high current setting. This reduces exposure times but considerably increases total operating times. The system used by Webb and Webbl6#l6 incorporates an automatic sample changer with multiple determinations, with the same crucible. Under these conditions out-gassing is economical, but the system described here necessitates single use of the crucibles. TABLE I WAVELENGTHS OF LINES USED FOR QUANTITATIVE STUDIES 269.806 .. 234.984 .. 289.797 .. 306.772 .. 326.106 ..31 7.933 .. 396.847 .. 294.364 .. 325.609 .. 283.307 .. 328.068 .. 238-576 .. 2 7 6.7 87 .. 2 8 6.3 3 3 .. 317.602 .. 255.796 a - .. Element Antimony .. .. Arsenic .. .. .. Bismuth .. .. Bismuth .. .. Cadmium . . .. Calcium .. .. Calcium .. .. Gallium .. .. Indium . . .. .. Lead . . .. .. Silver . . .. .. Tellurium .. .. Thallium .. .. Tin .. .. .. Tin . . .. .. zinc .. .. .. FACTORS AFFECTING SENSITIVITY- z Helium 294.610 Electrode geometry-Samples of graphite powder doped with oxides of impurity elements were pressed into pellets and exposed under constant conditions by using a variety of cathode shapes. Both the internal depths of the cathodes and the internal diameters were varied, the former from 15 to 50 mm and the latter from 4 to 8 mm. High purity graphite (Rings- dorff grade RWO) was used for the preparation of all electrodes and the wall thickness was kept constant.A total exposure time of 7 minutes was used, with a current sweep up to 1-40 A. Fig. 4 indicates the effect of variable electrode depth on sensitivity for a selection of elements. It is seen that for each element, maximum sensitivity is obtained at a specific cavity depth and, further, that the less volatile elements exhibit maximum sensitivity with short electrodes. By increasing the length of the electrode, the spectral emission of these elements is reduced and that of the more volatile elements increased. As it is often the latter group that produces deleterious effects in high temperature alloys, this property can be used to obtain selective enhancement. The effect of increasing the internal diameter of the cathode was more uniform.In all instances, sensitivity decreased sharply, probably as a result of a lower concentration of vapour on the viewing axis of the electrode. By using this information in the context of the specific problems encountered in our laboratories, it was concluded that optimum conditions would be achieved with electrodes of internal length 25 mm and internal diameter 4 mm. This gives reasonably low spectral density from most of the matrix elements, e.g., nickel, cobalt, chromium and iron, coupled with high sensitivity for the elements of interest, e.g., lead, zinc and antimony. Carrier-gas presszcre-The effect of pressure on sensitivity has been studied for the range 5 to 30 torr.Exposures were made by superimposing current steps up to a maximum of 1.20 A. Fig. 5 shows typical variations for line relative intensities with increasing gas pressure and, as before, it can be seen that volatile elements exhibit significantly different relationships from the less volatile elements. Unfortunately, lamp stability is not good at pressures in excess of 20 torr, this being the optimum value for good stability and sensitivity.962 THORNTON : USE OF A HIGH TEMPERATURE HOLLOW-CATHODE [ANaZyst, Vol. 94 I= 1.2 Y C w 0.4 0 Electrode depth, rnm Fig. 4. Effect of electrode depth on element line density: A, antimony, 259.8 nm; B, cobalt, 340.5 nm; and C, thallium, 276.7nm. (The more con- ventional relative intensity scale is replaced by line density measurements to emphasise the effect over a wide line density range) loot - - - - - 10 - - - - - - - I = ? 10 20 30 Carrier-gas pressure, torr Fig.5. Relative line intensity values for varying gas pressure: A, silver, 328.0 nm; B, bismuth, 289.8 nm; C, iron, 248.8 nm; and D, aluminium, 308.2 nm FACTORS AFFECTING PRECISION- Exposwe time-In the previous discussion of operating procedure, it was indicated that lamp current is varied during the course of an exposure, starting at low values and increasing until a current is reached that will effect distillation of all of the elements to be determined. This is achieved by adjustment of the auto-transformer setting. In practice, this setting is varied by constant pre-determined increments, the amounts of these being governed by general discharge stability.The exposure time of each step has been varied from 30 seconds to 2 minutes and coefficients of variation calculated for a series of elements. Sensitivity was virtually unaffected, but it can be seen from the mean coefficients of variation in Table I1 that precision deteriorates below step exposures of 1 minute. No improvement is obtained with longer exposures. Carrier-gas $ressure-No quantitative results are available , but instability increases outside the range 10 to 20 torr, with consequent reduction in precision.November, 19691 LAMP FOR THE SPECTROGRAPHIC ANALYSIS OF STEELS TABLE I1 EFFECT OF STEP EXPOSURE TIME ON ANALYTICAL PRECISION Step exposure time, Total exposure time, Mean coefficient of variation, minutes minutes per cent.0.5 3 16.3 1 6 10.2 2 12 10.6 963 ELEMENT DISTILLATION- Studies have been made of the distillation of elements from a graphite matrix by relating lamp current (and hence temperature) to element line intensity. Selective volatilisation occurs and can be used to separate the emission lines of elements to be determined from the spectra of the matrix elements. This provides an effective method of overcoming problems associated with high spectral background and line interferences often encountered with complex alloys. Thus, for instance, the tellurium line 238676nm is often unusable in the presence of chromium (238.576 nm) and cobalt (238.582 nm). However, with hollow- cathode analysis, it is possible to distil off the tellurium completely at a lamp current of 0.75 A (3 minutes’ exposure would be sufficient for low concentrations), with no significant evapora- tion of either of the other two elements and consequently no interference.TABLE I11 DISTILLATION OF ELEMENTS FROM GRAPHITE IN A HOLLOW-CATHODE DISCHARGE LAMP Lamp current, A Elements completely distilled Ag, Cd, Hg, T1, Zn Bi, In, K, Na, Pb. Sn, Te As, Cu, Ga, Mg, Mn, Sb Al, B, Be, Co. Cr, Fe. Ni, Si 0.5 0.75 1.0 1.25 Ca, Ge 2 1.5 The lamp currents at which some elements distil are shown in Table 111. Distillation starts at lower currents than those indicated, the values shown being those for effective complete evaporation. OCCURRENCE AND EXTENT OF MATRIX EFFECTS- In the literature on hollow-cathode excitation, efforts have been concentrated on des- cribing techniques to improve sensitivity (often by orders of as much as 1@), or on the analysis for elements normally considered difficult to excite, e.g., sulphur and the halogens.An exten- sive survey has shown that authors have considered either the analysis of single or closely related materials, or when diverse compounds have been analysed, standards were prepared for each type. The problem of inter-element effects has not been discussed in any detail and, indeed, little reference to this aspect can be found. It might be expected that, as distillation is closely controlled and excitation is achieved in the presence of an inert gas, such problems would be negligible or non-existent. To check the validity of this assumption, a sample of specially melted nickel containing many of the elements of interest was analysed. The concentrations of the impurity elements were first established by spectrochemical analysis involving two techniques, namely dissolu- tion - oxidation procedure by the well known Jaycox methodm and carrier-precipitation with hydrogen sulphide.2l Calibration graphs were obtained for a series of such nickel standards by using the optimum hollow-cathode conditions (see Table IV).One of the samples was then analysed in the presence of additional pure elements. The selected sample was placed in a graphite hollow-cathode electrode and an equal weight of the selected element added. The analysis was carried out under the same conditions as those for the preparation of the calibration graphs. The element concentrations found by using these calibration graphs (ie., based on pure nickel) are compared with the accepted values in Table V.Obviously, severe inter-element effects occur, particularly in the determination of the elements, gallium, tin, indium and antimony. The effect is less pronounced for some others, notably those with very low boiling-points.964 THORNTON: USE OF A HIGH TEMPERATURE HOLLOW-CATHODE [AutdySt, VOL 94 TABLE IV OPERATING CONDITIONS FOR HOLLOW-CATHODE LAMP Carrier gas . . .. .. .. .. .. Gas purification . . .. .. .. .. Gas pressure .. .. .. .. .. Internal length of graphite electrode . . .. Internal diameter of graphite electrode . . .. Exposure time (pre-burn) . . . . .. .. Exposure time (total) . . .. .. .. Anode-to-cathode distance. ... .. .. Spectrograph . . .. .. .. .. Slitwidth .. .. .. .. .. .. Wavelength range . . .. . . .. .. Source-to-slit distance . . .. .. .. Condensing system . . .. .. .. .. Mineral grade helium None 20 tom 25 mm 4 mm About 4 cm (not critical) Nil 1 minute at 0.60 A 1 minute a t 0.20 A 1 minute a t 0.40 A 1 minute a t 0.80 A 1 minute a t 1-00 A Hilger Large Quartz E.742 10 pm 236-0 to 320.0 nm and 280.0 to 496.0 nm 20 cm F.1026 quartz lens, 2 cm from slit Superimposed I ELIMINATION OF MATRIX EFFECTS- The most severe effect is produced by silicon (see Table V). It was reasoned that an excess of this element might, when added to both samples and standards, diminish the effect of other elements. The validity of this assumption is indicated in Table VI, which shows results for the analysis of a sample of nickel by using a complex high temperature alloy for standardisation, with a 30 per cent.addition of silicon to samples and standards. The silicon used was Specpure grade metal lump (Johnson Matthey and Co. Ltd.). TABLE V EFFECT OF ADDED ELEMENTS ON THE ACCURACY OF HOLLOW-CATHODE DISCHARGE ANALYSIS 7 Added element Ag Chromium .. . . 0.0043 Molybdenum . . 0.0034 Silicon .. . . 0.0047 Aluminium .. . . 0.0020 Titanium . . . . 0.0061 Copper .. . . 0.0021 Zn 0.021 0.019 0.021 0-027 0-018 0-032 Element concentration found, per cent. Sb T1 Pb Ga Bi 0.026 0.0060 0-0044 0.025 0.0030 0.022 0-0043 0.0040 0.023 0.0023 0-125 0.0054 0-0038 >Om25 0.0032 0.068 0.0054 0-0038 0.14 0.0023 0.071 0-0023 0.0044 0.11 0.0027 0.004 0.0051 0.0037 0.0021 0.0016 A Sn 0.003 1 0.0038 0.035 0.0038 0.013 0.0005 7 In 0.0086 0.0095 0.021 0.014 0.019 0~0010 Spectrochemical analysis .. . . 0.0036 0.023 0.010 0.0046 0.0040 0.0070 0.0025 0.0015 0.0050 For comparison, the values obtained for the analysis of this nickel sample from the same high temperature alloy standards are shown without the addition of silicon. The complex alloy standards used for calibration purposes contained significant concentrations of many of the matrix elements shown in Table V, namely, chromium, molybdenum, aluminium and titanium . TABLE VI ELIMINATION OF MATRIX EFFECTS BY THE ADDITION OF SILICON Analysis of unalloyed nickel by using complex high temperature alloy standards Element concentration found, per cent. Addition ~i Zn Sb Pb Ga Bi Sn No addition* .. . . 0-0028 0.020 0.0011 0.0032 0.0006 0.0022 0.0002 Silicon addedt . . . . 0.0034 0,024 0.010 0-0038 0-0076 0.0022 0.0017 Accepted analysis . . 0.0036 0.023 0-010 0.0040 0.0070 0.0025 0.0015 No additions made to either the nickel sample or the high temperature alloy standards. t 30 per cent. w/w of silicon added to both sample and standards. Results in Table VI show that the assumption that silicon would mask matrix effects caused by other elements is fully justified, and its addition provides a basis for the analysisNovember, 19691 LAMP FOR THE SPECTROGRAPHIC ANALYSIS OF STEELS 965 of a wide variety of materials. Further experiments with a larger selection of added elements have indicated that accurate results can be obtained on all matrix types except those that are highly volatile, eg., zinc, copper or aluminium base.The amount of silicon needed for optimum results has been studied, and it was found that 30 per cent., or more, of the weight of sample is required for total elimination of matrix effects. The effect of adding extra silicon is to produce increasingly unstable discharges, with consequent decrease in precision. On the basis of these results, an addition of 10 mg of silicon is made to the weight of sample normally taken (30mg). STANDARDISATION- In an attempt to eliminate the necessity for relying on chemically analysed samples for the preparation of working curves, samples of high purity nickel powder were mixed with calculated additions of compounds of the elements to be determined.The resultant graphs were found to be concurrent with those obtained with chemically analysed, melted samples. It was also found that graphs obtained by varying the weight taken of any one standard were concurrent with graphs obtained for constant weights of standards of varying element concentration. Standardisation is thus reduced to the mechanical blending of one sample, and the necessity for chemical analysis is eliminated. It seems that, in the presence of silicon, the chemical form of the element analysed has no effect on its resultant line intensity. RESULTS By using the conditions outlined in Table IV studies have been made of the precision and accuracy of the technique. PRECISION- Several samples of a particular nickel-base complex alloy were used to establish calibra- tions for the elements arsenic, antimony, bismuth, tin, lead and silver.A further sample of this material was analysed sixteen times on the same day with the same photographic plate. The results obtained are shown in Table VII. TABLE VII ANALYTICAL PRECISION FOR HIGH TEMPERATURE NICKEL ALLOY Element* As Sb Bi Sn Pb Ag 0.0014 0.0030 0.012 Content, per cent. . . . . 0.020 0.026 0.0009 Standard deviation . . .. 2.6 x 2-4 x 7-2 x 1.4 x 1-2 x lo-* 3.4 x lo-* Coefficient of variation, per cent. 12.5 9-2 8.0 11-6 8.6 11-4 * The wavelengths of lines used are given in Table I, together with that of the helium reference line. The mean precision, expressed as a coefficient of variation, is 10.2 per cent. A sample of doped nickel powder was used to prepare calibration graphs in the presence of 10mg of silicon.The weight of doped powder standard was varied to prepare graphs covering the appropriate element concentration range and the final weight adjusted to 40 mg (including 10mg of silicon) by the addition of Specpure nickel turnings. Samples were analysed by adding 10 mg of silicon to 30 mg of the sample in the graphite crucible, without mixing. The results obtained (averages of four exposures) are shown in Tables VIII, IX and X, together with accepted values. ACCURACY- DISCUSSION The results shown in Tables VIII, IX and X were all obtained by using calibrations from a single blended powder standard. Similar accuracy was achieved for high alloy steels and a variety of materials used in their manufacture, including chromium, cobalt, ferro-molyb- denum, nickel - niobium (40 + 60 per cent.), titanium and tungsten.In fact, the use of a blended nickel powder standard, together with the addition of silicon buffer to the standard and samples, will provide accurate results for most of the materials encountered in the high temperature alloy field.Sample No. Material R3393 Nickel R3394 Cobalt R3396 Iron R3396 Nickel 80%; chromium 20% R3400 Nickel 40% ; iron 40%; chromium 20% R3401 Iron 60%; nickel 30%; cobalt 20% *The “Accepted” TABLE VIII ANALYSIS OF VARIOUS SAMPLES WITH CONSTANT TRACE-ELEMENT ADDITIONS Element concentration found, per cent. Accepted value Hollow cathode Accepted value Hollow cathode Accepted value Hollow cathode Accepted value Hollow cathode Accepted value Hollow cathode Bi 0*0009 0.0012 0.0008 0~0010 0.0008 0.00 10 0*0009 0.001 1 0.0008 0.001 1 In 0~0010 0~0010 0.001 1 0-0009 0.0008 0.0009 0-0008 0*0009 0.0009 0.001 1 Ga* 0.001 0.00 12 0.001 0.0013 0.001 0.001 9 0.00 1 0*0009 0.001 0.0012 Sn 0.006 0.006 0.006 0.006 0.007 0.007 0.006 0.005 0.006 0.006 Pb 0.0014 0.0013 0.0014 0-0012 0-0009 0.001 3 0-0014 0.00 12 0~0010 0.0012 T1 0.00 14 0.0016 0.0004 0.0006 0.0008 0~0010 0-0008 0.0012 0.0008 0~0010 Sb 0.0013 0.0018 0.0020 0,002 1 0.0033 0.0038 0-0016 0.0018 0.0025 0.0028 Zn* 0.001 0.0009 0.001 0-0007 0.001 0-0008 0.001 0~0010 0.001 0.0009 Ag 0.001 1 0.0009 0*0010 0.0018 0~0010 0.001 1 0.0012 0~0010 0.0012 0-0013 Te* 0.00 1 0.0012 0.001 0.0014 0.001 0.001 1 0.001 0~0010 0.001 0.0012 Accepted value 0.0008 0.0009 0.001 0.007 0.0011 0-0007 0.0019 0.001 0.0011 0.001 Hollow cathode 0.0009 0.0010 0.0014 0.006 0.0011 0.0010 0-0026 0.0007 0.0010 0.0011 values given are the nominal additions ; all other elements were determined by spectrochemical methods.TABLE IX HOLLOW-CATHODE DISCHARGE ANALYSIS OF HIGH TEMPERATURE ALLOYS Sample No. Material R3385 Nickel-base alloyt Accepted value Hollow cathode R3386 Nickel-base alloyt Accepted value Hollow cathode R3387 Nickel-base alloyt Accepted value Hollow cathode R3388 Nickel-base alloyt Accepted value Hollow cathode Element concentration found, per cent. A I Bi In Ga* Sn Pb T1 Sb Zn* Ag Te* o*ooo 1 0.0002 0.0002 0.0003 0.0009 0.0012 0*0005 0.0006 0.000 1 0~0002 0~0002 0-0003 0.0009 0.001 1 0.0005 0-0006 0~0002 0.0003 0.0005 0.0006 0.002 0.0026 0.001 0.0012 0.009 0.01 1 0-0050 0.0056 0-0014 0.0015 0.0026 0.0025 0~000 1 0.0002 0.0002 0.0003 0.0009 0.0012 0*0005 0-0006 0~0001 0~0002 0~0002 0.0003 0.0009 0.001 1 0.0004 0.0006 0.0004 0-0006 0.0009 0.001 1 0-0048 0.0056 0.0019 0.0022 o*ooo 1 0~0001 0~0002 0.0002 0.001 0.0010 0.0005 0*0004 0~0001 0~0001 0.0002 0*0002 0.0009 0.0009 0.0005 0.0004 0~0001 0-0002 0.0002 0*0003 0.001 0-0013 0.0005 0*0007 * The “Accepted” values given are the nominal additions; all other elements were determined by spectrochemical methods.t Basic composition, per cent.: nickel 60, chromium 15, cobalt 15, molybdenum 5, titanium 24, aluminium 24. As’. 0.01 0.014 0.01 0.012 0.01 0.013 0.01 0.009 0.01 0.010 0.0 1 0.010 7 As* 0.01 0.0053 0.005 0.0051 0.001 0.00 1 6 0.002 0-0023 E X “aNovember, 19691 LAMP FOR THE SPECTROGRAPHIC ANALYSIS OF STEELS TABLE X ANALYSIS OF CERTIFIED STEEL STANDARDS Element concentration found, per cent. 967 Sample No.SS13 SS16 SS32 ss33 Material Mild steel Certificate value Hollow cathode Mild steel Certificate value Hollow cathode Carbon steel Certificate value Hollow cathode Carbon steel Certificate value Hollow cathode $b Sn A; 0-003(0) 0-06(5) - 0-OOS(5) * O.Ol(0) 0.0030 0-062 - 0-0062 0.010 - - - 0.003 - 0.0035 - 0.070 - 0.067 - - - - * Non-standardised element. Analysis of low melting-point materials, such as copper, manganese or aluminium alloys, requires the blending of separate standards based on the particular matrix type encountered and, in general, produces inferior sensitivities, particularly for those elements with boiling- points close to those of the matrix components.The determination of elements whose boiling-points are higher than those of the matrix elements is impracticable by direct means. TABLE XI LIMITS OF DETECTION FOR HOLLOW-CATHODE EXCITATION Limit of detection, p.p.rn. < 0.01 >0-01 to g0.1 >0*1 to G1.0 >1*0 to ,(lo As, Te E 1 erne n t Ag, Ca, Cd. In, Mg, Na Bi, Cu, Ga, Mn, Pb, T1, Zn Ba, Ge, K, Sb, Sn By using the conditions outlined previously, limits of detection for some elements have been determined for steels and nickel-base alloys, and are shown in Table XI. These limits refer to the optimum general procedure used, but it is obvious from the discussion on sensi- tivity that the limit for any particular element may be extended by reference to such factors as electrode geometry and gas pressure. The values in Table XI are estimates for direct examination, but the use of pre-concentration techniques may also be applicable, with consequent increases in sensitivity. I thank Mr. B. E. Balfour and Mr. D. Jukes for valuable editorial comment, and Messrs. Henry Wiggin & Co. Ltd., for permission to publish this paper. I am also indebted to Messrs. Ridsdale Ltd. for permission to quote the certified values given in Table X. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. REFERENCES Paschen, F., Annln Phys., 1916, 50, 901. -, Sber. preuss. Akad. Wiss., 1928, 32, 3. -, Annln Phys., 1923, 71, 142. McNally, J . R., Harrison, G. R., and Rowe, E., J . Opt. SOC. Amer., 1947, 37. 93. Birks, F. T., Spectrochim. Acta, 1954, 6, 169. Berezin, I. A., and Aleksandrovich, K. V., Zh. Analit. Khim., 1961, 16, 613. Berezin, I. A., Zav. Lab., 1961, 27, 859. Falk, H., Sfectrochim. pcta, 1965, 21, 423. Korovin, Y. I., and Lipis, L. V., Optika Spektrosk., 1958, 5, 334. Ivanov, N. P., Nedler, V. V.. and Andrikanis, E. N., Zav. Lab., 1961, 27, 836. Pevstov, G. A., and Krasil’shchik, V. Z., Zh. Analit. Khim., 1964, 19, 1106. MatiC, J. S., and PeSiC, D. S., Revue Roum. Chim., 1965, 10, 733. Rosen, B., Revue UnivZle Mines, 1953, 9, 445. Webb, M. S. W., and Webb, R. J., Analytica Chim. Acta, 1965, 33, 138. Mitchell, K. B., J . Opt. S O ~ . Amer., 1961, 51, 846. van Voorhis, C. C., and Shenstone, A. G., Rev. Scient. Instrum., 1941, 12, 267. Mitchell, G. P., and Harris, C. I., PYOC. SOC. Analyt. Chem., 1965, 105. Jaycox, E. K., J . Opt. SOC. Amer., 1947, 37, 159. Balfour, B. E., Jukes, D., and Thornton, K., Appl. Spectrosc., 1966, 20, 168. -- P , Appl. Spectrosc., 1968, 22, 63. , Ibid., 1966, 36, 403. J -- Received November 4th, 1968 Accepted April 23rd, 1969

ISSN:0003-2654    

DOI:10.1039/AN9699400958

出版商:RSC    

年代:1969

数据来源: RSC

摘要:

968 Analyst, November, 1969, Vol. 94, jbp. 968-975 The Determination of Trace Amounts of Calcium in Stainless Steels by Solvent Extraction Followed by Atomic-absorption Spectrophotome try BY J. B. HEADRIDGE AND J. RICHARDSON (Department of Chemistry, The University, Shefield. S3 7HF) A method is described for the atomic-absorption spectrophotometric determination of 2 to 60 pg g-l of calcium in stainIess steels after solvent extraction of most of the iron, chromium and nickel with acetylacetone and pyridine. For thirteen B.C.S. steels, the calcium contents varied from 4 to 51 pg g-l. The standard deviation in the errors from the means for the various steels (fifty-eight determinations) was 1.4 pg g-l. The results of applying the solvent-extraction procedure to twenty-five other elements of interest to the steelmaker are also reported.THE metallurgist is increasingly requiring information on the concentration of trace elements in stainless steel, as even minute traces of some elements can have detrimental effects on the desired properties of these alloys. In this paper a method is described for the determination of calcium in stainless steel at concentrations of 2 to 60 pg g-l. So that the interference of trace amounts of phosphate and silicate in the atomic- absorption spectrophotometric determination of calcium can readily be eliminated, a nitrous oxide - acetylene flame is used. With this flame the 1 per cent. absorption value for calcium is about 0.08 pg rnl-l for many commercial instruments in the presence of high concentrations of another ionisable metal.The standard deviation of the background “noise” is usually about one sixth of the 1 per cent. absorption value for cheaper atomic-absorption spectro- photometers. If the standard deviation of the sample “noise” is the same as the background “noise” (this is the minimum; it is usually appreciably greater), then the standard deviation in measured absorbance for a sample is twice the standard deviation in the background “noise,” or about one quarter of the 1 per cent. absorption value. Therefore, under the best conditions, the standard deviation in the error for a calcium determination would be about 0.02 pg ml-l with most British commercial instruments. For a 1 per cent. w/v solution of stainless steel this amounts to 2 pg g-l; in practice, the figure is likely to be nearer 5 pg g-1.However, a precision approaching 1 pg g-l was considered necessary for the determination of calcium in stainless steel and this could only be accomplished by using, in a direct method, a more concentrated solution of the steel, with the increased possibility of burner clogging, or by extracting the base elements from the steel into an organic solvent and concentrating the aqueous phase, which contained the calcium, before nebulisation. It was decided to adopt the latter method, for it has the additional advantage of allowing calibration graphs to be prepared from pure calcium solutions without the necessity of adding weighed amounts of Specpure base elements, which may contain calcium, to the standard solutions.This arises because the residual amounts of base elements, which are left in the aqueous phase after solvent extraction, have no interfering effect on the calcium signal. A literature survey was undertaken to obtain information on suitable solvent-extraction methods for iron, chromium and nickel. Acetylacetone, which forms a separate layer with aqueous solutions, is a suitable extractant for iron(II1) within the pH range of 1 to about 7.112 The neutral tris(acetylacetonato)iron(III) is extracted. With nickel,2 the maximum percen- tage extraction with 0.1 M acetylacetone in benzene occurs a t pH 10 but only to the extent of 25 per cent. The low percentage extraction of nickel could result because most of the nickel may be present in the aqueous phase as Ni(acac),-.Ni2+ + 3Hacac --+ Ni(acac),- + 3H+. Alternatively, the complex in the aqueous phase could be [Ni(acac),(H20),]0, and hydrogen bonding between water molecules on the complex and those in the bulk of the solution would keep the neutral complex in the aqueous phase. 0 SAC and the authors.HEADRIDGE AND RICHARDSON 969 To induce the nickel to extract more completely into the acetylacetone layer, it is necessary to add a neutral non-hydrogen bonding complexing agent that will oust the third acetylacetonate anion or the water molecules and give the neutral complex, [Ni(aca~)~L?]o. Pyridine was selected for this purpose. From a buffer solution of equal amounts of pyndme and its hydrochloride at pH 5-2, the extraction of nickel was now almost complete (see below).Since this work was started, Tanino and Kitahara have also reported that nickel can be extracted from aqueous solution with chloroform containing acetylacetone and ~yridine.~ Starvl states that chromium(II1) is only extracted from aqueous solutions into pure acetylacetone if the chromium(II1) solution is refluxed with acetylacetone. The dissociation of aquochromium( 111) complexes before complex formation is very slow at room temperature, hence the necessity for refluxing. However, when acetylacetone plus pyridine were used with chromium(II1) solutions that had been prepared by dissolving the metal in aqua regia and removing most of the acid by evaporation, it was found that 97 to 98 per cent. of the chromium was consistently extracted. This difference in behaviour of chromium( 111) is under further investigation, but we have verified that no extraction of chromium(II1) occurs at room temperature in the absence of pyridine.The solvent-extraction procedure for the removal of most of the iron, chromium and nickel from stainless steels is described below. Calcium is not extracted at pH 5.2 by acetylacetone and pyridine and is left in the aqueous phase for subsequent determination by at omic-absorpt ion spectrophot ome try. EXPERIMENTAL APPARATUS- a nitrous oxide - acetylene burner. Atomic-absorption spectrophotometry was carried out on a Unicam SP90, fitted with REAGENTS- Hydrochloric acid, sp.gr. 1-18. Nitric acid, sp.gr. 1.42. Hydrojuoric acid, 40 per cent. w/w. Potassium chloride. Calcium carbonate. These reagents were of analytical-reagent grade.Iron, chromium, nickel, manganese and titanium metals and molybdenum trioxide were of Specpure quality (Johnson, Matthey Ltd.). A cetylacetone-General-purpose reagent. Pyridine-General-purpose reagent. Re-distil both reagents before use. Freshly distilled water-Store in a polypropylene aspirator. Standard calcium solution A-Dissolve 0-624 g of dried calcium carbonate in the minimum amount of hydrochloric acid and dilute the solution to 1 litre in a graduated flask. Transfer the solution at once to a dry polythene bottle. Transfer the solution at once to a dry polythene bottle. 1 ml of solution B = 25 pg of calcium. 1 ml of solution A = 250 pg of calcium. Standard calcium solution B-Dilute 10 ml of solution A to 100 ml in a graduated flask.PRELIMINARY INVESTIGATIONS- It was hoped to develop a satisfactory method for the determination of calcium by dissolving 1 g of stainless steel in aqua regia, evaporating the solution to the first appearance of solid, dissolving this solid in water, diluting the solution to 50ml and extracting most of the iron(II1) into an organic phase by shaking this solution with 50 ml of acetylacetone in a separating funnel. On adding 10 ml of pyridine and re-shaking the solution, it was then expected that most of the chromium(II1) and nickel(I1) would also be extracted into the organic phase, and cakium ions remain in the aqueous phase. When acetylacetone reacts with metal ions to form complexes, hydrogen ions are released. M3+ + 3Hacac -+ M(acac), + 3H+.970 [Analyst, Vol.94 From 1 g of iron metal, 54 mmoles of hydrogen ions are produced. Another source of hydrogen ions in the solution is obviously the acid that remains in the solution when the steel solution has been evaporated to the first appearance of solid. It was established by titration with standard alkali that the 50 ml of steel solution, which were ready to be extracted with acetylacetone, always contained about 40 mmoles of hydrogen ions; 94 mmoles of hydrogen ions react with 7.6ml of pyridine to produce pyridinium ions and, as excess of pyridine in the aqueous layer is necessary for complexation of the nickel, it was decided to use 10 ml of pyridine in the solvent-extraction procedure. Therefore, the following tentative method was devised for preparing from stainless steel a solution suitable for nebulisation in the atomic-absorption spectrophotometer to determine the calcium content of the steel. HEADRIDGE AND RICHARDSON : DETERMINATION OF TRACE AMOUNTS TENTATIVE METHOD- Dissolve 1 g of steel in 20 ml of concentrated hydrochloric acid and 5 ml of concentrated nitric acid.Evaporate the resulting solution to the first appearance of appreciable solid material (about 7 ml). Take up the solid in water and dilute the solution to 50 ml. Transfer the solution to a separating funnel and shake the solution with 50 rnl of acetylacetone. Add 10ml of pyridine and re-shake the solution. Remove the lower aqueous layer containing the calcium ions and evaporate it to the first appearance of solid. Add 1 ml of potassium chloride solution (2.5 per cent.w/v in potassium ions) to suppress the ionisation of calcium atoms in the flame, take up the solid in water and dilute the solution to 25 ml in a graduated flask. Transfer the solution immediately to a dry polythene bottle. The usual procedure for a blank solution is to replace 1 g of stainless steel by 1 g of Specpure iron and to carry through the procedure exactly as for a steel. However, it cannot be assumed that this iron is completely free from calcium, which is a trace impurity in most materials. Therefore, it was decided to dispense with the use of 1 g of Specpure iron and replace it with 54 mmoles of hydrogen ions, added as hydrochloric acid. (The calcium content of this acid is negligible.) The tentative procedure with the blank was, therefore, as follows.TENTATIVE PROCEDURE WITH BLANK- Evaporate 20 ml of concentrated hydrochloric acid and 5 ml of concentrated nitric acid to 7 ml in a beaker. Add 10-8 ml of 5 M hydrochloric acid and dilute the solution to 50 ml. Transfer the solution to a separating funnel and proceed as for the steel solution. Conditions had now been devised to test the effectiveness of the solvent-extraction pro- cedure for iron, chromium and nickel. For convenience, the concentration of metal in the aqueous phase before extraction was taken as 0.2 per cent. w/v. The procedure was similar to that outlined above for steel except that 100 mg of Specpure metal were used. The con- centration of metal ions left in the aqueous phase after solvent extraction was determined by atomic-absorption spectrophotometry.Standard solutions for the calibration graphs were prepared by adding suitable aliquots of more concentrated aqueous standard solutions to the aqueous phase of a.n extracted blank solution. The results obtained by solvent extraction with an atomic-absorption spectrophotometric finish are shown in Table I. TABLE I DATA ON THE SOLVENT EXTRACTION OF IRON, CHROMIUM AND NICKEL Wavelength, Slit width, Lamp current, Percentage Inorganic species nm mm mA extracted Iron(II1) .. .. 248.3 0.10 15 98 Chromium(II1j . . .. 357-9 0.08 10 97 Nickel(I1) . . .. .. 232.0 0.10 15 95 The light path was 1 cm above the burner top, the acetylene flow-rate 3.0 1 minute-1 at 15 p.s.i. pressure and the nitrous oxide flow-rate 5.0 1 minute-l at 30 p.s.i.These results show that the iron, chromium and nickel are almost completely extracted with the tentative procedure. It was hoped that it would be possible to prepare calibration graphs for calcium in stainless steels by adding aliquots of standard calcium solutions to the extracted blank solution together with the potassium chloride solution. However, before this method was adopted,November, 19691 971 it was necessary to check that the low concentration of residual base elements had no effect on the calcium signal obtained in their absence, and that no calcium is removed into the organic phase in the extraction procedure. OF CALCIUM IN STAINLESS STEELS BY SOLVENT EXTRACTION Calcium concentration, pg ml” Fig. 1. Calcium calibration A calibration graph obtained by adding aliquots of standard calcium solution to the extracted blank solution is shown in Fig. 1.An identical calibration graph was obtained when 20-mg amounts of iron(II1) were added together with the aliquots of standard calcium solution. This proves that residual amounts of iron have no effect on the calcium signal, which is not unexpected, for the residual concentration of iron in the 25 ml of final solution is only 800pgml-1, but the concentrations of potassium ion and pyridinium chloride are 1000 pg ml-l and about 3.8 M, respectively. Similarly a calibration graph obtained by adding aliquots of standard calcium solution to blank solutions before extraction, followed by the solvent-extraction procedure and work- up, was identical with that shown in Fig.1. This proves that, as expected, no calcium is lost in the solvent-extraction step. In this instance the blank solution was prepared by using 1 g of Specpure iron per 50 ml of solution to be extracted rather than by adding 54 mmoles of hydrogen ions. Finally, it was necessary to investigate the possible interference effects of other con- stituents of stainless steels on the calcium signals. This was carried out by replacing x g of the 1 g of iron by x g of the other element, where lOOx per cent. is the maximum concentration of that element in stainless steels, adding aliquots of standard calcium solution, carrying through the solvent-extraction procedure and obtaining a calibration graph. The elements used with their respective concentrations are shown below.graph Element . . . . Chromium Nickel Molybdenum Manganese Titanium Percentage present . . 30 25 3 2 1 These calibration graphs were identical with that obtained for extracted solutions of 1 g of Specpure iron plus added amounts of calcium. These results showed that residual amounts of other elements likely to be present in the aqueous phase of a steel solution, after solvent extraction, will have no effect on the calcium signals obtained in their absence. However, it should be pointed out that the above remarks have only been substantiated for amounts of calcium not in excess of 6 pg ml-l in the solutions that are ready for spraying. This corresponds to a calcium concentration of 125 p g g-l in a stainless steel. For larger amounts of calcium there may be interference effects.The instrumental conditions for the atomic-absorption spectrophotometric determination of calcium are given later. MAIN STUDY A method was now available for the determination of calcium in stainless steels. How- ever, two slight modifications to the tentative procedure were considered desirable. When the tentative procedure was applied to stainless steels, the final solution usually contained972 [Analyst, Vol. 94 a small precipitate of hydrated silica. As this precipitate could possibly contain a trace of a calcium silicate mineral that had remained undissolved on dissolution of the steel, it was decided to bring all of the constituents into solution at an early stage in the analysis. This was achieved by dissolving the steel with aqua regia in a Teflon beaker, evaporating the solution to about 7 ml, adding 3 drops (0.15 ml) of hydrofluoric acid solution (40 per cent. w/w) and simmering the solution for 15 minutes, with the beaker partly covered with a Teflon cover.Secondly, the volume of pyridine used in the solvent extraction was increased from 10 to 15 ml. With 15 ml of pyridine the concentrations of pyridine and pyridinium ion in the aqueous phase are about equal and the buffering capacity of the solution is at a maximum. With 10 ml of pyridine the ratio of pyridine to pyridinium ion is about 1 : 3. The final method for the determination of calcium in stainless steels is, therefore, as follows. HEADRIDGE AND RICHARDSON : DETERMINATION OF TRACE AMOUNTS When this procedure was adopted the final solutions were always clear.METHOD Dissolve 1 g of steel in 20 ml of concentrated hydrochloric acid and 5 ml of concentrated nitric acid in a 150-ml Teflon beaker. Evaporate the resulting solution to the first appearance of appreciable amounts of solid material (about 7 ml). Add 3 drops (0.15 ml) of hydrofluoric acid (40 per cent. w/w) and simmer the solution for 15 minutes, with the beaker partly covered with a Teflon cover. Take up the solid in water and dilute the solution to 50 ml. Transfer the solution to a Pyrex separating funnel and shake the solution with 50 ml of acetylacetone. Add 15ml of pyridine and re-shake it. Remove the lower aqueous layer containing the calcium ions and evaporate it to the first appearance of solid. Add 1 ml of potassium chloride solution (2-5 per cent.w/v in potassium ions), take up the solid in water and dilute the solution to 25 ml in a Pyrex graduated flask. Transfer the solution immediately to a dry polythene bottle. Nebulise the solution in the Unicam SP90 atomic-absorption spectro- photometer by using the conditions given in Table 11, and determine the absorbance for this solution. TABLE I1 INSTRUMENTAL CONDITIONS FOR THE DETERMINATION OF CALCIUM Acetylene flow-rate a t 15 p.s.i., 1 minute-' . . . . .. 3.1 Nitrous oxide flow-rate a t 30 p.s.i., 1 minute-1 . . .. . . 5-0 Wavelength for use with the calcium lamp, nm . . . . 422.7 Slit width, mm . . .. . . .. .. . . .. 0-02 Lamp current, mA . . . . . . .. . . . . .. 12 Distance of centre of light path above burner, mm . . .. 10 Scale expansion .. .. . . .. .. .. .. x 3 Prepare the calibration graph for calcium in the following way. Evaporate in each of six 150-ml Teflon beakers, 20 ml of concentrated hydrochloric acid and 5 ml of concentrated nitric acid to 7ml. To each beaker add 3 drops (0.15ml) of hydrofluoric acid (40 per cent. w/w) and simmer the solutions for 15 minutes, with the beakers partly covered with Teflon covers. To each solution add 10.8 ml of 5 M hydrochloric acid and dilute to 50 ml with distilled water. Transfer the solutions to separating funnels and shake with 50 ml of acetyl- acetone. Add 15ml of pyridine and re-shake them. In all instances, remove the lower aqueous layers and evaporate them to volumes of about 10ml. To six 25-ml graduated flasks marked 0, 0.5, 1.0, 1.5, 2.0 and 2-5 pg ml-l of added calcium, add 0, 0.5, 1.0, 1.5, 2.0 and 2.5 ml, respectively, of standard calcium solution B.To each flask add 1 ml of potassium chloride solution (2.5 per cent. w/v in potassium ions). Then transfer the extracted and con- centrated solutions to each flask. Wash out the beakers with distilled water, add the washings to the appropriate flasks and make up to the marks. The flask marked 0 pg ml-l of added calcium is the blank solution for both the calibration graph and the steel solutions. It contains trace amounts of calcium from all of the reagents. Transfer the solutions immediately to dry polythene bottles. If a large Teflon beaker is available, a mixed acid stock solution, which has been extracted with acetylacetone and pyridine, can be prepared; 10-ml aliquots of this stock solution are then transferred by pipette into each 25-ml graduated flask.The solutions are now ready for nebulisation with the Unicam SP90 spectrophotometer. The following spraying procedure should be adopted. With a x3 scale expansion on the recorder, spray the 2.5 pgml-l calcium solution, then distilled water and finally theNovember, 19691 OF CALCIUM IN STAINLESS STEELS BY SOLVENT EXTRACTION 973 0 pg ml-l solution (the blank solution). Subtract the absorbance of the 0 pg ml-l solution from that of the 2-5 pg ml-l solution. The difference, which is the absorbance of the 2.5 pg ml-I solution corrected for the blank, should be about 0.08. If so, the instrument settings are correct for a satisfactory determination of calcium. Now spray the solutions in the following order: (i), distilled water; (ii), 2.5 pg ml-l calcium solution; (iii), distilled water; (iv), 2-0 pg ml-l calcium solution; (v), distilled water; (vi), 2.5 pg ml-l calcium solution; (vii), distilled water; (viii), 1.5 pg ml-l calcium solution; (ix), distilled water; (x), 2.5 pg ml-l calcium solution, and so on.Every second solution should be distilled water and every fourth solution the most concentrated standard solution. Continue in this way until all of the standard solutions, the blank solution and the steel solutions have been sprayed. The transmission of any solution containing calcium is determined by drawing the best lines through the “noise” on the recording for this solution and on the recordings for the distilled water immediately preceding and following that for the solution containing calcium, averaging the values for the distilled water and subtracting this average from the measured value for the solution containing calcium.This procedure corrects for slight drift in the base-line. The transmissions of every calcium-containing solution are converted into absorbances by using appropriate tables, and the absorbance of each solution is divided by the average absorbance for the two sprayings of the most concentrated standard solution that come immediately before and after it. This procedure corrects for any slight drift in flame tem- perature, rate of nebulisation, etc. Construct the calibration graph by plotting these absorbance ratios against concentration of the calcium solution. Read off the concentration of calcium in the steel solution from this graph. The graph does not pass through the origin because there is some calcium in the blank solution.It will be appreciated that the same stock solutions of acids, etc., must be used in the construction of the calibration graph and in preparing the steel solutions. A new calibration graph is necessary only when fresh stock solutions have to be used. RESULTS By using the above method the calcium contents of thirteen stainless steels were deter- mined. These results are shown in Table 111. The blank solution usually contained about 0.6 pg ml-I of calcium. The absorbance for a solution containing 2.5 pg ml-1 of calcium was always about 0.08, after correction for the blank.This gives a 1 per cent. absorption value of 0.12(5) pg ml-l. TABLE I11 RESULTS FOR THE DETERMINATION OF CALCIUM IX STAINLESS STEELS B.C.S. Steel 235/2 246 261 33 1 332 333 334 335 336 337 338 339 340 Calcium content, pg g-l, by the described method 4, 4, 5, 8 7, 7, 10, 11 49, 51, 52, 52 3*, 4*, 4, 4, 5, 6 4, 6, 6, 6*, 7*, 8 2*, 2, 3, 5, 5, 5* 21, 22, 23, 23 12, 13, 14, 16 7, 9, 10, 11 6, 6, 7, 8 12, 12, 14, 16 10, 11, 12, 14 6, 6, 8, 11 * These results were obtained after subtracting a standard addition of calcium equivalent to 25 pg g-’ of calcium in the steel. SOLVENT EXTRACTION OF OTHER ELEMENTS WITH ACETYLACETONE- The percentages of iron, chromium, nickel and twenty-six other elements of interest to the steelmaker, which are extracted under the final conditions used for the analysis of stainless steels, are given in Table IV, together with information on related extraction procedures.974 HEADRIDGE AND RICHARDSON : DETERMINATION OF TRACE AMOUNTS [Analyst, VOl.94 TABLE IV EXTENT OF EXTRACTION OF VARIOUS ELEMENTS FROM THE AQUEOUS PHASE Element Aluminium . . Antimony(V) . . Arsenic(V) . . Bismuth . . Boron . . .. Calcium . . Cerium (111) Chromium(II1j * Cobalt(I1) . . Copper(I1) . . Iron(II1) . . Lead . . .. Magnesium . . Manganese(I1) Molybdenum(V1) Nickel . . .. Niobium(V) . . Phosphorus(V) Selenium(V1) . . Sulphur(V1) . . Tantalum(V) . . Tellurium(V1) . . Thallium(1) . . Titanium(1V) . . Tungsten(V1) . . Vanadium(V) . . Zinc . . . . Zirconium . . Tin(1V) .. .. .. .. .. .. .. . . .. .. .. . . .. .. .... .. . . .. .. . . .. . . . . .. .. .. .. . . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . . . . . . .. .. . . .. . . .. . . .. . . . . . . . . Extraction system and percentage extraded I I1 1x1 96 96 96 0 0 0 0 0 0 96 100 100 2 2 2 0 0 0 90 90 91 98 98 99 96 96 96 97 98 98 97 98 99 81 81 82 0 . 0 0 70 70 70 53 85 87 96 98 99 93 93 93 0'. 0 0 1. 1 1 0 , 0 0 88 88 90 0 0 0 2 2 2 96 100 100 6 8 8 2 0 0 100 100 100 87 92 91 91 92 92 r A -% for calcium, magnesium, iron, chromium and nickel, 50-ml In all instances. exceDt aliquots of aqueous solutiin containing 50mg-or 0.5 mmoles of the element were used in the solvent-extraction procedure. With calcium, magnesium, iron, chromium and nickel, 250 pg, 250 pg, 1 g, 0.3 g and 0.2 g were used in 50 ml of solution, respectively.The 50-ml aliquots also contained 8.25 ml of concentrated hydrochloric acid and 2 drops (0.10 ml) of hydrofluoric acid solution (40 per cent. w/w). Separate aliquots were treated in the following way. Procedure I-Shake the solution with 50 ml of acetylacetone, then add 15 ml of pyridine and re-shake it. (This is the procedure used in the analysis of the stainless steels.) Procedure II-Carry out procedure I, remove the lower aqueous layer and re-shake it with 50ml of acetylactone. Procedure III-Carry out procedure I, remove the lower aqueous layer and re-shake it with 50 ml of chloroform - acetylacetone mixture (4 + 1 v/v). In all instances the extracted aqueous layers were evaporated to the first appearance of solid. This solid was dissolved in a little water and the solution diluted to 25 ml.Aliquots of these solutions were analysed for element content by well established atomic-absorption spectrophotometric, solution spectrophotometric, titrimetric and gravimetric methods. DISCUSSION The results for the determination of calcium in stainless steels are considered to be very satisfactory. The standard deviation in the errors from the mean for the various steels is 1.4 pg g-l. Less precise results for the determination of calcium in some of these stainless steels by two direct methods are available in a restricted report.4 In nearly all instances our resulls are in good agreement with these results. The results of the solvent-extraction studies (procedure I) indicate that a method similar to that used for calcium could probably be used for the determination of antimony, arsenic, magnesium, selenium, tellurium and thallium in steels. The 2 per cent. of thallium extracted can be attributed to aerial oxidation of a small amount of thallium(1) to thallium(III), whichNovember, 19691 OF CALCIUM IN STAINLESS STEELS BY SOLVENT EXTRACTION 975 is then extracted. This could be prevented by the presence of a suitable reducing agent. It will be noted in Table IV that re-shaking of the aqueous phase with more acetylacetone or with an acetylacetone - chloroform mixture results in more complete extraction with a few of the elements, but with most a second extraction makes little difference to the amount extracted by one shaking with acetylacetone and pyridine. We thank the BISRA/Inter-Group Laboratories of the British Steel Corporation for a grant towards this work, and Mr. P. H. Scholes of BISRA and Dr. M. S. Taylor of Firth-Brown Ltd. for discussions on the project. REFERENCES 1. 2. 3. 4. BISRA Report, MG/D/404/69. Starg, J., with the assistance of Irving, H., “The Solvent Extraction of Metal Chelates,” Pergamon Tanino, K., and Kitahara, S., Sci. Pap. Inst. Phys. Chew. Res., Tokyo, 1967, 61, 35. Press, Oxford, 1964, p. 56. - , op. cit., p. 53. Received March 18th. 1969 Accepted April 29th, 1969

ISSN:0003-2654    

DOI:10.1039/AN9699400968

出版商:RSC    

年代:1969

数据来源: RSC

摘要:

976 Analyst, November, 1969, Vol. 94, $@. 976-978 Determination of Tin in Geological Material by Neutron-activation Analysis BY 0. JOHANSEN AND E. STEINNES (Inslitutt for A tomenergi, Isotope Laboratories, Kjeller, Norway) A neutron-activation method for the determination of tin in geological material, based on tin-113 (half-life 112 days), is described. Tin is radio- chemically separated from the irradiated samples after alkaline fusion, by a method consisting in two sulphide precipitations with an intermediate extraction of tin from 5 M ammonium thiocyanate with diethyl ether. The sensitivity of the method at a neutron flux of 10la neutrons cm-2 s-1 is about 0-2 pg of tin, if irradiation is applied for 14 days. ALTHOUGH tin is not one of the elements exhibiting very high sensitivity in neutron-activation analysis, several investigations involving the use of the technique have been carried out to determine this element in various matrices.Applications of neutron activation for the determination of tin in geological material, however, seem to be few. Hamaguchi, Kawabuchi, Onuma and Kurodal have determined tin in silicate rocks by a method based on a number of sulphide precipitations and solvent-extraction steps. Kies12 determined tin simultaneously with selenium, arsenic, antimony and mercury in stone meteorites by distillation of the halides and subsequent anion-exchange separation. As there still seemed to be a need for simple neutron-activation methods for determining tin in rocks, an attempt was made to produce such a method, with the extraction of tin from thiocyanate solution into diethyl ether as the principal separation step.This extraction has previously been used by Hagebd, Kjelberg and Pappas3 for the separation of tin from fission-product mixtures, but, as yet, has found no application in neutron-activation work. TABLE I NEUTRON-ACTIVATION SENSITIVITIES FOR SOME TIN ISOTOPES Mass number of stable tin isotope 112 116 120 122 124 Half-life Abundance, of induced per cent. radioisotope 0-95 112 days 14-24 14.5 days 32-97 27.5 hours 4.71 40 minutes 5-98 10 minutes Cross-section reaction, mb4 1300 f 300 of @r) 6 + 2 140 f 30 160 f. 40 200 f 100 Type of Calculated radioactive sensitivity, * EC 7 IT 40 0.2 1.0 B- B-’ Y B - 8 Y 0.6 decay Pg y (113m In daughter) * The sensitivity calculations are based on the following assumptions: neutron flux, neutrons cm-O s-l; irradiation, 14 days or saturation, whichever is shorter; minimum measurable activity, 40 disintegrations s-l.SELECTION OF A TIN ISOTOPE- As several of the ten stable tin isotopes yield radioactive products on neutron irradiation, there exist several possibilities for the determination of tin. Relevant nuclear data for some of the stable isotopes are given in Table I. Figures calculated by the method of Meix~ke,~ showing the sensitivity for each isotope, are also included in Table I. These figures indicate an advantage in using the pure /3-emitter tin-121 (half-life 27-5 hours) for the determination, as carried out by Hamaguchi, Kawabuchi, Onuma and Kuroda. This, however, seemed unfavourable for two reasons.First, the chemical separation has to be performed with samples that are highly active because of sodium-24. Secondly, elaborate separations are generally required to obtain /3-emitters of sufficiently high purity for analysis. It was, therefore, decided to use the tin-113 (half-life 112 days) isotope. The sensitivity for this long-lived isotope, compared with the calculated figure in Table I, can be improved by a factor of about 50 by using y-ray spectrometry as described in this paper. The limit of detection is defined as the amount of tin giving rise to an integrated photopeak area of 1000 counts in 60 minutes of counting. 0 SAC and the authors.JOHANSEN AND STEINNES 977 EXPERIMENTAL APPARATUS- crystal was used for the activity measurements.REAGENTS- A 400-channel y-spectrometer with a well-type, 3 x 3-inch, sodium iodide (thallium) Reagents of analytical-reagent grade quality were used. Tirt carrier solzttion-Dissolve a known weight of tin metal in a few millilitres of aqua Zinc holdback carrier-Dissolve an appropriate amount of zinc chloride in water, and Tin standard-For tin standards 50-mg amounts of 0-l-mni tin foil (99.5 per cent., regia and dilute with water to give a concentration of about 50 mg ml-l of tin. dilute to give a concentration of about 5 mg ml-l of zinc. British Drug Houses Ltd.) were used. SELECTIONS OF SAMPLES- Six geological samples, the tin content of which had been previously investigated by emission spectrography and found to be in the range from 10 to lo00 p.p.m., were selected for analysis. The samples are listed in Table 11.In addition, the US. Geological Survey standard rock W-1, with a tin content of about 3 p.p.m.,B was included. TABLE I1 THE CONTENT OF W-1 AND SOME SELECTED MICAS Single values, Sample p.p.m. Standard diabase W-1 . . .. . . 2-81, 3.30, 3.07, 3.06, 3-39 Biotite, pegmatite, Rasvann, Norway 13.4, 14.8, 15.6, 13.6, 15-4 Muscovite, pegmatite, Korgfjell, Norway 23-1, 21.4, 24-1, 21.8, 18.9 Biotite, Toven granite, Mosjraen, Norway 33.9, 31.8, 35.7, 34-5, 31-3 Biotite, Gneiss, Svenningdalen, Norway . . 50-2, 53.8, 49.3, 51-9, 53.0 Zinnvaldite, Cinovec, Czechoslovakia . . 245, 275, 256, 274, 243 Muscovite, Trardal, Norway . . . . . . 1158, 1125, 1154, 1162, 1229 Average, p.p.m. 3.13 14.6 21-8 33.4 51-6 259 1166 Relative standard deviation, per cent.7.3 7.0 9.1 5.5 3.6 5.9 3-3 IRRADIATION- Finely crushed samples of about 100mg were accurately weighed and wrapped in aluminium foil. Samples and standards were irradiated for 14 days in the JEEP-I reactor (Kjeller, Norway), at a neutron flux of about 1 x 1012 neutrons cm-2 s-l. The irradiated samples were stored for 14 days to allow the decay of short-lived activities. (For W-1, 500-mg samples were used.) RADIOCHEMICAL PROCEDURE- Transfer the sample to a nickel crucible containing 1 ml of tin carrier and 1 ml of zinc holdback carrier, evaporated to dryness under a heating lamp in the presence of excess of sodium hydroxide. Fuse the sample with 2 g of sodium hydroxide pellets with an electro- thermal bunsen. After cooling, dissolve the fusion cake in water and transfer the solution to a 250-ml beaker.Acidify with 10 ml of 3 M sulphuric acid and add 50 ml of acid, in excess. Heat to obtain a clear solution. Add 500 mg of thioacetamide to precipitate the tin, centrifuge and wash the sulphide precipitate twice with 5 ml of 3 M sulphuric acid. Dissolve the precipi- tate in 5 ml of concentrated hydrochloric acid and 0.5 ml of concentrated nitric acid, add 20 ml of 5 M ammonium thiocyanate and allow the solution to stand for 10 minutes. Filter it through a funnel and discard the precipitate. Extract three times with 30ml of diethyl ether and wash the combined ether phases three times with 5 m l of a solution 1 M with respect to ammonium thiocyanate and 0-5 M with respect to hydrochloric acid.Evaporate the ether over 50 ml of 3 M sulphuric acid on a water-bath, filter the solution through a funnel and discard the precipitate. Make a second sulphide precipitation with about 500mg of thioacetamide. Centrifuge, wash the precipitate twice with 5 ml of 3 M sulpliuric acid, followed by 5 ml of water, and transfer it to a porcelain crucible after drying for a few minutes at 110" C. After cooling, transfer to tared counting vials, and weigh it as tin(1V) oxide. The yield is about 60 to 70 per cent. Ignite it for 30 minutes at 1000" C.978 JOHANSEN AND STEINNES ACTIVITY MEASUREMENTS- After 24 hours during which the transient equilibrium tin-113 - indium-113m is estab- lished, the y-spectra of samples and standards were recorded. The counting time was from 5 to 60 minutes, depending on the activity level.The analyses were based on the 0.39-MeV photopeak of indium-l13m, and the peak areas were evaluated according to the method of Covell.‘ CONTAMINATION FROM OTHER RADIONUCLIDES- In a preliminary stage of this work, contamination with zinc-65 was encountered in some samples. After the introduction of zinc holdback carrier, as indicated in the procedure, this interference disappeared, and the decontamination factor for zinc was found to be greater than lo3. In a few samples, slight contamination from antimony-122 - antimony-124 appeared, but did not disturb the tin measurement significantly. Other y-emitters were not identified in the present samples. As the half-life of tin-113 is too long to pennit purity checks by decay measurements, the results might be affected by other long-lived radionuclides emitting y-rays in the 400-KeV area.Inspection of the nuclear data of elements expected to be present in significant amounts indicated gold-198 (half-life 2-7 days) and selenium-75 (half-life 127 days) to be the most likely contaminants. Tracer investigations with these nuclides showed that both elements follow tin closely through the chemical separation steps, but are volatilised during the ignition. For selenium-75, the decontamination factor was found to be greater than 500; for gold-198, a factor of about 50 was found. Considering the low abundance of gold in rocks, and the relatively short half-life of gold-198, the interference from gold-198 should be unimportant in most samples.RESULTS AND DISCUSSION Results for the tin content of the selected mineral and rock samples are listed in Table 11. Five replicate analyses of each sample are reported. The results have been obtained in three independently treated series. The relative standard deviation for each sample is within the range 3.3 to 9.1 per cent., depending on the concentration level. For W-1, the present mean value of 3.13 p.p.m. supports the “recommended value” of 3 p.p.m. for this rock.s The agreement with the previously obtained neutron-activation value of 2.52 p.p.m., by Hama- guclii and Onuma,g is also reasonable. As none of the major elements in the actual samples is a strong neutron absorber] and as the neutron-absorption cross-section for tin is as low as 625 mb,4 possible shielding effects should be unimportant both in samples and standards.No interference from disturbing nuclear reactions is likely to occur in this case, because tin-113 is not produced from uranium fission or from nuclear reactions in neighbouring elements. If no impurities disturb the peak area measurements, the accuracy of the method should be similar to the precision. The limit of detection for tin by the present method, if the samples are irradiated for 14 days at a neutron flux of 1 x 1012 neutrons cm-2 s-l, and a counting time of 1 hour is used, was found to be about 0.2mg. We thank Professor I;. &I. Vokes, who provided samples for the present work. REFERENCES 1. 2. 3. 4. 6. 6. Fleischer, M., Geochim. Cosmochim. Acta, 1965, 29, 1263. 7. 8. Hamaguchi, H., Kawabuchi, K., Onuma, N.. and Kuroda, R., Analytzca Chim. Acta, 1964,30, 336. Kiesl, W., 2. andyt. Chem., 1967, 227, 13. Hagebar, E., Kjelberg, A., and Pappas, A. C., J . Inorg. Nucl. Chem., 1962, 24, 117. Hughes, D. J., and Schwartz, R. B., Report BNL-325, 1968. Meinke, W. W., Science, 1955, 121, No. 3137, 177. Covell, D. F., Analyt. Chem., 1959, 31, 1785. Hamapchi, H., and Onuma, N., unpublished work (1967), cited in Fleischer, M., Geochim. Cosmo- Received February 3rd, 1969 Accepted April 14th, 1969 chim. Acta, 1969, 33, 65.

ISSN:0003-2654    

DOI:10.1039/AN9699400976

出版商:RSC    

年代:1969

数据来源: RSC

摘要:

Analyst, November, 1969, Vol. 94, p p . 979-984 979 Determination of Lutetium, Ytterbium and Terbium in Rocks by Neutron Activation and Mixed Solvent Anion-exchange Chromatography* BY A. 0. BRUNFELT (Mincralogical-Geological Museum, University of Oslo, Savsgt. 1, Oslo 6) AND E. STEINNES (Institutt for A tomenevgi, Isotope Laboratories, Kjellev, Norway) Lutetium, ytterbium and terbium have been determined in some standard rocks from U.S. Geological Survey by neutron-activation analysis. A chro- matographic procedure involving anion exchange in a mixed solvent system has been developed to obtain selective separation of the required radioisotopes from other activities induced during the irradiation. Lutetium and ytterbium are eluted in the same fraction, while terbium is obtained in a separate fraction.Chemical yield determination is performed by re-activation. The precision of the method is found to be about A5 to 15 per cent. for the samples analysed. ALL of the fourteen stable rare earth elements can be determined by a combination of neutron- activation and chromatographic methods,ls2g3 but these methods are time consuming and analysis of a large number of samples would in many instances be too laborious. However, the geochemical behaviour of the rare earth elements will, for many purposes, be adequately described if only seven or eight members of this group are determined. Considerable efforts have therefore been made to develop simple methods for the determination of some of the rare earth elements. The application of sodium iodide (thallium) scintillation detectors in neutron-activation analysis for the instrumental determination of lanthanum and samari~m,~ as well as in destructive analysis for cerium and europium in silicate rocks, based on anion-exchange separation from nitric acid - ethanol media,K has been demonstrated in our earlier work.To facilitate also the study of some of the heavy rare earth elements, a similar neutron-activation method for the determination of lutetium, ytterbium and terbium has been developed. The conditions chosen for the method of separation used were based on the anion- exchange study of the rare earths made by Faris and Warton.6 Previously, Desai, Krishna- moorthy Iyer and Sankar Das7 have used similar separation techniques for lanthanum, sam- arium] scandium and ytterbium in activation analysis of geochemical samples.Neutron- activation analysis involving a rapid radiochemical separation based on cation exchange with a-hydroxyisobutyrate as eluting agent for some of the rare earths has been reported.8~~ Chase, Winchester and CoryelllO have also determined lanthanum, samarium and dysprosium in geological materials by activation analysis based on a simple chemical group separation of the rare earth elements. While the present work was in progress, high resolution german- ium(1ithium) detectors were reported to have been used in the non-destructive and destructive determination of several rare earth elements, with application to silicate rocks.11J2~~~14 EXPERIMENTAL REAGENTS- Reagents of analytical-reagent grade quality were used.Anion-exchange resins-Convert the resin Dowex 1-X8 (100 to 200 mesh; chloride form) into the nitrate form by treatment with 7 M nitric acid. Before use, pre-condition the resin with the mixture nitric acid - methanol (10 + 90 v/v). Carrier solutiouzs-Prepare stock solutions by dissolving amounts of terbium, ytterbium and lutetium oxides in dilute nitric acid, corresponding to about 1 mg of terbium, 5 mg of ytterbium and 1 mg of lutetium ml-l. * Paper presented a t the Second SAC Conference 1968. Nottingham. 0 SAC and the authors.980 BRUNFELT AND STEINNES : DETERMINATION OF LUTETIUM, [Analyst, Vol. 94 Standard solutions-Prepare stock solutions by dissolving terbium, ytterbium and lutetium oxides in dilute nitric acid to give solutions containing about 5pg of terbium, 50pg of ytterbium and 5 p g of lutetium ml-1.APPARATUS- tioned Dowex 1-X8 resin to a height of 8 cm of resin bed, sodium iodide (thallium) crystal was used. IRRADIATION- Finely crushed rock samples of about 100mg were accurately weighed, wrapped in aluminium foil and irradiated for 1 week at a thermal-neutron flux of about 1 x 10l2 neutrons cm-2 s-l. Standard solutions containing 100 p1 of each element were evaporated on aluminium foils and irradiated simultaneously. RADIOCHEMICAL PROCEDURE- Store the activated sample for 5 days to allow short-lived activities to decay. By pipette, introduce exactly 1.00 ml of each carrier solution of terbium, ytterbium and lutetium into a nickel crucible and evaporate carefully to dryness under a heating lamp.Unwrap the aluminium foil, pour the sample quantitatively into the crucible and fuse for 5 minutes with 2 g of sodium hydroxide (pellets). Cool the crucible, leach the melt with water, transfer the suspension to a 250-ml beaker, acidify with concentrated nitric acid and add concentrated ammonia solution until it gives an alkaline reaction. Centrifuge it and discard the supernatant liquid. Wash the precipitate with water and then with methanol. Dissolve the precipitate in 5 ml of concentrated nitric acid and 45 ml of methanol. Leave the solution to stand for half an hour, then filter it through a funnel placed on the top of the prepared ion-exchange column. When the solution has passed through the resin, remove the funnel and discard the precipitate.methanol (10 + 90) to remove the residual activity of scandium-46 (half-life 84 days). Discard the effluent, which also contains some lutetium and ytterbium. Elute the main fraction of lutetium and ytterbium with an additional 50-ml portion of the same solution and collect the eluate in a 100-ml polythene screw-capped bottle. Finally, elute terbium with three 20-ml portions of 7-0 M nitric acid - methanol (20 + 80) and again collect the eluate in a 100-ml polythene screw-capped bottle. A flow-rate of about 0.5ml minute-l should be used in the separations. TREATMENT OF STANDARDS- Transfer the activated aluminium foils with the evaporated standards to separate 100-ml glass beakers and dissolve by adding 30 ml of dilute nitric acid (1 + 2). Heat the beakers on a hot-plate and stir with a glass rod to ensure complete dissolution of the evaporated standard.Dilute the solutions to obtain the same counting geometry as for the eluted fractions. ACTIVITY MEASUREMENTS- The activities of the eluted fractions and standards were assayed by y-ray spectrometry. The relevant nuclear data are shown in Table I. Fig. 1 shows a typical y-spectrum of the eluate containing ytterbium and lutetium, compared with standard spectra of these elements. Ytterbium could be determined by the 396-keV photopeak of ytterbium-175 (half-life 4.2 days), while lutetium was determined by the 208-keV photopeak of lutetium-177 (half-life 6.74 days). In Fig. 2 the y-spectrum of an eluate fraction containing terbium-160 (half-life 72.1 days) is given. No extraneous activities were observed in the terbium fractions. The peak areas were calculated by the method of Covell,l5 except for the 208-keV peak of lutetium-177, which contains a contribution from the 177 plus 198-keV peak of ytterbium-169 (half-life 32 days).In this instance, the total area of the composite peak was calculated, and the contribution from ytterbium to the peak was evaluated from the standard and from the ytterbium content already calculated by means of ytterbium-175. Anion-exchange columns-Load columns of internal diameter 1 cm with the pre-condi- Counting eqztipment-A 400-channel pulse height analyser with a well type, 3 x 3-inch, After the sorption step, wash the column with an additional 50-ml portion of nitric acid Record the y-spectra of the eluted fractions.Transfer to a 100-ml polythene screw-capped bottle.November, 19691 YTTERBIUM AND TERBIUM IN ROCKS TABLE I DATA FOR NUCLEAR REACTIONS INVOLVED IN THE ANALYSIS 981 Half-life Abundance Cross- of radio- Nuclear of stable section, nuclide, y-Energy used in the analysis, Element reaction nuclide, per cent. b days keV Terbium 16@Tb (n,y) l6oTb 100 46 72- 1 87, 298,* 879 + 962 + 966 Ytterbium la8Yb (n,y) lsgYb 0.135 11,000 31.8 X-ray + 63 174Yb (n,y) 175Yb 97-41 60 4.2 396 Lutetium 176Lu (n,y) 177Lut 2.59 2100 6.75 208 * Preferred energy. P- t Also produced by the reaction 17aYb (n,y) 177Yb - 177Lu. 1.9 hours X-ray Energy, keV Fig. 1. y-Ray spectrum of eluate containing lutetium and ytterbium compared with spectra of standards of these elements: A, eluate; B, lutetium standard; and C, ytterbium standard DETERMINATION OF CHEMICAL YIELD- After measurement of the activities, dilute the fractions containing terbium, lutetium and ytterbium to 100 ml with distilled water. Seal about 1.2 ml of each solution in polythene tubes and activate for 1 hour, together with aliquots of the carrier solutions diluted in the same manner, at a thermal flux of about 2 x 10l2 neutrons cm-2 s-l.After the irradiation, allow the solutions to cool for 3 to 4 days. Introduce, by pipette, 1.00 ml of each solution into separate polythene vials and again assay the activity by y-ray spectrometry.982 BRUNFELT AND STEINNES : DETERMINATION OF LUTETIUM, [ A fid’ySt, Vol. 94 RESULTS AND DISCUSSION The experimental results obtained in four different irradiations (A, B, C and D) are presented in Table 11.In the first irradiation, ytterbium was measured by the low energy peak (X-ray plus 63-keV y-ray) of the ytterbium-169 nuclide. For some of the rock samples (AGV-1, BCR-1 and W-1) the results seem to be slightly higher than those obtained in the later experiments. This is probably caused by spectral interference from thulium-170 (half-life 129 days). 0962 Id - 0 I I 0 0.50 I -00 I -5 Energy, MeV Fig. 2. y-Ray spectrum of eluate containing terbium TABLE I1 RESULTS FOR ELEMENTS DETERMINED IN STANDARD ROCKS, P.P.M. Experimental results Andesite AGV-1 Basalt BCR-1 Diabase W-1.. Dunite DTS-1 Granite G-2 . . Element .. Tb Yb Lu . . Tb Yb Lu . . Tb Yb Lu . . Tb Yb Lu . . Tb Yb Lu r A 0.73, 0.76, 0.67 (2.23, 2.39, 2-00)* 1.20, 1.10, 1-06 (4.18, 4.01, 4.26)* 0.66, 0.69, 0.80 (2.75, 2.80, 2-48)* < 0.04 < 0.02 <0*03 0-50, 0-45, 0.47 (0.98, 0-70, 0*90)* - - - - B 0.97, 0-77 1.96, 2-11 0.29, 0.24 1.46, 1-34 4.14, 3-98 0.63, 0.60 2.11, 2.52 0.35, 0.39 - - - - 0-53.0-49 0.92, 1.03 0.097, 0-089 C 0.64, 0-63 1.98, 1-86 0-31, 0-30 0.93, 1.27 3-46, 3.41 0.68, 0-56 0.67, 0.69 2.32, 2.21 0.36, 0-38 - - - 0.61, 0.46 0.76 0.104 1 D 0.82 1-74 0.21 1-37 3.46 0.44 0-74 2-30 0.28 - - - 0-60 0.73 0.077 Mean value 0.75 1.91 0.27 1-22 3-69 0.54 0.71 2.29 0.35 < 0-04 < 0.02 < 0-03 0.61 0.86 0.092 Granodiorite GSP-1. . Tb 1.30, 1-34 1.38, 1-22 1.30, 1.23 1-41 1.31 Yb (2-10, 1.97, 2.12)* 1.95, 2-00 1-78, 1.96 1.97 1.93 Lu - 0.28, 0.26 0.30, 0.28 0-17 0.26 - < 0.04 Peridotite PCC-1 .. Tb < 0.04 - - - < 0.05 Yb (0.05 - - - < 0.03 Lu < 0.03 - - * Based on ytterbium-169. These results are not included in the mean values.November, 19691 YTTERBIUM AND TERBIUM IN ROCKS 983 The precision is about +5 to 15 per cent, for the samples analysed in this work. TABLE I11 COMPARISON OF REPORTED RESULTS OF TERBIUM, YTTERBIUM AND LUTETIUM IN DIABASE W-1, P.P.M. Mass spectrometry X-ray - Neutron activation with isotope Present Element enrichment Spark source dilution Previous results results fluorescence, Stable r A \ Terbium 2.018 1,17 0*60,18 0.49,'0 - 0-75,21 0 ~ 8 1 , ~ 0 5 7 , ~ ~ 0*601' 0.7 1 Ytterbium 3.018 1,17 1.618 2.0920 2.1," 2.23,' 2.1," 2-2,18 1.9lS 2-29 Lutetium 0.516 0 ~ 2 , ~ ' 0.29,18 0-2010 0.3420 0.33." 0.35,' 0-35,11 0 ~ 4 4 , ~ ~ 0~35,~' 0.36 In Table 111, the present mean values for W-1 are compared with results reported by other workers who used neutron activation, mass spectrometry and X-ray fluorescence.The results of the present work are in good agreement with other neutron-activation values. The agreement with the stable isotope mass-spectrometric results of Schnetzler, Thomas and Philpotts20 is also good. TABLE IV NEUTRON-ACTIVATION RESULTS OF TERBIUM, YTTERBIUM AND LUTETIUM FOR U.S. GEOLOGICAL SURVEY STANDARD ROCKS, P.P.M. Grano- Andesite Basalt Diabase Granite diorite GSP- 1 Element References AGV- 1 BCR- 1 w- 1 G-2 Terbium Gordon and co- workersle 0.77 f 0.4 1.0 f 0.1 0.57 f 0.16 0.52 f 0-05 1-3 f 0.1 Present work 0.75 f 0.11 1-22 f 0.18 0.71 f 0.05 0.51 f 0.06 1.41 f 0.07 Ytterbium Gordon and co- workerslZ 1.6 f 0.3 3.2 f 0.4 2.2 f 0.4 0.8 f 0.2 2.0 f 0.5 Present work 1.91 f 0.14 3.69 f 0.35 2.29 f 0.15 0.86 f 0-14 1-93 f 0.06 Lutetium Gordon and co- workerslZ 0.37 f 0.06 0.60 f 0.05 0.44 f 0.03 0.18 f 0.08 0.17 f 0.04 Present work 0.27 f 0.04 0.54 f 0.06 0-35 f 0.04 0.0092 f 0.01 0.26 f 0.05 The mean values obtained are compared in Table IV with the results of Gordon and co-workers,12 who analysed the same rocks by using instrumental neutron activation based on germanium(1ithium) detectors.The errors given for the results of the present work are single-value standard deviations of the observed results, while those in the instrumental work were estimated from counting data. The systematic errors of the present method should be rather low.Neutron-shielding effects are not likely to influence the results significantly. The only interfering nuclear reaction that would affect the results considerably is 17eYb (n,y) 17?Yb (p-) l77Lu. The yield of this reaction is fairly low, however, so that the interference would be 1 per cent. or less for all of the samples analysed. The agreement, in general, is good. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. REFERENCES Haskin, L. A., Wildeman. T. R., and Haskin, M. A., J. Radioanalyt. Chem., 1968, 1, 337. Schmitt, R. A., Smith, R. H., Lasch, J. E., Mosen, A. W., Olehy, D. A., and Vasilevskis, J.. Geochim. Cosmochim. Acta, 1963, 27, 577. Towell, D. G., Volfovsky, R., and Winchester, J. W., Ibid., 1965, 29, 569. Brunfelt, A. O., and Steinnes, E., Ibid., 1966, 30, 921. Faris, J. P., and Warton, J. W., Analyt. Chem., 1962, 34, 1077. Desai, H. B., Krishnamoorthy Iyer, R., and Sankar Das, M., Talanta, 1964, 11, 1249. Rengan, K., and Meinke, W. W., Analyt. Chem., 1964, 36, 157. Massart, D. L., and Hoste, J., Analytica Chim. Acta, 1968, 42, 16. Chase, J. W., Winchester, J. W., and Coryell, C. D., J. Geophys. Res.. 1963, 68, 567. Cobb, J. C., Analyt. Chem., 1967, 39, 127. Gordon, G. E., Randle, K., Goles, G. G., Corliss, J. B., Beeson, M. H., and Oxley, S. S., Geochim. , , Chem. Geol., 1967, 2, 199. -- Cosmochim. Acta, 1968, 32, 369.984 13. 14. 16. 16. 17. 18. 19. 20. 21. BRUNFELT AND STEINNES Tomura, K., Higuchi, H., Miyaji, N., Onuma, N., and Hamaguchi, H., Analytica Chim. Acta, 1968. 41. 217. Tomura, K.,kiguchi, H., Onuma, N., and Hamaguchi, H., Ibid., 1968, 41, 389. Covell, F., Analyt. Chem., 1959, 31, 1785. Aleksiev, E., and Boyadjieva, R., Geochim. Cosmochim. Acta, 1966, 30, 511. Brown, R., and Wolstenholme, W. A., Nature, 1964, 201, 598. Taylor, S. R., Geochim. Cosmochim. Acta. 1965, 29, 1243. Nicholls, G. D., Graham, A. L., Williams, E., and Wood, M., Analyt. Chem., 1967, 39, 584. Schnetzler, C., Thomas, H. H., and Philpotts. J. A., Ibid., 1967, 39, 1888. Haskin, L. A., and Gehl. M. A., J. Geophys. Res., 1963, 68, 2037. Received February 3rd. 1969 Accepted May 5th, 1969

ISSN:0003-2654    

DOI:10.1039/AN9699400979

出版商:RSC    

年代:1969

数据来源: RSC

摘要:

Analyst, November, 1969, Vol. 94, $9. 985-988 985 The Determination of Amosite and Chrysotile in Airborne Dusts by an X-ray Diffraction Method BY K. GOODHEAD AND R. W. MARTINDALE (Ministry of Technology, Laboratory of the Government Chemist, Cornwall House, Stamford Street, London, S.E. 1) A method for the determination of asbestos in airborne dusts is described, in which an X-ray diffraction technique is used with photographic recording. This paper describes the method as applied to amosite and chrysotile, but it is equally applicable to other types of asbestos. With a sample size of 30 mg the coefficient of variation is within 10 per cent. over the range 15 to 100 per cent. of asbestos. Smaller sample sizes can be used if lower accuracy is acceptable. IN recent years the increasing awareness of the health hazard presented by airborne asbestos dust, which can cause asbestosis1 and mesothelioma,2 has given rise to the need for an efficient and preferably quick method for the quantitative determination of this material in admixture with the many other compounds commonly found in dust.The chemical determination of asbestos in dust is complicated by the variety of asbestos types as well as the presence of the other constituents, which vary from sample to sample. Furthermore, certain types of asbestos, chrysotile in particular, are decomposed by acid treatment. The X-ray diffraction technique can easily distinguish between the different types of asbestos and, unless there is considerable interference from the other constituents of the dust, whose diffraction patterns will be superimposed on that of the asbestos, it can be used to determine the percentage present.The asbestos content of dusts has been determined for several years in this laboratory and a review of the method together with a thorough investigation of the quantitative reproducibility has recently been undertaken. The two most commonly encountered types of asbestos in our experience are amosite and chrysotile. The results obtained for these materials are reported in this paper, but work is continuing on the applicability of the method to other types of asbestos, particularly crocidolite, which is the third most commonly encountered variety. Recently published reports3y4 of an X-ray diffraction technique for the determination of asbestos in dusts require relatively large sample areas in order to use a diffractometer, as opposed to the small sample sizes and less elaborate equipment required for the technique we describe.In another recent method5 the infrared spectroscopic technique used is highly sensitive but is non-specific and requires confirmation by an X-ray diffraction method. EXPERIMENTAL APPARATUS- A Philips PW1009 X-ray generator is used with copper Koc radiation for iron- and cobalt- free samples and a cobalt Ka radiation for samples containing these elements. The diffraction patterns are normally recorded with a Philips 6cm diameter powder camera. If they give an insufficiently resolved diffraction pattern, then a Unicam 9cm diameter camera, which has been modified so that the film is in the Bradley-Jay position, is used.PROCEDURE- in warm acetone and the dust collected by centrifuging. The dusts are collected on cellulose acetate filters. The filter material is first dissolved 0 SAC; Crown Copyright Reserved.986 [Alzat?yst, vol. 94 A qualitative examination of the dust is then undertaken to determine the type of asbestos present and also the possibility of interference with the quantitative determination by the other constituents of the dust. For this examination the sample is hand-ground under light petroleum (boiling-point 40” to 60” C) in an agate pestle and mortar and is mounted in the camera in an 0.3 mm diameter lithium-glass tube. The diffraction pattern is recorded on Ilford Industrial G X-ray film. The other crystalline phases present are varied, but generally one or more of quartz, chalk, iron oxide or a clay or similar material is present.For the quantitative examination it is necessary to add an internal standard to the sample to compensate for variations in the exposure conditions and development of the film. The internal standard used for this determination is elemental silicon, 0-010g of which is added to 0.030 g of the hand-ground sample. The mixture is further ground for 10 minutes in an RIIC Grindex. Cyclohexane is found to facilitate grinding in this instance. The sample is mounted in the camera in an 0.3 mm diameter lithium-glass tube as previously, but the quantitative diffraction pattern is recorded on Ilford Industrial B X-ray film. If there is less than 0.030g of sample then these figures can be reduced, keeping the sample-to-silicon ratio constant, with a corresponding reduction in accuracy.We consider 0.005 g of sample to be the smallest that can be handled with reasonable accuracy. From the examination of the “qualitative film” it will be known not only which types of asbestos are present but also which of the diffraction rings of the asbestos are free from interference from the other components of the dust and can thus be used as analytical lines in the quantitative determination. The intensity of each of the analytical lines selected is then measured on the quantitative film by using a Joyce Loebl microdensitometer. The ratio of the intensity of the asbestos line to the internal standard line is then calculated and the value obtained referred to a calibration graph to determine the amount of asbestos present.GOODHEAD AND MARTINDALE : DETERMINATION OF AMOSITE AND The lines normally used for analysis are- Amosite . . . . 8.3 A, 2-77 A, 2-64 A, 2-51 A Chrysotile . . . . 7-36 A, 3-66 A Silicon . . . . 3.14 A, 1-92 A RESULTS Standard mixtures of asbestos and internal standard were weighed, with the ratios of asbestos to internal standard 0.5 : 1, 1.0 : 1, 1.5 : 1, 2.0 : 1, 2-5 : 1 and 3-0 : 1. Three standards of each value were weighed, two of which were given three exposures, and the third four exposures. The intensity of the analytical lines on both sides of the film were measured, giving twenty readings. The coefficient of variation of the readings was found to vary between 5 and 10 per cent.of the value in the range 15 to 100 per cent. of asbestos, being best in the middle of the range and falling off at both the low and high ends of the scale, as seen in Table I. TABLE I VARIATION OF THE ACCURACY OF THE METHOD WITH ASBESTOS CONCENTRATION Standard asbestos-to-silicon ratio 0-5: 1 1.0 : 1 1.5 : 1 2.0: 1 2.5 : 1 3.0 : 1 Coefficient of variation of standards r 1 Chrysotile Amosite 9 9 8 8 6 8 5 8 7 8 8 8 The calibration curves are slightly sigmoid, which is to be expected when a photographic process is involved and no correction made for the characteristics of the emulsion. The deviation from linearity is slight, however, as shown in Figs. 1 and 2, and the calculated best straight line through the points will give a result to within 10 per cent.of the correct value in most instances. This may suffice for survey work, but to obtain the highest accuracy one must take account of the deviation from linearity of the calibrations.November, 19691 CHRYSOTILE IN AIRBORNE DUSTS BY AN X-RAY DIFFRACTION METHOD 987 Like all minerals, asbestos obtained from different sources will have variable chemical composition, which may alter the diffraction pattern. Hence it is desirable to construct the calibration curves from material originating from the same source as that being analysed. In this laboratory we analyse samples obtained from many sources, so with each batch of samples a standard made from the raw asbestos used in the premises from which the dust originated is also examined and correction applied for any small deviations from the calibra- tion curves.In practice it is found that this procedure is justified in most instances, but when the deviation from the calibration is large a new calibration is constructed with the raw material that has given rise to the dust sample. Amoslte, per cent. Fig. 1. Chrysotile calibration: chrysotile 3.66 A - silicon 1.92 A Y O I Chrysotile, per cent. Fig. 2. Amosite calibration : amosite 8.3 A - silicon 3-1 A Synthetic mixtures of chrysotile and corundum were prepared containing 20, 30, 50 and 70 per cent. of chrysotile and analysed by this method to test its validity for actual samples. Each sample was given a single exposure and both sides of the film were read, giving a result which is the mean of two readings for each line.Corundum was chosen as diluent because of its strong line at 3.48 A, which is sufficiently close to the 3-66 chrysotile line to give some interference. The 7.3 A chrysotile line and both silicon lines are free from interference by988 GOODHEAD AND MARTINDALE the corundum. The results obtained are listed in Table I1 and are seen to be well within the limits quoted. Also it can be seen that there is no significant difference between the results given by the 7-3 A chrysotile line and the 3-66 A chrysotile line, even although the latter is overlapped slightly by the 3.48 A corundum line. TABLE I1 PRECISION OF THE METHOD IN SYNTHETIC SAMPLES 20 per cent. Chrysotile line 7.3 A Chrysotile - 3.1 A silicon . . Measured, per .. 18 20 3.66 A Chrysotile - 3.1 A silicon . . 3.66 A Chrysotile - 1.92 A silicon. . .. 21 .. 30 per cent. Chrysotile 7.3 A Chrysotile - 3.1 A silicon . . 3.66 A Chrysotile - 3.1 A silicon . . 3.66 A Chrysotile - 1-92 A silicon. . .. .. . . 27-5 28 31 48 49.5 3-66 A Chrysotile - 1-92 A silicon. . .. 51.5 7.3 A Chrysotile - 3-1 A silicon . . 68.5 3.66 A Chrysotile - 3.1 A silicon . . 64.5 3.66 A Chrysotile - 1-92 A silicon. . .. 64.5 50 per cent. Chrysotile 7.3 A Chrysotile - 3.1 A silicon . . 3-66 A Chrysotile - 3.1 A silicon . . .. .. 70 per cent. Chrysotile .. .. cent. Absolute error 2 0 1 2.5 2 1 2 0-5 1.5 1-5 5-5 5.5 Error, per cent. 10 0 5 We thank the Government Chemist for permission to publish this work. REFERENCES 1. 2. 3. 4. 5. Smith, K. W., A.M.A. Archs Ind. Hlth, 1955, 22, 198. Annual Report of H.M. Chief Inspector of Factories on Industrial Health, H.M. Stationery Office, Crable, J. V.. Amer. Ind. Hyg. Ass. J.. 1966, 27, 293. Crable, J. V., and Knott, M. J., Ibid.. 1966, 27, 383. Gadsden, J. A,, and Smith, \V. L., “The Determination of Chrysotile in Airborne Asbestos,” Received September 5th, 1968 Accepted May 7th, 1969 London, 1967, p. 15. Warren Spring Laboratory Report LR93 (PCS), 1968.

ISSN:0003-2654    

DOI:10.1039/AN9699400985

出版商:RSC    

年代:1969

数据来源: RSC

摘要:

Analyst, November, 1969, Vol. 94, $9. 989-991 989 A Comparison of Some Methods for the Determination of Tungsten BY BARBARA CROSSLAND AND T. R. F. W. FENNELL (Royal Aircraft Establishment, Farnborough, Hants.) Four methods, one titrimetric and three gravimetric, for the determination of tungsten in the 60 to 100-mg range have been compared, and a study of the effect of the presence of molybdenum was made. In the absence of molybdenum, precipitation with benzidine and ignition to tungsten trioxide is preferred. Precipitation from peroxotungstate solution prevents inter- ference by molybdenum at the level tested but recoveries are slightly low. GRAVIMETRIC methods predominate in the procedures proposed for the determination of tungsten in the 50 to 100-mg range, many of them involving the use of organic precipitants and subsequent ignition to tungsten tri0xide.l A more recent exception is the use of tri- butylamine to precipitate the tungstophosphate, which can be weighed directly.2~~ Dams and Hoste,4 quoted by Belcher and WilsonP6 reported that precipitation of tungsten trioxide from peroxotungstate solution gave recoveries superior to those given by methods involving the use of cinchonine or /3-naphthoquinoline as precipitants. Lukes investigated the use of various metals for reduction of tungsten so that a titrimetric method could be used and concluded that lead was the best reductant. We decided to examine these last three methods in comparison with precipitation with benzidine, the method adopted for assay of AnalaR sodium tungstate.' EXPERIMENTAL TUNGSTATE SOLUTIONS- The aqueous solutions used were prepared from sodium tungstate or tungsten trioxide, the AnalaR salt being taken for the former.Tungsten trioxide samples (analytical-reagent and Specpure grades) were weighed into polythene beakers and dissolved by warming in 5 ml of 2 N sodium hydroxide. TITRIMETRIC METHOD- Luke's methods was followed, with samples containing about 5Omg of tungsten. The whole procedure had to be carried out as rapidly as possible, and it was necessary to maintain a rigid procedure so that no delay could occur at any stage. Lead salts frequently precipitated in the receiving flask, but unless this happened early in the reduction process, it did not appear to affect the titration value. This precipitation could be minimised by rinsing out the column with hot water, then with hot hydrochloric acid (1 + 1) immediately before passing the tungsten solution through it.From time to time, when evolution of gas from the lead became particularly vigorous, the column was treated with hydrogen peroxide, as described by Luke. The lead reduction column was subject to periods of erratic behaviour, and it was never possible to be sure that it was functioning efficiently during a determination. 0 SAC; Crown Copyright Reserved. Reproduced with the permission of the Controller, H.M. Stationery Office.990 TRIBUTYLAMINE METHOD- The method of Miller and Thow2ss was followed, with samples containing 100mg of tungsten. The tributylamine was purified by distillation at a pressure of 3 mm of mercury, the fraction boiling between 56.5" and 60" C being collected.Before use, 1 ml of the purified material was shaken with 4-0 ml of 0.25 N hydrochloric acid. The reagent was added to the tungstophosphate solution in a dropwise manner, as recommended in the original paper2; more rapid addition tended to produce discoloured and contaminated precipitates. An overnight stand at room temperature before filtration was more satisfactory than cooling for 1 hour at 0" C. PRECIPITATION FROM PEROXOTUNGSTATE SOLUTION- The procedure given in the paper by Dams and Hoste4 was varied in detail after some experience. Tungstate solutions, containing 100 mg of tungsten, were neutralised to phenol red with 14 N nitric acid and made up to 25 ml; 25 ml of 14 N nitric acid and 1 ml of 30 per cent.hydrogen peroxide were then added. Complete precipitation was not achieved at the recommended temperature of 60" C, so the solution was maintained for 1Q hours at 80" C on a steam-bath. The precipitate was filtered on to a paper-pulp pad and dried at 120" C for 1 hour before being ignited at 700" to 750" C. This ignition temperature was chosen, rather than the 800°C recommended by Dams and Hoste, to prevent volatilisation of tungsten trioxide (cf. Norwitzs). BENZIDINE METHOD- Solutions, containing 100 mg of tungsten, were neutralised with hydrochloric acid (1 + 1) with phenol red indicator. The specified procedure' was followed, except that all volumes were halved. Filtration, drying and ignition were effected under the conditions outlined for the peroxo t ungs t at e met hod.CROSSLAND AND FENNELL: A COMPARISON OF SOME [APtdyst, VOl. 94 RESULTS The results, expressed as percentage recoveries, are summarised in Table I. TABLE I PERCENTAGE RECOVERY OF TUNGSTEN - Titrimetric r- Number of deter- Mean minations Sodium tungstate 98.71 4 Tungsten trioxide 98.28 12 Tungsten trioxide - - (AnalaR) (analytical-reagent grade) (Specpure) (pooled) Standard deviation 0.116 Method A Tributylamine Peroxotungstate && Number Number of deter- of deter- Mean minations Mean minations 100*18 4 99.49 6 100-06 12 99-31 18 100.24 4 99.52 6 0.065 0.093 1 Benzidine - Number of deter- Mean minations 99.86 5 99-44 14 99-96 6 0.070 The effect of the presence of molybdenum was briefly examined by adding 0.75 mg of The apparent molybdenum (as sodium molybdate) to the sodium tungstate solution.percentage recoveries of tungsten obtained are given in Table 11. TABLE I1 EFFECT OF MOLYBDENUM ON THE DETERMINATION OF TUNGSTEN Method Titrimetric Tributylamine Peroxotungstate Benzidine L I > Apparent percentage recovery of tungsten 99-52 101.48 99-46 100.46November, 19691 METHODS FOR THE DETERMINATION OF TUNGSTEN 991 DISCUSSION No exact figures for percentage purity were available for any of the materials studied. The AnalaR sodium tungstate was claimed to be not less than 99.0 per cent. pure, and the Specpure tungsten trioxide certificate indicated negbble metallic impurities. If we assume a purity of 100 per cent. for the Specpure tungsten trioxide, the results in Table I indicate that the benzidine method is without bias, whereas the tributylamine method has a positive bias (+0.24 t 0.06 per cent.) and the peroxotungstate method gives low recoveries (-0.48 & 0.08 per cent.).The last method is, however, the only one of those tested that is not affected (Table 11) by the presence of small amounts of molybdenum, even when a temperature of 80” C is used for precipitation (cf. Dams and Haste*). The titrirnetric method, besides being less precise than the precipitation methods, is clearly not stoicheiometric. The results confirm Luke’s conclusions that the reduction is only about 99 per cent. efficient. 1. 2. 3. 4. 6. 6. 7. 8. REFERENCES Chalmers, R. A., in Wilson, C. L., and Wilson, D. W., Editors, “Comprehensive Analytical Chem- istry,” Volume lC, Elsevier Publishing Co., Amsterdam, London, New York and Princeton, 1962, p. 698. Miller, C. C., and Thow, D. H., AnaZyst, 1959, 84, 440. Belcher, R., and Wilson, C. L., “New Methods of Analytical Chemistry,” Second Edition, Chapman and Hall Ltd., London, 1964, p. 346. Dams, R., and Hoste, J.. Tulunta, 1961, 8, 664. Belcher, R., and Wilson, C. L., op. cit., p. 242. Luke, C. L., AnaZyt. Chem., 1961, 33, 1366. “AnalaR Standards for Laboratory Chemicals,” Fifth Edition, British Drug Houses Ltd., and Hopkin and Williams Ltd., 1967, p. 333. Norwitz, G., AnaZyt. Chem., 1961, 33, 1266. Received March 21st, 1969 Accepted May 8th, 1969

ISSN:0003-2654    

DOI:10.1039/AN9699400989

出版商:RSC    

年代:1969

数据来源: RSC

摘要:

992 Analyst, November, 1969, Vol. 94, $9. 992-1001 The Size Distribution of Particles Contaminating Parenteral ,Solutions BY M. J. GROVES (Pharmacy Department, Chelsea College of Science and Technology (University of London), Manresa Road, London, S. W.3) Some problems in the formulation of a standard for the minimum accept- able level of particulate contamination in large volume parenteral solutions are discussed. The proposed Australian Biological Standards Laboratory standard, in which an electrical method is used to detect particulate matter, makes some assumptions about the nature of the particle-size distribution. These are critically examined from the published results. Experimentally, it is confirmed that the relationship between particle diameter D and number N is d N / d ( D ) = C P M where C and M are constants.A membrane filtration method for the determination of particle numbers is shown to have serious limitations for the type of solution at present available, as there are few particles present that are larger than the lowest practical limit of detection (10 to 15 pm). Examination of forty-five containers of sodium chloride injection B.P. from seven different batches with a modified Coulter Counter shows that, generally, the material in each container can be regarded as being unique to that container. The assumption that the size distributions are parallel within a batch is shown to be unsound. The main problem in setting an adequate standard is that of accurately defining the count of some threshold level in a solution that is clean and, therefore, relatively free of particles.The statistics of the particle-size distribution are examined and a simplified procedure suggested to define the count at a threshold and the slope of the particle-size distribution. There is no direct relationship between these two parameters, but materials that would have been regarded as unsatisfactory by the standard proposed earlier could be eliminated by a relationship of the form log N1.0 - 2.5 M 3 0.5 where Nl.o is the estimated number of particles of diameter 1.0p.m and M is the slope of the log - log distribution. FOREIGN particulate matter, definedl as mobile undissolved substances unintentionally present in parenteral solutions, i.e., for injection into the human body, has been a problem in the pharmaceutical industry since the introduction of injectable preparations.Although such particles are undesirable, they are, nevertheless, unavoidable. At present there are insufficient grounds for stating clearly what is an acceptable level of contamination. Relevant official or governmental standards are subjective and arbitrary, largely because of lack of objectivity in inspection methods. In recent years instrumental methods of quantitatively assessing this contamination have been described,2s39* p6 and standards based on these methods have been propo~ed.~ s4 Material larger than about 50 pm can be detected visually without opening the sealed container, if necessary with the aid of specially devised ap~aratus.~s* s9 ,lo However, the more sensitive test methods, such as membrane filtration or the Coulter Counter, are destructive so samples must be taken from a batch in much the same way as samples are taken for sterility or pyrogen tests.The problems of sampling have been discussed in detail in the provisional standard published by the Australian Biological Standards Laboratory.6 At present there is some doubt about the limits that should be imposed upon the producer of injectable solu- tions because of the absence of any substantial body of medical opinion on the hazard pre- sented to the patient by the numbers or size of the contaminating particles. The recent 0 SAC and the author.GROVES 993 suggestion that instrumental counts should be regarded as an index of good manufacturing practicell probably offers the best interim solution to this problem until further medical information becomes available.Nevertheless, it is desirable that this index should be set at a realistic level based on a better knowledge of the nature of the contamination, or, if this is not practicable, on the likely form of the size distribution of small numbers of particles arising from chance contamination. Considering a within-container situation as opposed to a batchwise variation, the main problem is that parenteral solutions are already “clean.” Hence it would be anticipated that the particle count at any threshold of particle size would be small and, therefore, difficult to establish with any reasonable degree of accuracy. Particulate contamination can be attributed12 to two main sources: (a) “background,” arising from material that has not been removed by the clarification process or “housekeeping” procedure designed to produce a clean solution ; and (b) accidental contamination, consisting of material left in the container before filling, introduced from the environment or detached from the container or the closure during the subsequent manipulative procedures such as sterilisation.Particles introduced by accidental Contamination are often large in size although small in number and can be detected by visual inspection. Appino and Robinson13 suggested that there are frequently only fifty particles or less per millilitre greater than 5 pm in size in good quality parenteral solutions and most contaminating particles are, therefore, sub-visual “background.” This material is readily detectable quantitatively by instrumental methods, such as the Coulter Counter. Instrumental counts might be used to indicate the use of unsatisfactory filtration materials or inefficient cleaning and “housekeeping” procedures.LIKELY FORM OF THE PARTICLE-SIZE DISTRIBUTION- In general terms it is possible to deduce, from distributions published to date,3s4s5 that parenteral solutions contain few particles per unit volume at a threshold of 5 pm, but often quite considerable numbers of particles at the 2-pm level. This is a generalisation drawn from Coulter Counter data but is also applicable to the data obtained by other quantitative procedures. Groves3 and Ho, Church and Lee5 plotted logarithms of the cumulative numbers of particles against the corresponding particle diameters (volumes) and obtained curves of an exponential type.Vessey and Kendal14 plotted log number against log diameter and obtained approximately straight lines, but gave no reason for this arbitrary choice of presenta- tion. In an alternative method, Marshall14 suggested that the logarithm of the count of a realistically filtered solution over the size range of 1 to 12 pm should not exceed a value of (2.5 - 0.2 D) where D is the particle diameter. The value of this observation is not clear. Particle-size distributions of particulate materials have been discussed by Herdan,ls among others, and often follow a log-normal function. However, when a particulate system has a very narrow size range about a mean, a log - log relationship between particle size and number is often found, although, admittedly, it is only an approximate distribution function.Nevertheless, for materials such as bacteria, starch grains, zinc oxide, calcium carbonate, carbon black or stearic acid, all of which have been reported in intravenous fluids,l6v17 this type of distribution might be preferred, provided they were not present to a great extent. This latter condition would probably be met in parenteral preparations, as even a badly contaminated solution contained only about 15 parts per 100 million by weight of extraneous solids.12 A log - log relationship between the particle size and number has been noted in analogous contexts. Cadlels observed that this relationship proved useful in describing the particulate material in both natural and polluted atmospheres, such as Los Angeles “smog.” Gilvarrylg and Gilvarry and Bergstrom20 examined the fracture of brittle solids and found, theoretically and experimentally, that small fragmented glass particles obeyed a log - log distribution function.This type of size distribution can be conveniently represented by an equation of the form where N is the cumulative number of particles at the threshold corresponding to diameter D, and C and M are constants. As D approaches zero dN/d(D) approaches infinity, which is experimentally observed with parenteral ~ontamination.39~ drv/d(D) = cD-M994 GROVES : SIZE DISTRIBUTION OF PARTICLES [Analyst, VOl. 94 The constant M, the slope of distribution, is of interest.For example, Junge21 reported the value of M was about 4 for particles found in natural atmospheres. R. W. Lines (in a private communication) pointed out that values of M obtained from both published and un- published data range from 1.6 to 5.2. The suggestion by Vessey and Kendall,4 and more recently by Kendall,82 that plots of log cumulative count against log particle size are nearly parallel forms the basis of the Australian standard6 and may have substance. However, in view of Lines' observation this clearly requires confirmation. EXPERIMENTAL MATE RIALS EXAM IN ED- Forty-five samples of sterile sodium chloride injection B.P. were obtained from a variety of sources, both industrial and hospital pharmacy laboratories being represented. At least five containers from each batch were sampled; batches A and G were from the same manu- facturer. Every bottle or container was given a code number on receipt and was sampled more or less at random after initial ranging experiments involving the counting of one sample from each batch.The samples are thought to be representative of good quality solutions available at the time (October, 1968) and included all-plastic containers and rubber-stoppered glass bottles. The characteristics of each batch are described in Table 111. MEMBRANE FILTRATION METHOD- A membrane filtration method was initially investigated as it was claimed23 to have a number of advantages. The experimental technique was based on that described by Trasen,23 as amplified in the Millipore Application Bulletin No. AR-2. This method involves filtration of the sample through a Millipore membrane and microscopic examination by reflected light.Difficulty was experienced in washing the membranes free from surface contamination before sampling. Crystals of sodium chloride were often observed on the surface after drying a sample of saline solution; the drying process itself caused the membranes to curl so that they could not be held flat on a microscope slide. The following method was, therefore, devised and found to offer improvement. The samples investigated were saline solutions taken immediately after examination with a Coulter Counter, as it was hoped to correlate results from the two methods. All work was undertaken in a clean atmosphere (Bassair Laminar Flow Cabinet). A Swinnex 25 filter holder (Millipore Ltd.) was washed in water containing laboratory detergent and then thoroughly rinsed in a stream of filtered distilled water from a hand-held Millipore Millijet solvent-filtering dispenser, fitted with a 0.46-pm pore diameter membrane filter.A white 0-45-pm membrane, overprinted with a black grid (Millipore No. GSWP), was rinsed in a stream of filtered isopropyl alcohol from the solvent-dispenser unit and fitted into the filter holder with a silicone rubber gasket, rinsed in the same way. The assembly was placed in a covered glass dish and dried overnight at 60" C. After Coulter analysis the saline container under examination was weighed and the filter- holder assembly attached to the lower end of the administration set (see next section).Most of the remaining contents were withdrawn from the container in a slow stream and the container re-weighed to enable the sample size to be determined. About 1 O O m l of filtered distilled water were then passed through the filter to remove water-soluble constituents of the parented solution remaining within the holder. The membrane filter was then removed from the holder and dried in a glass dish under a cover at 60" C. This drylng stage was found to be critical in subsequent examination and required at least five days to remove the final traces of moisture. The membrane was then placed in a few drops of filtered immersion oil (Microil, George T. Gurr Ltd., refractive index 1.517) on a clean microscope slide. The oil was allowed to penetrate through the membrane, which was then made to lie flat by gently pressing it down with a second clean slide, which was then held in place with transparent tape along the edges.Particles were counted by using a transmitted light microscope with partially (45") plane-polarised light. With a magnification of x 200 and a micrometer stage the membrane was systematically scanned from side to side. By keeping the inked grid lines in focus only those particles on the top surface of the membrane were counted.November, 19691 CONTAMINATING PARENTERAL SOLUTIONS 995 Control membranes were prepared and counted by the same technique but with the sampling stage omitted. Counts obtained from the samples were corrected for the “back- ground,” and for the fact that the effective area of filtration was only over the central part of the membrane.TABLE I PARTICLE COUNTS OBTAINED ON CONTROL (WASHED) MEMBRANE FILTERS Membrane number Number of particles counted over the whole surface 1 93 2 47 3 69 4 80 5 104 6 99 mean count = 82 TABLE I1 PARTICLE COUNTS OBTAINED BY MEMBRANE FILTRATION Volume passed, Batch Sample No. ml A 1 386 B 2 338 C 3 360 D 4 261 E 5 290 F 6 225 Count over whole surface 331 590 738 458 4504 1018 Count adjusted for area and Number background ml-l 177 <1 359 1 466 1 265 1 3130 18 664 3 Estimated number ml-l at 15 pm by Coulter Counter <1 5 4 12 20 12 Results obtained in this experiment are given in Tables I and I1 and illustrate the experimental difficulties of the method. It is clear from Table I that it is difficult to clean the membranes effectively before sampling from a solution.It is also evident (Table 11) that the results are only approximately comparable with Coulter counts at a particle threshold above about 15 pm. Experience suggested that only particles larger than about 10 pm could be distinguished against the granular background of the membrane material unless they were black, highly coloured, irregular in shape or strongly birefringent. Small particles with a refractive index approximately the same as that of the membrane are almost impossible to distinguish and are likely to be missed. A similar point has recently been made by Kenda11.22 COULTER COUNTER METHOD- Coulter Counter methods for determining particulate numbers have been described previously and vary from the use of an open vesse12J or an enclosed sampling t ~ b e .~ , ~ The former system is easier to use but can be contaminated by atmospheric dust. Although free from this defect, the latter system requires to be filled and flushed repeatedly before countings The present work was carried out in a dusty environment and it appeared desirable to eliminate extraneous particulate matter but to allow the orifice of the Counter to sample in much the same way as a patient would receive the solution, i e . , at the end of an adminis- tration set and as a continuous drip. The modification used in this investigation is shown in Fig. 1. The sample of parenteral fluid was taken from the container with a Capon - Heaton administration unit fitted with a filtered-air inlet and de-bubbling chambers.The outlet of the set was removed and the plastic tubing shortened before attaching it to a glass tube fitted with a tap. The lower end of this glass tube was bent upwards and opened out into a funnel shape. The orifice tube of the Counter fitted inside the funnel with the outer electrode wrapped around the tube as shown in the diagram. The whole assembly was clipped to the glassware stand of the Counter.with spring clips and immersed in a beaker fitted with a side-arm and a two-way tap. The flow-rate through the funnel was controlled by the tap at the bottom of the adminis- tration set and was adjusted to one or two drops s-l in the de-bubbler. In this way the orifice was kept immersed in a slow stream of material flowing directly from the container under examination.Experiments with coloured solutions showed that displacement of material in the “dead-space” of the administration set was almost complete provided the flow-rate996 GROVES SIZE DISTRIBUTION OF PARTICLES [Artalyst, VOl. 94 was slow. Air bubbles were effectively removed by the de-bubbler. This was confirmed by changing over two solutions with known levels of contamination. Only 25 to 30 ml of fluid, corresponding to the volume of the “dead-space,” were required before the counting level was established at the new rate. For this reason the two-way tap on the beaker was con- nected to the waste on one arm and to a calibrated receiver on the other, in which the first 30ml of solution were collected after fitting a new container to the sampling device.A = Sample bottle and Capon - Heaton D = Beaker with a two-way tap at the administration set with hollow out- outflow, one arm to a calibrated let and air inlet attached to an air receiver and the other to waste filter (not shown) E = Sampling tube 6 = De-bubbler chambers F = Coulter Counter, electrodes and C = Aluminium foil wrapped around the sample outlet tube and attached to earth Fig. 1. orifice tube The sampling device attached to the Coulter Counter As noted recently by Kendall,22 the extraneous electrical noise of the Counter was in- creased considerably by an increase in the volume in effective contact with the electrodes. However, this was reduced to a very low level by shielding the sampling unit from the top of the glass tube to the liquid level in the bottom of the de-bubbler unit, which itself made a contact break with the bulk of the sample solution.The shielding was made from domestic aluminium foil wrapped around the tubing and was connected to earth with a crocodile clip. It was also found necessary to shield the outflow from the sample beaker in the same manner. The Coulter used in this investigation was a Model A (Industrial) Counter fitted with a 50-pm narrow orifice tube calibrated with polymer latices of known diameter. Four replicateNovember, 19691 CONTAMINATING PARENTERAL SOLUTIONS 997 counts were taken at 0-2-pm intervals between 1-0 and 3.0 pm and increasing intervals from 3-0 pm to 15 pm. Counts were corrected for coincidence and switching pulses. Some typical results are shown in Fig.2 and manipulated data are given in Table 111. Particle diameter, p m Fig. 2. The particle-size distribution of six of the samples of sodium chloride solution selected from three batches. Samples taken from each batch represent the extremes of slope (M) within that batch. Experimental points are the mean counts at that threshold. The curves are the lines of best fit determined from the statistical parameters (Table 111): 0, sample 17, batch E; e, sample 5, batch E; 0, sample 35, batch A; ., sample 1, batch A; A, sample 33, batch B; A, sample 2, batch B MANIPULATION OF THE DATA Although the Coulter Counter is claimed to give replicate counts within 5 per cent. or less, the counts are insufficiently large for adequate statistical sampling to be obtained except at the lowest size thresholds, when counts become much larger.The raw data obtained from the machine were in accordance with the general observation that there was an increase in count as the lower thresholds were approached. When plotted on log - log paper, as shown in Fig. 2, the data were more or less linear and lines could be fitted by the method of least998 GROVES : SIZE DISTRIBUTION OF PARTICLES TABLE I11 CHARACTERISTICS OF BATCHES OF SODIUM CHLORIDE INJECTION B.P. DETERMINED BY COULTER COUNTER [Autalyst, VOl. 94 Batch A 500-ml all-plastic ampoule B 500-ml rubber- plugged glass bottle C 500-ml plastic bag D 1000-ml rubber- plugged glass bottle E 500-ml rubber- glass bottle plugged F 500-ml rubber- glass bottle G 1000-ml all-plastic ampoule plugged Sample number 1 7 8 35 36 37 2 33 34 43 44 45 3 9 10 11 12 13 4 28 29 30 31 32 5 17 18 19 20 21 22 23 24 25 6 14 15 16 26 27 38 39 40 41 42 Correlation coefficient between log dN/d(D) and log D, v 0.985 0.981 0.992 0.994 0.994 0.975 0.985 0.983 0.975 0.988 0-974 0.989 0-985 0.907 0.998 0.984 0.988 0.996 0.990 0.990 0-994 0-995 0.988 0.990 0.981 0.992 0-993 0-993 0.996 0.997 0.982 0.998 0-993 0.993 0.920 0.990 0.975 0-994 0-984 0-993 0-994 0.988 0.994 0-993 0.994 Slope of the curve, -M 2-76 3-19 3-32 5.77 4-19 2.83 2.26 5.02 4-47 4.66 4.01 4.86 3-44 2.28 3.69 2.69 4-05 4.01 2-13 3-40 4-38 4.40 2-48 2.80 2.92 3.71 3-52 3.39 3.45 3.70 3.10 3.33 3.45 2.93 2.69 3-68 2.84 4-00 3-17 4.07 4-02 4-38 4.36 4.76 3-29 log - log Intercept estimated no.of particles at D = 1.0 pm 3.3537 3.8365 4.0399 4.4497 4.1303 3.8983 3.2585 3-7113 4.3217 3.8945 3.4714 3-6822 4-2390 3-0 100 4.2916 2.5520 4.7720 4.6164 3.5217 3-5524 3.5274 3.8631 3-5726 3-4192 4.7225 4.7900 4.1373 4.4732 4.5626 4.8648 4-537 1 4.4469 4-5394 4.4406 4.2923 4.4464 3.6668 3.9802 3.9222 3-9807 3.7401 3.8968 343230 3.6990 4-1721 (log N1.0) S Estimated value no.of particles log NI-o - 2.5 at 3-5 pm M 70 0.3099 100 0.4195 0.4642 205 15 0-3378 35 0-3407 300 0.4937 0.3352 105 10 0.2413 0.4072 75 35 0.2994 6 0.2423 10 0.2435 175 0.5054 56 0.2237 180 0.4858 95 0.3905 350 0-5532 250 0.5273 0.4788 215 50 0.3099 11 0.2339 35 0-3098 180 0.4325 80 0.3282 0.9311 600 0.6178 160 0.4649 450 0.5822 450 0.5984 600 0-6400 600 0-6567 320 0.5843 400 0.5906 0.6632 550 560 0-6673 280 0.5296 110 0.41 10 60 0.3704 160 0.4494 60 0.3638 0.3082 30 30 0-3188 30 0-3033 10 0.25 17 220 0.5089 1350 squares.Particle diameters (instrument thresholds) were tabulated and converted into logarithms (values of X). The corresponding counts (dN/d(D)) were also tabulated and converted into logarithms (values of Y). The correlation coefficients for each sample were then calculated as: CXCY CXY - - Correlation 9t coefficient = Y where n = number of thresholds at which particle counts are taken.November, 19691 CONTAMINATING PARENTERAL SOLUTIONS 999 As will be seen from Table 111, there is a highly significant relationship between X and Y, and this can be written in terms of the familiar relationship y = mx + c, i.e., The two characteristics of this equation are the slope, M, and the constant, C, which is the intercept on the Y axis at the point X = 0.Le.. the cumulative count at a threshold log dN/d(D) = C - Mlog D. corresponding to a particle diamete; The slope is readily found from of 1.0 pm (log film0). the relationship CXCY CXY - - n ZY cx n n log Nls0 = P + MX where P and 1 are the mean values, i.e. , - and - , respectively. The values of r, M and log Nle0 were calculated and are tabulated for each of the samples examined (Table 111). DISCUSSION As noted, the principal difficulty in quantitatively assessing particulate contamination of injectable solutions is the problem of adequately defining a situation in which counts per unit volume are low and the sampling error can be large. From these considerations alone the membrane filtration technique suffers from the disadvantage that only particles larger than about 10 to 15 pm diameter can be distinguished, and at these particle thresholds solu- tions are generally so clean that only a few particles per millilitre are present.l3 This requires the sampling of an inordinate volume of the solution and a considerable increase in counting accuracy by a microscope technique.Nevertheless , the membrane filtration method remains the only means at present available for identifying the nature of the contamination and hence locating its source. From an instrumental point of view the counts at thresholds of 5 pm and below are of greater interest, as the numbers of contaminating particles increase as the particle threshold decreases and statistically the chances of increasing the counting accuracy are improved.Nevertheless, some variation in count would be anticipated at any one instrument threshold. Thus a repeatability of about 5 per cent. is as much as could be expected with the Coulter Counter, especially when the numbers of particles are small. This implies that there is yet another difficulty in specifying that a solution should not contain more than an arbitrary number of particles determined at a single particle-size threshold as a standard of cleanliness. VALIDITY OF THE UNDERLYING BASIS FOR THE PROVISIONAL AUSTRALIAN STANDARD- The provisional Australian Standard6 specified that the mean count from ten containers should not exceed 250 particles greater than 3.5 pm d-l, and that the sum of the mean P h s twice the standard deviation should not exceed 500 particles ml-l.The background to the provisions of this standard has recently been discussed in detail by Kendall.22 For example, the particle count was fixed at 3.5pm because in this region the (instrument) threshold can be set reasonably accurately; there are usually enough particles to give counts of reasonable accuracy; and one of the Dow latices has particles of about 3.5 pm, thus making an ideal calibration point. As the Coulter Counter has a substantially linear potentiometer, the point at which calibration is carried out is largely irrelevant. In addition, Dow latices of this size range are no longer commercially available (Dow latices are supplied by Serva A.G. , Heidelberg).How- ever, the underlying basis for the suggested standard is the observation that the slopes of the size distributions of nearly all the solutions were similar in spite of the great over-all differences in counts. Plots of log count against log size were noted as being not far from linear and nearly parallel; few exceptions to this were noted. The procedure used in this paper allows this assumption to be examined critically. Here the counts are taken at small increases of instrument threshold over a wide range of particle sizes. Any variation in count at one threshold is effectively smoothed out by combining1000 GROVES : SIZE DISTRIBUTION OF PARTICLES [Analyst, VOl. 94 with the counts at other thresholds and by using a statistical procedure to determine the line of best fit.As seen in Table 111, there is good circumstantial evidence for choosing a log - log relationship between particle size and number and although this is an insensitive means of presenting a particle-size distribution the characteristic parameters of each distri- bution can be defined with reasonable precision. It is immediately clear that the slopes (M) of the distributions within a batch of material differ, sometimes within wide limits, and the distributions are not parallel. This is shown in Fig. 2, in which the two limits of slope are plotted from t h e e of the batches of saline examined. The points are the mean experimental counts at each instrument threshold and the lines are the statistical lines of best fit. Even in batch E, in which the extremes of slope are only 2.92 and 3.71, the differences are im- mediately apparent when the experimental results are plotted on a log - log scale.Thus the underlying basis for the provisional Australian standard is unsound and an alternative method of expressing the data is required. 29 0 0 36 0 43 0 410 45 0 33 2.5 3.0 3.5 4.0 4 5 5 0 5 5 M Fig. 3. The relationship between log Nl.,, and M. The samples are numbered (Table 111) and blocked points are those samples which would have been excluded by the provisional Australian specification (> 250 particles per ml >3.6 pm). The earlier full Australian specification (> 1000 at 2.0 pm, 250 at 3.5 pm, 100 at 5-0 p m and 25 at 10-0 pm) is shown as X A POSSIBLE ALTERNATIVE METHOD OF EXPRESSING THE STANDARD- Kendal122 noted that ideally the particle-size distribution should be determined and a,n earlier version of the Standard laid down limits at four different threshold levels (2.0, 3.5, 5.0 and 10.0pm).This is cumbersome to express and manipulate and it is desirable to simplify the test. Allowing for the arbitrary nature of any standard, the present results were examined to determine whether there was a relationship between the two characteristic parameters of the size distributions, viz., the slope M and the intercept log iVle0. The results are shown in Fig. 3 and no direct correlation could be drawn. However, it was noted that most of the individual containers which would have exceeded the requirements of the Australian standard were congregated in an area of the graph arbitrarily defined by a line of equation: log Nla0 = 0.5 M + 2.5.November, 19691 CONTAMINATING PARENTERAL SOLUTIONS 1001 This suggested that the size distribution could itself be characterised by a single parameter in the form log Nle0 - 2.5 M = specific value (S) .If this value was used as an index of cleanliness, solutions with a value of S > 0.5 could be regarded as unsatisfactory, on the same basis as materials which would have been excluded by the Australian standard. This suggestion would have the effect of removing some of the uncertainty that inevitably arises if the slopes of the distributions are not parallel between bottles taken from the same batch. The provisions made in the Australian standard for the number of samples taken from a batch and allowing for batchwise variation could be adapted to this numerical index without difficulty. CONCLUSIONS Although the underlying basis for the methods of presenting a standard for the cleanliness of injection solutions selected for the Australian Biological Standards Laboratory standard is shown to be an over-simplification, it offers a basis for discussion.The method outlined in this paper appears to offer some improvement and is not confined either to saline or to the Coulter Counter method of measurement used in the present investigation. Recently acquired (unpublished) results lend support to this contention, although additional collaborative work is required to establish the method with greater certainty. I would like to thank Mr. R. U. Robinson and Mr.C. E. Kendall for making the type- scripts of their papers to the New York Academy of Sciences available to me in advance of publication. I am also grateful to Messrs. J. A. Baker (Westminster Hospital), J. P. Curtis (Barnet General Hospital) and J. W. Hadgraft (Royal Free Hospital) for the provision of samples for this investigation, and to Mr. R. W. Lines (Coulter Electronics Ltd.) for his helpful discussion at the early stages. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. REFERENCES Krueger, E. O., and Riggs, T. H., Bull. Parent. Drug Ass., 1968, 22, 99. Groves, M. J., and Major, J. F. G., Pharm. J.. 1964, 193, 227. Groves, M. J., J . Pharm. Pharmac., 1966, 18, 161. Vessey, I., and Kendall, C. E., Analyst, 1966, 91, 273. Ho, N. F. H., Church, R. L., and Lee, H., Drug Intelligence, 1967, 1, 356. Australian National Standards Laboratory, Canberra A.C.T. Kendall, C . E., Analyst, 1966, 91, 280. Hudson, H. E., Pharm. J., 1961, 186, 201. Boucher, C. C., and Sloot, €3. A., Pharm. Weekbl. Ned., 1965, 100, 253. Wynn, J. B., Bull. Parent. Drug Ass., 1968, 22, 13. Goddard, J. L., Ibid., 1966, 20, 183. Groves, M. J ., in “Proceedings of the Symposium on Filtration Medical Engineering,” Filtration Appino, J. B., and Robinson, R. U., Ann. N . Y . Acad. Sci., in the press. Marshall, K., Paper presented at a Meeting of the Particle Size Analysis Group of the Society Herdan, G., “Small Particle Statistics,” Second Edition, Butterworths, London, 1961. Garvan, M., and Gunner, B. W., Med. J . Aust., 1963, 2, 140. -- , Ibid., 1964, 13, 1. Cad;, R. D., “Particle Size; Theory and Applications,” Reinhold, New York, 1965. Gilvarry, J. J., Afifil. Phys. Q., 1961, 32, 391. Gilvarry, J. J., and Bergstrom, B. H., Ibid., 1961, 32, 400. Junge, C., J . Met., 1955, 12, 13. Kendall, C. E., Ann. N . Y . Acad. Sci., in the press. Trasen, B., Bull. Parent. Drug Ass., 1968, 22, 1. Society, London, 1969, in the press. for Analytical Chemistry, London, 1968. Received May 6th 1969 Accepted June 5th 1969

ISSN:0003-2654    

DOI:10.1039/AN9699400992

出版商:RSC    

年代:1969

数据来源: RSC

摘要:

1002 Analyst, November, 1969, Vol. 94, $$. 1002-1005 Quantitative Analysis of Mixtures of Alkylphenol Isomers Fluorine-I9 Magnetic Resonance Spectroscopy BY KAZUO KONISHI, YOSHIHIRO MORI AND NOR10 TANIGUCHI (Industrial Research Laboratories, Kao Soap Co. Ltd., 1334 Minato-yakushubata , Wakayama-shi, Jafian) To develop a convenient method for determining 0- and p-alkylphenols in mixtures of these isomers the application of the nuclear magnetic resonance technique has been investigated. The mixed alkylphenols were converted into the corresponding trifluoroacetates and the fluorine magnetic resonance spectra measured. The differences in chemical shifts between the alkylphenol isomers were large enough to permit their quantitative determination by comparing the relative intensities of the resonance pairs.The standard deviation for the peak height measurement method was 0.34 per cent. The method can be applied satisfactorily to the determination of 0- and p-alkyl- phenols in which branching of the alkyl chains is slight. A FEW reports occur in the literature on the quantitative analysis of mixtures of dkylphenol isomers by nuclear magnetic resonance spectr0scopy.l sa Crutchfield, Irani and Yoder have reported that characteristic patterns for the aromatic proton resonance, and for the resonance of the protons on the carbon atoms in the or-position relative to the aromatic ring, can yield information on the ortho-to-para ratio in a1kylphenols.l Lindeman and Nicksic prepared acetate derivatives of alkylphenols that were not greatly hindered, and analysed the mixtures of the isomers by using the distinctive acetate methyl resonances corresponding to the 0- and $-isomers.In these methods, however, the extent to which the ortho-to-para ratio can be determined is limited, either because the resonance signals for the 0- and$-isomers overlap or the difference in chemical shifts for both isomers is insufficient to permit the quantitative analysis of mixtures in which one isomer is present in high concentrations of the other. When the alkylphenol is converted into the corresponding trifluoroacetate, the analysis is much more reliable because the extent of the 19F chemical shifts is greater than those of protons, and there is no interference from the resonance of the protons of the alkylphen~l.~s~s~ The present paper describes the quantitative analysis of mixtures of alkylphenol isomers by fluorine magnetic resonance spectroscopy following their conversion into the corresponding trifluoroacet at es (TFA-alkylphenols) .EXPERIMENTAL APPARATUS- All fluorine spectra were measured at 56.4 MHz by using a JNM-3H-60 type, nuclear magnetic resonance spectrometer (Japan Electron Optics Laboratory Co. Ltd.) equipped with an integrator. 0.5" C and the spectra recorded at a sweep rate of 1.69 Hz s-l. The chemical shifts between the signals were measured by the side-band method, with a standard frequency of 100 Hz. REAGENTS- Measurements were carried out at a temperature of 20" Tri$uoroacetic anhydride-Analytical-reagent grade. Analytical-reagent grade ortho and para isomers of cresol, ethylphenol and s-butylphenol were used without additional purification.Propyl-, hexyl-, nonyl- and dodecylphenol were prepared from phenol and acid chlorides, followed by the Fries reaction and Clemmensen reduction. TRIFLUOROACETYLATION- About 400 mg of alkylphenol (up to hexylphenol) and an excess (700 mg) of trifluoro- acetic anhydride are mixed in a 10-ml Erlenmeyer flask, with a ground-glass stopper, and allowed to stand for 5 hours at room temperature. The reaction is then complete, as shown by no further change occurring in the spectrum when longer reaction times are used. For nonylphenol and dodecylphenol, the time required for trifluoroacetylation must be extended 0 SAC and the authors.KONISHI, MORI AND TANIGUCHI 1003 to 24 hours.The excess of trifluoroacetic anhydride and the trifluoroacetic acid formed by the reaction do not interfere with the measurements and can be used as solvents. The nuclear magnetic resonance spectrum of the resultant solution is then measured directly. Compounds with large substituents in the ortho position, as with di-t-butylphenol, cannot be trifluoroacetylated under the conditions described above. PROCEDURE- As the resonance signal heights of the TFA derivatives of the o- and $-alkylphenols are used as the basis of this method of analysis, the fluorine magnetic resonance spectrum is recorded five times for each sample mixture. The five pairs of signal heights are measured to an accuracy of 0.1 mm and the average percentage of the o-isomer then calculated. C d a b 6 Fig.1. lSF nuclear magnetic resonance spectra a t 56.4 MHz of TFA derivatives of alkylphenols: a, o-isomer; b, $-isomer; c, trifluoroacetic anhydride; and d, trifluoroacetic acid: 1, cresol; 2, ethyl- phenol; 3, propylphenol ; 4, s-butylphenol; 5, hexylphenol; 6, nonylphenol; and 7, dodecylphenol RESULTS AND DISCUSSION FLUORINE MAGNETIC RESONANCE SPECTRA- The fluorine magnetic resonance spectra of TFA derivatives of some alkylphenols are presented in Fig. 1. The signal at a low field was assigned to TFA-o-alkylphenol and that at a higher field to TFA-$-alkylphenol by comparing with standard samples. The chemical shift differences for some TFA-alkylphenols are given in Table I, from which it can be seen TABLE I CHEMICAL SHIFTS OF TRIFLUOROACETYL GROUPS IN SOME ISOMERIC TFA-ALKYLPHENOLS AND THEIR DIFFERENCES Chemical shifts, Hz* Chemical shift Alkylphenol o-Isomer $-Isomer differences, Hz Cresol .. . . .. . . 28-6 18.0 Propylphenol . . .. . . 26-4 21-6 s-Butylphenol . . . . . . 30.0 21-6 Hexylphenol . . . . . . 33.2 22.4 Nonylphenol . . .. . . 39.4 25-4 Dodecylphenol . . . . 40.8 24-2 Ethylphenol . . . . . . 26-2 204 10.6 5.8 8.4 10.8 14.0 16-6 4-8 *The spectra were recorded for a mixed solution of trifluoroacetic anhydride and trifluoroacetic acid a t 56.4 MHz and 20' C. The chemical shifts are in a low field when trifluoroacetic anhydride is used as a standard.1004 KONISHI, MORI AND TANIGUCHI QUANTITATIVE ANALYSIS [AndySt, VOl. 94 that the extents of the differences in the chemical shifts between the alkylphenol isomers are large enough to permit quantitative determination of these by comparing the relative intensities of the resonance pairs.CHOICE OF RADIOFREQUENCY LEVEL- In quantitative analysis by nuclear magnetic resonance spectroscopy, it is important to choose a suitable radiofrequency level, H,. This can satisfactorily be obtained by the following procedure. Spectra are recorded at various radiofrequency levels and the intensities are divided by each level. The logarithm of this ratio is plotted versus radiofrequency level, as shown in Fig. 2, from which the radiofrequency level of -51 db was chosen as the most satisfactory compromise between signal intensity and saturation. 1031 I I I I -30 4 0 -50 -60 Radiofrequency level (H,), db Fig. 2. Relationship between signal intensity (I) and radiofrequency level (H,) RESULTS The results obtained by the method described here for the isomers of cresol, ethylphenol From these results it is apparent that the and s-butylphenol are presented in Table 11.agreement between found and calculated values is excellent. TABLE I1 RESULTS ON MIXTURES OF o- AND $-CRESOL, ETHYLPHENOL AND S-BUTYLPHENOL OF KNOWN COMPOSITION o-Isomer, per cent. w/w Cresol 1 .. 2 .. 3 .. 4 .. 5 I . 6 .. 2 .. 3 .. 4 .. 5 .. 6 .. 2 . . 3 . . 4 . . 5 .. 0 . . Ethylphenol 1 .. s-Butylphenol 1 . . 1 Calhated Found* .. 79.0 79-2 .. 70.5 70.1 .. 60.3 59.9 . . 41.3 41.7 .. 29.8 29-7 .. 22.1 21.9 .. 78-7 78.5 .. 70.0 70-5 .. 60.3 59.5 .. 38.5 39.2 .. 38.5 38.5 .. 19.6 20.3 .. 79.9 78.4 . . 70.0 69-9 .. 59.7 58.9 .. 42.3 42.2 ..33.9 33.4 .. 22-0 21.7 Average of five analyses. Difference, per cent. + 0.2 - 0.4 - 0.4 + 0.4 - 0.1 - 0.2 - 0.2 + 0.5 - 0.8 + 0.7 0 + 0.7 - 1.5 - 0.1 - 0.8 - 0.1 - 0.5 - 0.3November, 19691 OF MIXTURES OF ALKYLPHENOL ISOMERS 1005 PRECISION- There are two methods for measuring fluorine magnetic resonance signal intensities. In one the peak height is measured and in the other the area. Eleven successive measurements for a sample mixture (o-cresol 58.3 per cent. and p-cresol 41-7 per cent.) by both methods showed some difference in repeatability. The standard deviation calculated for the former was 0.34 per cent., while for the latter it was 1-15 per cent. Consequently, the peak height measurement method was adopted for the quantitative determination of the proportion of a component in isomer mixtures. When the signals have the same line shapes and the same widths at half-height, it appears that the height measurement method is more precise than that for measuring peak area by an integrator (area measurement method). When the nonylphenol and dodecylphenol, of which the alkyl groups of both are synthe- sised from polypropylene, used in detergent industries are analysed for their isomers by this method the sharp resonance signal appears with the TFA-$-isomer, whereas with the TFA- o-isomer the resonance signal broadens. Consequently, in these circumstances, the method cannot be used. REFERENCES 1. 2. 3. 4. 5. Konishi, K., and Kanoh, Y., Ibid., 1968, 40, 1881. Crutchfield, M. M., Irani, R. R., and Yoder, J. T., J. Amer. Oil Chem. SOC., 1964, 41, 129. Lindeman, L. P., and Nicksic, S. W., Analyt. Chem., 1964, 36, 2414. Manatt, S. L., J- Amer. Chem. SOC., 1966, 88, 1323. - , Analyt. Chem., 1966, 38, 1063. Received December 5th, 1968 Accepted March 18th, 1969

ISSN:0003-2654    

DOI:10.1039/AN9699401002

出版商:RSC    

年代:1969

数据来源: RSC

摘要:

1006 Analyst, November, 1969, Vol. 94, fifi. 1006-1009 Quantitative Analysis for Enan tiomeric Purity of Alkan-2-01s by Fluorine-19 Magnetic Resonance Spectroscopy BY KAZUO KONISHI, YOSHIHIRO MORI AND NOR10 TANIGUCHI (Industrial Research Laboratories, Kao Soap Co. Ltd., 1334 Minato-yakushubata, Wakayama-shi, Japan) Fluorine magnetic resonance spectroscopy has been investigated as a means of determining the enantiomeric purity of alkan-2-01s. These were converted into the I-mandelate diastereomers, which were allowed to react with trifluoroacetic anhydride. The fluorine magnetic resonance spectrum of the trifluoroacetyl derivatives shows a pair of signals, and the relative signal heights of a resonance pair were compared to permit the quantitative determination of the proportions of the components in the enantiomeric mixture.Agreement between the results given by the nuclear magnetic resonance method and gas chromatography was satisfactory. SEVERAL reports occur in the literature on the determination of the optical purity of a mixture of enantiomers by the use of magnetic anisotropy in nuclear magnetic resonance. Raban and Mislow1s2 and Jacobus, Raban and Mislow3 used nuclear magnetic resonance spectroscopy in the analysis of a mixture of diastereomers as a method of determining optical purity. In this method, alcohols and amines were converted irreversibly into the diastereomers with 2-hexylpropanoyl chloride or o-methylmandelyl chloride. Pirkle and co-workers* 76 96 9 7 have reported that proton and fluorine magnetic resonance spectra of enantiomeric mixtures of alkylarylcarbinols are dissimilar when optically active a-phenethylamine or a- (1-naphthy1)- ethylamine is used as solvent. A notable feature of this method is that the determination of the optical purity of the enantiomer is absolute and a standard of known optical purity is not therefore required.However, at present, the applicability of this method to the analysis of enantiomers is restricted to the alkylarylcarbinols. On the other hand, Raban and Mislow's method, in which diastereomers are prepared irreversibly with an optically active reagent, is applicable to the analysis of enantiomers that do not contain aryl groups in the molecule. The present study has been made to extend the scope of Raban and Mislow's method to enable the enantiomeric purity of alkan-2-01s to be determined, by the preparation of their 1-mandelates followed by trifluoroacetylation of the diastereomers.EXPERIMENTAL REAGENTS- All reagents should be of analytical-reagent grade. TriJEuoroacetic anhydride. p-Toluenesulfihonic acid. 1-Mandelic acid-Merck's reagent. Brucine. Benzene. A cetone. Diethyl ether. Phthalic anhydride. Analytical-reagent grade racemic alkan-2-01s (butan-2-01, hexan-2-01 , octan-22-01 and 0 SAC and the authors. Assay, 99 per cent. ; melting-point, 132" to 135" C; specific rotation ([a]&o, c = 2, water), - 154" 3"; sulphated ash, 0-1 per cent. dodecan-2-01) were used.KONISHI, MORI AND TANIGUCHI 1007 APPARATUS- All fluorine spectra were recorded at 56.4 MHz by using a JNM-3H-60 type, nuclear magnetic resonance spectrometer (Japan Electron Optics Laboratory Co.Ltd.). The measure- ments were carried out at a temperature of 20" &- 0.5" C and the spectra recorded at a sweep rate of 0.846 Hz s-l. The chemical shifts between the signals were measured by the side-band method, with a standard frequency of 100 Hz. The optical purity of the diastereomers was measured with a GCG 550 type, gas chromatograph (Yanagimoto Seisakusho Co. Ltd.), with a thermal-conductivity detector. SYNTHESIS OF ALK-2-YL z-MANDELATE- A mixture of the alkan-2-01 (0.002 mole), I-mandelic acid (0.0025 mole) and p-toluene- sulphonic acid (50 mg) in 35 ml of benzene was refluxed with an H-shaped reflux condenser for 2 hours. Refluxing was then stopped momentarily and the water that had deposited in the con- denser removed by washing the condenser with ethanol followed by ether. The esterification was then continued for another 2 hours.Refluxing was again stopped momentarily and the water removed from the condenser as before. After discarding 10 ml of the benzene solution, the esterification was continued for a further 2 hours. The residual solution was then diluted with ether and washed with 1 per cent. sodium hydroxide solution followed in turn by 10 per cent. sodium chloride solution to remove the 9-toluenesulphonic acid and the excess of Z-mandelic acid. The alk-2-yl I-mandelate was obtained on evaporating the ether solution. TRIFLUOROACETYLATION- About 400 mg of the alk-2-yl Z-mandelate and an excess (700 mg) of trifluoroacetic anhydride were mixed in a 10-ml conical flask at room temperature.The reaction was complete in 2 hours, as shown by no further change occurring in the spectrum when longer reaction times were used. The excess of trifluoroacetic anhydride and the trifluoroacetic acid formed by the reaction did not interfere in the measurement and could be used as a solvent. The nuclear magnetic resonance spectrum of the resultant solution was thus measured directly. RESOLUTION OF OCTAN-%OL- To assign the conformation for a pair of signals of the trifluoroacetylated oct-2-yl I-mandelate, a racemic mixture of octan-2-01s was resolved, via their brucine salts, into d- and I-octan-2-ols.8 Their optical purities were established by gas chromatography to be 95.1 and 90.1 per cent., respectively.Low - High Fig. 1. l*F nuclear magnetic resonance spectrum a t 56.4 MHz of TFA-oct-2-yl I-mandelate : a, 11-diastereomer ; b, dl-diastere- omer ; c , trifluoroacetic anhy- dride; and d, trifluoroacetic acid1008 KONISHI, MORI AND TANIGUCHI : QUANTITATIVE ANALYSIS FOR [Analyst, Vol. 94 As it was required to know the enantiomeric purity of the d- and Z-octan-2-01s in the resolved and known mixtures, gas chromatography was carried out for diastereomeric Z-mandelates. The conditions used were as follows: column packing, 25 per cent. Apiezon L on Chromosorb W (60 to 80 mesh) ; column, stainless steel (3 mm id., 150 cm long) ; column temperature, 194" C ; injection temperature, 260" C ; detector temperature, 260" C ; carrier gas, helium; carrier gas flow-rate, 20 ml minute-l; and sample size, 1 pl.CALCULATION- GAS CHROMATOGRAPHY OF OCT-2-YL z-MANDELATE- Calculations were made by the method previously rep~rted.~ RESULTS AND DISCUSSION The fluorine magnetic resonance spectrum of trifluoroacetylated oct-2-yl Z-mandelate (TFA-oct-2-yl Z-mandelate), prepared from the racemic octan-2-01s and optically pure Z-mandelic acid, is presented in Fig. 1, in which the diastereomeric TFA-oct-2-yl Z-mandelate shows a pair of signals with equal intensity, as expected for the racemate. The signal at a TABLE I CHEMICAL SHIFTS AND CHEMICAL-SHIFT DIFFERENCES OF DIASTEREOMERIC TRIFLUOROACETYL GROUPS IN TFA-ALK-2-YL I-MANDELATES Chemical shifts, Hz* Chemical-shi f t differences, Alkan-2-01 l l - d i a s n e o mer A8Hz Butan-2-01 .. .. .. 27-9 26-4 1.5 Hexan-2-01 . . .. .. 31.5 29-2 2.3 Octan-2-01 . . .. .. 31.8 28.8 3.0 Dodecan-2-01 .. .. 32-2 29.7 2.5 * The spectra were recorded for a mixed solution of trifluoroacetic anhydride and tnfluoroacetic acid at 56.4 MHz and 20" C. The chemical shifts are in a low field when trifluoroacetic anhydride is used as a standard. low field was assigned to TFA-Z-oct-2-yl Z-mandelate (ZZ-diastereomer) and at a higher field to TFA-d-oct-2-yl Z-mandelate (dZ-diastereomer) , by comparing with standard samples pre- pared from optically active d- and Z-octan-2-01 and Z-mandelic acid. From this, it is considered that assignments for other TFA-alk-2-yl Z-mandelates also follow this rule. The chemical shifts and the chemical-shift differences for some TFA-alk-2-yl Z-mandelates are given in Table I, from which it can be seen that the extent of the differences of chemical shifts between diastereomeric pairs is large enough to permit quantitative determination of the optical purity of the enantiomeric alkan-2-01s by comparing the relative intensities of the resonance pairs.TABLE I1 QUANTITATIVE ANALYSIS OF KNOWN MIXTURES Enantiomeric purity, per cent. Known mixtures of d- and E-octan-2-01s Calculated 95-1 4.9 70.7 29.3 50.0 50.0 34.2 65-8 9.9 90.1 {a {Id {fl {i" Nuclear magnetic resonance method 95-0 5-0 70-7 29.3 49-7 50-3 34.6 65-4 10.4 89.6 Gas-chromatographic method 95.1 4.9 71.7 28.3 50-3 49-7 34.4 65.6 9.9 90.1November, 19691 ENANTIOMERIC PURITY OF ALKAN-ZOLS 1009 To confirm the reliability of the method, known samples, prepared by mixing d- and Z-octan-2-ols, were analysed. The results, presented in Table 11, show that the agreement between those found and calculated is excellent.In the determination of optical purity of the enantiomers by preparation of the diastereo- mers, it is important that the proportions of original enantiomers do not change. It is, therefore, necessary to ensure that the esterification of the alkan-2-01 with Z-mandelic acid reaches completion. Accordingly in this work, as much as possible of the water formed by the reaction was removed. Consequently, the yield of oct-2-yl Z-mandelate was 98.6 per cent. No racemisation was observed during the preparation of diastereomers, as the agreement between the found and calculated results was satisfactory. Moreover, to establish that the analytical results obtained are reliable, the identical Z-mandelate samples that were used for nuclear magnetic resonance analysis were also analysed by gas chromatography.The results given by the two methods are compared in Table 11, and it can be seen that they are in satisfactory agreement. It is, therefore, possible to determine the enantiomeric purity of alkan-2-01s by the nuclear magnetic resonance method described. The use of a commercially available optically pure Z-mandelic acid to prepare the diastereomers and the introduction of fluorine nuclei into the diastereomers by means of trifluoroacetylation are considered merits of the proposed method. 1. 2. 3. 4. 5. 6. 7. 8. 9. REFERENCES Raban, M., and Mislow, K., Tetrahedron Lett., 1965, No. 48, 4249. Jacobus, J,, Raban, M., and Mislow, K., J. Org. Chem., 1968, 33, 1142. Pirkle, W. H., J. Amer. Chem. SOC., 1966, 88, 1837. Burlingame, T. G., and Pirkle, W. H., Ibid., 1966, 88, 4294. Pirkle, W. H., and Burlingame, T. G., Tetrahedron Lett., 1967, No. 41, 4039. Pirkle, W. H., and Beare, S. D., J . Amer. Chem. SOC., 1967, 89, 5486. Ingersoll, A. W., “Organic Reaction,” Volume 2, John Wiley & Sons Inc., New York, 1944, p. 400. Konishi, K., and Kanoh, Y., Analyt. Chem., 1968, 40, 1881. Received March 3rd, 1969 Accepted April 25th, 1969 , , Ibid., 1966, No. 33, 3961. --

ISSN:0003-2654    

DOI:10.1039/AN9699401006

出版商:RSC    

年代:1969

数据来源: RSC