II.5. Metal Analysis

Elemental analysis determines the proportion of different chemical elements present in a particular sculpture’s casting alloy. The elemental analysis may be qualitative or quantitative. The former simply lists the elements detected, with only a rough idea of their relative amounts; the latter gives a measurement of the amount of each element present as a percentage of the total, usually reported as a weight percent (wt%). Quantitative analysis should also provide an estimate of the uncertainty associated with the measurement (the error).¹

Some researchers also use the term “semi-quantitative,” often in circumstances where the error of the measurement is relatively large, or unknown. In such circumstances, the numerical measurements are unlikely to be reproducible by other analysts and are thus not useful for comparison with the results of other laboratories.

Any quantitative analytical method can be characterized in terms of accuracy, precision, and sensitivity. In scientific terms, an accurate method is one where the average of many measurements is likely to be very close to the true value. This is sometimes called trueness. An analytical method is considered precise if the results of many repeated measurements are likely to be very nearly the same. Given these definitions, an analytical method can be accurate, precise, neither, or both. A common way of visualizing these four possibilities is shown in figure 413. The term “sensitivity” can be used in different ways in reference to chemical analysis, and there have been vigorous debates on the subject. Here, the term is used in its common sense, as “a device’s ability to detect a small amount or slight change in the measured quantity.”²

Open viewer Figure 413

A basic alloy-type determination should be standard practice for the technical study of bronze sculpture for purposes of documentation and typology. This may be achieved with qualitative analysis. Quantitative analysis provides additional important information that may help to refine the possible date of production or confirm that a sculpture’s alloy is consistent with securely identified examples from a certain time period, region, workshop, or even artist (see I.2§4.1).

In the end, the information that may be derived from elemental analysis of any given sculpture depends to a large degree on the quantity and quality of reference data available from comparable, well-documented examples. Conducting new and rigorous analysis of securely attributed works is of great importance to expand the body of knowledge usable for comparative purposes. Some strides have been made in the development of large, shared databases of copper alloy compositional data,³ which facilitate sophisticated statistical analysis that will certainly lead to new discoveries with regard to workshop, regional, and temporal characterization of bronze sculpture.

Different methods of elemental analysis may be invasive, noninvasive, or minimally invasive. The terms “destructive” and “nondestructive” are also commonly used, although in analytical sciences, “destructive” refers specifically to the sample, not to the item: the analysis is destructive if the sample is destroyed after analysis, nondestructive if not. Each type of analysis has its advantages and disadvantages.

X-ray fluorescence analysis (XRF) is a relatively accessible analytical method employed to characterize elemental composition. It is nondestructive and offers rapid, multi-element analysis with good sensitivity, often down to a few hundredths of a percent in copper alloy samples (equivalent to a few hundreds of parts per million [ppm]). A beam of X-rays is directed at the sample material to stimulate the emission of fluorescent X-rays with specific energies characteristic of the elements present.

The emitted X-rays are detected and converted into a spectrum, either with an energy dispersive (ED-XRF) or a wavelength dispersive (WD-XRF) detector. ED-XRF is far more common than WD-XRF and is used in all of the portable and handheld XRF instruments that are now widely available in museum and university laboratories. ED-XRF offers the possibility of giving a quick qualitative assessment of the type of alloy present. Newer silicon drift detectors (SDDs) take measurements in a minute or less, making it possible to take many measurements in the course of an examination.

To optimize results for historic copper alloys, voltage on the spectrometer should be set relatively high (around 40–50 kV). In most cases, moderate to heavy filtration of the beam is advantageous, using aluminum, titanium, and/or copper filters. In the case of modern sculptures, if light elements such as silicon, phosphorous, or aluminum are to be analyzed, filters should be removed, the voltage turned down to about 15kV, and a helium flush or vacuum should be used if available. These operating parameters also allow for the analysis of sulfur in both historic and modern bronzes, which may be very informative.

Due to the complex interactions between X-rays and the sample material, quantitative alloy analysis with ED-XRF requires complex calibration procedures. Many ED-XRF instrument manufacturers provide software and calibrations for metals that will generate quantitative results. Unfortunately, the use of manufacturers’ proprietary calibrations comes with some difficulties: the results from different manufacturers (and even different models from the same manufacturer) are often not in agreement; it can be difficult or impossible for the user to select the list of elements to be quantified; error estimates are often unrealistically small; and the spectrum processing and calibration methods used are often not transparent to the user.⁴ These and other factors make it difficult to share quantitative results confidently between researchers, and manufacturers’ results should generally be considered semi-quantitative. Some protocols designed to enhance reproducibility and facilitate sharing have been developed that may be useful to researchers wishing to share results and build collaborative databases.⁵

Elemental mapping may be done using XRF. Dedicated XRF scanners are available that are capable of scanning flat (or nearly flat) areas of a square meter or more at high resolution. For three-dimensional objects, smaller areas are scanned (typically a few square centimeters) with many individual analyses of the elemental composition at definite points, together defining an elemental map (fig. 414).⁶ The objective may be to detect and characterize inlays and overlays; brazing, welding, or soldering; and/or repairs. Mapping requires specialized equipment and software.

XRF is essentially a surface analysis technique, with an effective penetration in copper alloys of around 30–80 µm, or less than a tenth of a millimeter. This means that the technique is very sensitive to surface alteration or contamination. For qualitative use on historic materials, this is usually not a significant problem. But for quantitative analysis, surfaces to be measured should be clean, flat, and free from patina or corrosion.⁷ It is possible to encounter both surface enhancement and surface depletion of selected elements that may result in misleading results, particularly with archaeological materials. Working with an XRF analyst who has significant experience with copper alloys can help produce reliable and useful results for comparative purposes.

In 2023, portable ED-XRF instruments cost between US$25,000 and $50,000. Commercial laboratories may be able to provide analysis for US$50 or less per sample.

WD-XRF offers significantly higher spectral resolution and sensitivity than ED-XRF, but is less common and much more expensive to own and operate. Samples for WD-XRF analysis are typically extracted from the sculpture, embedded, cut, and polished prior to analysis. Sample size requirements may vary, but are often from 1 × 1 mm to 5 × 5 mm. These instruments are found primarily in commercial and research laboratories focusing on earth sciences. In theory, WD-XRF instruments offer the possibility of more precise quantitative analysis of alloy composition than ED-XRF instruments, although many earth sciences labs may not routinely analyze copper alloys, and therefore may not have calibrations prepared in advance. Costs per sample are normally less than US$100.

Open viewer Figure 414

Particle-induced X-ray emission (PIXE) commonly utilizes a highly collimated beam of protons to excite X-ray emission from a sample. Other charged particles may be used as well, but protons are by far the most common. It is possible but not necessary to place the sample in a vacuum chamber or flush it with helium to improve detection of light elements. X-ray spectra are collected, usually by an ED detector, for processing into a quantitative elemental composition. PIXE offers better sensitivity than XRF⁸ and can detect very small quantities of many materials typically present in casting alloys, often as little as a few tenths of ppm. Moreover, PIXE offers better spatial resolution (a few tenths of microns) thus enabling the analysis of very small features such as inlays (fig. 350). In the case of point analysis, no specific preparation is required, though as with XRF, PIXE analysis is a surface technique, so the sample site should be clean, flat, and free of patina or corrosion. The depth of penetration of PIXE using protons is somewhat less than for XRF. Unfortunately, few facilities provide access to ion beam analysis techniques.

As for XRF, chemical mapping can be carried out with PIXE.⁹

Open viewer Figure 350

Atomic spectroscopy refers to a group of analytical techniques where sample material is introduced into a flame, furnace, electric arc, or plasma and breaks down into individual atoms and ions. In this excited state, the atoms emit and absorb specific frequencies of light (from near-infrared to ultraviolet, including the visible light range) that are characteristic of the specific elements present. The elemental composition of the sample is thus determined by analyzing the spectral features of the emitted light (atomic or optical emission spectroscopy, or AES/OES) or of light transmitted through the excited sample material (atomic absorption spectroscopy, or AAS).

Atomic spectroscopy was at the very origin of the rise of archaeometallurgy as a discipline.¹⁴ In particular, flame AAS and inductively coupled plasma AES (ICP-AES) have provided the large majority of published analysis on bronze sculpture so far.¹⁵ ICP refers to a highly controlled and reproducible method for the atomization/ionization of sample material using plasma made from an argon stream.

AAS and AES are sensitive to trace quantities of elemental components, from several hundreds of ppm to the ppm level in a copper matrix depending on the element. This allows precise quantification of a large number of trace elements, which may be useful, notably when addressing provenance questions. Preparation of the sample for analysis involves complex chemical digestion procedures to dissolve all components of the sample. The complexity of atomic spectroscopy techniques means that analysis is best performed by private or university laboratories specializing in this analysis, and with staff experienced in cultural heritage applications. Small changes to the analysis protocol can make significant differences in the quality of the outcome.

Analysis by atomic spectroscopy techniques typically requires a sample of about 10–50 mg. At the time of this publication, costs per sample range from around US$100 to $350 for academic clients.

In recent years, ICP mass spectrometry (ICP-MS) has largely superseded ICP-OES as the method of choice for the analysis of trace elements in bronze alloys. As with ICP-AES, the sample is atomized and ionized using an inductively coupled plasma. But instead of characterizing the light emitted during excitation, ICP-MS relies on direct detection of the ions produced in the plasma stream. These are sorted and counted by the spectrometer according to their mass-to-charge ratios. ICP-MS are complex instruments and are made in several different configurations, but all measure the abundance of each ion type with a high degree of sensitivity, leading to limits of detection typically 10–1000 times lower than ICP-AES (ppm to ppb), depending on the element and the instrument configuration.¹⁶ ICP-MS can also detect different isotopes of the same element, allowing for isotopic analyses (see below).

ICP-MS systems are significantly more expensive to purchase and operate than ICP-OES systems. As for atomic spectroscopy techniques, ICP-MS analysis is best conducted in specialized laboratories dedicated to the technique and with staff experienced in the analysis of cultural heritage materials.

ICP-MS has had a disadvantage in comparison to ICP-AES, which is a difficulty in analyzing, with one single sample, a wide range of element concentrations (from sub-ppm to percent levels). When optimized for trace elements, major elements (tin, lead, and zinc) have often been measured with relative precision equal to or lower than by ED-XRF, and copper has rarely been quantified at all. Significant improvements in instrument design since the 2010s have largely mitigated this problem.

ICP-MS is sometimes used in conjunction with laser ablation (LA), a technique that uses a focused, high-intensity, pulsed laser to vaporize tiny spots of sample material that are then drawn into the ICP-MS instrument. LA-ICP-MS is generally considered a micro-destructive technique because the ablation spots are extremely small, but it is generally applied to fragments or samples that are small enough to be placed onto an instrument’s analysis stage, often only several inches across. Laser spot sizes often fall in the range of 10–300 µm, so multiple sites should be analyzed within the sampling area to ensure that micro-inhomogeneities in the metal structure do not yield misleading results.

Various so-called matrix effects can complicate quantitative analyses produced via LA-ICP-MS. While the technique has been used extensively for the quantitative analysis of glass, it has only been applied relatively recently to quantitative analysis of copper alloys and is not yet widely considered an established technique. As with LIBS, its ability to repetitively ablate the material may give access to sound metal below the intentional patina and/or corrosion layers.

Neutron activation analysis (NAA) is based on the principle that if a sample material is bombarded with relatively low-energy “thermal” neutrons (usually produced at a nuclear reactor), the atoms in the sample will capture neutrons into their nuclei, forming radioactive isotopes. As these isotopes decay over a period of days or weeks, they emit gamma rays characteristic of each element present in the sample. The gamma rays may be detected and identified, leading to a quantitative estimate of the sample’s total elemental composition. The high penetrating power of neutrons and gamma rays means that NAA provides information from the interior as well as the exterior of the sample material. Neutron activation is a highly sensitive method for chemical analysis. Data for a wide range of elements can be obtained with sensitivity at the low- and even sub-ppm level. No specific preparation of the sample is needed.

NAA may be used as a noninvasive technique if the entire object to be studied is analyzed, though often samples are taken in order to reduce cost and transportation issues for valuable sculptures. A drilled sample also allows for removal of surface layers and corrosion product prior to analysis. Reliable results require about 50–100 mg of sample material. Sensitivity for some elements is less than for ICP-MS, and the analysis requires more time to complete than other techniques (since the rate of decay is variable, gamma-ray collection often happens over a period of days).

The restricted availability and cost of this nuclear method make it increasingly difficult to access, and so it has been largely supplanted by ICP-MS.

In scanning electron microscopy (SEM), a small sample is placed into a vacuum or near-vacuum and bombarded with a highly focused beam of electrons that is continuously rastered, or scanned, across the surface. Multiple detectors are placed inside the chamber that sense different types of emissions from the surface of the sample as the beam is scanning.

For elemental analysis, SEM may be coupled with X-ray detectors for energy-dispersive spectroscopy (SEM-EDS or SEM-EDX) or wavelength-dispersive spectroscopy (SEM-WDS or microprobe). These detectors yield X-ray spectra that are useful for elemental analysis.¹⁷ Areas can be scanned to obtain bulk composition measurements, or spot analysis can be made of individual metallic phases or inclusions. Both techniques are capable of quantitative elemental analysis, although WDS offers substantially higher precision and sensitivity of measurement than EDS.

Samples for quantitative elemental SEM analysis of copper alloys should be at least 1 × 1 mm, but preferably 5 × 5 mm or more to take into account natural inhomogeneities in copper alloys, particularly inhomogeneity of lead (which tends to form globules in copper) and elements associated with it (such as arsenic, silver, and bismuth). Samples are normally embedded in resin, cut, and polished to provide a clean, flat surface for analysis.

Today, many major museum and academic laboratories are equipped with SEM-EDS. These instruments may be valuable for metallographic analysis (see below), but for high-precision, high-sensitivity quantitative elemental analysis, other (unfortunately less accessible) methods such as ICP-AES, ICP-MS, PIXE, SEM-WDS, or NAA (see II.5§3.3) may be preferable.

Sampling for elemental alloy analysis (atomic spectroscopies and NAA techniques) is most often done by drilling (fig. 416). Naturally, this is usually carried out in hidden areas of the base or in the interior of the sculpture (fig. 156). A target sample weight of 50 mg of copper alloy represents approximately 5–6 cubic millimeters of metal. If a 1.5 mm–diameter drill bit (steel or tungsten carbide preferred is used, this corresponds to a drilling depth of about 3 mm. If greater depth of drilling is possible, the size of the drill may be reduced correspondingly; for instance, a 1 mm drill bit will provide enough sample if drilled to a depth of about 7 mm, while a 0.6 mm drill bit would require a drilling depth of nearly 20 mm. In practice, this last situation would almost certainly mean drilling multiple holes. It should also be noted that smaller-diameter drills are more likely to break while drilling and, particularly if a porosity is encountered, recovery of the broken end may be difficult or impossible.

When sampling bronze sculpture, drilling is often done on an edge, such that the drill penetrates parallel to the metal wall, avoiding surface effects (see II.5§2.1). For thin-walled sculptures, the thickness of the metal wall may constrain the diameter of drill that can be used; the drill should not be larger than half the wall thickness. Normally the first 0.5 mm of sample material is discarded (drilling depth adjusted accordingly) in order to remove any possibility of surface enhancement or depletion effects. Carefully clean or replace the drill bit between uses.

Preparation of sampled material for analysis is a particularly important step that is usually carried out by the laboratory in charge of the analysis; errors may result if it is not well carried out. Once collected, drillings should be carefully examined under a microscope to remove unwanted materials (corrosion, fibers from clothes, fragments of the drill, et cetera). A magnet may prove useful to remove steel fragments from the drill. For atomic spectroscopies and ICP-MS where the sample will be digested in acid, precise weighing is necessary, and digestion solutions specific to historic copper alloys should be used.¹⁸

Sampling for SEM analysis usually follows protocols used for metallography (see II.5§6.2).

Open viewer Figure 416

Open viewer Figure 156

Atoms are made of a nucleus around which a “cloud” of electrons spins. The nucleus is made of two main types of particles, protons and neutrons, and the number of protons defines the element. Copper, for example, has 29 protons, tin has 50, and lead has 82. The number of neutrons in the nucleus of any given element can vary within a certain range, and this number of neutrons determines the isotope. Isotopes are designated according to the combined number of protons and neutrons in the core (the atomic weight). So, for example, a lead atom (Pb) with 82 protons and 122 neutrons is designated ²⁰⁴Pb. Some isotopes are stable and do not change over time. Others are not stable (these are radioactive) and will decay at a predictable rate into other elements by loss of nuclear particles and energy. Carbon-14, or ¹⁴C, is a well-known example of an unstable isotope that is used for radiocarbon dating purposes (see II.8§2).

Isotope analysis for bronze sculpture focuses on the quantification of different stable isotopes.²⁰ The total amount of any given isotope in a copper alloy may be extremely small. For instance, in unalloyed copper, an important lead isotope, ²⁰⁴Pb (see II.5§5.2), may be present in quantities around 0.0001 wt% (1 ppm). A sensitive technique is thus needed, typically mass spectrometry. In addition, the difference in isotope ratios between two copper deposits is often very small, thus requiring mass measurements with very high precision (around 0.1%) that only certain types of advanced mass spectrometers can achieve. Typically, so-called multicollectors (MC) are used and the technique is referred to as MC-ICP-MS.

Lead, a well-studied isotope series, has four stable isotopes as well as scores of unstable isotopes. The stable isotopes, however, make up nearly all the lead on Earth and are ²⁰⁴Pb, ²⁰⁶Pb, ²⁰⁷Pb, ²⁰⁸Pb. Depending on the age, geological location, and nature of the ore from which it is derived, lead metal contains different proportions of these different isotopes. Small amounts of lead are systematically present in copper as a natural impurity, so lead isotope studies on copper artifacts are often designed to study the geographic origin of the copper. Conventionally, the ratios ²⁰⁶Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb, and ²⁰⁸Pb/²⁰⁴Pb are used for describing results.

Few isotope analyses have thus far been carried out on bronze sculpture. The main applications have been lead isotope studies on ancient copper alloy artifacts—mainly protohistoric collections of weapons, tools, and/or everyday items.²¹ Tin isotopes have recently become a focus of greater study in the context of bronze artifacts, while zinc and copper isotopes have potential applications, but have not been widely used.²² In the long run, the study of several isotope series in combination will probably be more informative than an examination of lead alone.

Using MC-ICP-MS analysis, very small sample sizes are needed, often less than 10 mg. Samples may be obtained by drilling as described in II.5§3.5.

Isotopic analysis for copper alloys is mostly carried out in earth science laboratories. In 2023, the cost of one analysis ranged from US$100 to $300.

The provenancing of copper based on lead stable isotope analysis alone is far from straightforward.²³ Isotope analysis is probably best used as a complementary approach together with elemental analysis, archaeology, and/or history.

Another major issue in applying isotope analysis to bronze sculpture is that lead is commonly added to the copper by founders or even by the metal supplier, be it through direct addition or recycling. In this case, the lead isotope signature will reflect not the provenance of the copper but rather that of the lead. In bronze sculpture, it may not always be evident whether low levels of lead are naturally occurring or added. Rather than focusing on geographic provenancing of metal, isotope analysis may be better suited to authentication and attribution, by characterizing expected isotope ratios for different regions and time periods. This will only be possible if large numbers of well-provenanced sculptures are studied in the future.

Metallography is the study of metal microstructure using microscopic imaging techniques.²⁴ Microstructure, its interpretation, and its importance for the study of bronze sculpture is discussed in I.2§3. The reader may consult that section to learn about the types of information to be gleaned from metallographic study and evaluate whether such study is worth undertaking.

In the field of cultural heritage, metallographic analysis is often confused with elemental metal analysis, but the two are quite different. In addition to the study of the metallurgical state of the metal, metallography might determine the different phases and their distributions. When two or more metals are mixed together to form an alloy—for example, copper and tin—different phases may form depending on the relative proportions of the different ingredients and the cooling temperature. Phase diagrams are useful to infer what phases may form and the nature and variety of the properties of the alloys, including melting temperatures and freezing ranges (fig. 132).²⁵

A metallography study involves two main steps: sampling and preparation, followed by observation and analysis.

Open viewer Figure 132

In situ characterization of microstructure without sampling is rarely relevant, for several reasons. First, the microstructure of the bronze’s surface is rarely what requires observation. A cross section of the metal wall is often much more informative, as it may show joints, repairs, et cetera. Also, the original surface may be severely altered by wear, use, later interventions, and/or corrosion. Second, a proper polishing is difficult to achieve in situ. (Several attempts have been carried out to develop pre-polishing and diamond or alumina polishing using rotary abrasive heads.) Third, in situ observation is very limiting in terms of magnification and image quality. Fourth, the surface to be prepared and etched must be large enough to encompass the heterogeneity of the microstructure (at least 1 × 1 cm).

Metallography on an extracted sample is the standard method. For hollow bronzes, a transverse section is usually required, extending over the whole thickness of the wall. The section can be as narrow as possible (approx. 1 mm), and its length depends on the questions to be addressed (5–20 mm, for instance a weld joint), such that enough surface is covered to enable a fully representative study. The choice of sample site is obviously crucial. Bronze is hard, and metal walls on a sculpture are often several millimeters thick. A diamond saw mounted on a drilling machine is generally the best tool (figs. 417, 418). Great care must be taken when sampling to preserve the structure of surface layers (see II.6§3.1).

The preparation of the cross section may be as follows. The cross section is embedded in resin (often epoxies or acrylates), pre-polished using silicon carbide papers of different grades, then polished on cloths to a mirror surface, either with alumina or diamond paste. Many microstructural features may be observed optically, directly on a polished surface. SEM examination is also usually conducted directly on the polished sample.

In order to reveal particular features, a solution made of various acids and/or bases may be applied to the surface of the cross section. This is called etching. Etching solutions react chemically with some specific parts of the microstructure such as grain boundaries, and thus reveal these features by either removing them or coloring them. The optimal etching solution depends on both the type of copper alloy and the features under investigation. Etching is discussed at length in Scott 1991.²⁶ Before etching, make sure to observe and record all the microstructural features that are best observed without etching, or else repolishing will be necessary.

Open viewer Figure 417

Open viewer Figure 418

The polished cross sections should first be examined using an optical microscope both to get an overview of the microstructure and/or surface layers (fig. 146) and to estimate their homogeneity or heterogeneity. There are advantages to the specificities of this technique. Two viewing modes are possible. In bright field conditions (specular light), metallic and nonmetallic inclusions can be visible, as can different metals, such as gilding layers. This is usually done before etching. Bright field mode also enables the clear visualization of relief generated by etching as well as the colors of the microstructure generated by etching (figs. 142, 145, 419, 420). Dark field (diffuse reflectance) is rarely used for microstructural observation, but rather for the investigation of surface layers, patina, gilding, and coatings, similar to stratigraphic cross sections used in the study of paintings.

Scanning electronic microscopy (see II.5§3.4) may be required for higher magnification and for elemental analysis of inclusions, metallic phases, and/or surface layers (figs. 421, 422, 423, 424). Some SEM instruments may allow for electron back-scattered diffraction (EBSD) structural analysis as well.²⁷

Metallography is a highly invasive procedure and (as explained in I.2§3.3) should only be considered if it will answer very specific questions. As with most other analytical techniques mentioned in these Guidelines, a minimum of skills and techniques (for sampling, preparation, observation, and interpretation) is mandatory to ensure that the most valuable information is gathered.

Open viewer Figure 146

Open viewer Figure 142

Open viewer Figure 145

Open viewer Figure 419

Open viewer Figure 420

Open viewer Figure 421

Open viewer Figure 422

Open viewer Figure 423

Open viewer Figure 424