JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL. 9 737 INTER-LABORATORY NOTE Multi-element Analysis of Archaeological Bronze Objects Using Inductively Coupled Plasma Atomic Emission Spectrometry Aspects of Sample Preparation and Spectral Line Selection I. Segal and A. Kloner lsrael Antiquities Authority P.O. Box 586 Jerusalem 9 1004 Israel 1. B. Brenner* Geochemistry Division Geological Survey of Israel 30 Malkhe lsrael Street Jerusalem 9500 I lsrael Inductively coupled plasma atomic emission spectrometry has been applied to the determination of trace and minor elements in archeological bronzes. Trace element spectral lines were selected on the basis of minimum interference from major element matrix components i.e. Cu Pb Fe and Sn. Using Sc as the internal reference element provided a significant improvement in the relative standard deviation. Accuracy evaluated using standard reference alloys of similar composition was satisfactory.Keywords lnductively coupled plasma atomic emission spectrometry; archaeomefry; bronze; spectral line interference; sample preparation The use of trace and minor elements to fingerprint bronze archaeologic hoards has been reported widely in the litera- ture.14 These studies indicated that the major and trace element composition of the artifacts can reveal additional information on the metallurgical procedures (smelting alloying and casting) used to produce these materials and the source of the raw materials. Most of the analytical data in the literature were obtained using flame atomic absorption spec- trometry.This technique has several disadvantages namely single-element capability inadequate limits of detection and accuracy. Inductively coupled plasma atomic emission spec- trometry (ICP-AES) is now a mature technique for multi- element analysis. However the number of reports describing the use of this technique for the analysis of bronze archaeolog- ical objects is limited. Trampuz-Ore1 et al.' used ICP-AES to determine the provenance of bronze objects from Slovenia. Gluimlia-Mairlo analysed Cu-based objects by ICP-AES and compared artifacts from several archaeological sites although the number of elements determined was limited. Merkel" used ICP-AES for the analysis of Cu ores from Timna located in southern Israel however trace metals were not determined. In the present paper the analytical procedure and perform- ance for the multi-element analysis of archaeological bronzes using ICP-AES is described.Spectral lines for the determi- nation of As Mo Zn Sb Bi Co Cd Ni Au Mn Fe Cr and V are recommended together with their true limits of detection in Cu-Sn-Pb-Zn matrices. The presence of these elements at high concentrations were taken into account in the sample preparation and decomposition procedures. The accuracy of the method was evaluated by analysing standard reference bronzes and related Cu alloys. In the future it is planned to correlate the concentrations of the elements with the provenance and the sources of the archaeological objects using pattern recognition techniques. The aim is to determine the composition and source of the raw materials purity of the metallurgical processes and ulti- mately fingerprint the objects.~~~ ~ * To whom correspondence should be addressed. Experimental Sample Location The bronze objects were collected from two archaeological excavations from Central Israel of the Maresha site mainly Hellenistic in age; and at Bet Guvrin Roman Byzantine and Medieval in age. The archaeology of these sites has been described by K1oner.l' Biblical Maresha consists mainly of three periods of occupation of the Iron Age Late Judean Monarchite Persian Period and Hellenistic. This habitation was terminated by massive destruction in 113 BCE. The archaeo- logical objects studied consisted of bronze and silver coins bronze (occasionally with iron wire bindings) statuettes figur- ines bronze domestic cooking vessels bronze buckles and furniture.The objects analysed from Bet Guvrin were disco- vered in a Roman amphitheatre (2-3 centuries CE) a cave a Crusader church and a medieval fort and consist of a variety of iron bronze and copper objects from different periods. Sample Preparation and Decomposition Owing to the great archaeological value of the objects special attention was given to the pre-preparation and chemical dissolution procedures. Prior to acid dissolution oxidative layers were mechanically removed by scrubbing them with a steel brush. Samples were then cleaned with dilute HCl acetone and de-ionized water. Samples for analysis were removed by micro-drilling in order to minimize damage. The first portions were discarded.In certain cases objects were analysed semi- quantitatively by X-ray fluorescence (XRF) in order to deter- mine major elements and to ascertain the approximate ranges of their concentrations. The dissolution procedure was similar to that described by Hughes et al.l3 Samples weighing 25-50 mg were dissolved in aqua regia (hydrochloric plus nitric acid 3 + 1) in 25-50 ml Pyrex beakers. Samples containing high amounts of Ag and Pb were decomposed in nitric acid. In this procedure Au was not solubilized. In order to determine Au 2ml of aqua regia were added to the sample. The contents of the beakers were heated to 60 "C in order to accelerate dissolution. After cooling 10 ml of de-ionized water were added. Diethylenetriamine was used to avoid precipitation of Ag.14 In most cases complete dissolution was obtained.Samples containing residues were738 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL. 9 fused with sodium peroxide and consisted mainly of alumino- silicate impurities. Solutions were made up to volumes of 25-50 ml. Scandium was employed as an internal standard the final concentration amounting to 5 mg 1-I. Table 2 Details of spectrometer instrumentation Monochromator:- Sequential system Grating Dispersion Jobin Yvon JY 38 3600 grooves mm-' range 180-490 nm 0.27 nm mm-' 6 pm in the first order (25 pm slit-widths) Multi-element Calibration The concentrations and range of the elements in the calibration standards were determined on the basis of the data cited in the literature.'-'' These standards and the expected concen- tration ranges are listed in Table 1.It should be noted that although the Cu contents in the sample usually exceeded 90% a standard containing the equivalent of 100% was prepared. The multi-element standards were prepared from concentrated single-element stock solutions (Merck Darmstadt Germany) and were matrix matched with respect to the acid concentration of the samples. The standard for Ag was prepared separately in nitric acid in order to avoid precipitation of AgCl. The elements contained in a particular standard were dic- tated by chemical stability and mutual spectral interference considerations. Thus Ag occurred on its own in order to avoid precipitation. The standard used for the determination of Zn and Bi did not contain Cu owing to mutual spectral line interferences. Po1ychromator:-- System Grating 170-450 nm Dispersion 0.35 nm mm-' R.f. generator Torch Jobin Yvon Ryton de-mountable Jobin Yvon JY 48 1 m 2550 grooves mm-' spectral range Plasma Therm 2.5 kW < 10 W reflected Table 3 ICP operating conditions Pneumatic nebulizer Meinhard TR-C-20 45 psi* 1.2 1 min-l Plasma gas 14 1 min-l Intermediate gas 0.2-0.4 1 min - ' Sheath gas Trassy-Mermet pneumatic nebulizer 0.2 1 min-' Aerosol carrier gas 0.95 ml min-' Washout period Meinhard nebulizer 30 s Integration period Polychromator 10 s monochromator 0.5 s * 1 psi = 6894.76 Pa.Instrumentation and ICP Operating Conditions The major minor and trace elements were determined using a Jobin Yvon JY 48 polychromator and a JY 38 sequential system.The latter instrument was employed for the determi- nation of elements using alternative interference free spectral lines and for the determination using lines that are subject to interference in the polychromator but are suitable for use in the sequential system owing to its superior resolution. Details of the instrumentation and ICP operating conditions are listed in Tables 2 and 3. ference of Zn I 213.856 nm and Cu I 213.851 nm Fe I 226.505 on Cd I1 226.502nm mutual interference of As 1228.812nm on Cd I1 228.802nm Pb 197.272nm on As I 197.297 Sn I 206.858 nm on Sb 206.833 Cu 328.068 nm on Ag 328.068 nm (an elevated linear background was also noted) and Cu I1 223.008 on Bi I 223.061 nm (Figs. 1-7). Consequently inter- ference free Zn I 206.200 As I 193.699 and Ag I 338.289nm were determined using a high-resolution JY 38 sequential system.A spectral line interference correction due to Sn was made for Sb 206.833nm. Tin is a major element in the specimens analysed and the correction was made for concen- trations amounting to 500 mg 1 of Sn (50% in the solid samples). Similarly an inter-element correction for the effect of Fe I 226.505 on Cd I1 226.502nm was made for Fe concentrations amounting to 1000 mg 1-' (up to 100% in the solid samples). In addition a spectral correction was made for the interference of Cu I1 223.008 on Bi I 223.061 nm up to 800 mg 1-' of Cu (80% in the solid sample). It should also be mentioned that in cases of high As content (Lower Bronze Results Selection of Spectral Wavelength The spectral wavelengths and background compensation pos- itions were selected by studying spectral scans in the regions of interest.These are listed in Table 4. Spectral line selection was made on the basis of minimum spectral line interferences and maximum sensitivity. Significant spectral interferences were observed for the following spectral lines mutual inter- Table 1 factors 500-1000 Calibration standards for the analysis of bronzes and copper alloys. Concentrations are given in mg I-' in solution. Dilution Standard Element and line Sn I1 As I Mo I1 Zn I Pb I1 Bi I c o I1 Cd I1 Ni I1 Au I Mn I1 Fe I1 Cr I1 v I1 c u I Ag 1 Sb I s c I1 Wavelength/ nm 189.980 193.699 202.030 206.200 220.353 223.061 228.616 226.502 23 1.604 242.795 257.610 259.940 267.716 3 10.230 324.754 338.289 206.833 361.384 Blank 1 0 20 0 0 0 0 20 0 0 0 0 0 0 0 20 0 0 0 0 - - - - - - - - - - - 800 - - - 5 mg 1-l internal standard 2 3 4 200 - 5 20 - - 20 - 20 - - 400JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL.9 739 7213.856 nm Cu 1213.851 nrn Table 4 Analytical ranges and spectral lines used for the analysis of Cu-Sn-Pb and Ag objects Element and line Sn I1 As I Mo I1 Zn I Pb I1 Bi I c o I1 Cd I1 Ni I1 . Au I Mn I1 Fe I1 Cr I1 v I1 c u I Ag 1 Sb I s c I1 Wavelength/nm 189.980 193.699 202.030 206.200 220.353 223.061 228.6 16 226.502 231.604 242.79 5 257.610 259.940 267.716 310.230 324.754 338.289 206.833 361.384 Background/nm 0.03 1 0.058 0.03 1 0.047 0.03 1 0.05 1 0.05 1 0.03 1 0.03 1 0.03 1 0.03 1 0.03 1 0.05 1 - 0.037 - 0.037 __ -0.037$ - Instrument* SIM SEQ SIM SEQ SIM SIM SIM SIM SIM SIM SIM SIM SIM SIM SIM SEQ SIM SIM Integration time/s 10 0.5 10 0.5 10 10 10 10 10 10 10 10 10 10 10 10 10 0.5 LODt/pg 1-' 349.0 31.0 10.0 2.6 22.0 16.0 5.7 1.7 11.0 3.5 0.7 8.3 2.0 3.5 0.7 1.2 17.0 __ * SIM polychromator JY 48; SEQ sequential JY 38.t LOD limit of detection. $ When Fe is >5% the background for Sb is 0.051 nm. 2 13.806 213.856 213.906 Wavelengthlnm Fig. 1 Spectral scan of Zn I 213.856 nm (corrected for up to 800 mgl-l Cu) age objects) Cd I1 228.802 nm is not suitable owing to the interference of As I 228.812 nm. Additionally when samples contain high Pb contents As 197.297nm is subject to inter- ference from the intense Pb 197.272 nm line. Thus the use of a high-resolution sequential spectrometer was an advantage and allowed alternative spectral lines to be employed.It is evident that the alloy elements of archaeological bronzes cause significant spectral line interferences that must be taken into consideration when selecting spectral lines for the determi- nation of trace elements. Limits of Detection The 2a limits of detection were determined using ultra-pure matrix solutions as blanks. These are listed in Table 4. 226.552 226.452 226.502 Wavelengthlnm Fig. 2 Spectral scan of Cd I 226.502 nm (corrected for up to 1000 mg 1-1 Fe) Precision and Accuracy The precision [relative standard deviation (YO RSD)] of deter- mination without the use of Sc I1 as the internal standard varied from about 0.2 to 6%. The use of Sc I1 as the internal standard resulted in a significant improvement both in the precision (from <0.1 to about 3%) and accuracy (Table 5).Internal standardization compensated for physical interference effects in the sample introduction system. High RSD values observed for Bi could be attributed to the low signal-to- background ratio and the inadequacy of the Sc I1 line to compensate for variations in these intensities. The accuracy of the analytical protocol was determined by740 I Cu1328.068nm I I JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL. 9 Cd I I I ! i As1228.812nm 228.712 228.802 228.892 Wavelengthlnm Fig.3 Spectral scan of Cd I 228.802nm. Note As I 228.812 interference t > v) C a C 4- .- 4- - ~~ . . . . . . :,Pb 197.272 nm . . . * . . . . . . @ a . . . .'.*. .... .. . . . : ..'. . . -.-../I.- .-.-. -. . . . I I I 197.247 197.297 197.347 Wavelengthhm Fig. 4 Spectral scan of As I 197.297 nm. Note Pb 197.272 interference analysing certified reference BCS 207 Bronze MBH GM 50 MBH LB 30 and a simulated synthetic standard containing a wide range of trace elements of interest. A comparison of the data obtained with the recommended values for the major minor and several trace elements indicated that the accuracy obtained is satisfactory (Table 5 ) . t 1. v) C a C 4- .- 4- - 1 . .... Sn i 206.858 rimy' Sb 1206.833 nm I I j.783 206.833 206.883 Wave1 engt hln m Fig.5 Spectral scan of Sb I 206.833nm. Note Sn I 206.858nm interference t > v) S a C .id .- 4- - I\ I ' I 'I/ Ag 1328.068 nm I I \ I 1 1 1 I 1 1 1 1 1 1 1 I I I I 328.018 328.1 18 328.068 Wavelengt hln m Fig.6 Spectral scan of Ag I 328.068nm.Note Cu I 328.068 nm interference and significant elevation of the background Conclusions In developing a multi-element method for the determination of trace and minor elements in Cu-Sn-Pb archaeological objects using ICP-AES there are two main concerns selection of the appropriate spectral lines and the dissolution procedure ensuring that all the components (Pb and Sn) are present in the solution and that they remain stable over a reasonable period of time. The addition of diethylenetriamine resulted in prolonged stabilization of Ag bearing solutions. The study indicated that significant spectral line interferences can occur owing to the presence of high concentrations ofJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL.9 Table 5 Analytical data for certified reference bronzes and similar materials with and without internal references 741 Element Sn As Mo Zn Sb Pb Bi c o Cd Cd Ni Mn Fe Cr V c u Ag Element Sn As Mo Zn Sb Pb Bi Co Cd Cd Ni Mn Fe V cu Ag Wavelength/ 189.980 193.699 202.030 206.200 206.8 3 3 220.353 223.061 228.616 226.502 228.802 231.604 257.610 259.943 267.716 310.230 324.754 338.289 nm Instrument SIM SEQ SIM SEQ SIM SIM SIM SIM SIM SEQ SIM SIM SIM SIM SIM SIM SEQ Wavelength/ 189,980 193.699 202.030 206.200 206.833 220.353 223.061 228.616 226.502 228.802 231.604 257.610 259.943 310.230 324.754 338.289 nm In strument SIM SIM SIM SIM SIM SIM SIM SEQ SIM SIM SIM SIM SIM SEQ SEQ SEQ Simulated bronze SRM 207t With IR? Without IR Recommended/ Determined/ RSD mg 1-' mgl-' (%) 100 5 5 5 5 15 5 5 5 5 5 15 5 5 200 - 99.9 4.78 4.82 4.81 4.75 13.8 - 4.86 4.82 4.72 4.87 4.86 4.81 4.92 13.4 205.6 0.47 1.40 0.67 0.3 0.71 0.58 0.55 0.03 0.04 0.50 0.06 0.02 0.10 0.06 0.68 - MBH GM 50 Gun Metal? With IR Recommended Determined RSD (% m/m) (X m/m) (%) 4.73 5.1 0.2 0.05 0.046 2.3 5.08 5.13 1.2 0.08 0.07 2.3 4.12 5.4 0.12 0.05 < 0.01 - - - - - - - 1.45 1.56 0.3 0.05 0.066 0.1 0.23 0.21 0.1 Determined/ RSD Recommended mgl-' (%) (% m/m) 113.2 0.40 5.35 0.58 5.02 0.37 5.14 0.49 3.64 1.80 15.1 1.70 5.10 0.52 5.11 1.6 5.05 0.90 5.05 0.52 5.08 0.70 14.9 0.67 5.05 0.65 5.15 0.56 222.3 0.15 - - - ~- 9.80 0.05 2.50 0.41 <0.078 - - - - - 0.09 0.06 - - - 86.84 0.02 MBH LB 30 Leaded bronze With IR Recommended (% m/m) 10.3 0.02 < 0.01 0.04 9.4 0.025 - - - - 1.52 <0.01 <0.01 - __ - Determined (% m/m) 11.4 < 0.05 <0.1 0.03 10 < 0.01 - - - - 1.66 < 0.02 < 0.05 - .- - With IR Determined (X m/m) 9.70 0.06 2.45 0.38 - - t 0 .1 - - - 0.097 0.06 - - - 85.9 0.02 RSD (%) 0.70 1.50 0.70 0.90 -. - - - __ - 2.20 0.51 - - - 0.29 2.4 Without IR Determined (% m/m) 10.70 0.06 2.48 0.43 0.10 - - - - - 0.097 0.06 - - - 87.3 0.02 RSD (%) 0.25 1.90 0.11 0.26 5.80 - - - - - 0.66 0.51 - - - 0.32 2.4 * IR internal reference. t Bureau of Analysed Samples (Newby Middlesbrough UK) Certificate No. 207 Bronze. 2 MBH Analytical (Barnet Hertfordshire UK) Certificate of Analysis C 33 X No. NGM 50. Q MBH Analytical Certificate of Analysis C 32 X No. LB 30. ~ Cu II 223.01 nm I ' Bi 1223.061 nm r/ I I I I I I I I I I -.-.--.?K... .. _. . . . ..'. .. ... .. .'.:--.z-.- d- I 223.021 223.061 223.101 Wavelengthhm Fig. 7 Spectral scan of Bi I 223.061 nm. Note Cu TI 223.01 nm interference alloying elements such as Cu Pb Sn and Fe. In certain cases unusual concentrations of As in Lower Bronze age implements could interfere with the sensitive line of Cd. The use of a high- resolution sequential spectrometer facilitated the use of these spectral lines and in cases where overlap occurred the selection of alternative lines for analysis. References Ping Y. -D. Recent Advances in the Conservation and Analysis of Artifacts University of London Institute of Archaeology London Craddock P. T. in The Egyptian Mining Temple at Timna ed. Rothenberg B.Institute For Archaeometallurgical Studies London 1988 pp. 169-181. Rothenberg B. The Ancient Metallurgy of Copper Institute For Archaeometallurgical Studies London 1990. Craddock P. T. and Glumlia-Mair A. R. Bronze Working Centers of Western Asia 1000-539 BC ed. Curtis J. Kegan Paul International British Museum London and New York 1988 Shalev S. Goren Y. Levy T. E. and Northover J. P. Archaeometry 1992 34 63. Leese M. N. Craddock P. T. Freestone I. C. and Rothenberg B. Wiener Berichte iiber Naturwissenschaft in der Kunst 198511986 213 90. Key C. A. The Cave of the Treasure ed. Bar Adon P. Israel Exploration Society Jerusalem 1980. Telecote R. F. A History of Metallurgy Metals Society London 1976. Trampuz-Orel N. Milic Z. Hudnik V. and Orel B. in Archaeometry 1990 33 267. 1987 pp. 119-124. pp. 317-326.742 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JUNE 1994 VOL. 9 10 Gliumlia-Mair A. R. Archaeometry 1992,34 107. 14 Dixon K. Russell G. M. Wall G. J. Eddy B. T. Mallett R. C. 11 Merkel J. J. MASCA 1985 3 164. and Royal S. J. The Analysis of Anode Sludges and Their Process 12 Kloner A. Ancient Maresha (In Hebrew) Kadmoniot 1991 Solutions and Beneficiation Products National Institute for p. 9596. Metallurgy Randburg South Africa 1979 Report 201 1. 13 Hughes M. J. Cowell M. R. and Craddock P. T. Archaeometry 1976 18 19. Paper 310691 2E Received November 11 1993 Accepted March 16 1994