Introduction

Museum professionals who work with collections or design enclosures and mounts for the storage, transport, or exhibition of art must regularly decide which materials to select for their projects1,2,3,4. Aside from cost, ease of use, and aesthetic, the material’s propensity to off-gas reactive volatile organic compounds (VOCs) and semi-volatile compounds (SVOCs) should be considered. These emissions can affect the material’s long-term stability or degrade adjacent artworks. However, the evaluation of long-term stability or potential interaction with the museum collections remains challenging for museum professionals. Safety Data Sheets (SDSs) and product literature are often absent or lack detailed chemical composition. As a result, many museum employees base their decisions on microchemical tests, such as pH analysis or sulphur and chlorine detection, the Photographic Activity Test (ISO 18916), or accelerated corrosion tests like the Oddy test5,6.

The Oddy test is the most widely used accelerated corrosion test in the field of preventive conservation. It exposes three metal coupons (silver, copper, and lead) to volatile compounds emitted by the testing material in a closed vessel during 28 days at 60 °C and 100% RH5. The Oddy test has undergone several revisions and improvements throughout its history7,8,9,10. Nevertheless, the Oddy test has some limitations that have been widely acknowledged by researchers and practitioners11,12,13. The key limitations can be summarized as: (1) Experiment duration. It requires 28 days for the completion of the test. Additionally, coupon preparation, photography, and interpretation of the results are steps that demand an extra time investment; (2) Limited substrate reactivity. The test primarily focuses on the corrosion of silver, copper, and lead, making it a selective test assessing the potential risks to these metals. However, it may not effectively assess the risk to other artists’ materials present in a museum environment such as paper, paints, or plastics; (3) Elevated temperature and humidity conditions. The test exposes materials to elevated environmental conditions (60 °C and 100% RH), which do not accurately simulate the nuanced and varied conditions within a museum; (4) Subjectivity in interpretation. The interpretation of the corrosion or tarnish on the Oddy test metals is based on subjective visual inspection, that relies on the judgement and expertize of individual testers. During the evaluation, materials are classified into three categories of use (permanent (P), temporary (T), and unsuitable (U)), depending on the degree of color or lustre changes observed on the metals. This subjectivity introduces variability which impacts the reliability of the conclusions drawn from the test; (5) Lack of standardization. More than 40 versions of the Oddy test are known which vary in vessel type, metal purity or preparation, water to vessel volume ratio, method of suspending metals or sealing vessels, or method evaluation criteria. Standardization would serve to reduce variability between test versions. This need has been highlighted in a recent interlaboratory comparison of the Oddy test11, where many variables that could affect test results were compared and discussed. The outcomes of this study indicated that the main discrepancies arose from coupon preparation inhomogeneity, evaluator inexperience, and, especially, from inconsistent evaluation of copper and lead coupons classified as temporary. This encourages further investigation of reliable corrosion test alternatives, which would allow conservators and other museum professionals to select and advocate for more inert storage, display and transport materials.

The Oddy Test has been implemented at many institutions as a tool to screen for reactive chemicals. The vast majority have not reported damage to collections caused by materials that have passed the Oddy test, though this may be an issue with under-reporting. A recent publication from English Heritage, however, corroborates the utility of the Oddy test by describing 600 instances of corrosion that could not be linked to materials that passed the Oddy test13, and The British Museum reported only two instances of silver corrosion on displayed objects in over twenty years when employing the Oddy test14.

An example of Oddy tested materials causing issues with collections has been reported15,16, where a white crystalline efflorescence was observed on objects housed in multiple display cases at several different institutions all manufactured by the same company. This issue was due to the lack of an obvious response of the Oddy test to a particular chemical species. Different analytical approaches showed that the efflorescence was caused by a reaction between the volatile 2,2,6,6-tetramethyl-4-piperidinol (TMP-ol) and acidic compounds originating from other materials and objects within the cases. The source of TMP-ol was identified as the structural adhesive, Terostat-9220, applied by the casework manufacturer from 2009 through 201417,18. The adhesive passed the Oddy test conducted by museum staff because it does not react corrosively with the metal substrates, yet the material still caused significant issues when used in a display case.

To greatly decrease the time required for analysis, reduce subjectivity, and potentially identify non-corrosive chemicals such as TMP-ol, conservation laboratories are evaluating mass spectrometry (MS) tools and methods, mostly employed with chromatographic separation, to qualitatively identify VOCs and SVOCs released by museum construction materials. For example, The Indianapolis Museum of Art-Newfields has implemented the use of Direct Thermal Desorption coupled to Gas Chromatography-Mass Spectrometry (DTD-GC-MS), the authors formerly referred to this approach as Evolved Gas Analysis (EGA)-GC-MS19,20,21, to measure the composition of plastic materials used during the construction of display cases. The Metropolitan Museum of Art has investigated Solid-Phase Micro Extraction coupled to GC-MS (SPME-GC-MS) to correlate the Oddy test results with the VOCs released by the materials9,22,23. These approaches present notable advantages concerning analysis speed, and they offer the potential for a chemically informed evaluation of the potential risks associated with materials compared to those generated by the Oddy test. While the Oddy test does not allow direct identification of individual pollutants, MS analysis can provide targeted and untargeted analysis of VOCs emitted by materials.

In addition, the molecular data obtained via MS analysis can be compared with the visual evaluation of the Oddy test performed on the same materials. This comparison could facilitate the identification of materials that pass the Oddy test but are potentially harmful to collections. With this aim, in this study, the volatile and semi-volatile emission profiles of ten materials were assessed using three different mass spectrometry approaches available at the Rijksmuseum, The Metropolitan Museum of Art, and the Smithsonian’s Museum Conservation Institute. This work seeks to probe the potential of MS to predict the results of the Oddy test by correlating volatile chemicals and observed corrosion behavior, using chemical information to reduce the time required to select materials for use in museums.

A novel Headspace-Solid Phase Micro Extraction-trap-enrichment (HS-SPME-GC-MS) approach has been tested for the first time here to evaluate the volatile and semi-volatile emissions of each material. Sorbent-based, single extraction headspace methods such as SPME have been used in conservation laboratories to characterize the emission of materials intended for use near art objects18,22. However, SPME-based approaches can suffer from issues associated with site competition within the sorbent24,25,26. In this work, the impact on the sampling time and the effect of cumulative extractions were evaluated, while keeping constant the extraction temperature, volume, and sample weight. This new approach allows automated sample enrichment/pre-concentration by multiple desorption steps onto a cryogen-free trap system. The performance of this method has been compared with an established DTD-GC-MS protocol, a method that has also been used to assess the volatile profiles from commercial materials19. In addition, Direct Analysis in Real Time coupled to High Resolution MS (DART-HRMS) and Pyrolysis (Py)-DART-HRMS have been used for the first time as rapid techniques to characterize the composition of museum exhibit construction materials. By performing a comparison of the same materials, this investigation aims to demonstrate the potential of different mass spectrometry techniques while highlighting the need to define and standardize the analysis criteria. It is important to note that, similar to international standards such as EN ISO 16000-6:2021 (Determination of organic compounds (VVOC, VOC, SVOC) in indoor and test chamber air by active sampling on sorbent tubes, thermal desorption and gas chromatography using MS or MS FID)27, 16000-9:2006 (emission test chamber method)28, and 16000-10:2006 (emission test cell method)29, the MS approaches presented here cannot foresee unwanted reactions caused by long term degradation products of the materials. Finally, to help evaluate the information generated by these MS-based approaches, the results were compared to the Oddy test of the same materials11.

Methods

Investigated samples

To evaluate the MS approaches presented in this study, ten potential display case construction materials of relevance for cultural institutions, were previously evaluated by Oddy test (see Table 1).

Table 1 Potential display case construction materials tested by Oddy test and MS
Full size table

Oddy testing

An interlaboratory comparison of different versions of the Oddy test performed on the same materials was recently published11. Oddy tests reported here were conducted by the National Centre for Metallurgical Research (CENIM). Experimental details are outlined in a previous publication under institution IV11. The Oddy test results presented in this study correspond to the visual examination of the coupons conducted by independent expert judges from the CENIM.

HS-SPME-trap and HS-SPME-trap-enrichment sampling and GC-MS analysis

For single HS-SPME-trap sampling, 1 g of material was placed in an 18 mm diameter, 20 mL screw top glass vial (23082, Restek) capped with a SPME compatible septum (C-HSVSCS, Markes International). Extraction was automated in a Centri® autosampler (Markes International). The sample was equilibrated for 1 min at 80 °C. A Carbon/WR/PDMS SPME-Arrow fiber (phase thickness 120 µm, PAL System) was selected due to its capacity to trap highly volatile analytes within a wide molecular weight range (m/z 30-225)30. In addition, the use of the SPME-Arrow offers a larger sorption volume and greater fiber robustness compared to traditional SPME fibers25. The conditioned SPME Arrow fiber was then exposed to the headspace for a total of 10 or 20 min at 80 °C. After extraction, the fiber was thermally desorbed (2 min, 250 °C, 50 ml min−1) to a cryogen-free sorbent-packed focusing trap (U-T15ATA-2S, Markes International). The trap was Peltier-cooled to −10 °C during fiber desorption. Finally, a 1 min helium purge of the trap was performed at 20 ml min−1 before desorption (3 min, 300 °C, 5 ml min−1 split flow) onto the GC column.

HS-SPME-trap-enrichment sampling was performed using the same conditions described above with a 5 min delay between enrichment extractions. Enrichment was achieved by repeating the sampling and desorption into the cryogen-free focusing trap as described above for a total of two to four cumulative extractions for each vial (see Table 2). Then, the trap was purged and injected into the GC as described above.

Table 2 Sampling design for SPME-trap-enrichment
Full size table

Injections were performed using a Centri® autosampler (Markes International) integrated to a Thermo Scientific Trace 1310 Gas Chromatograph (Thermo Fisher Scientific, USA). The GC separation was achieved using a Supelco SLB5 MS (20 m × 0.18 mm × 0.18 m) capillary column and a constant flow of 1 mL min−1 helium carrier gas was applied. The initial oven temperature of 40 °C was held for 5 min and a ramp rate of 10 °C min−1 was applied to increase the temperature from 40 °C to 250 °C, and the final temperature of 250 °C was held for 10 min.

Mass spectra were acquired on a single quadrupole mass spectrometer (Thermo Scientific ISQ LT) in EI mode at 70 eV. The scan range was set to 40–600 m/z with a scan time of 0.2 s. System temperatures were the following: ion source 240 °C, and transfer line 270 °C.

Data analysis was performed with Xcalibur software (Thermo Fisher Scientific USA) and the NIST library 2011 Mass spectra Library V.2.0 was accessed to qualitatively identify the chromatographic peaks. No quantitative analyses were performed in this study, however, normalized relative peak areas were used to semi-quantitatively compare the concentration of specific chemicals observed in test materials by both GC-MS techniques.

Only compounds with a match factor (MF; a software processing method combining forward and backward library comparison and probability weighting) greater than 850 were accepted as putatively positively identified. A MF value of 850 was selected as the best value to optimize the quantity and quality of potential matches31. As a further control, all compounds were manually examined, and only compounds that were present in all triplicate samples and gave consistent retention times were included in the analysis.

DTD sampling and GCMS analysis

Cut samples (~10 mg) were added to small vial inserts (2406-1025, GL Sciences) and placed inside fritted thermal desorption (TD) tubes (2414-1101, GL Sciences). Method blanks were performed with empty TD tubes after each sample. Sample introduction was automated using a PAL RTC autosampler.

Direct thermal desorption of TD tubes was performed using an OPTIC-4 inlet with cryotrap (GL Sciences) connected to a gas chromatograph (Agilent 7890B) equipped with a DB-5MS Ultra Inert column (Agilent, 30 m × 0.25 mm × 0.25 μm). Helium was used as the carrier gas, with a column flow rate of 1.5 ml min−1, and an initial split flow of 3 ml min−1.

Immediately after the TD tube was added to the inlet, the inlet was purged with He for 70 s through the split vent, and the cryotrap was cooled to −60 °C. To transfer volatiles to the cryotrap, the inlet temperature was raised to 180 °C at 60 °C s−1 and held for 1 min. Thermal desorption conditions at 60 °C, 90 °C, 115 °C, and 180 °C with cryotrapping have been investigated at The Metropolitan Museum of Art and other institutions through the collaborative AIC Materials Working Group initiative32. The Met’s results have shown that all desorption temperatures typically yield similar peaks, with signal-to-noise ratio generally increasing with increasing desorption temperature. However, methods at 60 °C and 90 °C require lengthy desorption and associated cryotrapping times to generate signal above the GC-MS limit of detection. Methods at 115 °C and 180 °C yield comparable emission profiles, though peak intensities tend to be greater for samples desorbed at 180 °C. This elevated desorption temperature was selected to maximize the amount of chemical information obtained from each analysis. After the desorption period, the GC-MS run was initiated by raising the cryotrap temperature at a rate of 60 °C s−1 to 300 °C, and the inlet was cooled below 35 °C. The split flow was increased to 200 ml min−1 for 800 s before a reduction to 100 ml min−1 for the remainder of the run. The GC oven temperature was ramped from 35 °C to 40 °C at a rate of 20 °C min−1 and held for 3 min before a ramp to 300 °C at a rate of 10 °C min−1 and held for 2.5 min.

Mass spectra were acquired on a single quadrupole mass spectrometer (Agilent 5977B) equipped with an Inert EI Ion Source (70 eV). Spectra were recorded in profile mode with a scan range of 29–450 m/z, a threshold of 100, and scan speed of 781 (N = 3). System temperatures were the following: ion source, 230 °C, quadrupole, 150 °C, and transfer line, 280 °C.

Chromatogram deconvolution and library searching of mass spectra was automated using MassHunter Unknowns Analysis (Agilent, version 10.2). The identity of the deconvoluted components was confirmed by library searching against the NIST 23 library, using the software’s internal search algorithm & default parameters. Retention time indexing was used in a first round of library searching; RTs were compared to a C7-C30 alkane mix (Millipore Sigma) injected prior to the first sample run. Hits were limited using a 9 s RT match filter (trapezoidal, penalty-free) and a MF greater than 800. If no hit was identified using the first search criteria the library search was repeated without the RT match criteria.

DART-HRMS analysis

A DART-100 probe operated by an SVP controller (IonSense, Saugus, MA) was operated in transmission mode in front of an Orbitrap Elite mass spectrometer (Thermo Fisher Scientific, Waltham, MA) fitted with a differentially-pumped Vapur interface. The DART probe was custom-mounted from above with multi-dimensional translation stages in front of the mass spectrometer (Fig. 1A). A 40 mm stainless steel transfer tube was used in place of the typical ceramic transfer tube on the Vapur interface. The DART helium was set to 300 or 400 °C. Material sections were mounted in forceps on translation stages and manoeuvered between the DART probe and transfer tube for tens of seconds. MS data from desorbed and ionized material was acquired at 120 k resolving power with a maximum ion trap fill time of 100 ms. The DART probe was operated in positive mode or negative mode while the Orbitrap was scanned in positive or negative polarity, respectively.

Fig. 1: Transmission mode DART sampling configurations.
figure 1

A DART mounted in transmission mode, with the sample positioned between the ion source and the transfer line. B Py-DART mode with the pyroprobe orthogonal to the post-plasma flow between the DART and transfer line.

Full size image

Data was analyzed by targeting compound masses for molecules previously identified in SPME-GCMS or DTD-GCMS analyses.

Py-DART-HRMS analysis

Sub-milligram samples were loaded into a quartz sample tube and inserted in the coil of a Pyroprobe 1000 (CDS Analytical Inc., Oxford, PA). The pyrolysis assembly was mounted orthogonal to the post-plasma flow between the DART and transfer tube (Fig. 1B). Samples were pyrolyzed by flashing the quartz tube at a setting of 650 °C for 1 s, and the vaporized sample was ionized in the DART stream with the helium set to 50 °C.

Results

Evaluation of different MS approaches for material analysis

The first part of this study aims to compare the emission profiles obtained with the two GC-MS methods tested in this study, HS-SPME and DTD, followed by a comparison with DART-HRMS.

Figure 2A shows the comparison of acetic acid detected by the four HS-SPME-trap extraction methods tested in this study (see experiment design in Table 2). The acetic acid response was normalized to that of the material which shows the greatest acetic acid peak area. The trap-enrichment SPME method, where multiple extractions were collected on a single trap and then analysed in one chromatographic run, was optimized to maximize acetic acid response due to its ability to corrode lead coupons and, to a lesser extent, copper coupons, even at very low vapor concentrations22,33. In terms of sensitivity, HS-SPME-trap-enrichment produced a greater peak area yield (on average two to six times greater) than a single HS-SPME-trap extraction. The addition of the enrichment step also allowed successful detection of acetic acid in materials 3, 5, and 6, whereas acetic acid was not detected during single enrichment HS-SPME-trap sampling. This suggests that the trap-enrichment step can minimize the issues associated with phase competition in the headspace, while enhancing target chemical detection relative to single extraction methods by loading extractions onto a focusing trap. These results indicate that significant improvements in the performance can be achieved by optimizing the sampling parameters.

Fig. 2: Influence of extraction parameters on the detection of acetic acid.
figure 2

A Acetic acid peak areas (n = 3), using two different extraction methods (SPME-trap and SPME-trap-enrichment) at 10 and 20 min durations. B Comparison of normalized acetic acid peak areas obtained via DTD or SPME-trap-enrichment. The acetic acid response for DTD and SPME have been normalized for each method to material 7 which shows the greatest acetic acid peak area for both techniques. C Silver, copper and lead coupons (from left to right) after the Oddy test. P means permanent use (no corrosion observed relative to control), T, temporary use (slight corrosion observed relative to control) and U, unsuitable use (significant corrosion observed relative to control).

Full size image

Given the success of the HS-SPME-trap-enrichment extraction method, it was selected for further comparison with DTD extraction, a method that has also been used previously to assess the volatile profiles from commercial materials19. HS-SPME-trap-enrichment and DTD were compared through the analysis of the ten studied materials.

The results shown in Fig. 2B were generated by normalizing the data within each method to the chromatogram showing the greatest acetic acid peak area. HS-SPME-trap-enrichment and DTD extraction have comparable performance in terms of acetic acid detection for materials 3, 5, 6, 7, and 8, where normalized acid levels were less than 0.1 a.u. for materials 3, 5, and 6 and greater than 0.7 a.u. for materials 7 and 8. DTD extraction, however, identified acetic acid in four materials (materials 1, 4, 9, and 10) where HS-SPME did not detect it. For both HS-SPME-trap-enrichment and DTD, material 7 produced the greatest acetic acid peak areas, and the normalized results for the two techniques produced much larger relative peak areas for acetic acid for materials 7 and 8 than for the rest of the materials.

Reactive compounds other than acetic acid and surface or bulk modifiers (e.g., plasticizers) were observed using both HS-SPME and DTD. Table 3 summarizes the classes of select observed chemicals for each material. Table SI.1. lists tentative compound assignments based on DART-HRMS measurements, where observed isotope ratios combined with knowledge of the compounds that are likely to be present in the samples were used to inform compound identification. It is important to acknowledge the limitation of these techniques on detecting other compounds of interest, such as highly reactive peroxides, ozone, and free radical generators capable of causing degradation reactions in museum objects. Due to their thermally labile nature or/and reactivity, these compounds are not amenable to detection with any of the techniques in this study.

Table 3 Volatile species detected by HS-SPME-trap-enrichments and DTD with a match factor (MF) greater than 850 and 800, respectively
Full size table

A visual comparison of the chromatographic outputs from HS-SPME-trap-enrichment and DTD analyses and the mass spectra from DART-HRMS analysis are shown in Fig. 3 for material 5 (black rubber material). The chromatograms include the volatiles emitted by the material during the thermal sampling processes at 80 °C for HS-SPME-trap-enrichment and 180 °C for DTD sampling (Fig. 3A, B). Figure 3C shows the DART-HRMS mass spectrum after exposing the same material at 300 °C. Py-DART-HRMS data was also acquired at higher temperature (650 °C) to vaporize synthetic materials more effectively. A comparison of the spectra of material 4 obtained with both temperatures is shown in Fig. SI.1. The Py-DART-HRMS spectrum of material 4, includes hundreds of additional masses with relative abundances below 10. In addition, there are differences in the observed masses with relative abundances above 10, which were generally greater for Py-DART-HRMS. These differences may be due to the use of a greater thermal extraction temperature in Py-DART-HRMS. Additionally, vaporization and ionization events are separated in space, creating a momentary lag during which other reactions could conceivably occur. Unassigned peaks observed in the spectrum with higher pyrolysis temperatures may also result from thermal breakdown of additives or unidentified components.

Fig. 3: Analysis of Material 5 (rubber).
figure 3

A Total Ion Chromatogram of SPME-GC-MS (5 min extraction at 80 °C with 4 sequential enrichment steps) and B Total Ion Chromatogram of DTD-GC-MS. C DART-HRMS mass spectrum with the protonated molecules of the sample components without any separation.

Full size image

Overall, while chromatographic methods were used in this study as untargeted screening techniques, DART coupled to Orbitrap high resolution mass spectrometry was useful for confirming chemical identifications with greater confidence for targeted compounds.

Correlation of the MS outcome with the Oddy test evaluation

Results of previously published Oddy tests of these materials are shown in Table 311, where ratings for each metal coupon along with the overall Oddy test result are included. Of the ten materials, two were rated as permanent (P), four as temporary (T) and four as unsuitable (U).

The two materials that passed the Oddy test (materials 1 and 9) produced a similar emission profile via HS-SPME-trap-enrichment and DTD analysis. Each contained mostly hydrocarbons (alkanes and alkenes), aldehydes, and phthalates (esters of phthalic acid used as plasticizers). Although these chemical families are unlikely to cause corrosion and thus are often classified as low or no-risk for use with museum objects, the presence of plasticizers such as phthalates may have non-corrosive impacts. It has been previously shown that phthalates can migrate from the source and be incorporated into adjacent objects causing plasticization or softening of the object, which can affect object integrity and have a negative visual impact20,34,35. Of all materials analyzed, material 9 (MS-35 Sealant) showed the largest number of phthalates in its emission profile, suggesting that despite passing the Oddy test, this material must be carefully considered before it is used in a museum setting. This insight would be missed without GC-MS analysis to supplement the Oddy test results. In addition, DTD analysis detected acetic acid was in both materials 1 and 9 (Fig. 2B). However, the normalized relative low peak areas of this compound might suggest that the emitted concentration is not enough to produce corrosion, as will be discussed below.

Materials 2, 3, 4, and 6, which all received temporary (T) ratings, emitted a number of VOCs that could potentially be linked to the corrosion observed on the metal coupons. For example, carboxylic acids detected by HS-SPME and DTD, were emitted by all four materials. Carboxylic acids have been previously associated with unwanted changes of artworks including corrosion, protrusions, color change, and bleaching1. Glycols, which are commonly used as co-solvents or plasticizers in adhesives can contribute to the softening of other plastic materials in their proximity19, were detected by HS-SPME, DTD and DART in material 3 and by DTD in material 8. TCPP, a phosphate flame retardant, emitted by material 6 and detected by HS-SPME and DTD, and by material 5, 7 and 8 only detected by DTD, must also be considered as reactive species, since flame retardants have been associated with corrosive reactions involving metals33. 2-ethylhexanol, a known degradation product of the plasticizer diethyl-hexylphthalate (DEHP)36, was detected by HS-SPME and DTD in materials 3 and 4. In addition, these four temporary materials also emitted phthalates. As discussed above, plasticizers can contribute to the softening of other plastic materials in their proximity19.

Materials 5, 7, 8, and 10 all received an unsuitable (U) rating when subjected to Oddy testing. Each of these unsuitable materials also exhibits at least one compound in their MS emission profiles that has been previously classified as a potential hazard for museum collections, and therefore may be linked to the unsuitable ratings. Material 5 was unsuitable for both copper and silver. Materials 7, 8, and 10 produced an unsuitable level of lead corrosion. Material 10, in addition to corroding the lead, produced an unsuitable level of copper corrosion.

Material 5, a natural rubber, was the only material that received an unsuitable score for the silver coupon (Table 3). As shown in Fig. 3, all three MS techniques revealed the emission of sulfur compounds from material 5. For material 5 the observation of sulfur-containing compounds may be linked to the corrosion observed on the silver and copper coupons as black tarnish (see Fig. 2C). Sulfur is incorporated into many rubber products as a vulcanizing agent37. While the manufacturing process used to produce material 5 is not known, if it is a vulcanized rubber, this may explain the presence of sulfur.

In addition to material 5, DTD or HS-SPME also detected sulfur containing compounds in materials 1, 2, 3, and 9, and these materials did tarnish the silver coupons during the Oddy test (Table 3). So, while GC-MS can alert collection stewards to the presence of reactive compounds, more work is needed to relate the amounts and types of sulfur-containing compounds identified by GC-MS to risk levels for collections.

Lactic acid and its lactide were detected by HS-SPME and DTD in material 10 (polylactic acid filament, PLA) (Fig. 4). PLA has been shown to accelerate corrosion of Fe and Zn surfaces due to its breakdown into lactide/lactic acid, creating localized acidic environments38. In addition, this material also emits other carboxylic acids, which are known to corrode copper and lead (the two coupons classified as unsuitable). These corrosive reactions may explain the unsuitable Oddy test score for the copper and lead coupons for material 10.

Fig. 4: Analysis of material 10.
figure 4

A Total Ion Chromatogram of SPME-GC-MS, silver, copper and lead coupons (from left to right) after Oddy test. B Total Ion Chromatogram of DTD-GC-MS.

Full size image

GC-MS analysis of the two wood-based materials (7 and 8) showed evidence of high acidic emissions. Organic acids are known agents of corrosion; they are also readily emitted from wood products, and their use in museum environments is inadvisable1,39. It is particularly noteworthy to compare the acetic acid peak area of these two wood-based materials with other materials that emit acetic acid but in smaller amounts (Fig. 2B). Figure 2C includes images of the silver, copper, and lead coupons for both the material tests and controls for those materials in which acetic acid was detected using HS-SPME and DTD analysis.

Only the two materials with normalized acetic acid peaks greater than 0.7 (materials 7 and 8) caused the most severe tarnish on the corresponding lead coupons, severe enough to receive an unsuitable rating (U). In contrast, materials 1, 3, 4, 5, 6, 9, and 10 generated much smaller acetic acid peak areas, and the corresponding lead coupons were comparatively less corroded than those aged with wood. These seven materials produced normalized acetic acid peaks areas below 0.2 a.u. and lead coupons that were classified as either permanent (P) or temporary (T). However, there was no correlation between normalized peak areas of acetic acid and T or P ratings. Overall, this might suggest that with a reference set that more completely covers the range of normalized acetic acid levels between 0 and 0.75 a.u. may serve as an indication of expected degree of Oddy test corrosion22. Testing of additional materials is needed to firmly establish what acid levels can be used to predict the degree of Oddy test corrosion.

While acetic acid may be useful as an indicator for whether a material should be used with cultural heritage objects, additional organic acids may also contribute to the observed corrosion. For example, DTD analysis of the wood-based materials 7 (birchwood) and 8 (MDF) showed the presence of other organic acids, such as formic acid, to which copper and lead are very sensitive (Fig. 2C)40.

Although acetic acid is known to be highly reactive toward metals and other materials even at low concentrations, the results presented here show that the simple presence or absence of acetic acid as detected via MS methods should be carefully considered. In addition, when utilizing materials in cultural heritage settings, the impact of acetic acid on heritage objects will depend on other factors such as the material nature of the object, number of emissive materials used, enclosure volume, air exchange rate or leak rate, and emission rate of acetic acid within the exhibition case, storage, or transportation enclosure. To allow the practical interpretation of MS results, it would be necessary to run standard materials with known corrosion behaviors, such as materials that do not corrode, corrode slightly, moderately, and significantly corrode the metal coupons. By normalizing the data from these standard materials, it may be possible to estimate the expected levels of lead and possibly copper corrosion produced by other materials to these standards.

Discussion

As stated in the introduction, HS-SPME-GC-MS is commonly used for untargeted profiling of volatile organic compounds (VOCs) emitted by historical artifacts or present in museum environments41,42,43,44,45,46. However, the technique can suffer from limited sensitivity due to competition between individual VOCs for adsorption on the fiber phases25,26. To overcome the limitations of traditional single-SPME sampling, the enrichment step into the cryogen-free focusing trap presented in this study improves the quantity of sample collected in the SPME fiber, enhancing MS signal strength, which in turn increases the total number of interpretable datapoints for an analyte and generally improves confidence in chemical identification. This has been demonstrated here during the extraction of acetic acid, a low molecular weight VOC, where greater signals were observed when multiple extractions were combined and analyzed. A complementary technique also used to analyze the VOC emissions from museum materials is DTD-GC-MS. In the DTD method, materials are placed in a helium-purged oven at elevated temperatures under isothermal conditions to expedite the release of VOCs. Emitted VOCs are directly transferred onto a cryo-cooled capillary column for analysis. Notably, unlike other approaches, there is no concern regarding sorbent selectivity or pre-concentration equilibrium. This means that all analytes compatible with GC separation and MS detection can be thoroughly analyzed. Furthermore, the method enables the analysis of a large volume of evolved gases, resulting in enhanced detection limits of low molecular weight VOC, such as formic and acetic acid (Table 3), and the number of detected compounds, at least for the materials investigated in this study, was greater for DTD than the HS-SPME-trap-enrichment method. This variation in the number of detected compounds might be explained by the different sampling temperatures between both methods. In contrast to HS-SPME-enrichment, DTD offers the direct transfer of volatile analytes onto the column, which eliminates the potential for site competition on the SPME fiber. However, to avoid cryo-trapping for extended periods of time, higher temperatures are used to rapidly desorb volatiles during the DTD extraction (180 °C (DTD) vs 80 °C (SPME)), which may result in the development and evolution of volatile degradation products that are less likely to be observed at room temperature47.

Quantitative experiments are needed to determine the concentration of reactive compounds, for example acetic acid, required to induce corrosion on the coupons. Although quantification measurements have previously been reported using DTD20 and SPME26, quantitative analysis is not feasible in this study due to the lack of control over sampling conditions. In addition, the challenge of using mass spectrometry techniques to assess display, transport, or storage materials lies in the selection of target compounds and calibration standards that cover the broad range of VOCs emitted by complex samples.

The two chromatographic methods implemented in this study require less than 1 h to complete the sample preparation and analysis as illustrated in Fig. 5. This offers a significant advantage over the 28 days required for Oddy testing, however, a scientist must currently rely on “chemical intuition” to link the identified compounds to corrosive effects when vetting materials using MS19. On the other hand, DART-HRMS can analyze each sample in situ, without sample preparation, and within a few seconds (Fig. 5). Although, without chromatographic separation, low-abundance analytes among a great variety of background species may be detected by the instrumentation but overlooked by the analyst unless the ion of interest is known in advance or can be found by manually analyzing the spectra, ion by ion. Competitive ionization from compounds in very high abundance may decrease the detection efficiency of analytes of interest at low abundance. As a result, data interpretation requires a higher level of expertize than with traditional mass spectrometry data where chromatographic separation of analytes is implemented. As a result, DART-HRMS is effective in confirming the presence of specific additives, while GC-MS has the potential to serve as a screening tool for material classification18,48.

Fig. 5
figure 5

Timeline schematic for the analysis of 10 materials via the Oddy test and three MS approaches, DTD-GC-MS, SPME-GC-MS, and DART-HRMS.

Full size image

The aim of this study was to show the potential of three MS-based methods to evaluate exhibition case materials by characterizing the emitted volatile compounds at the molecular level. The use of HS-SPME-trap-enrichment along with the optimized instrumental parameters improved not only the sensitivity but also the level of information that can be extracted from a single material analysis relative to single SPME extraction experiments. DTD extraction was shown to be sensitive to similar classes of chemicals and multiple chemical types, identifying their presence more often than SPME-trap-enrichment due to the temperature reached during the analysis, as explained in the discussion section. DART-HRMS, which requires minimal sample preparation, confirmed with accurate mass measurements the presence of compounds that were tentatively identified by the chromatographic methods.

The results of the two chromatographic approaches were compared with the traditional Oddy test conducted on the same materials. The outcome showed a loose correlation between the instrumental analysis and Oddy test results, suggesting that more work is needed to utilize MS-based methods to accurately predict the utility of a material for use with heritage objects. However, MS-based methods can serve as a valuable supplement to traditional accelerated corrosion testing in evaluating museum construction materials.

While the analysis time using MS-based approaches is substantially decreased compared to Oddy test, the data interpretation requires a higher level of expertize. The chemical mixture emitted by an exhibition material must be cautiously evaluated for its potential to harm artwork, relying on existing literature and the analyst’s chemical intuition. This, in turn, entails the need to create a database that collates a list of unwanted chemicals and the concentrations at which they are capable of damaging collections.