176 177 GLANCING INCIDENCE XRF FOR THE ANALYSIS OF EARLY CHINESE BRONZE MIRRORS Robert W. Zuneska, Y. Rong, Isaac Vander, and F. J. Cadieu* Physics Dept., Queens College of CUNY, Flushing, NY 11367. ABSTRACT The composition of ancient Chinese bronze mirrors has been determined by X-ray fluorescence measurements. The mirrors were archived to be from either 200 BCE-200 CE and from approximately the 800 CE period. Glancing incidence X-Ray excitation proofed highly useful for obtaining the elemental composition from the fluorescence measurements. Measurements of the tin L /K X-ray fluorescence ratio as a function of the glancing incidence angle showed that this ratio became independent of the glancing angle for angles less than 5. Glancing X-ray fluorescence measurements made at 2 were then calibrated to known standards to give highly accurate compositions for major elements in the Chinese bronze mirrors. X-Ray diffraction measurements were used to test for composition segregation or crystal structure changes near the surface of the mirrors. The X-ray diffraction angles were found to be independent of the glancing angle indicating that surface composition segregation was not observed. Changing relative intensities of the observed diffraction reflections indicated that the crystal texturing changed near the surface. INTRODUCTION X-ray fluorescence (XRF) has become a popular method to investigate archaeological and historical materials because it provides nondestructive elemental analysis measurements. An attractive feature of XRF is that it can successfully employ a number of spectrometers including newly utilized portable systems. In this study XRF has been used to study the elemental composition of 6 ancient Chinese mirrors dating from approximately 200 BCE 800 CE obtained courtesy of the Godwin-Ternbach Museum (Winter 2003) located at Queens College of the City University of New York. In certain cases X-ray diffraction has been used to also determine the crystal phasing of the mirrors. A key part of these studies is that the irradiation angle between the X-ray beam and the mirror surfaces, defined as, has been varied systematically to aid in calibrating the measured XRF elemental intensities to sample compositions. A key problem for XRF is that converting the measured elemental intensities to actual element compositions is not a simple problem and depends on many factors. A major problem for the study of bronzes is the determination of the tin content. Measurements of the Sn L kev, intensity are often used to determine the Sn content. One of the things shown in this study is that the measured Sn L /K intensity ratio is a strong function of the irradiation angle. Standard bronze samples with certified mass spectrograph compositions obtained from the Concast Corporation were used as calibration standards. It is obviously not possible to directly do mass spectrograph composition measurements on the Chinese mirrors since that would permanently mark the museum pieces.
This document was presented at the Denver X-ray Conference (DXC) on Applications of X-ray Analysis. Sponsored by the International Centre for Diffraction Data (ICDD). This document is provided by ICDD in cooperation with the authors and presenters of the DXC for the express purpose of educating the scientific community. All copyrights for the document are retained by ICDD. Usage is restricted for the purposes of education and scientific research. DXC Website www.dxcicdd.com ICDD Website - www.icdd.com
177 178 With this same diffractometer geometry it was also possible to do glancing incidence XRD to study the crystallographic properties of the mirror surfaces. These relatively new techniques have proven very useful for the study of growth and segregation effects in thin films. (van Brussel and De Hosson, 1994) In this variable incidence angle X-ray diffraction the angle between the substrate plane and the incident beam,, is held constant for a series of different 2 scans, with 2 defined as the angle between the incident beam and the diffracted beam. Such glancing incidence XRD studies have proven particularly useful for studying changes in crystallographic structures for metal film systems. (Cadieu 2010) EXPERIMENT The Chinese bronze mirrors and composition standards were analyzed in a PANalytical Materials Research Diffractometer. An Amptek, XR-100T, Peltier cooled energy dispersive X-ray detector was added to this system to facilitate XRF measurements. The samples were irradiated by the Cu X-ray tube operated at a beam voltage of 45 kv with a 20 µm thick Ni Cu K filter (99% Cu Kâ reduction). Beam currents ranging from 5 to 40 ma were utilized to limit the Amptek detector dead time to 15%. A programmable divergence slit, PDS, was used to set the irradiated sample length to 5 mm independent of the angle. XRD data was collected using a programmable receiving slit, PRS, with a Cu K monochrometer. With this setup the irradiation angle could be reliably set to a small fraction of 1 degree. The Amptek fluorescence data were collected using a multichannel analyzer with the ADC gain set to 1024 channels. Figure 1 shows the basic setup inside the diffractometer. The standard compositions for the three bronze samples determined by Concast Corporation by mass spectrographic are designated ST932, ST903, and ST907. Our measurements on these standards are designated by CT932, CT903, and CT907 where the dots stand for a measurement angle. Fig. 1. Setup inside X'Pert Pro Overhead View, is the incoming angle. RESULTS and DISCUSSION
178 179 The Chinese bronze Mirrors were loaned by the Godwin-Ternbach Museum at Queens College. The museum designations for the mirrors are, Mirror A (#66.45) is 8.9 cm in diameter (Fig. 2A), made around C. 206 BCE-220 CE; Mirror B (#63.18) is 11.4 cm in diameter (Fig. 2B), made around C. 618-906 CE; Mirror C (#63.19) is 8.9 cm in diameter (Fig. 2C), made around C. 206 BCE-220 CE; Mirror D (#63-17) is 10.9 cm in diameter (Fig. 2D), made around C. 206 BCE-220 CE: Mirror E (#66-46) is 10.6 cm in diameter (Fig. 2E), made around C. 206 BCE-220 CE: Mirror F (#66-47) is 10.7 cm in diameter (Fig. 2F), made around C. 206 BCE-220 CE. Composition measurements were made at several locations on the mirror side in regions most free of corrosion or discoloration. Glancing and non-glancing X-ray excitation angles were investigated. Figure 3 shows the ratios of Sn/Cu and Pb/Cu fluorescence count percentages at glancing angle 2 and a more normal excitation angle of 60. The chart shows that the intensity ratios measured at a glancing angle of 2 are clearly higher than those measured for 60. Furthermore,
179 180 Mirror B has a higher composition of lead which implies that the mirror was made in a later period as lead content was increased in the making after the Song dynasty. [Chou 2000] Figure 4 below shows lead Lâ/Lá ratios at a normal angle of 60 and a glancing angle of 2. The motivation is to show that the normal incidence and the glancing excitation angles give consistently different results. The lead Lâ/Lá ratios in Fig. 4 are clearly higher for 60 excitation than for 2. These differences in fluorescence ratios as a function of excitation angle illustrate that it not obvious how to translate the fluorescence counts to actual compositions. To solve this problem data was collected for a bronze standard, A932, which approximates the mirror compositions. Mirrors Normal and Glancing 8 7 6 Sn/Cu Pb/Cu 5 count ratio % 4 3 2 1 0 A1017G A1017N B1015G B1015N C1017G MC1017N Fig. 3. Fluorescence counts for Sn and Pb relative to Cu are shown for mirrors A, B, and C as measured at glancing angles of 2, G, and 60, N.
180 181 1 0.8 0.6 0.4 0.2 0 Fig. 4. Lead Lâ/Lá Ratios at = 60, N, and Glancing, =2, G (A, B, C are ratios for respective Chinese Mirrors, CT932 is a bronze standard, Pb is pure Lead) The Amptek amplifier gain was set to either collect X-rays up to 16 or 30 kev. With the gain set to collect up to 30 kev, fluorescence counts for both the Sn L as well as the Sn K X-rays were collected. Figure 5 shows the fluorescence spectrum collected for the Concast bronze standard CT932 for energies up to 30 kev. Figure 6 shows the collected ratios for Sn L /K versus the angle for the Concast bronze standard CT932. This was a smooth machined surface so that the angles could be precisely defined. The Sn L X-rays have a relatively low energy at 3.444 kev so that the observed Sn L intensity is a function of both the depth of sample irradiated and of the probability for such a low energy X-ray to escape from the sample and be counted. The Sn K X-rays have a relatively high energy at 25.193 kev and can consequently escape from a greater depth of the sample than the Sn L X-rays. The depth of sample irradiated decreases as the angle decreases. The Sn L /K intensity ratio is expected to increase with decreasing until the L X-rays have a greater probability of exiting than being absorbed within the sample. For this sample the Sn L /K intensity ratios become independent of for 5. The relative intensities of the Sn L and K X-rays illustrate the problems inherent in calibrating X-ray intensities at different energies to percent composition values. It is desired that the Sn concentration determined from the Sn L and Sn K X-ray intensities should be the same and it is also a non-obvious problem in determining compositions from X-rays at different energies for other elements. This result indicates that calibration comparisons should be made for angles that are less than 5. Compositions were determined from glancing measurements for = 2.
181 182 1000000 100000 Cu Pb Pb Counts for 1000 s 10000 1000 Sn Sn Sn 100 Zn 10 0 5 10 15 20 25 30 Energy (KeV) Fig. 5. Spectrum of Concast sample CT932 at low gain with 2 glancing angle. 1.3 Sn Intensity Ratio Lá/Ká 1.2 1.1 1 0.9 0.8 0.7 1 10 100 Angle Omega degrees Fig. 6. The intensity ratio for the Sn L /K X-rays versus irradiation angle is shown for the Concast Bronze CT932 sample.
182 183 A key result of Fig. 6 is that the relative XRF counts for spectrum collected at = 2 is independent of for angles near 2. XRF spectrum collected at = 2 for the Concast sample CT932, which closely matched the element content and compositions for the Chinese mirrors, were ascribed to wt.% for the elements of interest by the following formula. ܭ 100% = %.ݐݓ ݏݐܥ σ ܭ ݏݐܥ In this expression Cts i is the peak counts above an assumed flat background in the vicinity of the peak. Background corrections were normally small here but the same expressions could also be used in cases where the background is non-negligible. A sum over 5 channels centered on the peak was used for the i th element as measured at the angle = 2, K i is a constant that makes the wt.% composition for the i th element in the Concast standard CT932 match for the elements Sn, Fe, Ni, Cu, Zn, and Pb in the Concast composition with the constraint K Cu =1.00. The remaining K i for the elements Sn, Fe, Ni, Zn, and Pb are determined by a what if iterative procedure in Microsoft Excel. The peaks used for elements were the Sn L, Fe K, Ni K, Cu K, Zn K, and Pb L. The wt.% composition for the mirrors are determined by using the K i values applied to XRF spectrum collected at = 2. This expression can be checked by determined composition values for the other Concast standard known compositions. Calibration for the relatively weak Zn K intensity compared to the Cu K peak was made by subtracting a scaled Cu Kâ peak for Concast standard CT907 that was certified to contain only 0.047 wt.% Zn. With the set of coefficients determined the measured Zn content for the Concast standard CT903 agreed to within 0.1 wt.% and for Concast standard CT907 within 0.04 wt.%. The average Zn wt.% for the six mirrors was 0.32 ±0.72 wt.% for a 2 tolerance. This value certainly straddles zero. Zn is expected to be the element with the largest tolerances. The compositions of the mirrors are indicted graphically in Fig. 7 and then tabulated in Table 1. Coefficients were set by ST932. Also shown in Table 1 are the deduced compositions for the two standards ST903 and ST907 furnished by Concast and their mass spectrograph measured compositions performed by Concast. For both of these other standards the determined and Concast certified compositions for Sn and Pb agree within 0.4 wt.% for each standard. The determined Cu content for the Concast standards CT903 and CT907 agree within 0.2 wt.% with measured Concast values. Mirror B that was archived on the basis of design to be from C. 800 CE was found to have much lower Sn composition than the 5 other mirrors, but much greater Pb concentration.
183 184 Wt.% 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 AG2 BG2 CG2 DG2 EG2 FG2 Sn Cu Zn Pb Fig. 7. Mirrors A to F major metal content as calibrated from glancing angle at 2 are indicated. Table 1. Wt.% compositions for the mirrors A to F are indicated plus measured and known compositions for Concast certified standards ST932, ST903, and ST907. wt.% ST932 CT932G2 AG2 BG2 CG2 DG2 EG2 FG2 CT903G2 ST903 CT907G2 ST907 Sb 0.032 0.03 0.2 0.1 0.1 0.1 0.1 0.1 0.0 0.0 0.1 0.0 Sn 6.32 6.32 36.7 7.3 24.2 24.1 29.2 21.7 8.2 7.8 11.0 10.7 Fe 0.107 0.11 0.0 0.1 0.1 0.1 0.1 0.4 0.1 0.0 0.1 0.0 Ni 0.373 0.37 0.2 0.2 0.3 0.3 0.3 0.2 0.4 0.0 0.3 0.0 Cu 81.69 81.97 56.5 66.4 67.0 66.5 66.6 57.4 87.4 87.6 89.0 89.2 Zn 4.23 4.23-0.02 0.48 0.87 0.49-0.05 0.15 4.39 4.29 0.01 0.05 Pb 6.97 6.97 6.4 25.3 7.4 8.4 3.8 20.0-0.4 0.1-0.5 0.0 Sum = 99.7 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 99.8 100.0 100.0 Since at glancing angles a thinner surface region of the mirrors is sampled, it may be possible that the surface region is different from the interior because of surface segregation of one or more components. Since this is an alloy system it is expected than changes in composition should result in expanded or contracted lattice parameters. Figure 8 shows X-ray diffraction traces for different glancing angles for mirror D and a trace for the sample CT932. The fact that the diffraction angle remains constant for mirror D as a function of the glancing angle means that this mirror does not exhibit surface composition segregation. This figure also shows that the lattice parameter for mirror D is shifted from that of the standard CT932 because of alloying effects.
Copyright JCPDS-International Centre for Diffraction Data 2011 ISSN 1097-0002 184 185 1400 DG10 1200 DG2 x 10 1000 DG22 800 CT932G 30 600 400 200 0 41 42 43 44 2-theta for different glancing angles Fig. 8. X-Ray diffraction data for the main bronze diffraction reflection for different glancing angles for mirror D and the bronze standard CT932 are shown. Figure 9 shows that the lattice parameters for mirrors D, E, F are different and different from that of the standard. In each case the position of the diffraction peak does not shift with changing glancing angle values which shows that surface composition segregation is not observed. The bronzes as well as Cu exhibit a cubic structure with the major peak corresponding to the (111) reflection occurring at about 43 2-theta and the next most intense peak, the (200) at about 50 2-theta. Figure 10 shows the ratio of the (200) intensity to the (111) intensity as a function of glancing angle for mirror D. This figure shows that the interior as sampled at large glancing angles exhibits a diffraction pattern similar to that for a random oriented sample, but that the surface region as sampled at small glancing angles exhibits a predominant (111) texture. Note that as shown above that this texture change is not accompanied by a surface composition change.
Copyright JCPDS-International Centre for Diffraction Data 2011 ISSN 1097-0002 185 186 2000 1800 1600 Counts 1400 CT932G30 1200 FG22 1000 DG22 800 EG22 600 400 200 0 41 42 43 44 2-theta degres Fig. 9. X-Ray diffraction data for the mirrors D, E, F, and the bronze standard CT932 are shown. Int. Ratio (200)/(111) 0.3 0.25 Mirror D 0.2 0.15 0.1 0.05 0 0 5 10 15 20 25 Glancing Angle Omega degrees Fig. 10. The intensity ratios of the bronze (200) to (111) reflections for mirror D are shown versus the glancing angle.
186 187 CONCLUSIONS It was shown that the relative XRF intensity of the Sn La/Ka was independent of the glancing excitation angle for near 2. The XRF intensities were then calibrated by a set of coefficients to give agreement with the composition of mass spectrographically analyzed known standard composition. The coefficients could then be used to determine the major element composition for XRF excitation at a glancing angle of 2 for the six Chinese bronze mirrors and for two other bronze standards to a high precision of 0.3 wt.% for Sn and Pb. Even though a Cu X-ray tube was used in the analysis, a bronze sample with a similar Cu content to that in the mirrors was used for the composition calibration. Any spurious effects that might affect the Cu peak such as Rayleigh scatter are expected to affect the concentration standard and mirror samples equally and thus cancel out. XRD results showed that the diffraction angle for the principal (111) reflection was independent of the glancing angle w. This result shows that composition segregation was not a problem and gave validity to the compositions determined at a glancing angle of 2. The relative intensity of the XRD (200)/(111) bronze reflections did change as a function of implying that the outer surface regions were (111) oriented crystallites. The XRF composition analysis showed that mirror B that the museum had ascribed to the about 800 CE time period was substantially different in Sn and Pb concentration than the other five mirrors. REFERENCES * Author for Correspondence: F. J. Cadieu, cadieu@qc.edu Cadieu, F. J., Vander, I. Rong, Y. and Zuneska, R. W. (2010) Glancing XRD and XRF for the Study of Texture Development in SmCo Based Films Sputtered Onto Silicon Substrates, Denver X-Ray Conference 2010, Paper F-11. Chou, J. (2000). Circles of Reflection: The Carter Collection of Chinese Bronze Mirrors, Cleveland, The Cleveland Museum of Art. van Brussel, B. A., and De Hosson, J. Th. (1994), Glancing Angle X-Ray Diffraction: A Different Approach, Appl. Phys. Lett. 64, 1585. Winter, Amy H. (2003), The Light of Infinite Wisdom, Asian Art From the Godwin- Ternbach Museum, NYC, The Godwin-Ternbach Museum, Queens College of CUNY.