The Gel, the Colour, and the Complexing Agent: An Investigation of a Rigid Gel Application of Benzotriazole for Verdigris-Damaged Paper

1.1 Verdigris

Verdigris as a substance is copper acetate, of which there are various forms (Scott 2002, 271). It is prepared by exposing copper plates to acetic acid. Verdigris as a pigment is put through processes such as dissolving in vinegar and allowing large crystals to form, or mixing with other ingredients such as the juice of plants, similar to other ways of creating pigments (Merrifield 1849, clxvii, ccxvii, cccxi, 66). It can turn brown when exposed to light, to heat, or when paired with other substances, causing many works to be kept in the dark (Scott 2002). However, there are some cases when it retained its colour quite well (Eastaugh et al. 2004, 391). The mercurial nature of the pigment has added to the challenges of researchers and conservators regarding original appearance and efforts to preserve its current state.

1.2 Damage Caused by Verdigris Pigment

One of the hallmark signs of verdigris possibly causing damage to the paper substrate is the pigment’s colour change. The pigment may turn from the original blue or green to more yellowish or brownish colours. The paper may also discolour beneath the area where the pigment was applied and cause discolouration onto the opposing page (Carlson 1997). The pigment does not necessarily change colour before discolouring the paper beneath it or vice versa. This can be seen in the top row of images in Figure 1. The top left image is a detail of the recto from the Languedoc map. There are green areas around the title and a large brown area in the bottom right corner. These two painted areas were confirmed to be copper-based pigments by X-ray fluorescence (XRF) analysis. The green areas around the title have not changed colour while the large brown area was likely once a green colour and now changed to brown. The top centre image is the verso of the same area. Where the pigment was applied on the recto, the paper has turned brown on the verso. This shows it is necessary to check both the recto and the verso for discolouration. While the discolouration of the pigment or the paper may not necessarily equate to damage, known damage seems to follow after this discolouration has occurred. This suggests that discolouration of the pigment or the paper might be indicative of damage.

Figure 1:

Examples of verdigris discolouration, paper discolouration by verdigris, and verdigris damage. Top row, details from the Languedoc map: Left, detail of recto showing discolouration of pigment; Center, detail of verso showing discolouration of paper; Right, detail of verso under UV showing the verdigris pigmented areas appearing darker than surrounding areas. Bottom row, detail from the Lancashyre map: Recto, area titled “Upper”, showing brittleness and detaching damage.

The degradation of cellulose by verdigris is not completely understood due to the many factors which affect each case. However, well-studied degradation processes caused by metal ions such as the Fenton reaction are discussed in depth in other research (Banik and Ponahlo 1982; Daniels 1989; Porck and Castelijns 1991; Selih et al. 2007).

Brittleness of the paper substrate where verdigris pigment was applied is a type of damage. The paper has lost some of its original structure and flexibility and is unable to bend without breaking. Detached areas are another type of damage. In areas where concentrated verdigris pigment was applied there can be some material loss as the chemical reactions have caused the paper to degrade completely. Particularly when there is a border around an area such as on a map, this can cause a piece to become detached from the rest of the paper. See Figure 1 for examples of damage in the bottom image. Where borders were executed with verdigris, some of the paper has become brittle and has detached.

Under ultraviolet wavelengths (UV), areas with verdigris may appear drastically darker than the surrounding paper or other pigments (see Figure 1, top row, right image for an example). UV may reveal darkened halos outside the areas where the verdigris pigment was applied. This may indicate that the copper ions migrated beyond where the pigment was applied and possibly affected the surrounding paper similar to other metal-based pigments such as iron gall ink (Reissland, Scheper and Fleischer, 2007). This phenomenon was also observed by Henniges et al. (2006) on artificially aged paper with copper pigment. UV light was used during the test series in order to ensure that both the painted area and the possibly migrated copper ions were treated throughout all the phases.

1.3 Treatment of Verdigris Damage

Due to the various types of degradation of paper caused by verdigris, different mechanical or chemical treatment options have been developed. The selection of a treatment option is always case specific. This study focussed on chemical treatment using the copper ion complexing agent BTA.

1.4 Benzotriazole

Benzotriazole was originally a copper inhibitor used in industry and then adapted for archaeology conservation in the early 1960s (Sease 1978). As mentioned earlier, BTA is a complexing agent and forms insoluble CuBTA when in contact with free copper ions. BTA was researched as a treatment option for paper items with verdigris damage by Ahn et al. (2014), which was expanded through findings published in the two subsequent years (Hofmann et al. 2015, 2016). It was found that BTA was more effective than some other copper ion inhibitors, but caused discolouration of paper in accelerated light aging tests especially when the concentration of BTA increased (Ahn et al. 2014). In Ahn et al.’s research various BTA concentrations from 0.5 to 3% w/v in ethanol were applied. A 3% BTA solution in 96% ethanol applied by brush proved to be the most effective treatment (Hofmann et al. 2016). However, it should be noted that the efficacy of BTA may be dependent upon the condition of the item and other influencing factors.

1.5 Gellan Gum Gel

Gellan gum is the product of a specific type of bacteria after fermentation (Kelco 2019). It has been used in food production, medicine, and pharmaceutical developments. Because of its properties, gellan gum has been adapted and applied to paper conservation for cleaning, backing removal, and stain reduction (Angelova et al. 2017). In order to evaluate if gellan gum gel is compatible with BTA, the gel must be miscible with the solvent for BTA and the pore size of the gel must also allow BTA to move through the matrix of the gel into the paper. Gellan gum gels have been mixed and used successfully with soluble chelating agents (Prestowitz 2017). This is useful to understand that a soluble substance was successfully put into a gel matrix and then migrated out of the gel matrix onto the paper. Previously, the chelating agent ethylenediaminetetraacetic acid (EDTA) was incorporated into gellan gum. EDTA is a larger molecule than BTA (Bonaldo 2015). This suggested that successful incorporation of BTA into gellan gum gel was possible. It should be noted again that BTA is a complexing agent and not a chelating agent to avoid confusion.

For this study, gels were applied to the verso of the samples, similar to the brush application method by Hofmann et al. (2016), in all test series in order to complex copper ions in the paper and inhibit further interacting. Application to the verso was carried out in order to limit negative effects such as possible unintended removal of printing or original pigments. The application of gels was not intended to remove free copper ions. Washing after the gel treatment was not considered due to the possible migration of free copper ions which may not have been complexed by the BTA.

There have been some concerns regarding gel residues remaining on the paper surface when using gels without subsequent washing treatment. A study conducted by Sullivan et al. (2017) showed that they do leave minimal residues. However, it should be noted that small amounts of residue may not harm the paper substrate as polysaccharides are similar to cellulose on a chemical level and they have a known stable aging behaviour (Sullivan et al. 2017). Additionally, in Sullivan et al.’s study (2017), it was found that sized papers had lower amounts of residue than un-sized papers.

This research was structured in 4 phases.

• Phase 1: testing varying concentration of gel.

• Phase 2: testing varying concentration of gel, BTA, and BTA/IMS solution.

• Phase 3: selection of the three best gel recipes from phase 2, further testing in order to choose one recipe for testing in phase 4.

• Phase 4: Comparison of brush and gel applications in situ.

2.1 Sample Preparation

Historical samples were used throughout all the phases because a realistic treatment-scenario result was to be gained and the abundance of material available. In order to ensure the choice for this type of sampling was ethical, the Icon Heritage Science Group’s Ethical Sampling Guidance was consulted (Quye and Strlič 2019). The author was not aware of a published method of artificial aging which mimics all the hallmark effects of aging on historical items with verdigris damage (pigment colour change without still being soluble, brittleness, etc.). Though many researchers have worked with artificially aged samples to study copper ion induced degradation of paper, an established method of artificial aging has not been developed yet. Using historical samples came with benefits. One of these benefits was the research reflecting a possible treatment scenario for a similar item, since the historical samples have had numerous factors affecting them. However, inconsistencies in the samples may have affected the results.

Through phases 1–3, samples from a historic map referred to as the Languedoc map were used. It is a mid-17th century French map made from handmade, laid rag paper approximately 109 g/m². Figure 2 shows areas used for samples (10 × 10 mm) in phase 1 and 2 highlighted in red and areas used for samples (30 × 20 mm) in phase 3 highlighted in blue. The presence of copper in areas showing signs of copper corrosion was confirmed using both XRF and Hulthe’s indicator paper.

Figure 2:

Sampling areas from Languedoc map. Red, phase 1 and 2; Blue, phase 3 samples.

In phase 1 and 2, one gel recipe was applied to three 10 × 10 mm sample areas. They were randomised as much as possible to avoid skewed results from unknown factors which may have affected specific areas of the pigmented region. This was done because of the unknown history of the map’s creation, storage, and unknown verdigris pigment recipe.

In phase 3, ca. 30 × 20 mm samples were cut from edges of the areas coloured by the verdigris. This was done in order to assess tideline formation possibly caused by various gel recipes. The bottom two-thirds of each sample (about 20 mm) was covered by verdigris paint, while the top third (about 10 mm) was not covered by verdigris paint.

For phases 1, 2 and 3, gels were applied on the verso of the samples with a piece of glass (∼16 g) and a light weight (∼106 g) placed on top for the duration of the treatment. For a schematic image and picture of the gel application in these phases see Figure 3.

Figure 3:

Schematic and picture of how phase 1, 2, and 3 samples were tested.

In phase 4, a second historical map, referred to as the Lancashyre map was used. It is a mid-17th century English map of handmade laid paper, about 125 g/m². The presence of copper in areas showing visible signs of copper corrosion were confirmed using XRF and Hulthe’s indicator paper. Two verdigris-damaged areas were chosen for the comparison of two application methods.

2.2 Gel

For phases 1–4, gellan gum gels were mixed at 2, 3, and 4% weight-to-volume ratio using deionised water. The concentrations were chosen based on concentrations used for paper treatments (Angelova et al. 2017). The mixture was then brought to boil until no further granules were seen.

The pH of each of the gels ranged between 6 and 7. The pH was recorded to provide information for possible future work as it has been found that adjusting the pH of gels may improve treatment results (Angelova et al. 2017). Adjusting the pH was not part of the scope of this research and therefore was not performed. The gels were used within 30 min after solidification.

2.3 Assessment Methods

In order to assess the effects of the various gel recipes, two colour-changing indicator papers were used, Hulthe’s bathocuproine reagent paper (Hulthe 1970) and MQuant™ Test Cu indicator strips. Hulthe’s paper was previously used in research on copper corrosion treatment by Hofmann et al. (2016). The indicator paper was created using the following instructions: Whatman 1 filter paper was cut into strips and dipped into 0.2 M solution of NaCl in deionised water. The paper was hung and allowed to dry for a minimum of 18 h. A solution was prepared of 100 ml of 50/50 vol acetone/chloroform and was mixed with 0.2 g hydroquinone and 0.15 g bathocuproine. The strips were then dipped into this solution and left to dry for 48 h. Every strip of paper was immersed the same way from the same mixed solution and only one solution was created in order to ensure as much uniformity among the strips as possible. It should be noted that mixing acetone and chloroform is potentially explosive and therefore was performed within an appropriate lab environment.

Hulthe’s indicator paper was used by dampening it with 1–2 drops of deionised water and blotting it to remove excess water. It was then gently, but firmly pressed onto the sample using a fingertip while wearing gloves in order to avoid contamination. It has been noted that changes in pressure by the hand among the samples could cause slight differences in results. If free copper ions are present, the indicator paper changes to an orange colour almost instantly, and the intensity of the colour varies depending on the amount of free copper ions present (Hulthe 1970). Overall, this is a practical, qualitative method to quickly determine if copper ions have been complexed by the BTA treatment by observing almost immediate visual differences of the indicator papers before and after treatment.

Because of the use of historical samples and to anticipate some variance in the amount of free copper ions in them, every sample had an initial reference piece of Hulthe’s indicator paper before treatment. After every treatment using the gel, each subsequent piece of Hulthe’s indicator paper was compared to the initial, before treatment paper and to unused Hulthe’s paper to judge the colour change. This was done to determine if a visually noticeable complexation of the copper ions had occurred. Hulthe’s indicator paper was used in all phases of this research. It should be noted the colour of unused Hulthe’s indicator paper was slightly yellow compared to pure Whatman paper and this hue varied very slightly among the indicator papers created.

MQuant™ Test Cu indicator strips were included to perform a semi-quantitative evaluation of the gel recipes with modified directions. It also provided corroborative data in assessing the effectiveness of Hulthe’s indicator paper. The MQuant™ Test Cu indictor strips were originally developed for water testing and have been adopted for use in paper conservation (Banik and Brückle 2018). As stated in Paper and Water (2018), MQuant™’s indicator strips should not be used in direct contact with an artwork since the coloured components of the strips will stain. Instead it should be used by placing a drop of water onto the item, soaking it up with filter paper and placing the filter paper onto the indicator strip. This method was tested a few times on the historical samples, and it gave inconsistent results. Instead, it was decided to place the indicator strip in direct contact with the sample. This was done by placing a drop of water onto the sample and pressing the indicator strip on the wetted area for no more than 2 s. This provided more consistent results.

The indicator strip did stain the samples and placing the water droplets directly onto the samples caused tidelines. This approach is not suggested when treating artefacts. The MQuant™ Test Cu indicator strips were only used in phase 3 of this research due to the limited number of them available at the time. It should be noted that both strips require the limited use of moisture, either directly onto the strip or onto a piece of Whatman paper, in order to interact with the sample and transfer copper ions. This may cause some migration of free copper ions in the item being treated.

3.1 Phase 1: Testing Pure Gellan Gel on Verdigris Areas

Control samples were treated with pure gellan gum gel mixed at 2, 3 and 4% of 20 ml deionized water in a 60 × 12 mm Petri dish. Each type of gel was tested on three samples.

3.1.1 Phase 1 Results and Discussion

It was found that for some samples, gellan gum gel may have slightly reduced the free copper ions present. However, the effect was negligible. This meant that there was little to no effect of the gellan gum gel itself on the samples’ free copper ions and that in the following phases the gross effects would be caused by BTA (see Figure 4).

Figure 4:

Phase 1 results of applying 2, 3, and 4% gellan gel onto samples from Languedoc map. The results showed negligible effect of the gel on the free copper ions in the samples.

3.2 Phase 2: Gel Recipes with BTA

The variables for gel recipes were gel concentration, complexing agent concentration, complexing agent/solvent solution concentration, and treatment time. In order to determine the concentrations of gel, complexing agent, and the complexing agent/solvent solution, two factors needed to first be determined 1) the baseline concentration of BTA necessary to complex the copper ions for phase 2 samples and 2) the volume of the gel to be used. For 1) BTA testing was performed on the Languedoc map samples. 1% BTA weight-to-volume in solvent was found to effectively complex the copper ions. 1, 2.5 and 5% concentrations of BTA were then chosen in order to see possible differences between different gels. For 2) because there is no previous research of treating items with BTA in a gel, 5, 12.5, and 25% of BTA/solvent: gel (volume-to-volume percentages) were chosen. This was done in order to be able to see differences between the various gels. In order to determine the amounts of BTA, solvent and gel for the chosen recipes, calculations were done for all the gel recipes as found in Table 1.^[1]

Since the amounts of the BTA and BTA/solvent solution are based on predetermined ratios, the recipes are presented as percentages in the following phases. Exact amounts will not be recounted since not every phase used the same volume of gel.

To avoid the BTA/solvent solution evaporating off the gel while it was being incorporated, the gel was allowed to cool until it reached temperatures between 50° and 55 °C. Visually, this was about 5–10 min after the steam stopped evaporating from the surface of the gel. After the BTA/solvent solution was stirred into the gel, the gel was then poured into a 60 × 12 mm Petri dish.

During initial testing of materials, ethanol was used as the solvent for BTA, in phase 2 the solvent was replaced by industrial methylated spirits (IMS). Industrial methylated spirits are a denatured alcohol which is about 95% ethanol and about 5% methanol (SciChem 2014). This was considered an appropriate adjustment due to the better accessibility, the amount of gel samples and the precedent of use of 96% ethanol for BTA treatment (Hofmann et al. 2016).

The samples were tested using Hulthe’s indicator paper at the following intervals: before treatment (untreated), 30 s, 1 min, 3 min, 5 min, 10 min, 15 min, 20 min, and 30 min. There was one more test a minimum of 2 h after the 30-min treatment time to observe if there were copper ions which were not complexed. Table 1 displays the gel recipes below.

3.3 Phase 3

The three gel recipes from phase 2 considered the most effective were chosen based on the criteria listed in Table 2. They are numerically ranked by desired outcome. The lower the total score the better the gel recipe. Additionally, the criteria are listed in descending priority. For example, it is more important to use a BTA concentration as low as possible than to use a gel that it easy to handle during application. See Table 2 for the comparison of the gel recipes from phase 2. Gels with the best properties are highlighted in bold.

After three gel recipes were selected, they were compared against one another in a new series of testing. Each gel was applied to five samples from the Languedoc map in the blue highlighted areas (Figure 2). The gels were applied on the verso of the samples on the damaged areas with about 3–4 mm overlapping onto the non-verdigris area to ensure that possible migrated copper ions would be complexed. A piece of glass and two light weights were placed on top.

The gel recipes were compared by criteria listed in the comparison criteria for Table 3. They are numerically ranked by desired outcome. The lower the number the better the gel recipe (1 = best properties). Additionally, the criteria are listed in descending priority. Table 3 displays the compared recipes with the best recipe highlighted in bold.

The interval of testing with the indicator strips for both the Hulthe’s indicator paper and the MQuant™ Test Cu indicator strips was adjusted to before treatment, 5 min, 10 min, 15 min and 20 min. This change was due to the fact that in phase 2, no significant changes could be observed between results before treatment and after 3 min, or between 20 and 30 min. Therefore, it was concluded to test at 5-min intervals instead. When using MQuant™ Test Cu indicator strips, a drop of water was placed directly onto the sample every time, saturating the sample area with water. Therefore, tideline formations could no longer be observed, and this criterion was no longer considered appropriate.

3.4 Phase 4

The purpose of this phase was to compare BTA applications by brush and gel. The most effective gel recipe from phase 3 was compared to Ahn et al.’s (2014) method of BTA application by brush. The BTA gel was applied to the area designated “Upper”. BTA was applied by brush to the area designated “Lower”. Both of these areas were chosen because they were similar in visual appearance and damage such as brittleness and areas of loss. Since this was an in-situ testing of the gel and not a real-time treatment, decisions regarding placement of the gel were made for consistency and comparison of results to previous phases. The gel was not applied to the entire pigmented region, since a previous tideline was visible in this area. In order to reduce the number of variables for treatment and assessment afterwards, the tideline was avoided, and the gel was placed on the area highlighted in red only (Figure 2). This was the area with the most homogenous appearance. Only the area outside the verdigris-pigmented section was evaluated for tidelines, since in a treatment scenario the entire verdigris-pigmented area would be covered by the gel.

The gel was 5 mm thick and cut about 5 mm larger than the verdigris pigmented area to accommodate damage beyond the pigmented area. It was applied on the verso in the following arrangement: object – gel – Melinex^® barrier – piece of glass placed – two light weights. The gel was left to interact for an initial 20 min which was the amount of time needed to complex copper ions (see phase 3). If ions were not fully complexed after this time, the gel could be reapplied at 10 min intervals until the copper ions were effectively complexed.

When applying BTA with a brush, Hofmann et al.’s (2016) method was followed. The BTA solution was applied with a synthetic 3/0 sized brush to the verso in as many coats as needed to effectively complex the copper ions. The two methods of application were evaluated based on the following criteria:

1. Overall effectiveness at complexing free copper ions by monitoring three specific regions (labelled Lower 1, 2, and 3 and Upper 4, 5, and 6) on the verso. Each area was tested before, during and after the total treatment time with Hulthe’s indicator paper.

2. The overall effect on Upper and Lower during treatment:

   • Distortion

   • Tidelines (excluding the anticipated tideline in Upper caused by the gel in the homogenous pigmented area)

   • Disturbance of media

3. The amount of time to perform the treatment.

4. Changes in overall appearance measured by a spectrophotometer. Spectrophotometer measurements were taken before and after treatment in chosen areas of Upper and Lower on both the recto and the verso. Three areas were chosen for an average reading before and after treatments to determine the overall colour change, dE*ab.

3.5 Gel Application on Upper

Recipe 3 was applied on the verso for 20 min and tested by Hulthe’s indicator paper and then reapplied for three 10-min intervals until there was slight to no detection of free copper ions, for a total of 50 min of application time.

Before treatment there were no areas of loss to the paper carrier of the Lancashyre map. However due to an accidental bump of the map while being moved from one lab space to another, areas that were vulnerable became detached. This occurred particularly in Upper as the entire verdigris-damaged area detached from the map. This damage was not caused by the gel treatment.

1. The treatment was effective in complexing the free copper ions. This is indicated by the reduced intensity of the colour of the indicator strips, comparing the untreated area (before treatment) to the treated are (after 50 min treatment) (see Figure 6).

2. After the Upper was fully dried, tidelines were apparent in various areas outside the pigmented area (excluding the anticipated tideline in Upper caused by the gel in the homogenous paint area) (see Figure 5).

3. The treatment lasted for a total of 50 min.

4. There were changes in the overall appearance. Visibly the colour shifted from a brownish to a greenish hue. The spectrophotometer showed an overall difference of 4.67 dE*ab on the recto and 6.59 dE*ab on the verso. Differences noticeable to the naked eye must be dE*ab = 1 or greater. The readings were taken as an average of three measurements, one in each of the three designated areas in Upper. This visibly different appearance may have been caused by the reintroduction of moisture. However, it should be noted that this phenomenon did not occur in the samples taken from the Languedoc map. Hofmann et al. (2016) did not note colour changes of the paper when BTA was applied by brush. Additionally, the gel drew out some paper discolouration from the area as expected. The data from the spectrophotometer for Upper is found in Table 4. The gel removed some of the yellow/brown discolouration from the verso of the paper. This can be seen in Figure 5, where the normally clear gel exhibits yellow/brown areas.

Figure 5:

Lancashyre verso, detail of Upper. Comparison of before (left) and after (right) gel treatment. The area where the gel was placed is highlighted by the red dotted line. It was decided to place the gel in this area because of homogenous verdigris-discolouration due to taking spectrophotometer data and monitoring the complexing of the free copper ions in various areas by the indicator paper order to avoid potentially skewed data from original tideline and beyond. The original tideline can be seen in both images. The new tideline directly to the right of the gel on the verdigris-discoloured area was anticipated and does not reflect the assessed comparative effects of a gel placement for a treatment scenario. The inset image is of the gel after treatment showing the removal of some yellow-brown paper discolouration products.

Figure 6:

Results of phase 4 testing. On the right are images Upper and Lower verso, marking the areas where the indicator paper was consistently applied to monitor the complexing of free copper ions.

3.6 Brush Application on Lower

The 2.5% BTA in BTA/IMS solution was applied three times to the verso using a 3/0 synthetic brush.

1. The treatment was effective in complexing the free copper ions. This can be seen in the reduction of the colour intensity of the indicator paper from the untreated (before treatment) to the final (after 3 coats) in Figure 6.

2. There were no visible negative effects to the area during treatments.

3. The treatment time lasted for 45 min. Each application took about 10 min and was left to dry for 5 min before testing and applying the next coat.

4. There were slight changes in the overall appearance as measured by the spectrophotometer. However, there were unforeseen factors to the treatment and the measurements. Similar to the colour change observed following gel application, when the slightly humidified Hulthe’s indicator paper was placed in repetitive contact with specified areas of Lower, those areas turned slightly green.

5. When the spectrophotometer was placed on those same areas, the resulting data no longer reflected the effects of the application method. Instead, the resulting spectrophotometer data would rather reflect the changes caused during application of the indicator paper. This showed one of the risks in using Hulthe’s indicator paper, in that the humidity required to use the paper may alter the appearance of the item. Colour change from using the Hulthe’s indicator paper was not seen in the samples from the Languedoc map. Two other areas not touched by the indicator paper were chosen for spectrophotometry which were visibly negligent in colour difference, but these better reflected the effect of the brush application method. The spectrophotometer showed an overall difference of 2.63 dE*ab on the recto and 2.66 dE*ab on the verso. Application of BTA with a brush did cause colour changes visible to the naked eye, but these are less pronounced than those caused by the gel application. The data from the spectrophotometer for Lower is shown in Table 4.