Time of flight neutron diffraction

Two sets of copper tinned tokens were analysed using time-of-flight neutron diffraction (Error! Reference source not found.): (a) as-tinned and annealed copper tokens tinned by the wiping method which were examined by BSE and XRD reported above, and (b) a series of as-tinned and annealed tokens tinned by the dipping method which produced thicker metal coatings (section 3.7). As neutron diffraction is a bulk analyses method, because information derives from the bulk volume rather than the surface of a sample in the beam path, there was a concern that the concentration volume of diffracting IMCs in the wipe-tinned tokens may be too small for the method to detect. The double-sided and increased metal coating thickness in dip-tinned tokens offered an opportunity to analyse sample volumes with higher IMC concentrations to assess the efficacy of the method. Rietveld analysis of the diffraction data could provide quantitative information of IMC concentrations to aid this task and preliminary results are reported below. The incident neutron beam size was in the range of 15x25mm;

experimental details are in section 3.8.2.1.

The ROTAX and GEM diffractometers used have the ability to investigate crystal preferred

the IMC phases in the tokens which could influence the quality of the interpretation of XRD results overall. This is achieved by comparing the intensity of specific diffraction lines in data collected from different detector banks because the detectors are positioned at different fixed scattering angles with regards to the orientation of the diffracting volume in an object (Figure 3.5). Examination of neutron diffraction patterns of S15 shows that the ε-Cu3Sn diffraction line at 2.39Å is present in GEM-bank 5 but is absent in data collected from GEM-bank 4, which is evidence of crystallite preferred orientation in this phase (Figure 4.20). XRD data from this sample showed a dramatic change in the intensity ratios of ε-Cu₃Sn diffraction lines, which validates the XRD analyses. A simple comparison of diffraction patterns does not constitute a full texture analysis in terms of pole figure determination which can be done with GEM data, but offers a good indication of the degree of texture in this phase.

Figure 4.20: Neutron diffraction patterns of S15 collected by GEM detector banks 4 (top) and 5 (bottom), which are positioned at different fixed 2θ angles. The 2.39Å is present in diffraction data from bank 5 but is absent in bank 4, indicating crystallite preferred orientation. The line represents reference data for ε-CuSn.

Neutron diffraction patterns of analysed tokens are recorded in Figures 4.23 to 4.28 and details of annealing conditions are in (Table 3.8). The normalised intensity of the diffraction lines from IMCs is much smaller (1-2%) compared to Cu reflections deriving from the substrate, and compared to the XRD patterns of the same tokens. This is because neutrons scattered from the tokens travelled a way through the copper substrate before detection. The low intensity of IMCs influences qualitative interpretation of the diffraction patterns.

Diffraction patterns of S1 and S5 collected from ROTAX-bank 3 show mainly reflections from Sn (Figure 4.21). The indicative η-Cu6Sn5 peak at 2.96Å (PDF65-2303) is absent or hindered by the relatively high noise at this end of the pattern. A small peak may be present at 1.71Å but the evidence is not definitive. Data from the same tokens and S9 from GEM-bank 5 (Figure 4.22) equally show the presence of Sn. Because data from GEM bank 5 are recorded up to 2.5Å, the major η-Cu₆Sn₅ peak at 2.96Å is absent and η-Cu₆Sn₅ cannot be confirmed on these tokens._. Figure 4.23 shows data of S7 and S11 collected at GEM-bank 5. S7 includes clear reflections from ε-Cu3Sn and a small amount of Sn. The intensity ε-Cu3Sn peaks in S7 is higher than in S11. The presence of η-Cu6Sn5 could perhaps be indicated by peaks at 1.72 and 1.98Å including those annotated in Figure 4.23. Reflections from ε-Cu₃Sn are also visible in S15, which includes reflections from δ-Cu₄₁Sn₁₁ (Figure 4.24). δ-Cu₄₁Sn₁₁ is the dominant IMC phase in S19 and includes some α solid solution phase (Figure 4.24), as revealed previously by XRD. The lattice parameter of this α-bronze is 3.69Å, which corresponds to a concentration of approximately 13wt% Sn as estimated by comparison to laboratory α-bronze standards.

Figure 4.25 and Figure 4.26 show diffraction patterns of dip-tinned copper samples compared with wipe-tinned copper tokens reported above. The increased thickness of dip-tinned samples resulted in higher intensity reflections from Sn, ε-Cu3Sn and δ-Cu41Sn11 making easier assignment of phases to individual major reflections. Assigned peaks are annotated on the diffraction patterns (Figure 4.25, Figure 4.26). Despite the higher intensity of phases in the diffraction patterns of the dip-tinned tokens, which resulted in unambiguous identification of Sn, ε-Cu₃Sn, δ-Cu₄₀Sn₁₁ and α-bronze, the lack of evidence for the presence of η-Cu6Sn₅ on samples which are previously shown to have a substantial η-Cu6Sn5 content is surprising.

Figure 4.21: Diffraction patterns of S1, S4 and S5 of data collected at ROTAX bank 3.

Figure 4.22: Diffraction patterns of S2, S5, S9 of data collected from GEM bank 5.

Figure 4.23: Diffraction patterns of S7 and S11 of data collected at GEM-bank 5.

Figure 4.24: Diffraction patterns of S15 and S19 of data collected from GEM-bank 5.

Figure 4.25: Diffraction patterns of S12, S13, S14 and S15 of data collected from ROTAX-bank 3.

Figure 4.26: Diffraction patterns of S17, S18, S19 and S20 of data collected from ROTAX-bank 3.

Data were scrutinised using GSAS Rietveld phase composition analysis to validate qualitative interpretation of the diffraction patterns, investigate the presence of η-Cu6Sn5 and provide an indication of the relative concentration of phases. This was performed by averaging data collected from several banks of GEM and from ROTAX bank 3 and using published crystal structure models (η-Cu6Sn5 ICSD 158248, ε-Cu3Sn ICSD 1-3102, PDF65-5721, δ-Cu40Sn11

CRYSTMET AL4034, ICSD 800). Data collected at GEM are more reliable due to the large number of detectors surrounding the sample considerably reducing texture effects and providing lower detection limits due to its higher resolution and count rates. Texture effects were taken into account during GSAS analyses of data from ROTAX-bank 3. Results are in Table 4.5.

Sample S1 S3 S5 S7 S9 S11 S13 S15 S17 S19

As tinned As tinned 250°C 250°C 350°C 350°C 450°C 450°C 550°C 550°C

Phase 5 min 60 min 5 min 60 min 5 min 60 min 5 min 60 min

As tinned As tinned 250°C 250°C 350°C 350°C 450°C 450°C 550°C 550°C

Phase 5 min 60 min 5 min 60 min 5 min 60 min 5 min 60 min

As tinned As tinned 250°C 250°C 350°C 350°C 450°C 450°C 550°C 550°C

Phase 5 min 60 min 5 min 60 min 5 min 60 min 5 min 60 min

Table 4.5: Semi-quantitative wt% phase composition analyses of tokens analysed at ROTAX-bank 3 and quantitative wt% phase composition analysis of tokens analysed at GEM. The remaining wt%

composition is copper and trace amounts of oxides. The incident neutron beam size was c. 15x25mm and the thickness of the copper tokens 3.25mm. α-bronze observed on S19 is of c. 13wt% Sn.

GSAS analysis results show that both diffractometers have some ability to detect ε-Cu₃Sn to about 0.5 wt%, δ at c. 1 wt% and α-bronze at c. 0.5wt% fraction of the diffracting volume in tokens which appear to be at the limit of detection of these phases at GEM. Results show that ε-Cu3Sn is present in S19, which was not definitive in XRD data due to the complexity of the XRD pattern due to the complex nature of the X-ray diffraction pattern. GSAS refinement did

neutron diffraction patterns. Its absence from the neutron diffraction patterns is surprising.

This shows that the different IMC phases have very different volume detection limits for neutron diffraction and there is no obvious reason why this should be the case (Kockelmann, pers. com.). Understanding why η-Cu6Sn5 was not detected by neutron diffraction, despite being present in substantial quantities in several samples, requires further investigation and is beyond the scope of this thesis.

In summary, as η-Cu6Sn5 is the major phase on as-tinned copper and bronze tokens neutron diffraction method cannot effectively be used to identify tinning on archaeological bronzes.

This inability of neutron diffraction to detect η-Cu6Sn5 must be contrasted with its increased sensitivity to detect small concentrations of ε-Cu₃Sn, δ and α-bronze by the high resolution diffractometer GEM, which may be present on archaeological tinned bronzes exposed to heat, depending on the thickness of the metal coating and its volume fraction on an object.

Considering these limitations, neutron diffraction can be a useful non-destructive diffraction method for analysis of archaeological tinned bronzes. XRD shows advantage over neutron diffraction in this context of application.