January, 19661 ASPINAL 33 Vacuum Fusion Analysis with a Mass Spectrometer BY M. L. ASPINAL (Associated Electrical Industries Limited, Central Research Laboratory, Rugby) The suitability of a mass spectrometer for vacuum fusion analysis is discussed; the equipment and its mode of operation and calibration are described. The objections to using an oil pump for extracting the gases liberated from the metal samples have been overcome with this apparatus. Results for oxygen determination in steels, molybdenum and zirconium have been independently checked by using a fast neutron-activation tech- nique, and the results of two methods are shown to be in good agreement. Oxygen levels of 116 p.p.m. have a standard deviation of 6 p.p.m. and coefficient of variation of 5 per cent. Although most of the work has been carried out on oxygen determinations, nitrogen results are quoted for two standard iron samples showing that the method is also acceptable for nitrogen determinations.Kitrogen levels of 31 p.p.m. have a standard deviation of 3 p.p.m. and a coefficient of variation of 10 per cent. The limits of detection for the equipment are 0.1 p g of oxygen, 0.1 ,ug of nitrogen and 0.01 pg of hydrogen. VACGUM fusion analysis is a well known method for the determination of oxygen, nitrogen and hydrogen in metals, and a comprehensive, up-to-date review of the literature has been given by James.l Most papers, however, describe equipment capable of determining oxygen down to 10 p.p.m. This is no longer adequate as the metallurgical requirements for metals sucl as copper, molybdenum, tantalum and tungsten may be below this level; for example, coppir that is to be used in vacuum devices is required to have an oxygen content of less than 0.5 p.p.m.The detection limit of some equipments could be lowered by increasing the sample weights, but large samples are not always available, and an equipment capable of determining gas contents on small samples, in particular oxygen contents below 10 p.p.m., was required. Techniques for the analysis of small amounts of gas were considered, such as gas chromatography and mass spectrometry, and mass spectrometry was thought to be most suitable for vacuum fusion analysis. Mass spectrometers have been used by Taylor2 for the analysis of gas obtained by vacuum fusion; in this method the gas was collected in a bottle and then transferred to the spectro- meter.Martin et aL3 used a mass spectrometer directly coupled to their vacuum fusion equipment. This equipment, however, required large sample weights, 6 to 14 g, and blanks had to be carried out for 1 hour so that sufficient blank gas was present for analysis. The object of the work described in this paper was to produce an equipment suitable for routine analysis of materials with high gas contents, and in particular capable of analysing material with low gas contents for specialised purposes. EXPERIMENTAL DISCUSSION ( a ) THE SUITABILITY OF A MASS SPECTROMETER FOR VACUUM FUSION ANALYSIS There are three main reasons why a mass spectrometer was chosen for this work. First, because it is capable of measuring small amounts of gas quantitatively.Secondly, because it allows positive identification of each gas, whereas some of the earlier equipments could only label the residue gas as nitrogen after the extraction of carbon monoxide and hydrogen. The spectrometer, for example, would give differentiation between argon and nitrogen if this was required. Thirdly (an important advance on some earlier equipments), by coupling the output from the spectrometer to a recorder, a complete record of the rate at which one gaseous component is liberated from the sample, and also the time needed to restore the original blank rate can be observed. The gas liberated from the sample therefore appears as a step on the blank rate, the height of which can be measured from the trace, so making it unnecessary to carry out blanks at regular intervals between sample additions. One difficulty that arises from using a mass spectrometer is caused because nitrogen and carbon monoxide both have the same mass number.Gregory, Mapper and Woodward4 suggested that this problem could be overcome, but with considerable loss of accuracy, by measuring the mass 29 peak that is principally the monoxide of 13C, and multiplying this34 ASPINAL: VACUUM FUSION ANALYSIS WITH A MASS SPECTROMETER [A?Z&?J.’St, VOl. 91 by the 12C/13C ratio to obtain an estimate of the total monoxide present in the mass 28 peak. The presence of organic compounds that also produce a peak at mass 29 made this method unsatisfactory. If the spectrum of a gas is studied it is found that it consists of a parent peak at the mass equivalent of the molecule, a major peak that may be the parent peak and some small peaks that bear fixed ratios for a particular instrument and source condition to the major peak.The characteristic cracking patterns of carbon monoxide and nitrogen are shown in Table I, in which it is seen that the parent peak is also the major peak. It will also be seen TABLE I CHARACTERISTIC CRACKING PATTERNS Carbon monoxide- m/e . . 12 13 14 14.5 15 16 28 29 30 Relative intensity 4.49 0.048 0.61 0-007 0.001 0.95 100.00 1-13 0.21 Nitrogen- 14 - 15 - 28 29 30 m / e . . - - Relative intensity - - 7.18 - 0.021 - 100.00 0-77 0.002 that mass 12, the second largest peak, could be used to measure carbon monoxide] and similarly mass 14 could be used to measure nitrogen.Carbon monoxide also produces a significant mass 14 peak due to the presence of (C0)2+, but allowance is made for this by calibrating the spectrometer with carbon monoxide and measuring the mass 12 and 14 peaks. The amount of mass 14 coming from the carbon monoxide can be deducted from the “total” mass 14 peak before calculating the amount of nitrogen present. Bottles of “Specpure” carbon monoxide, nitrogen and hydrogen are used for calibrp ,ing the spectrometer. Small amounts of these gases are introduced into the gas-handling section, their pressure is measured by a McLeod gauge and the appropriate peak height by the spectrometer. The calibration of the spectrometer does not change by more than 2 per cent. over several months of continuous operation] and a calibration check once a week is satisfactory.The filaments of the spectrometer, which are replaceable] have a life of 6 months under normal operating conditions. Replacement of these filaments can be accom- plished with no appreciable change to the calibration. In order to check that mixtures of carbon monoxide and nitrogen could be satisfactorily separated with the spectrometer, varying amounts of the two gases were mixed in the gas-handling section and then analysed with the spectrometer. Table I1 shows the amounts added and found, from which it can be seen that satisfactory separation has taken place. TABLE I1 SEPARATION OF CARBON MONOXIDE AND NITROGEN Analysis of mixture with 3I.S. 10 Gas mixture added & ---7 Carbon Carbon CO added Nitrogen, monoxide, Nitrogen, monoxide, Nz added per cent.per cent. per cent. per cent. N, found CO found 15.4 84.6 15.5 84.5 0.99 0.99 9.4 90.6 9.8 90.2 0.98 1.01 13.0 87.0 12.7 87.3 1.02 0.99 53.2 47.8 49.5 50.5 0.98 1.08 80.0 20.0 79.0 21-0 1.02 0.95 71.0 29.0 70.0 30.0 1.02 0.97 (b) THE USE OF OIL-DIFFUSION PUMPS FOR VACUUM FUSION ANALYSIS- One criterion for good vacuum fusion is that a high-speed pump is used to extract the gases from the furnace section. This is to remove the gases quickly so they are less likely to be “gettered” by the metal films that form around the cooler parts of the furnace walls. A pumping speed of 30 litres second-l at the mouth of the crucible is usually acceptable. Many vacuum fusion equipments use a mercury-diffusion pump to transfer the gases from the furnace section to the analysing section.When a mercury-diffusion pump is used, the pressure in the furnace section is limited by the vapour pressure of mercury, which isJanuary, 19661 ASPINAL: VACUUM FUSION ANALYSIS WITH A MASS SPECTROMETER 35 3 x 10-3mm (of mercury) at room temperature. At this pressure the mean free path in the furnace section will be about 1 cm and there is a greater chance that “gettering” will occur. A cold trap could be introduced between the diffusion pump and furnace section to reduce the pressure in the furnace section, but this is not desirable as it will also reduce the pumping speed at the crucible. Fig. 1. Comparison of carbon monoxide blanks against temperature; curve -1 = filer- cury pump and low-pressure analyser (I’alladium thimble and “Hopcalite”) ; curve I3 = Oil pump and mass spectrometer One commercial equipment manufactured by Balzers incorporates an oil-diffusion pump.5 This is claimed to produce a pressure of mm of mercury above the crucible, and reduce “gettering” of the liberated gas as the films formed around the cooler parts of the furnace walls are now less porous than those formed with a mercury system.The main objections to the use of an oil pump are that any back-streaming of the oil could increase the blank rate by being cracked on the hot graphite crucible, and also that some of the gas liberated from the sample might be absorbed in the pump oil. Neither of these objections has been found to applv in this investigation, and the use of an oil pump has in fact provided significantly lower blank rates.This is shown in Fig. 1, which compares carbon monoxide blank rates when using an 033C oil-diffusion pump with values obtained over several years with a mercury pump and a low-pressure analyser. There has also been no evidence for “gettering.” TIIE VACUCM FUSION EQUIPMENT The general arrangement of the equipment is shown in Fig. 2 and can be divided into three parts ( a ) furnace, (b) gas-handling and (c) mass spectrometer sections. ( a ) FUKNACE SECTION- This is shown in Fig. 3. A small silica crucible and pedestal are supported by the silica furnace tube. The graphite crucible (14 in. x 8 in. ad., wall thickness & in.) is placed in the silica crucible and the space between the two crucibles is packed with graphite powder that has passed through a 100-mesh sieve.A slotted graphite funnel is placed on top of the graphite crucible. All the graphite parts and powder are made from Morganite EYSA graphite. The top rim of the silica crucible is turned in so that there is the minimum gap between the silica and graphite funnel that prevents graphite powder from being blown out of the crucible during evacuation or outgassing. The silica furnace tube is joined to a stainless-steel manifold by means of a viton O-ring fitting into a tapered groove and compressed with a screwed ring.36 ASPINAL: VACUUM FUSION ANALYSIS WITH A MASS SPECTROMETER [Analyst, VOl. 91 F = o = c = L = E = s = v = P = T = E M.S. Furnace Oil diffusion pump Cold trap Leaks Expansion volumes Spectrometer vacuum pumps Vacuum pumps Pirani gauge Three-way taD 10 M.S.10 = Mass spec;rograph Fig. 2. General arrangement of the apparatus A h L = Vacuum lock S = Sample arm M = Stainless-steel manifold J = Water jacket F = Quartz furnace tube C = RF Coil 0 = 033C Oil-diffusion pump Fig. 3. Furnace sectionJanuary, 19661 ASPINAL: VACUUM FUSION AKALYSIS WITH A MASS SPECTROMETER 37 A Pye 3-1<W, 2-Mc second-l induction heater is used to heat the graphite assembly, and the bottom of the stainless-steel manifold is cooled by water to reduce the temperature of the O-ring seal during operation. The 2B-in. bore stainless-steel manifold connects together the furnace tube, sample-handling system and an A.E.I. O33C oil-diffusion pump, that has an unbaffled speed of 130 litres second-l.The pump uses 704 silicone oil as the pump fluid, and has a simple copper-foil baffle inserted on top of it to prevent back-streaming. The manifold also supports the funnel for guiding samples into the crucible. The top flange of the manifold supports the mechanism for operating the stopper. This consists of a graphite stopper connected to an iron slug with fine molybdenum wire. The slug is moved along the side arm with a magnet, so raising or lowering the stopper which is guided on to the crucible by means of the sample-guide tube. During sample additions and temperature measurements the stopper can be drawn out to one side of the guide tube to give an unobstructed view of the crucible. The temperature is measured with the optical flat on top of the glass-ware, a prism and optical pyrometer.A vacuum lock is fitted to the sample side arm to allow samples to be introduced while the system is under vacuum. ( b ) GAS-HANDLING SECTION- The gas-handling section is connected through a liquid-nitrogen cold trap to the backing side of the oil-diffusion pump, and also to the mass spectrometer by capillary leaks. The trap is used to remove small amounts of unwanted condensable gases such as water and organic compounds that complicate the mass spectrum, and are liberated from the graphite and the cooler parts of the furnace section. The section is pumped by a cold-trapped glass - mercury diffusion pump and A.E.I., D.R.l rotary pump. These pumps can either be connected in series with the oil-diffusion pump during outgassing, isolated from the gas-handling section during analysis or can be used to evacuate the gas-handling section.The three-way tap T provides the necessary control. In addition the section consists of three expansion volumes used to adjust the pressure of the extracted gases to a suitable value and a McLeod gauge that is used during calibration. Litre bottles of "Specpure" carbon monoxide, nitrogen and hydrogen that are obtained from the British Oxygen Company Ltd., are used during calibration, and connected to the system through double &-in. A .E. I. polyethylene-diaphragm valves so that small amounts of gas can be introduced for calibration. There are two capillary leaks in parallel connecting this section to the mass spectrometer.The leaks are made from precision-bore glass tubing and have conductances of 1.7 x litres second-l and 0-5 x The leak with the lower conductance is used for the hydrogen determinations. Greased taps are always a possible source of leaks and only one tap, T, is used in this equipment, the remainder are either metal - polyethylene diaphragm valves or greaseless glass taps with a viton diaphragm. These latter taps have a higher conductance than the &in. metal valves and are to be preferred where a high conductance is required. litres second-l. (C) 1'rASS SPECTROMETER SECTION- The mass spectrometer used is a small gas mass-spectrometer, type M.S. 10, manufactured by A.E.I. Instrumentation Division, and has mass ranges 2 to 4, 12 to 45 and 18 to 200.The output from the analyser tube is passed through electronic circuits where it is amplified and displayed either on the output meter or pen recorder. The spectrum can be scanned by varying the accelerating voltage, The spectrometer tube is mounted horizontally and pumped by its own vacuum pumps, which consist of a cold-trapped A.E.I. 033C oil-diffusion pump backed bv an A.E.I., D.R.l rotary pump. The oil used in the diffusion pump is Apiezon B.W. as there is some danger of coating the source unit with silica if silicone oils are used. Provision is made for baking the analyser tube while outgassing by using band heaters clamped around the flanges. These raise the temperature to 200" C with an unlagged system. Higher baking temperatures up to 400" C can be achieved with an oven, but for this particular application of the spectrometer baking at 200" C is adequate.The magnet, which weighs 35 lb, is easily removed during baking and can be replaced in the same position after baking by using guides. The spectrometer is connected to the capillary leaks by a small inlet flange mounted on the side of the spectrometer away from the source unit.3s ASPINAL: VACUUM FUSION ANALYSIS WITH A MASS SPECTROMETER [AWalyd, Vol. 91 METHOD OF OPERATIOX- The graphite crucible is placed in the silica crucible and the space between the two crucibles is loosely packed with graphite powder. The assembly is placed on its support in the furnace tube that is fitted to the furnace manifold. All the taps in the gas-handling section, which has been under vacuum from the previous analysis, are closed.The three-way tap, T, is turned so that the whole system can be slowly evacuated to avoid graphite powder from being blown out of the crucible. \$'hen the Pirani gauge, P, shows that the pressure has fallen to below 100 microns the oil-diffusion pump can be switched on. Pressure continues to fall in the gas-handling section until it has reached a few microns. Outgassing of the crucible can now start by raising the temperature in about ten steps within half an hour up to a maximum of 2100" C ; at higher temperatures there is a tendency for the stopper to stick to the crucible. When the temperature has reached the maximum of 2100" C it is maintained for 1 hour. This has been found to be sufficient to outgas the crucible to a suitably low level for use with samples with oxygen contents greater than 50 p.p.m.If, however, the sample is expected to have a lower gas content, the outgassing procedure is continued for a second hour. This produces slightlq? lo\ver blank rates; typical values are 19 pl hour-1 at 1900" C (carbon monoxide, 32.5 per cent.; nitrogen, 13.5 per cent.; hydrogen, 54 per cent.), and 8 pl hour-' at 1700" C (carbon monoxide, 23.7 per cent.; nitrogen, 25.7 per cent.; hydrogen, 50.6 per cent.). The outgassing rate can be monitored with the spectrometer set at mass 12, and the minimum outgassing time can be determined. Once this has been determined it is usual to allow the spectrometer to cool from its overnight bake, and prepare the samples while the outgassing of the crucible is taking place.The sample weights can vary in the range 4 g to 80mg depending on whether the expected oxygen content is below 1 p.p.m. of oxygen or above 1000 p.p.m. of oxygen, respectively . The preparation of the sample's surface is usually carried out by abrading it with a file and then by vapour de-greasing it with carbon tetrachloride. The optimum size of sample for low gas contents and ease of manipulation is one, or more, 4-mm cubes. If a suitable etch is available this is preferred to the abrasion, for example, copper samples can be etched in 1 to 1 nitric acid, then by 1 to 1 hydrochloric acid, washed with distilled water and dried with acetone and ether. After the samples have been weighed, they are placed as quickly as possible in the sample arm of the furnace section via the vacuum lock.If a bath material is being used, such as iron for molybdenum or platinum for zirconium, this is also stored in the sample arm. Liquid nitrogen is placed around the cold traps and the blank rate is checked for the equipment. This should be less than 20 pl hour-l at 1900" C. The temperature is lowered to the operating temperature, for example, 1900' C for materials such as zirconium requiring a platinum bath, 1750" C for materials requiring an iron bath. If a bath material is to be used it is introduced and outgassed. Tap T is turned to connect the gas-handling section to the furnace section and a sample is introduced to the crucible. The extracted gases are collected in the gas-handling section and adjusted to a suitable pressure with the expansion volumes.These are used to keep the spectrometer reading on scale when large amounts of gas are collected. The gas is monitored during collection by the spectrometer set either at mass 12 for carbon monoxide or mass 14 for nitrogen. The rate at which the gas leaks from the gas-handling section into the spectro- meter does not have any appreciable effect on the total volume of gas collected. The mass spectrometer is coupled to the recorder which shows one of the extracted gases as a step on the blank rate, Fig. 4. The collection is continued until the blank rate has returned to its original rate shown by the recorder. This usually takes between 2 to 8 minutes. The step height is measured and converted to pressure with the appropriate calibration graph and hence to oxygen, nitrogen or hydrogen depending on whether mass 12, 14 or 2 has been recorded.The gases which are not recorded are measured after completion of the extraction by scanning the spectrum manually. DISCUSSION Peaks at masses 12, 13, 14, 15, 27, 43, 44 and 2 were measured manually on the early analyses for both samples and blanks. It was found that in addition to the peaks at masses 12, 14 and 2, corresponding to carbon monoxide, nitrogen and hydrogen, only mass 15 (which is derived from methane) is producedJanuary, 19661 ASPINAL : VACUUM FUSION ANALYSIS WITH A MASS SPECTROMETER 39 Time, minutes Fig. 4. Typical extractions of carbon monoxide: ( a ) sample wcight 2.27 g, expansion volume 1170 ml ; (b) sample weight 2.96 g, expansion volume 740 ml in significant amounts by the sample.The other peaks 27, 43 and 44 were all present in the blank and sample to an equal extent and are attributed to the presence of hydrocarbons. All of the mass 13 could be attributed to the methane. As a result for routine analysis only masses 12, 14, 15 and 2 are measured. With samples containing large amounts of oxygen the mass 12 correction from mass 15 is small, but for samples with low oxygen contents it becomes significant. For nitrogen calculations two correction factors are applied, one is the mass 14 contri- bution from carbon monoxide determined during calibration, and the other is the mass 14 contribution from methane. Both correction factors for masses 12 and 14 from methane are taken from the manufacturer's results and correspond to 2.8 per cent.and 18.2 per cent. of the mass 15 peak, respectively. Although the correction factors obtained from the cracking patterns of methane will vary slightly between spectrometers, no appreciable error will occur in the results unless the variation is greater than 20 per cent. of the quoted results. A run of twelve steel samples can be analysed in 8 hours, including assembly and outgassing times. RESULTS Most of this work has been carried out on oxygen determinations only, as it is difficult to obtain material with a known nitrogen content. There is also some doubt expressed in the literature as to whether nitrogen from certain materials, for example, zirconium, is liberated quantitatively. The standard iron samples BOL 16, BPL 8 and CRL 17 were obtained from the British Iron and Steel Research Association. TABLE I11 OXYGEN RESULTS O F STANDARD IRON M.S.10 A r \ Number Mean of oxygen Standard deter- content, deviation, Sample minations pp.m. p.p.nl. BOL 18 . . 29 116 6 BPL 8 . . 8 52 8 CRL 17 . . 8 54 3 (100 p.p.m.) (40 p.p.m.) (35 p.p.m.) * BPL 15 B.I.S.K.A. Radiochernical c I Number Mean of oxygen Standard deter- content, deviation, ininations p.p.m. p.p.m. 6 119 4 6 62 8 6 84 5 quote 30 p.p.m. Coleman, Mean oxygen content, p.p.m. 120 (BOL 25) 37 (BPL 15)* 54 (CRL 15)40 ASPINAL: VACUUM FUSION ANALYSIS WITH A MASS SPECTROMETER [Analyst, VOl. 91 They had also been analysed by ColemanJ6 who used a fast-neutron activation technique. The numbers on these samples denote the bar used.It will be seen that Coleman used different bars of the iron, but all the bars are supplied with the same nominal gas content, except BPL 15. Analysis was also carried out on three samples of stainless steel, one sample of molybdenum requiring the use of an iron bath and four samples of zirconium that required the use of a platinum bath. The results for these materials are shown in Tables I11 and IV. I .oo Blank 0.75 - Sample Mol ybdenum-disc Stainless s teel-C, c3 C6 Zirconium-Zr 66 Zr 83 Zr 92 Zr 115 Copper sample Ig _- B TABLE IV OXYGEN RESULTS ON OTHER MATERIALS M.S. 10 Radiochemical h r 7 r .-h------, Number Mean Number Mean of oxygen Standard of oxygen Standard deter- content, deviation, deter- content, deviation, minations p.p.m.p.p.m. minations p.p.m. p.p.m. .. 6 32 2 4 38 3 .. 2 36 - 2 34 - . . 2 44 - 2 39 - . . 5 58 4 2 48 - . . 2 850 - 2 870 - . . 2 780 - 2 770 - . . 2 720 - 2 750 - .. 2 1440 - 2 1350 - ,4n independent check has also been carried out by this laboratory’s radiochemistry department on these samples by using the fast-neutron source at Wantage. The results obtained by use of this method have been quoted for comparison, and it can be seen that in general the agreement between the two methods is good. TABLE V LOW OXYGEN LEVELS IN COPPER Kumber of Mean oxygen Sample determinations content, p.p.m. O.F.H.C. copper . . .. 6 1.27 Copper - 1 per cent. silver . . 4 0.38 Copper - iron alloy . . . . 4 0.24 Vacuum-melted copper . . 6 0-07 ALLOYS Standard Coefficient deviation, of variation p.p.m.per cent. 0.126 9.9 0.048 12.5 0.042 17.5 0.013 18.6 One main advantage of this method lies in the determination of oxygen contents below 2 p.p.m. in metals. Table V shows oxygen levels found in a number of copper and copper- alloy samples. Fig. 5 shows the trace obtained for the lowest of these materials, from which it can be seen that the step that represents 0-15 pg of oxygen is still well above the background. n N ll p W Fig.January, 19661 41 These show a greater variation in the reproducibility, and although little work has been carried out at present on nitrogen determinations the results show that this method can be used for its determination. However, Table VII shows a range of hydrogen results obtained for various materials with this equipment.ASPINAL: VACUUM FUSION ANALYSIS WITH A MASS SPECTROMETER Nitrogen results are quoted for two standard iron samples (Table VI). No standard samples for the determination of hydrogen were available. Sample BNL 15 .. .. BOL 16 . . . . (160 p.p.m.) (45 p.p.m.) TABLE VI NITROGEN RESULTS Number of Mean determinations nitrogen content, Standard deviation, p.p.m. p.p.m. 15 180 30 29 31 3 TABLE VII HYDROGEN RESULTS Number of Mean hydrogen Standard Coefficient of Sample determinations content, deviation, variation, p.p.m. p.p.m. per cent. O.F.H.C. copper . . .. 6 0.7 1 0.10 14 Tantalum sheet . . . . 3 86.6 1.5 1.7 Zirconium . . . . .. 2 346 5 1.4 Vacuum-melted copper . . 6 0.01 0.01 100 CONCLUSIONS The method of vacuum fusion with a mass spectrometer, M.S.10, as the analyser has been used in this laboratory for the past 12 months. During this time the method has been completely reliable and the operation of the equipment is simple and rapid. The method can be applied satisfactorily over a large range of gas contents, ranging from 1400 to 0.05 p.p.m. of oxygen, 150 to 20 p.p.m. of nitrogen and 400 to 0.02 p.p.m. of hydrogen. The main advantage of this method is in determining oxygen contents of metals below 2 p.p.m., and the display of the gas-evolution rate on a recorder. The limits of detection of this equipment are at present 0.1 pg of oxygen, 0.1 pg of nitrogen and 0.01 pg of hydrogen. These limits can be improved if necessary by reducing the pumping speed of the spectrometer, but a major source of error will be surface con- tamination in the form of surface films, which must be removed before any lowering of the detection limit is justified. I am indebted to Mr. P. Jones, who carried out the fast neutron-activation analysis and to Mr. J. A. James for reading the manuscript and making helpful suggestions and criticisms, REFERENCES 1 . James, J . h., Metall. Rev., 1964, 9, 93. 2. 3. 4. 5. Kraus, T., Arch. Eisenhuttenw., 1962, 33, 527. 6. Taylor, K. E., Analylica Chim. A d a , 1959, 21, 6, 549. Martin, J . F., Friedline, J . E., hlclnick, L. M., and Pellissicr, G. E., Trans. Metall. Soc. A.I.M.E., Gregory, J. N., Mapper, D., and Woodward, J. A., Analyst, 1953, 78, 414. Coleman, R. F., Analyst, 1962, 87, 590. 1958, 212, 514. Received January lst, 1965