Determination of impurities in antique silver objects for authentication by laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) W. Devos, Ch. Moor* and P. Lienemann Swiss Federal Laboratories for Materials Testing and Research (EMPA), U� berlandstrasse 129, CH-8600 Du�bendorf, Switzerland Received 4th January 1999, Accepted 25th February 1999 In addition to visual characteristics, a less manipulable criterion for authenticity verification of silver antiques is given by trace and minor element patterns in the silver alloy. The analytical method used to analyse precious silver antiques should not visibly damage the object and should enable the determination of impurities in the ppm–0.5% range.Using laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS), visible damage can be restricted to an acceptable minimum. Because most antique silver objects are too large to fit into a normal laser ablation cell, an alternative cell design was used that allows a direct, virtually non-destructive analysis of entire antique silver objects.This cell is placed upon the object to be analysed. A micro-amount of the object is then ablated through an aperture in the bottom of the cell. The 100 mm wide craters are almost invisible on an antique silver object. The analytes Zn, Cd, Sn, Sb, Au, Pb and Bi were measured. Signals were normalized to the Ag signal and silver standard materials were used for external calibration.The crater-to-crater repeatability of the normalized signals in a homogeneous silver sample was below 10% RSD (n=3) for most elements. Detection limits lie within the sub-ppm to 2 ppm range. The accuracy was validated with comparative ICP-MS measurements after digestion and with X-ray fluorescence (XRF) measurements. The analysis of eight antique silver objects, including one forgery, illustrates the application of the method. Classical methods for direct solid sample analysis, such as Introduction X-ray fluorescence (XRF) analysis, scanning electron Various criteria are available to verify if an antique silver microscopy with an energy-dispersive X-ray detection system object is authentic or a forgery.A trained expert in silver (SEM-EDXRF) or electron probe microanalysis (EPMA), antiques may be able to identify forgeries by a careful visual are fully non-destructive, but lack sensitivity in the study of stylistic characteristics, hallmarks, fabrication tech- lower mg g-1 range.Furthermore, most antique silver objects nique and signs of wear. However, a malevolent silversmith are too large to be fitted into the sample chambers commonly with suYcient expertise can manipulate these criteria so as to used with these techniques. make a forgery look real, even to a trained eye. An additional Wet analytical techniques, e.g., atomic absorption speccriterion much more diYcult to manipulate is given by the trometry (AAS) and inductively coupled plasma mass specchemical composition of the silver alloy.In addition to the trometry (ICP-MS) of solutions, require acid digestion of the main constituents of silver (75–95%) and copper (5–25%), a Ag–Cu alloy. This is particularly troublesome if Au also has silver alloy contains minor or trace impurities typical of the to be determined, as traces of Au are only partially soluble in metallurgical manufacturing and refinement process of the nitric acid, whereas Ag will precipitate as silver chloride in silver, such as Zn, Cd, Sn, Sb, Au, Pb and Bi.Silver refinement aqua regia, giving rise to losses of trace elements due to has undergone fundamental changes during the past centuries, coprecipitation. It is possible to keep Ag in solution as chloride especially in the 19th century, as the introduction of the Parkes complexes with a carefully balanced mixture of nitric and process around 1850, followed by electrolytic refinement in hydrochloric acid, but this demands quite a lot of experience 1884, led to much lower contents of Zn, Sn, Sb, Au, Pb and and time.To overcome this problem, Hinds3 used solid sam- Bi. Cd, on the contrary, only discovered in 1817, has tempor- pling graphite furnace AAS (GFAAS) for the determination arily been used as an additive in 20th century silver solders of Au, Pd and Pt in high purity silver, and Moor et al.4 and special silver alloys and is a clear indication of a modern reported the application of solid sampling electrothermal silver alloy.1,2 As it was not until the latter part of the 19th vaporization ICP-MS (ETV-ICP-MS) to determine Au and century that collecting antique silver came into vogue, most other impurities in silver alloys.In order to allow practical fake silver objects stem from that period on. Therefore 1850 sample handling, both techniques require in practice at least is a significant date to distinguish forgeries from earlier genuine 50–100 mg of sample.Coupling solid sample introduction silver antiques. techniques to ICP-MS has a powerful advantage over solid It is evident that any analytical method used for authenticity sampling GFAAS in that it oVers a quasi-simultaneous multiverification of often precious antique silver objects should be element analysis in one signal acquisition run. either non-destructive or require an extremely small sample In recent years, laser ablation (LA) has gained increased amount.Furthermore, it should allow the determination, popularity as a solid sample introduction technique to preferably simultaneously, of the relevant elements in a concen- ICP-MS, thanks to its low detection limits, its wide linear dynamic range, its spatial resolution and its low consumption tration range from a few mg g-1 to about 0.5%. J. Anal. At. Spectrom., 1999, 14, 621–626 621of sample material.5–9 The benefits of LA-ICP-MS in trace objects, a flat rubber sealing-ring can be used instead.For the analyses in this study, the bottom disc with a 5 mm aperture analysis of silver and gold materials for assessing the provenance of stolen gold,10 for prospecting purposes11 and for was used. A supporting platform, adjustable in three dimensions and the determination of impurities in pure coinage silver and gold12 have been reported. mounted on the Laser Sampler Model 320 supporting block with its three stepping motors, allows the aperture of the cell In LA-ICP-MS, a sample is usually mounted into a relatively small, closed laser ablation cell, in which the ablation process and, consequently, the area of interest on the object to be exactly positioned under the focusing lens.The autofocus takes place and through which a carrier gas stream flows to transport the ablated material to the ICP-MS. Objects larger system then automatically adjusts the surface of the object within the focal plane of the laser beam.than a few centimetres, however, e.g., the silver antiques under study here, require the removal of a fragment of the object in order to fit into a normal laser ablation cell and can thus be Laser analysis and signal processing analysed only indirectly. For a high quality piece of antique The carrier gas flow rate was optimized for maximum inte- this is unacceptable. Special cell designs have been proposed grated signals with a silver standard material.A short bivariate by Arrowsmith and Hughes13 and Gu� nther14 which consist of optimization study was carried out for the number of laser a cell that encloses the plume of ablated material but not the pulses per analysis point and the excitation lamp energy in sample. In this work, an alternative cell design was used in order to obtain craters of a suitable size. By increasing the combination with a home-built autofocus system, allowing the excitation lamp energy, which directly influences the laser direct analysis of entire antique silver objects by LA-ICP-MS.beam energy, wider and deeper craters are produced, whereas Quantitative information was obtained using matrix matched increasing the number of shots only produces deeper craters. standards. The operating conditions of the ICP-MS and the laser sampler are listed in Table 1. The laser was focused on the sample Experimental surface. Analysis points were chosen a few millimetres apo correct for diVerences in ablation eYciency between Instrumentation analysis points, 107Ag was used as an internal standard.This requires the silver content of the samples and external stan- An Elan 6000 ICP-MS (Perkin-Elmer SCIEX, Thornhill, Ontario, Canada) was used in combination with a Laser dards to be approximately known. The external standards used in this work (see below) contained between 99.4 and Sampler Model 320 (Perkin-Elmer SCIEX, Thornhill, Ontario, Canada). Measurements were performed in the dual detection 100% silver.Unless given by other sources, a silver content of 84% was used as an approximation of the silver content in the mode of the Elan 6000, which oVers the measurement of pulsecounting and analogue signals simultaneously.15 The Laser alloy of antique silver objects. Some commonly used antique alloys were 12/16 (i.e., m/m Ag/alloy ratio), 13/16, 14/16 and Sampler Model 320 was modified by adding a home-built autofocus system.16 The laser used in this work is an Nd5YAG 15/16 silver (1/16=1 ‘Lot’), Sterling silver (92.5% fine silver) and 9/12, 10/12 and 11/12 silver (1/12=1 ‘denier’ or laser (Spectra-Physics, Darmstadt, Germany), operated at 532 nm.The area of ablation can be observed by a video ‘dinero’).17–19 Generally, the silver content of antique and modern alloys for silverware lies between about 75% and 95%. camera. An alternative laser ablation cell was used to analyse entire Thus, using an approximation of 84% silver for internal standardization introduces a supplementary relative uncer- antique silver objects directly, i.e., without the need to remove a fragment from the object.This cylindrical borosilicate glass tainty of, at most,±12%, which is still acceptable for authentication purposes. cell (r=2.1 cm; h=1.7–2.4 cm) has a flat exchangeable Plexiglas bottom with a central aperture of 5 mm, 10 mm or Based on the literature1 and preliminary experiments with solid sampling ETV-ICP-MS and conventional ICP-MS,4 the 20 mm, and is placed upon the object to be analysed (Fig. 1).Part of the object is then ablated by the laser beam through following isotopes were chosen for analysis: 66Zn, 107Ag, 111Cd, 117Sn, 121Sb, 197Au, 208Pb and 209Bi. The transient signals were the aperture. The cell is continuously flushed with a carrier gas (argon) via a lateral inlet and outlet tube to transport the ablated material to the ICP-MS. To prevent the Ar carrier Table 1 Operating conditions for the ICP-MS and the laser sampler gas, and therewith the ablated material, from escaping out of ICP-MS the cell through the aperture, modelling plasticine is applied Rf power/W 1050 between the cell and the object, around the aperture.For flat Gas flow rates/l min-1 Plasma gas 15 Auxiliary gas 0.8 Carrier gas 1.0 Lens setting Autolens mode (variable) Points per peak 1 (at peak maximum) Measuring mode Peak hopping Isotopes measured 66Zn, 107Ag, 110Cd (dummy), 111Cd, 117Sn, 121Sb, 197Au, 208Pb, 209Bi Dwell time/ms 10; 111Cd and 121Sb: 50 Sweeps per reading 1 Readings per replicate 91 Detector mode Dual (pulse counting and analogue) Laser sampler Mode Q-switch Q-switch time/ms 240 Excitation lamp energy/J 45 Pulse frequency/Hz 10 Fig. 1 The alternative laser ablation cell used in this work, placed Number of pulses/analysis point 60 upon a silver object: 1, laser ablation cell; 2, focusing lens; 3, autofocus; Focus On sample surface (autofocus) 4, supporting platform. 622 J. Anal. At. Spectrom., 1999, 14, 621–626integrated over a period of 20 s after the laser had started firing, thus covering the entire analyte signals, while allowing the silver signal to decrease to about 1% of its maximum intensity. A blank correction was carried out by subtracting the Ar gas blank signal, integrated over the same period of time and measured with the laser running but the laser beam path blocked.Calibration To obtain similar ablation conditions for external standards and samples, matrix-matched standards or standard materials with chemical and physical properties very similar to the samples are preferred in laser ablation analysis. Three silver calibration standards were purchased from VEB Bergbau- und Hu�ttenkombinat (Freiberg, Germany), containing the elements of interest at known concentration levels of around 500 mg g-1 (‘Ag500’), 100 mg g-1 (‘Ag100’) and 10 mg g-1 (‘Ag10’) respectively.As these standards were in the form of 10 cm rods with a diameter of only 7 mm, and were therefore too small to place the cell on, a cross-section was made of each standard after embedding in a Plexiglas resin (RESINAR F, Wirtz Buehler, Du� sseldorf, Germany). Analysed objects and cleaning procedure Eight silver objects, provided by a private antique silverware dealer, were analysed in order to investigate their authenticity. Before analysis, the objects were consecutively cleaned with silver polish to remove any AgS patina, pro analysi grade ethanol (Merck, Darmstadt, Germany) and ultrapure water (Milli-Q, Millipore, Bedford, MA, USA).Comparative analyses by solution ICP-MS and WD-XRF For comparative analysis by ICP-MS of solutions, four samples of 3–5 mg each were taken from an antique silver alloy and submitted to the following open acid digestion procedure. Fig. 2 Details of the silver ‘deer’ from Fig. 1 after laser ablation After dissolving the silver alloy with 1 ml concentrated HNO3 analysis.(a) Optical microscopy view. The arrows indicate the ablated for ca. 10 min in a 100 °C hot-water bath, 3 ml concentrated craters. (b) Scanning electron microscope (SEM) view of one crater. HCl was added (causing precipitation of Ag as AgCl) to The diameter of the crater is ca. 100 mm. The not quite circular crater dissolve the remaining undissolved Au, and the solution was shape is possibly due to imperfect optical alignment of the home-built left to boil for another 30–60 min.After cooling, the solution frequency doubling upgrade. was made up to 10 ml with deionized water. Comparative analyses by wavelength dispersive XRF Calibration and detection limits (WD-XRF) were carried out on a Philips PW1404 (Philips, The three calibration standards ‘Ag10’, ‘Ag100’ and ‘Ag500’ Almelo, The Netherlands) with a Cr tube for the determination were analysed to investigate whether a linear calibration could of Sn and Sb and a Philips PW2400 with an Rh tube for the be obtained with the alternative cell design.The linearity of other elements of interest. A sample with a flat geometry was the calibration graphs is excellent, with correlation coeYcients prepared from a piece of an antique object which could be of 0.996 or better (Fig. 3). Furthermore, the fitted calibration destroyed for analysis. The measurements were calibrated with lines go almost perfectly through zero, which confirms that standards ‘Ag100’ and ‘Ag500’.the subtraction of an Ar gas blank is an adequate method for blank correction for the given experimental set-up and this Results and discussion type of sample. The 111Cd signal initially gave rise to a problem. When Quasi-non-destructiveness 111Cd was measured immediately after 107Ag in the peak hopping sequence, a steady state signal instead of the normal The laser parameters shown in Table 1 resulted in ablation craters of about 100 mm across (Fig. 2b).Three craters per transient signal could be observed at mass 111, which completely masked the transient 111Cd signal for the standard with sample were shot. The craters are small enough to be almost invisible with the naked eye (Fig. 2a) on antique silver, and the lowest Cd concentration (‘Ag10’: 8.3 mg g-1). This resulted in far too high a signal for Cd in the Ag10 standard. The thus are virtually non-destructive. Although the crater size could easily be reduced further, smaller craters would decrease constant 111Cd signal abruptly dropped to background level at the point where the 107Ag signal had decreased to about the relative sensitivity and would be less representative of the silver alloy.From the crater width and d, the ablated 107 counts per second (Fig. 4a). This phenomenon was reproducible and was probably caused by a memory eVect in the mass per crater was estimated to be roughly about 3 mg.After ablation, sometimes a black discolouration of the silver could discrete dynode electron multiplier or the electronics of the detection system, due to the high count rates for 107Ag. be observed around the craters. This discolouration could easily be wiped oV with a silver polishing cloth. Introducing a ‘dummy’ mass between 107Ag and 111Cd, e.g., J. Anal. At. Spectrom., 1999, 14, 621–626 623110Cd, solved the problem and resulted in a normal transient signal (Fig 4b) and a linear calibration curve for 111Cd (Fig. 3). The detection limits for the elements measured, calculated as 3s from three Ar blank replicates, lie between 0.06 ppm (Pb and Bi) and 1.6 ppm (Zn). Accuracy and repeatability The linearity of the calibration lines based on the solid silver calibration standards suggests a good accuracy, limited by the quality of the standards. Furthermore, an antique silver sample, indicated here as ‘Alloy A’, was analysed both by LA-ICP-MS and by ICP-MS using a cross-flow nebulizer.As shown in Table 2, the results of the two approaches were in fairly good agreement, considering that only three craters were shot. The diVerences between the corresponding mean values obtained by the two methods are statistically insignificant (95% confidence level t-test) for all elements except Au and Bi. The somewhat lower values for these elements with the solution method may be due to incomplete dissolution. Additionally, the results of laser ablation analysis of another antique silver sample (’Alloy B’) were compared with XRF measurements (Table 2). The concentrations of Sb and Cd were below the detection limits of XRF for these elements in this type of matrix.A statistically significant diVerence between the two methods was observed for the Pb values. The reason for this diVerence is not clear, but it might be attributed to surface contamination eVects during sample preparation for XRF, as the comparison of LA-ICP-MS with solution ICP-MS did not reveal any systematic deviation of LA-ICP-MS for lead.The agreement between LA-ICP-MS and XRF for the other elements was satisfactory in view of the objectives of this study. The crater-to-crater repeatability (n=3) of the analyte signals (ratioed to the Ag signal ) is summarized in Table 3 for the measurements of the standards and the two aforementioned antique silver alloys. The uncertainties shown in Table 2 are larger than the RSDs in Table 3 as they were calculated on a Fig. 3 Calibration graphs of the integrated signal intensity of the 95% confidence level and include the propagated uncertainty isotope monitored ratioed to the integrated Ag signal intensity for the of the calibration. For a homogeneous silver alloy, like ‘Alloy three solid silver calibration standards (VEB Bergbau- und Hu� ttenkombinat, Freiberg, Germany). B’, RSDs (n=3) well below 10% could be observed. A larger relative uncertainty was found for Cd in this alloy because its signal was only just above the background noise.In other samples, sometimes a spreading up to 30% RSD was seen for some elements, possibly due to inhomogeneity. Analysis of the antique silverware objects Finally, eight antique objects dating from the 16th to the 20th century were directly analysed by LA-ICP-MS using the alternative cell design (Table 4). Their provenance is mainly West European, except for one small spoon from the USA.For gilded objects, an area was analysed where the gilding was not applied or where it had worn oV. It can be observed from Fig. 5 that the small spoon from around 1900 and the 20th century meat fork are characterized by relatively low impurity levels, except for Cd in the meat fork. The relatively high Cd concentration in the meat fork is a clear indication (but not a prerequisite) of a 20th century silver alloy (see Introduction). With the exception of the ‘deer’ and its matching socle, all objects from before 1850 show generally much higher concentrations of the measured impurities.For the ‘deer’ and socle, however, the impurity levels are much too low to be consistent with a silver alloy from before 1850. In view of these results, this object is to be regarded as a forgery, probably from the late 19th century. The high value for Au in the ‘deer’ is most Fig. 4 Transient 111Cd signal intensity for standard ‘Ag10’ measured probably due to remaining traces of the gilding.A fine silver without 110Cd (a) and with 110Cd (b) as dummy isotope. The 107Ag and the 117Sn signal profiles are given for comparison. determination by ICP-OES after digestion of a 3 mg sample 624 J. Anal. At. Spectrom., 1999, 14, 621–626Table 2 Comparison of LA-ICP-MS (n=3) with ICP-MS from solutions after digestion (n=4) for silver Alloy A and with WD-XRF (n=1) for Alloy B. All values are in mg g-1. The uncertainties, given as 95% confidence intervals, are propagated values, including the uncertainty of the calibration Alloy A Alloy B Analyte LA-ICP-MS Solution ICP-MS LA-ICP-MS WD-XRF Zn 7100±2300 7500±2400 1700±500 1400±200 Cd 34±10 26±10 <2 <200 Sn 290±90 290±60 400±70 580±200 Sb 300±100 300±70 40±20 <200 Au 2800±500 2200±500 1400±100 1600±200 Pb 2200±1200 1900±300 1900±500 2900±300 Bi 200±100 110±50 120±20 200±100 from the bottom of the socle additionally confirmed a younger Table 3 RSDs (%) for the normalized signal ratios (n=3) measured age, as a silver content of almost exactly 80% was found.This in the external standards and in two antique silver alloys. The corresponds to ‘silver 800’, an alloy that has been used only concentrations (mg g-1) of the analytes are given in parentheses since the 19th century. RSD (%) (and concentration/mg g-1) Analyte ‘Ag10’ ‘Ag100’ ‘Ag500’ ‘Alloy A’ ‘Alloy B’ Conclusion Zn 17 (9.8) 4 (120) 3 (424) 13 (7100) 6 (1700) The presented method has been shown to be suYciently precise Cd 8 (8.3) 3 (95) 5 (349) 12 (34) 33 (<2) and accurate to allow an ante quem/post quem dating (before Sn 11 (11.5) 13 (101) 2 (497) 12 (290) 4 (400) or after 1850) of antique silverware. The main advantage of Sb 4 (11) 5 (89.2) 3 (463) 18 (300) 5 (40) this method over wet techniques or electron probe microanal- Au 31 (13.6) 4 (101) 11 (506) 7 (2800) 2 (1400) ysis is that relatively large silver objects can be analysed Pb 5 (15.8) 11 (97.6) 1 (489) 22 (2200) 6 (1900) Bi 5 (14.7) 5 (107) 3 (566) 21 (200) 3 (120) directly without having to take a visible fragment of the object for digestion or in order to fit into a vacuum sample chamber.Table 4 Description of the antique silverware objects analysed by LA-ICP-MS Object Provenance Date Bern (Switzerland) 2nd half 18th century Fork Fork Paris (France) ca. 1780 Meat fork SchaVhausen (Switzerland) 20th century Small spoon USA ca. 1900 Spoon Amsterdam (The Netherlands) 2nd half 18th century Deer with matching socle (gilded) Southern Germany Mid-17th century (?) Pomegranate beaker (partially gilded) Southern Germany 1580 Polychromed wooden sculpture of a man (with gilded silver Schwa�bisch-Gmu�nd (Germany) 1620 decoration) Fig. 5 Impurity patterns in the antique silver objects analysed by LA-ICP-MS. For security reasons, only relative values are shown. The concentration of each element has been normalized to its maximum concentration measured in these objects.The low impurity levels in the ‘deer’ and the matching socle suggest a post-1850 origin for the objects. J. Anal. At. Spectrom., 1999, 14, 621–626 6258 E. F. Cromwell and P. Arrowsmith, Anal. Chem., 1995, 67, 131. As the craters obtained after firing the laser are almost invisible 9 W. T. Perkins, N. J. G. Pearce and J. A. Westgate, Geostandards with the naked eye, the proposed method can be re as Newsletter.The Journal of Geostandards and Geoanalysis, 1997, being virtually non-destructive. 21, 175. 10 R. J. Watling, H. K. Herbert, D. Delev and I. D. Abell, Spectrochim. Acta, 1994, 49B, 205. Acknowledgements 11 P. M. Outridge, W. Doherty and D. C. Gregoire, J. Geochemical Exploration, 1998, 60, 229. The authors wish to thank Mr. Martin Kiener for providing 12 V. V. Kogan, M. W. Hinds and G. I. Ramendik, Spectrochim. the antique silverware and the very helpful background Acta, 1994, 49B, 333. information. 13 P. Arrowsmith and S. K. Hughes, Appl. Spectrosc., 1988, 42, 1231. 14 D. Gu¡§ nther, Doctoral Thesis, Martin-Luther-University Halle Wittenberg, Halle, Germany, 1990. References 15 U. Vo¡§ llkopf and K. Barnes, At. Spectrosc., 1995, 16, 19. 16 B. Wanner, Ch. Moor, P. Richner, R. Bro¡§nnimann and B. 1 E.-L. Richter, Altes Silber, Imitiert-Kopiert-Gefa¡§lscht, Keyser, Magyar, Spectrochim. Acta, 1999, 54B, 287. Munich, 1983, ch. 15, pp. 245.249. 17 E.-L. Richter, Altes Silber, Imitiert-Kopiert-Gefa¡§lscht, Keyser, 2 R. D. Mushlitz, in McGraw-Hill Multimedia Encyclopedia of Munich, 1983, ch. 11, p. 166. Science and Technology.¡®Silver Metallurgy¡�, McGraw-Hill, CD- 18 J. Bly, Discovering Hallmarks on English Silver, Shire Publications, ROM Version 1.0, 1994. Buckinghamshire, 8th edn., 1997, pp. 3.13. 3 M.W. Hinds, Spectrochim. Acta, 1993, 48B, 435. 19 J. Divis¢§, Silber-Stempel aus aller Welt, Battenberg Verlag, 4 C. Moor, P. Boll and S. Wiget, Fresenius¡� J. Anal. Chem., 1997, Augsburg, 1995, p. 39 (original title: Znac¢§ky Str¢§©¥¢¥ bra, Aventinum 359, 404. Verlag, Prague, 1976). 5 P. Arrowsmith, Anal. Chem., 1987, 59, 1437. 6 E. R. Denoyer and K. J. Fredeen, Anal. Chem., 1991, 63, 445A. 7 L. Moenke-Blankenburg, Spectrochim. Acta, 1993, 15, 1. Paper 9/00073I 626 J. Anal. At. Spectrom., 1999, 14, 621