Determination of Titanium in Fine Gold by Electrothermal Atomic Absorption Spectrometry MICHAEL W. HINDS*a , IAN L. SHUTTLERb AND CYNTHIA PRIEST BOSNAKc aRoyal Canadian Mint, 320 Sussex Dr., Ottawa, Ontario, K1A 0G8, Canada bBodenseewerk Perkin-Elmer GmbH, Postfach 10 17 61, D-88647 U� berlingen, Germany cPerkin-Elmer Corporation, 2000 York Road, Suite 132, Oak Brook, IL 60522-3608 The transversely heated graphite atomizer (THGA) proved to expected that the absence of lower temperature sites would eliminate or substantially reduce analyte carry-over.be better suited to the task of determining Ti in gold owing to the reduction in analyte carry-over compared with a Massman This paper presents a brief comparison between THGA and HGA Massman atomizers for the atomization and the determi- type atomizer. Carry-over was also reduced by using a second maximum power atomization step (2600 °C) as a clean-out nation of Ti in a gold matrix, mainly in terms of analyte carryover.Results of experiments to establish optimal conditions step and cycling through the temperature program twice with no sample between solution samples. Platform atomization for titanium atomization (THGA) are presented and accuracy is assessed by analyses of gold reference materials with certified with the THGA was preferred because of the slightly higher observed sensitivity, especially as increasing amounts of gold concentration values for the titanium content.matrix per aliquot attenuates the analytical signal. The tube lifetime was lower owing to the high temperatures and long EXPERIMENTAL atomization times (10 s). The limit of detection (3s) was Apparatus estimated to be 5 mg l-1 based on 20 ml aliquots containing 40 mg of gold. The accuracy of the method was verified by Experiments were conducted on two dierent spectrometer/ determining titanium in gold reference materials with known graphite electrothermal atomizer combinations.Titanium was concentrations of titanium. Typical precision was determined in a Massman atomizer using a Perkin-Elmer approximately 4% (RSD) for 0.2% gold solutions containing HGA 500 atomizer in conjunction with an AS 40 autosampler 42 mg l-1 Ti (20 ml aliquot). (Perkin-Elmer Bodenseewerk, U� berlingen, Germany). Only wall atomization was measured from non-grooved pyrolytically Keywords: Atomic absorption spectrometry; electrothermal coated graphite atomizers (P-E part no.B010-5197). Atomic atomization; precious metal analysis; titanium absorption measurements were made on a P-E Model 5000 spectrometer with continuum source background correction. High purity gold (99.99%) refined at the Royal Canadian Mint The spectrometer was interfaced to a personal computer is assayed by determining the concentration of a suite of through a DAS8, 12 bit analogue-to-digital converter (Keithley elements that are known to be common impurities and sub- Metrabyte, Taunton, MA, USA).Data collection software was tracting these values from 100% to arrive at a purity value. similar in design to the work of Allen and Jackson.5 The Because of the monetary value of gold, rapid assays are atomizer was a pyrolytic graphite coated electro-graphite tube. required to maintain the flow of gold between the Mint and The other spectrometer employed was a Perkin-Elmer its customers. Solid sample atomic spectrometric techniques Model 4100 ZL atomic absorption spectrometer which such as spark ablation ICP-AES and laser ablation ICP-MS, included a transversely heated graphite electrothermal atomare routinely used for this purpose.However, solid sample izer and longitudinal Zeeman-eect background correction. calibration standards (gold matrix) are required for these Samples were dispensed by a Model AS 70 autosampler which methods. A set of gold reference materials were manufactured is an integral part of the system. Both platform and wall with dierent concentrations of the elements of interest and atomization were used.Platform atomization occurred from analysed by the Mint for the purpose of calibrating solid the graphite tube with an integrated platform (P-E part no. sample atomic spectrometers. Titanium is one of the elements B050-4033) and wall atomization occurred from the same monitored in the refining process and was therefore included atomizer with the platform manually removed. The temperain the suite of elements added during the manufacture of these ture programs used from both atomizer designs are summarreference materials.ized in Table 1. Atomic absorption measurements were eected Titanium was determined in plant materials1 and coal fly by PEAALABS software, version 6.2, running on an IBM ash2 by ETAAS with Massman type atomizers. In both cases, Personal System 2 computer, Model 70 386. high temperatures were used for atomization and clean-out In both instruments, a titanium hollow cathode lamp was steps.Because of the refractory nature of titanium, multiple used with a lamp current of 15 mA. The most sensitive clean-out steps1 or a set of multiple empty temperature wavelength, 364.3 nm, was used with a slit width of 0.2 nm. program cycles2 were used to minimize carry-over. The analyte atomic signal, from a Massman type atomizer, Reagents was variable mainly owing to the high carry-over between sample injections. It has been shown by Frech et al.3 that the High purity hydrochloric and nitric acids (Seastar Chemicals, Sidney, BC, Canada) were used for dissolving gold samples.cooler ends of the end-heated Massman atomizers (HGA) act as condensation sites for more refractory elements, from where Water used in these experiments was distilled and de-ionized using a Nanopure II system (Barsted/Thermolyne, Dubuque, re-atomization can occur. The transversely heated graphite atomizer (THGA) is virtually isothermal over its length owing IA, USA).Calibration standards were diluted from 1000 mg ml-1 stock solution made from high purity titanium to the electrical contacts at either side of the atomizer.4 It was Journal of Analytical Atomic Spectrometry, August 1997, Vol. 12 (833–836) 833Table 1 Temperature programs for the determination of Ti in dier- any possible analyte carry-over interfering with the measureent atomizers ment of the next sample. For the determination of Ti in FAU6 the method of standard Part A Massman T ype Atomizer— additions was used by adding a 10 ml aliquot of 50 ng ml-1 Ti Step Temperature/°C Ramp Time /s Hold Time /s solution to the 30 ml sample aliquot. The temperature program 1 200 10 10 outlined in Table 1, Part A was used for the determination, 2 20 1 5 although two more clean-out steps at 2700 °C were added to 3 2700 0 10* minimize analyte carry-over. 4 20 1 10 5 2700 1 5 RESULTS AND DISCUSSION Part B THGA atomizer (two maximum power heating step)— Step Temperature/°C Ramp Time /s Hold Time /s Atomizer and Temperature Program Studies 1 120 1 30 Analyte carry-over: Massman versus transersely heated 2 130 10 40 atomizers 3 1500 10 20 4 2600 0 10† The analyte carry-over in a Massman-type atomizer is illus- 5 20 1 30 trated in Fig. 1(a). The integrated absorbance observed for 6 2600 0 7 7 20 1 30 each pair of empty firings increases as Ti is introduced into the atomizer. It is also evident that the precision for samples * BOC sequence started 1 s before maximum power heating; argon with measureable amounts of Ti is variable.As noted in the gas flow stopped (otherwise at 250 ml min-1). introduction, analyte carry-over can be explained by the work † BOC sequence started 2 s before maximum power heating; argon done by Frech et al.,1 who showed that analyte atoms condense gas flow stopped (otherwise at 250 ml min-1). along with the metal (Pd) matrix modifier at the cooler tube ends of a Massman type atomizer.During the next atomizer (99.99%) dissolved in hydrochloric acid (High Purity heating cycle the condensed analytes at the tube ends are Standards, Charleston, SC, USA). Calibration solutions also re-atomized because, at the early stage of the atomization contained 0.1 g of high purity gold (Metalor, North cycle, the high temperature zone extends almost the length of Attelborough, Massachusetts) in 50 ml (0.2) and 1% the atomizer.8 However, as the atomization cycle progresses hydrochloric acid to match the sample solution matrix.the high temperature zone moves back from the tube ends which sets up another cycle of analyte/matrix condensation at the tube ends. Titanium does form refractory titanium carbide Samples and Sample Preparation through interaction with the graphite upon pyrolysis and Samples were fine gold reference materials produced and atomization. The titanium carbide formation9 particularly near analysed by the Royal Canadian Mint.Each of these reference the atomizer ends and the refractory nature of titanium contribmaterials was made by adding set amounts of 16 elements to utes to analyte carry-over. molten gold. Further treatment of each material resulted in Results from a transversely heated atomizer are shown in the trace elements being homogeneously distributed in the gold matrix. A more complete description is summarized by Kogan et al.6 The added elements are the same as the elements required in the American Society for Testing Materials specifi- cation for high purity gold,7 plus other elements that are of interest to the Mint.Samples for analysis were prepared by taking 1 g of turnings which were taken from dierent areas of the reference material bar and combined. Weighed samples were dissolved in 10 ml aqua regia (3 parts concentrated hydrochloric acid and 1 part concentrated nitric acid) in a closed vessel microwave system (Model MDS81D, CEM, Matthews, NC, USA).Prior to sealing, each Teflon vessel was flushed with argon gas to minimize the risk of igniting the hydrogen released during the dissolution process. A pressure controlling device maintained the pressure at 100 psi (1 psi#6.895 kPa). The dissolved gold solution was transferred quantitatively into 50 ml precalibrated plastic volumetric flasks and diluted to volume with water. A 10-fold dilution was performed prior to analysis by taking 100 ml of the gold solution and mixing with 900 ml of water directly in the autosampler cups.Analytical Procedures The Ti signals for each sample and standard were measured with duplicate injections and in a few instances in triplicate. Aliquot volumes varied between 20–30 ml depending on the analyte concentration. In half the determinations an additional 10 ml of water was also injected with the sample to assist in Fig. 1 Integrated absorbance observed for a series of atomizer firings spreading the sample on the atomizer surface.There did not with 20 ml aliquots of 0.2% Au solutions with titanium present. Two appear to be a dierence with or without this added water. empty atomizer firings occur between each duplicate sample firings: Between each sample or standard, two temperature program (a) Massman atomizer, with 1 ng Ti per aliquot; (b) THGA with 0.8 ng per aliquot. cycles were performed without any sample in order to minimize 834 Journal of Analytical Atomic Spectrometry, August 1997, Vol. 12Fig. 1(b). There is a marked decrease in the amount of carry- of a gold matrix. The HGA atomizer was observed to have a higher lifetime (90–110) but this was partially oset by the over observed for the empty firings between samples containing 0.8 ng Ti per aliquot and improved reproducibility for repli- high number of atomizer firings required to clean out the atomizer after the injection of solutions containing more than cates.The integrated absorbance from the second empty firing nearly returns to the values observed for the Au blank. 50 mg l-1 Ti. Eect of Amount of Gold on Integrated Absorbance Comparison between wall and platform atomization in a transversely heated atomizer As noted previously, the integrated absorbance observed for titanium decreases as the amount of gold per aliquot increases Titanium can be atomized eciently from a platform owing to the faster heating characteristics of the integrated platform (Fig. 2). A sharp background absorbance peak occurs about one second into the atomization step which corresponds to a of the transversely heated atomizer. As with Massman atomizers, wall atomization can also be considered, especially since rapid evolution of vapour (or smoke) through the dosing hole. The Ti signal just begins to be observed above the baseline as atomization of titanium does not occur until most of the gold matrix vaporizes and near the time when the atomizer tempera- the background signal peaks.This is similar to the eect observed in a paper by Frech et al.,10 in which the phenomenon ture is isothermal. A comparison of sensitivity and analyte carry-over for wall was explained by the gas phase analyte atoms adsorbing on metal matrix particles which form in the gas phase by cooler and platform atomization in the THGA is given in Table 2. For aqueous solutions of Ti there does not appear to be a atmospheric gas entering the atomizer.The amount of analyte carry-over in the atomizer is approxi- substantial dierence between the two signals produced from the atomization sites. However, the use of the platform results mately 7% of the analyte signal without the presence of the gold matrix and slightly increases with the amount of gold in slightly higher integrated absorbance (Qa) when titanium is determined in a gold matrix. This is particularly important matrix atomized. This is denoted from the titanium peak area measured for the first empty atomizer firing immediately because of the reduced sensitivity of titanium in the presence of large amounts of gold matrix.Consequently, platform following a gold sample containing a fixed concentration of titanium. Since there are no cooler ends in the transversely atomization is preferred despite the lower carry-over demonstrated by wall atomization. Carry-over can be minimized (for heated atomizer, it is likely that carry-over occurs as a result of titanium carbide formation within the atomizer graphite or either wall or platform) by including empty firings between samples.on the surface because of surface and graphite lattice imperfections. The increase of Ti carry-over in the presence of gold From Table 2, the characteristic mass (mo) value for titanium without the presence of gold obtained by platform THGA is may occur owing to the increased carbide formation from a greater number of active sites (surface imperfections) formed lower (more sensitive) than the instrument manufacturer’s expected mo value of 70 pg.This occurs because of the higher from more material interacting with the graphite. Another explanation comes from the work of Eloi et al.11 who found atomization temperature and the longer atomization/read time used in this study. that analytes in aqueous solution migrate into the graphite. It is probable that for a 0.2% Au solution, gold will migrate into the graphite and will take Ti with it into the lattice.The T emperature program for the transversely heated atomizer portion of Ti in the graphite lattice is less likely to be totally removed after the normal temperature program cycle. It was thought that analyte carry-over could be reduced by having two clean out steps in the HGA temperature program at the maximum temperature of 2600 °C (Part A Table 1) as previous demonstrated at 2700 °C with a Massman type atomizer. 1 The carry-over was about 14% of the signal obtained from 200 mg l-1 Ti in 0.1% m/v gold. Temperature programs with three or four clean-out steps did not substantially reduce the carry-over and often caused the furnace power supply to overheat, drawing more current than the internal circuit breakers permit, causing an interuption in the analysis run. A second maximum power heating step is permitted by the P-E 4100 ZL software and firmware. This was used as indicated in Table 1 (Part B).Although the inclusion of a second maximum power heating step (as a clean out step) does not eliminate analyte carry-over, it does require less time and energy, and does not appear to tax the power supply as is the case for multiple clean out steps. The atomizer lifetime is quite low at 60 temperature cycles for the THGA. This is not Fig. 2 Eect of the amount of gold matrix on the integrated unexpected because of the high atomization temperature, long absorbance from gold solutions containing 100 mg l-1 Ti (20 ml aliquot) atomization time (10 s), and second maximum power ramp using the THGA (platform): &, Ti integrated absorbance from gold step used for clean-out.This appears to be a reasonable trade- solutions; +, Ti integrated absorbance from first empty atomizer firing after the gold containing solution. o for the higher sensitivity of Ti determinations in the presence Table 2 Comparison between platform and wall atomization from THGA atomizers for titanium Sample Atomizer type Mean Qa/s m0/pg Residual Qa/s 2 ng Ti Platform 0.152 58 0.013 Wall 0.166 53 0.013 2 ng Ti+100 mg Au Platform 0.066 130 0.010 Wall 0.053 170 0.003 Journal of Analytical Atomic Spectrometry, August 1997, Vol. 12 835Table 3 Comparison of the determined concentration of Ti (mg g-1) Recent work by Goltz et al.12 has shown that 0.3% Freon in gold reference materials by ETAAS to the reference concentration 23 (CHF3) in Ar was eective in reducing uranium carbide values formation (as monitored by ETV-ICP-MS).A similar approach may be useful in reducing analyte carry-over due to titanium Reference material Determined concentration* Reference value† carbide formation. FAU9 6.1±1.1 (n=7) 5.9±1.1 FAU11 18.2±1.4 (n=9) 16.5±3.1 FAU12 21.6±0.7 (n=8) 22.29±0.25‡ Analytical Results * ±Standard deviation. Assessment of accuracy and precision † ±95% confidence interval. Titanium was determined in three fine gold reference materials ‡ ICP-AES determination±standard deviation.and the results are presented in Table 3. The determined concentration value with noted standard deviation overlaps has been verified with reasonable precision and limits of with the reference concentration values. Although analyte detection. carry-over was minimal, duplicate empty firings occurred between each sample and standard. As noted previously, this REFERENCES decreased the atomizer lifetime but this ensured the lowest amount of carry-over. 1 Lopez Garcia, I., Vinas, P., and Herna�ndez Co�rdoba, M., J. Anal. Typically, the relative standard deviation for two replicates At. Spectrom., 1992, 7, 529. was found to be approximately 4% for a 20 ml aliquot of 2 Bhattacharyya, S. S., Chakraborty, R., and Das, A. K., Anal. 42 mg l-1 Ti within a 0.2% gold solution. L ett., 1993, 26, 341. 3 Frech, W., Li, K., Berglund, M., and Baxter, D. C., J. Anal. At. Spectrom., 1992, 7, 141. L imit of detection 4 Sperling, M., Welz, B., Hertzberg, J., Reick, C., and Marowsky, G., Spectrochim.Acta, Part B, 1996, 51, 897. The limit of detection for titanium in a 0.2% gold solution 5 Allen, E., and Jackson, K. W., Anal. Chim. Acta, 1987, 192, 355. was estimated to be 5 mg l-1 based on 10 replicate determi- 6 Kogan, V.V., Hinds, M.W., Ocampo, G., and Valente, G., Precious Metals 1993, ed. Mishra, R., International Precious Metal nations of gold blank solution (k=3). For a 20ml aliquot Institute, 1993, p. 101–116. (40 mg Au), this corresponds to 2.5 mg g-1 in solid. This limit 7 Annual Book of ASTM Standards, Nonferrous Metal Products, B of detection appears to be acceptable to determine the concen- 562–86, vol. 02.04, American Society for Testing and Materials tration of Ti at higher levels of between 6 and 10 mg g-1 which Philadelphia, PA, 1991, p. 425. was a requirement of this research project. 8 Falk, H., Glismann, A., Bergann, L., Minkwitz, G., Shubert, M., and Skole, J., Spectrochim. Acta, Part B, 1985, 40, 533. 9 Weast, R. C., Editor, Handbook of Chemistry and Physics, 68th CONCLUSION edition, CRC Press, Boca Raton, FL, p. B-140. 10 Frech, W., L’vov, B. V., and Romanova, N. P., Spectrochim. Acta, It is evident from the reduction in analyte carry-over that the Part B, 1992, 47, 1461. THGA atomizer is preferred for the determination of Ti in a 11 Eloi, C., Robertson, J. D., and Majidi, V., J. Anal. At. Spectrom., gold matrix. High atomization temperatures (2600 °C), longer 1993, 8, 217. atomization times and a second maximum power step maxim- 12 Goltz, D. M., Chakrabarti, C. L., Gre�goire, D. C., and Byrne, J. P., Spectrochim. Acta Part B, 1995, 50, 803. ize sensitivity and minimize carry-over, but unfortunately reduce atomizer lifetimes. This appears to be an excellent trade-o in view of the diculty of determining such a refrac- Paper 7/00268H Received January 10, 1997 tory element in a relatively large amount of gold matrix that acts to suppress the sensitivity. The accuracy of the method Accepted April 15, 1997 836 Journal of Analytical Atomic Spectrometry, August 1997, V