Introduction

Historical background

Cellulose nitrate was one of the first cellulose derivates and was already discovered in 1833 [1]. A first description of synthesis of acylated or alkylated cellulose derivates was given by W. Suida in 1905 [2]. Subsequent research on methyl- and ethyl derivates of cellulose was done in the 1910s [3]. Pioneering patents for nonionic cellulose ethers were submitted by Hubert in 1920 [4], Leuchs in 1912 [5] and Lilienfeld in 1916 [6]. Hubert [4] for example described the synthesis of hydroxyethyl cellulose, whereas Jansen [7] introduced the sodium carboxymethyl cellulose (Na-CMC). CMC was used in Germany for improving detergency in low grade synthetic detergents during the second world war [8]. For example, the product Tylose® HBR was recommended for use in fatty acid soap products [8]. Hydroxypropyl cellulose (HPC) was developed in the late 1940s by Eugene Klug and coworkers from Hercules Powder Company in the USA and was patented in 1951 [9]. HPC was introduced as Klucel® on the market. Methyl hydroxypropyl cellulose was developed even later by researchers from Dow Chemical Company and was patented in 1968 [10].

Industrial production of cellulose ethers started in Germany in the early 1920s and in 1936/37 in the USA [11]. Tylose®, one of the first commercial brands of cellulose ethers in Germany was registered in 1926 (1936 USA). It was developed by Chemische Fabrik Kalle & Co. that became part of I.G. Farben AG in 1925. The production palette included methyl cellulose (MC), and carboxymethyl cellulose [8, 12]. After the second world war the Tylose® business became part of Hoechst AG. In 1997 it was sold off to Shin-Etsu Chemical being manufactured by SE Tylose GmbH & Co. KG. Other familiar brands in conservation are for example Klucel®, Culminal®, Methocel® or Blanose®. Klucel® was patented in 1951 and became a registered trademark in the USA in 1963, and in Germany in 1968. Klucel® had been manufactured by Hercules Inc. (business division: Aqualon group) until it was sold off to Ashland Global Specialty Chemicals Inc. in 2008. Culminal® was registered as trademark by Hercules Inc in Germany in 1951 (1963 USA). Initially, it was a trademark for methyl cellulose (MC) [3]; however, in the following decades Culminal® product lines got extended and include genuine methyl cellulose (Culminal® MC types), hydroxypropyl methyl cellulose (Culminal® MHPC types), and hydroxyethyl methyl cellulose (Culminal® MHEC types) [13, 14]. Methocel® products have been manufactured by Dow Chemical Company and include genuine MC (Methocel A types) as well as HPMC (Methocel® E, F, J, K types). It was registered as trademark in the USA in 1938 (1959 in Germany). Blanose® is a trademark for sodium carboxymethylcellulose (Na-CMC) that was developed by Hercules Inc. (Aqualon group). It was introduced in the USA in 1946 and in Germany in 1947 [15].

Cellulose ethers in conservation

Methyl cellulose (MC) was used already in the 1920s as a consolidant for corroded leaden bullae [16]. In paper conservation water soluble MC has a long record of usage as adhesive, size and as consolidant [3, 17,18,19,20,21,22,23]. MC solutions have been tested on wall paintings [24, 25]. They have been used as a medium for pigments [26,27,28,29,30], as a consolidant for waterlogged wood [31] and for basketry [32], for relining of canvas [3], as an adhesive for textiles [33, 34] and wallpapers [35,36,37]. MC among other cellulose ethers have been added as additives (rheology modifiers, protective colloids) to starch adhesives and polymer dispersions [38,39,40,41,42,43] to improve working properties. Ethyl cellulose (EC) is hardly used in conservation [44] but is sold as film-forming agent for paints and for hot melt coating [45]. Hydroxy ethyl cellulose (HEC) has also a limited use in conservation [44] and mainly occurs as emulsifier, stabilizer, thickener or protective colloid in dispersions [39, 41, 46, 47]. Only few references document the single use of ethyl hydroxyethyl cellulose (EHEC), methyl hydroxyethyl cellulose (MHEC) and methyl hydroxypropyl cellulose (MHPC) in conservation. EHEC has been used for textile consolidation [44] and for formulations for jute consolidation [48]. HEMC was suggested for a consolidation treatment for canvas [49] and for canvas lining [50]. HPMC is mainly used as a thickening agent, stabilizer, emulsifier and film forming agent. HPMC and MC are the most common gelling agents used in aqueous conservation cleaning preparations [51]. Moreover, HPMC was used as consolidant of an ethnographical object [52]. HPC, especially Klucel® G have been used as leather consolidant [19, 53, 54], in textile [55, 56] and paper conservation [23, 57,58,59,60], for the consolidation of a wax sculpture [61], of archaeological cartonnage [62] and of herbarium specimen [63]. Klucel® J and G have been used in general for pigment consolidation where a non-aqueous treatment is required [19, 64, 65]. Na-CMC has many applications among conservators and include for example the use as adhesive, consolidant and detergent in paper conservation [44, 66], as relining agent for canvas [50], as paper or textile size, as cleaner for stones, murals, leather and textiles [3]. However, for wall paintings cellulose ethers with less impurity content (NaCl, a by-product from the etherification process) are recommended [25].

Oddy testing of cellulose derivates

Conservation materials like adhesives, consolidants or coatings stay on treated objects and enter with them display cases. Therefore, they need to be Oddy tested like materials for storage or display [67]. Due to its simplicity, the Oddy test is a suitable method for mass screening of materials for harmful emissions.

Oddy tests of cellulose derivates are scarcely found in the published literature. Comprehensive results are only available for cellulose nitrate [67,68,69]. All tested products of this type failed the Oddy test dramatically. Heavy corrosion occurred on all coupons and the analyzed phases include rouaite (Cu2(NO3)(OH)3), litharge (PbO), cerussite (PbCO3), hydrocerussite (Pb3(CO3)2(OH)2) and a silver cyanide–silver nitrate phase [67, 68]. The authors concluded that nitrous fumes were generated during the testing causing the massive corrosion. The results are particularly important as cellulose nitrate is still used as adhesive [44] and can occur as admixture in acrylic adhesives [70, 71]. Nel and Lau [71] showed that the manufacturer changed the composition of a ready-to-use adhesive based on the very stable Paraloid® B72 by adding small amounts of the unstable cellulose nitrate. It can be assumed that the aging characteristics and the corrosive gas activity of this new formulation is much worse compared to pure Paraloid® B72. This case demonstrates the importance of testing not only the primary polymers as commercial products often contain undeclared additives.

To the best of our knowledge Oddy test results of only three cellulose ethers are published. Korenberg et al. [72] tested Klucel® G (HPC) and Culminal® MC 2000 (MC) and both products passed the Oddy test. Recently, the Metropolitan Museum of Art (MMA) published also an Oddy test result of Klucel® G, rating it for temporary use as slight corrosion occurred on the silver and lead coupon [73]. Moreover, they reported the results for Methocel® A4C Methylcellulose, assigning it as suitable for use (P rating) [74].

Methods/experimental

Oddy test

The “3 in 1” Oddy tests were prepared according to the procedure of the British museum [75] that was revised in 2018 [72]. A detailed description of the work flow can be found in these publications, hence, only a quick summary will be given here. Two grams of the sample material are placed together with a small test tube with 0.5 ml ultra-pure water and stoppered with cotton wool in a 55 ml test tube (DURAN®, Carl Roth GmbH + Co. KG). Three high purity (> 99.9%) metal coupons (0.8 × 3.5 cm) made of silver (Carl Roth GmbH + Co. KG), copper (Carl Roth GmbH + Co. KG) and lead (Merck KGaA) are polished and cleaned in HPLC-grade acetone. They are fixed in a silicon stopper (Versilic™ Silicone Stopper, Th. Geyer GmbH & Co. KG) that seals the test tube. One blank test was included for every test series. They were placed in the oven at 60 °C for 28 days with a relative humidity of 100% inside the tubes. All materials were running in duplicate and were repeated if the results differ. The tests are evaluated visually by comparing them with the blank coupons and rated as P (permanent use, no corrosion), T (temporary use, slight corrosion) and F (fail, unsuitable, heavy corrosion). An overall rating is assigned to indicate a material’s general suitability. Please note that a P rating does not mean that the tested substance does not release any volatile components. There can be still VOC present, but they are not visibly corrosive to the three metals that are used in the Oddy test. According to the experience at the British Museum, the results for these metals can be generalized to all materials. ‘Since [the Oddy test] was implemented at the British Museum, there has been virtually no corrosion caused to objects by indoor pollutants’ [72].

The metal coupons are photo-documented using a camera (Canon EOS 7D Mark II) on a copy stand in diffuse light. Corroded coupons are further examined by optical microscopy and Raman spectroscopy.

Raman spectroscopy

Raman measurements of corroded Oddy test coupons allow to identify the corrosion products which may give hints to the corrosive gaseous agents. They were conducted with a Renishaw inVia™ Raman system equipped with two lasers (632 and 785 nm) and a Peltier-cooled CCD detector. The 632 nm laser (grating 1800 lines/mm) was focused through a 50× lens from a Leica DM 2550 M microscope. The spectral range was set to 100–3700 cm−1 using various measurement conditions (10–60 s; 2 accumulations, 1–50% laser power). The Origin(Pro), Version 9.0 (OriginLab Corporation, Northampton, MA, USA) software was used for spike removal and multi-point baseline subtraction.

BEMMA

Determination for the BEMMA (‘Bewertung von Emissionen aus Materialien für Museumsausstattungen’ = Evaluation of emissions from materials for museum equipments) evaluation is carried out in microchambers (µ-CTE, Markes company), six small cylinders with a volume of 45 ml each. The following parameters apply when loading the microchambers with the samples to be tested: Flow: 28 ml/min; 23 °C ± 2 K, ≈ 50% relative air humidity, synthetic air, surface-specific air-flow rate depends on the tested material. With these parameters the concentrations of VOCs in exhaust air of the microchambers can be calculated. Extract samples with Tenax® for VOCs (volatile organic compounds) and SVOCs (semi-volatile organic compounds) were analysed (in accordance with ISO 16000-6 [76]). Quantification is performed with a thermal desorption GC–MS combination. The evaluation of the individual peaks is carried out with the total ion chromatogram, which quantifies by toluene calibration. The summation is carried out for all components whose quantification through toluene equivalents equals or is greater than 5 µg/m3 (threshold for consideration). The resulting main components (≥ 5 µg/m3) are listed separately. Aldehydes and ketones were sampled with DNPH (dinitro-2,4-phenylhydrazine)-cartridges overnight (in accordance with ISO 16000-3 [77]). At least formaldehyde and acetaldehyde should be listed separately. Additionally, aldehydes beyond a concentration of 5 µg/m3 should be noted. Sample extraction for volatile, short-chain, organic acids, like formic and acetic acid was performed on silica gel cartridges with the second overnight sampling. The concentrations were quantified with ion-chromatography [78]. The sums of very volatile organic compounds (VVOC), a subset of the last two procedures, were also used for the evaluation.

Assessment criteria: Substances with high contamination potential, such as formic acid, acetic acid, formaldehyde and oximes must not be detectable. That means these compounds must be lower than the limit of detection. The limit of sum emissions for Σ VVOCs is 100 μg/m3, for Σ VOCs is 500 μg/m3 with the exception of sealing materials with Σ VOCs 2000 μg/m3 due to significantly smaller application surface and for Σ SVOCs is 100 μg/m3. All listed assessment criteria must be fulfilled, otherwise the product fails the BEMMA scheme.

Samples

Two types of samples are included in the study: (1) recent materials, that are currently used in conservation and that were freshly purchased and sampled; and (2) older materials from the 90 s and 00 s that had been used in the past and that may be part of art works nowadays. These samples are obtained from the materials’ collection of the Institute of Conservation Sciences, Stuttgart State Academy of Art and Design. Detailed information about the sample materials can be found in Table 1.

Table 1 Comparison of the tested commercial cellulose ether products
Full size table

Results and discussion

From 60 samples 55% (i.e. 33 samples in total) passed the Oddy test and get a rating for permanent use. 33% (20 samples) are for temporary use as slight corrosion occurred on at least one metal coupon. Only 11% (7 samples) of the tested materials are unsuitable for use as they fail the Oddy test producing significant amount of corrosion due to a high concentration of harmful volatile components. Obviously, by checking the table of results (Table 1) no group of cellulose derivates entirely passes the Oddy test as there are always some products that release harmful volatile components.

Hydroxypropyl cellulose (HPC)

A total of 18 Klucel® HPC samples had been tested whereas 11 were freshly purchased and seven were historical or undated products from the materials’ collection. All historical samples failed the Oddy test with a T or and F rating (Table 1). Klucel® MF (Hercules, 2006, F rating) generated significant amount of corrosion on both copper and lead coupon. The phases were identified by means of Raman spectroscopy and include cuprite (Cu2O), massicot (PbO), hydrocerussite (Pb3(CO3)2(OH)2) and lead formate (Pb(HCOO)2). Cuprite yields Raman bands at 147, 215, 494 and 623 cm−1 which are often only visible as weak bands or shoulders (cf. Fig. 2a) [79, 80]. Massicot and its polymorph litharge are strong Raman scatterer. Both have their strongest band around 145 cm−1 and an additional main feature at 285 (massicot) and 340 cm−1 (litharge) [81, 82]. The main bands of hydrocerussite are located at 1048, 1051 cm−1 (shoulder) and c. 3540 cm−1 and can be used to differentiate it from plumbonacrite (Pb5O(OH)2(CO3)3) (see further discussion in this section and [83]). The Raman spectrum of lead formate is shown in Fig. 2c. Lead formate can be easily differentiated from lead acetates by its typical band pattern. The C–H stretching band doublet is located at 2843 and 2873 cm−1, the strong C–O stretching is centered at 1345 cm−1 and various spectral features occur at 760, 1075, 1375 and 1530 cm−1 [84, 85]. The identification of lead formate may not clearly proof the presence of formic acid as formaldehyde can be also oxidized during the test conditions, inducing the growth of metal formates [86, 87].

Regarding the fresh HPC products, Klucel® M and H passed the Oddy test with no visible corrosion on all metal coupons. The low viscosity type E generated little corrosion on the lead coupon leading to a T rating. Klucel® G was tested multiple times as ambivalent results occurred, causing at first a veritable confusion. A sample from 2017 (purchased from GMW) passed the Oddy test with no corrosion on any coupon (Fig. 1a). That was not surprising as Korenberg et al. [72] reported the same result for Klucel® G. However, the Oddy test from a freshly purchased product from 2021 (purchased from Deffner) shows some amount of corrosion on the lead coupon, yielding a T rating (Fig. 1c). The same result was achieved from Klucel® G purchased from Kremer (Fig. 1b). In both cases Raman spectroscopy revealed the presence of massicot/litharge (PbO), hydrocerussite and basic lead acetate (Pb3(CH3COO)2(OH)4). Basic lead acetate yields a plenty number of bands, with several bands (e.g. 370, 447, 612, 641, 648, 666, 912, 922, 929, 1345, 1410, 1428, 2923, 2970, 2996 cm−1) being diagnostic for this phase (cf. Fig. 2d). A detailed discussion and band assignment can be found elsewhere [84]. Especially the spectral feature at 370 cm−1 and the bands between 910 and 930 cm−1 can be used to differentiate it from lead acetate trihydrate (Pb(CH3COO)2·3H2O) [84]. Generally, lead acetates indicate the presence of acetic acid in the test tube. One gram of Klucel® G (Deffner) was more detailed tested utilizing the BEMMA scheme with a quantification of the outgassing VOC and VVOC. A high amount of acetic acid can be detected (about 4000 µg/m3). Furthermore, fragments of the hydroxypropyl cellulose can be detected with the VOC Tenax method and a small amount of formaldehyde was analyzed. This material does not fulfil the BEMMA scheme. A sum of VOC higher than 500 µg/m3 and the high acetic acid concentrations and the following high amounts of VVOCs were the reasons.

Fig. 1
figure 1

Lead coupons after the Oddy test of various Klucel® G samples. The lower part of the coupon was stuck into the stopper and may not be used for rating purpose. a Klucel® G 2017 (GMW), b Klucel® G 2021 (Kremer), c Klucel® G 2021 (Deffner), d Klucel® G IND (Ashland), e Klucel® GF (Ashland), f Klucel® GF Pharm (Ashland). Please note that the metallic reflections in f may lead to misleading interpretations regarding the metal corrosion

Full size image
Fig. 2
figure 2

Raman spectra of common corrosion phases. Offset was applied for better comparison. a Cuprite and tenorite (Tn) (Cu coupon of the Hydroxyethylcelluose test); b anglesite (Pb coupon of the Blanose® 7H4Xf test); c lead formate (Pb coupon of the Klucel® MF test); d massicot (M), plumbonacrite (Pl), basic lead acetate (Pb coupon of the Hydroxyethylcelluose test); e plumbonacrite (Pl), lead acetate trihydrate (Pb coupon of the Hydroxyethylcelluose test)

Full size image

In a next step, Klucel® G samples with different purity grades were obtained directly from the producer. The testing of the industrial quality (i.e. the same quality as for the Kremer and Deffner sample) shows unsurprisingly the same amount of corrosion as the two samples before (Fig. 1d). For Klucel® GF (food grade, i.e. higher purity) hardly any corrosion is visible, only some small white spots can be found on the edge of the coupon (Fig. 1e). Raman measurements show the presence of litharge and plumbonacrite. Plumbonacrite can be differentiated from hydrocerussite by means of Raman spectroscopy. It is characterized by a strong band at 1048 cm−1 with two additional sub-bands at 1052 and 1056 cm−1 that are visible as shoulders [83]. Additionally, among other band shifts in the fundamental modes the OH stretching results in two bands for plumbonacrite, whereas there is only one band for hydrocerussite [83]. Klucel® GF Pharm (pharmaceutical quality) passes the test with no corrosion on the lead coupon (Fig. 1f). The coupon even kept its metallic gloss after testing. The results for Klucel® G show that the amount of corrosion depends on the purity grade, hence, not the product itself releases the harmful volatile components but an impurity or a leftover from the production process causes the problems. The producer states upon request, that acetic acid is used in the production process for neutralization and that the acetate is removed during purification. Unfortunately, there is no data available how much acetate is left in the products after the purification. However, it is not surprising that the acetate amount is the highest in the industrial grade product. Moreover, according to the producer, there was an expand in production capacity for Klucel® in 2016 that may explain the different results for the older and new Klucel® G of industrial quality.

Hydroxyethyl cellulose (HEC), ethyl cellulose (EC), methyl cellulose (MC)

Only few HEC and only one EC sample were included in this study. EC (Ethylcellulose ET 200 purchased in 17.11.2015) failed the Oddy test, generating significant amount of corrosion on the lead coupon. The corrosion phases include massicot, litharge, hydrocerussite and basic lead acetate indicating the presence of acetic acid. For HEC, 2 of 5 samples passed the test, whereas one got a T rating (Tylose® H 100000 YP2, purchased in 11.2016) and two failed the test (Natrosol 250 MHBR purchased in 05.03.2004, Hydroxyethylcelluose (unknown date of purchase) (Table 1). The corrosion phases of the lead coupon from Hydroxyethylcelluose (Fluka) are particularly interesting as they include not only massicot and plumbonacrite but also basic lead acetate and lead acetate trihydrate. The Raman spectra of these phase associations are shown in Fig. 2d, e. Both lead acetates are good Raman scatterers and a direct comparison of their spectra reveal the differences. Lead acetate trihydrate yields its typical bands at 464, 617, 657, 930, 1345, 1429, 2930 and c. 2980 cm−1 [84]. The 370 cm−1 band is completely missing and there are significant differences in the 600–660 cm−1 and the 910–950 cm−1 ranges respectively. Both lead acetates indicate the generation of acetic acid during the Oddy test. Most of the MC products passed the Oddy test. Four freshly purchased samples were tested, three of them passed the test with no sign of corrosion. The only fresh sample that got a T rating was Methyl cellulose (Sigma) as little corrosion occurred on the lead coupon. Among the fresh samples, special emphasis was laid on Methocel® A4M as it is the most common cellulose ether in conservation practice. It was tested as industrial grade and passed the test. The same result was obtained from historical equivalents from 1997 and 2003. Generally, 9 historical or undated samples were included in this study of which 7 passed the Oddy test. Two low viscosity grade products got a different rating. Metolose® SM-15 (Shin Etsu) generated slight corrosion on the copper and lead coupon, yielding a T rating. Raman measurements reveal the presence of cuprite; litharge and hydrocerussite as main phases. Culminal® MC 15 S is the only tested MC that truly failed the Oddy test as heavy corrosion occurred on the lead coupon. Cuprite, massicot, hydrocerussite, plumbonacrite and lead formate could be identified by means of Raman spectroscopy.

Methyl hydroxyethyl cellulose (MHEC), Methyl hydroxypropyl cellulose (MHPC), sodium carboxymethyl cellulose (Na-CMC)

Nine MHEC products were included in this study of which five were freshly purchased. The results show that MHEC products mainly pass the Oddy test. Only two of the nine samples did not pass the test with a T rating for Tylose® MH 50 (Hoechst) and Tylose® MH 30000 YP4 (Shin Etsu). Raman measurements on the copper and lead coupon of the historical Tylose® MH 50 sample reveal the presence of cuprite, litharge, hydrocerussite and basic lead acetate. The corrosion phases of the high viscosity MHEC Tylose® MH 30000 YP4 include cuprite; massicot, hydrocerussite and lead formate. Most of the MHPC products got a P rating as no corrosion occurred during the tests. Culminal® MHPC 20000 P was tested as fresh and historical (2005) sample and both passed the Oddy test. A temporary rating was assigned to two historical products. The Oddy test of Methocel® K4MFG from 2003 yielded slight corrosion on copper (cuprite) and lead (hydrocerussite). Methocel® E5 Premium LV from 2000 created slight corrosion on the lead coupon with hydrocerussite as main phase.

Eight Na-CMC products were tested in this study and most of them passed the test. The two freshly purchased samples (Cekol® 700, Blanose® ref CMC 7M65) passed the Oddy test with no sign of corrosion. Slight corrosion on the lead coupon was observed from three historical or undated products. Raman measurements reveal the presence of hydrocerussite (Blanose® 7LF), anglesite (PbSO4) (Blanose® 7H4Xf; cf. Fig. 2b) or massicot, hydrocerussite and anglesite (Tylose® C6000) as main phases. Anglesite yields a characteristic Raman spectrum with a main band at 976 cm−1 and some weak to medium intensity bands at 439, 451 and 609 cm−1 [88].

Conclusion

60 commercial cellulose ether products have been Oddy tested for their VOC activity. Raman measurements of the corrosion products identified oxides like massicot, litharge, cuprite and tenorite among carbonates (hydrocerussite, plumbonacrite) and acetates like basic lead acetate, lead acetate trihydrate as well as lead formate as main phases. The results show the strong need of testing conservation materials for their ability of emitting harmful volatile organic compounds. In all tested cellulose ether classes, we found products that do not pass the Oddy test, generating a T or even a F rating. From the overall point of view, methyl cellulose, Na-CMC, MHEC, and MHPC gave the best results, whereas HEC and HPC yielded some bad surprises. Hence, there is no clear answer for the raised title question as there are many “good” examples that passed the test but there are also several products which failed. The example of Klucel® G shows, that even though the composition of a product did not change in recent time, some changes in the manufacture process can increase a leftover from the process in the product that can release VOC during aging. The Klucel® G sample from 2017 is rated for permanent use, whereas a freshly purchased sample from 2021 got a T rating, as slight corrosion occurred on the lead coupon. A regular retesting of such commonly used materials is absolutely mandatory. On the other hand, the popular methyl cellulose Methocel® A4M proved to be a product for permanent use. Both samples, the fresh one from 2021 and the older one from 1997 passed the Oddy test with no sign of corrosion. A further combination of Oddy mass screening and direct emission analysis of selected products utilizing the BEMMA scheme may also yield fruitful results in other product classes relevant for conservation.

Future work will expand the testing procedure to other product classes relevant for conservation (e.g. acrylics, polyvinyl acetate, etc.). Moreover, more extensive comparisons between Oddy tests and BEMMA results will be performed.