Rigid Solvent-Gels in Paper Conservation: A New Approach to Sticky Problems

ABSTRACT

Pressure-sensitive adhesives (PSA) present in pressure-sensitive tape or self-adhesive labels applied on paper pose substantial problems for the long-term conservation of the substrate because they degrade upon ageing, leaving embedded stains. This paper presents the development of a method to prepare agar-based solvent-gels with high organic solvent, and low water content. The objective is to provide conservators with a versatile, cheap, and easy solution to make solvent-gels for PSA removal, to be implemented as sole treatment or included in a multi-step treatment protocol. The treatment method proposed in the case-study aims at removing aged, rubber-based adhesive residues from the printed paper cover of a book. It includes the use of silicone-based solvent cyclomethicone D5 as a treatment-aid applied just before the application of the solvent-gel. The efficiency of D5 to prevent the formation of stains and tide-lines during local gel treatment is demonstrated, and potential side-effects are also pointed out.

ZUSAMMENFASSUNG

Auf Papier aufgebrachte druckempfindliche Klebstoffe (PSA) in Haftklebebändern oder Selbstklebeetiketten stellen erhebliche Probleme innerhalb des langfristigen Substanzerhalt dar, da sie sich im Zuge der Alterung zersetzen und Flecken bilden. In diesem Beitrag wird die Entwicklung einer Methode zur Herstellung von Lösungsmittelgelen auf der Basis von Agar mit hohem Gehalt an organischem Lösungsmittel und niedrigem Wassergehalt beschrieben. Ziel ist es, Restaurator*innen eine vielseitige, kostengünstige und einfache Lösung zur Herstellung von Lösungsmittelgelen für die PSA-Entfernung aufzuzeigen, die sowohl als alleinige Behandlung als auch in eine mehrstufigen Behandlung angewendet werden können. Die in der Fallstudie vorgeschlagene Behandlungsmethode zielt darauf ab, gealterte, kautschukbasierte Klebstoffreste von dem bedruckten Papiereinband zu entfernen. Die Studie umfasst den Einsatz des silikonhaltigen Lösungsmittels Cyclomethicone D5, welches unmittelbar vor dem Auftragen des Lösungsmittel-Gels eingesetzt werden kann. Die Wirksamkeit von D5 zur Vorbeugung einer möglichen Bildung von Flecken und Schmutzrändern während der lokalen Behandlung wird dargestellt und mögliche Nebenwirkungen aufgezeigt.

KEYWORDS:

• Solvent-gel
• agar
• cyclomethicone D5
• ATR-FTIR analysis
• pressure sensitive tape adhesive removal

SCHLÜSSELWÖRTER:

• Lösungsmittelgel
• Agar
• Cyclomethicone D5
• ATR-FTIR Analyse
• Entfernung druckempfindlicher Klebestoffe

Introduction

General issue

Every library, archive or museum collection has paper objects repaired with pressure-sensitive tape (PST), all-purpose glue, or identified with self-adhesive labels. These synthetic adhesives are ubiquitous and pose substantial problems for the long-term conservation of paper heritage because they generally degrade with time, leaving stains embedded in the paper.

Conservators faced with the problem of removing synthetic adhesives from paper often rely on organic solvents. One solution is to use solvent-gels: gels applied precisely to the area to treat can act as a reservoir to slowly deliver a selected solvent in a controlled fashion. As an added benefit, a small quantity of organic solvent confined in a gel lowers the risk of personal exposure and spillage.

Gels are mostly composed of liquid but they do not flow because their liquid portion is held in an internal network structure that confers to the gels some properties of a solid. Most gels used in conservation are hydrogels: their network structure is hydrophilic. Gels can have a soft, paste-like texture, easy to spread with a brush; or they can have a rigid, gummy bear-like texture, easy to cut into slabs of specific shape with a scalpel. In both cases, the liquid portion of the gel conveys the primary active treatment ingredient to the surface to be treated.

Rigid gels are advantageous compared to soft gels because they can be applied directly to the area to be treated and then removed in one motion with little risk of gelling material residues embedded in the paper. Used in combination with or in lieu of already existing synthetic adhesive removal methods, solvent-gels would be a valuable addition to the paper conservator’s tool-kit. Unfortunately, commercially available gel formulations can only be combined with a narrow range of solvents, and self-made gel formulations are geared towards the use of water-based treatment solutions. Implementing solvent-gel conservation treatment also presents some risks for the paper support, such as unwanted solubilization or migration of solvent-sensitive additives or media, and the formation of tidelines.

Objectives of the study

This study is about the formulation and application of rigid agar-based solvent-gel for the treatment of paper. The aim was to identify combinations of agar-based gels and polar and non-polar solvents that release no or minimum amounts of water when in contact with paper. The solvent-gel preparation method outlined here for the first time successfully replaces most of the water present in agar hydrogel with low or high-polarity solvents: the method is simple, versatile and effective. It is a useful advancement for the treatment of paper and could be of great interest for the treatment of other water-sensitive surfaces.

When used with aqueous solutions, the capillary network of the agar gel ensures treatment effectiveness: capillary transfer of water-soluble compounds from the paper to the slab of hydrogel applied onto the paper surface promotes cleaning. Replacing water with organic solvents in the gel, without compromising its capillarity, produces a gel capable of removing solvent-soluble matters from the paper by simply applying solvent-loaded agar gels onto the paper. The choice of agar as gelling material also responds to the goal of developing treatment methods accessible to private conservators. To that end, a simple preparation process requiring minimal equipment and versatility are paramount. The aim is to propose conservation solutions ultimately available to conservators operating on limited budgets to complement existing solvent application methods.

Preparing solvent-gels made of agar or agar combined with PVA-Borax can be achieved by adding organic solvents to the agar-based hydrogels before gelation, during the cooling phase. Pieces of hydrogel can also be soaked in organic solvents to promote solvent exchange: water migrates from the gel to the soaking solution while organic solvents ingress the slabs of gel. However, the first method results in a potentially undesirable large portion of water in the gels, and the second in unknown solvent content. In particular, soaking agar hydrogels in low-polarity solvents might result in insufficient ingress of soaking solvent in the gel because of its low miscibility with water. It can also affect the gel shape: the soaking solvent might extract water from the gel without replacing it, hence altering the gel’s structure. In this study, direct observation, handling and use of the gels led to optimizing the introduction of organic solvents in the gel matrix. In the method devised to prepare agar-based solvent-gels, ethanol was used to ease the ingress of lower polarity solvents in the gel. An experimental protocol was designed to test the role of alcohol on the ingress of low-polarity solvent in gels and to mitigate the effects of these solvents on the shape of the gels. The proportion of solvents in the gels was estimated with attenuated total reflection Fourier-transform infrared (ATR-FTIR) spectroscopy. This method is fast and easy to apply and has sufficient accuracy to distinguish the different preparation protocols of the gels.

Local treatment of paper objects comes with the risk of unwanted formation of tidelines and stains. Devising local treatment application strategies to mitigate risks without compromising efficiency is complementary to developing new products. The proposed application method includes the pre-treatment with a cyclic volatile methyl siloxane, thereafter referred to as cyclomethicone D5 or D5, applied on the paper substrate to mitigate the risk of tidelines. Using D5 as a treatment-aid raises questions related to efficiency, and potential long-term risks for paper. Two observation-based experiments addressed these questions. In the first one, photography under UV radiations and visible light before and after artificial ageing of paper samples provided image-based evidence of D5’s efficiency to mitigate the risk of stain when applying rigid gels on pure cellulose paper. The second test aimed at assessing the effects of D5 on the affinity of paper for water.

A case-study illustrates the use of agar-based solvent-gel to remove residual adhesive from pressure-sensitive tape along the spine of a book that could not be treated with a suction table.

Literature review – what are the limits of existing solvent application methods for the removal of PST adhesive from paper?

Solvent-gels: a complement to other PST adhesive removal methods

The removal of pressure-sensitive tape from paper has been a topic in paper conservation literature since the 1980s. In their 1983 article, Smith described and classified different solvent application methods: immersion, poultice, rolling and suction, and advocated that the most effective treatment might derive from the use of several techniques on the same object (Smith et al., 1983). Forty years onwards, immersion in solvent is not commonly implemented anymore, but combining poultices, suction and swabbing is still regarded as essential to lessen the risk of side effects while increasing treatment efficiency. In particular, the accretion of adhesive from the surface can be carried out by alternating between swabs (or rolling), poultice and suction disk (Müller et al., 2022). Treatment protocols can also include the application of solvent vapour using a solvent chamber, as a preparation step to soften surface adhesive accretion before mechanical removal or application of a solvent-gel (Lenning, 2010). Subsequent treatment steps to remove adhesive absorbed in the paper can be carried out by alternating between a poultice and suction disk (Müller et al., 2022). The object typology can limit the range of possible solvent application methods: for instance, suction cannot be implemented for the removal of adhesive from a multi-layered paper support, or from the spine of a book.

Possibilities and limits of existing solvent-gels for the treatment of paper

Paste-like solvent gels

Stemming from poultices, the use of soft but highly viscous cellulose ether gel combined with solvent has been used for the removal of aged pressure-sensitive tape adhesive (Lennig, 2010). Lenning selected a high-viscosity, low-adhesion cellulose ether, cetyl-hydroxyethylcellulose (Polysurf ^TM 67 CS) that was first mixed with water, before introducing organic solvents in the gel with a pipette. By selecting a cellulose ether that can be dispersed directly in the pure organic solvent, such as Klucel H, it is possible to prepare a water-free solvent-gel (Dupuy, 2012).

The major drawback in using paste-like gels like cellulose ethers, even highly viscous ones, is the risk of gel residues embedded in the paper fibres or sticking to its surface. A barrier layer is necessary to prevent residues (Warda et al., 2007; Lennig, 2010) but lessens treatment efficiency. Using rigid gels alleviates this risk.

Rigid solvent-gels

Rigid polysaccharides such as gellan gum, agar and agarose combined with water-based solutions have been used in paper conservation for decades (Iannuccelli & Sotgiu, 2010; Maheux, 2015; Petrella et al., 2016; Hughes & Sullivan, 2016; Van Dyke, 2004). Their combination with organic solvents can pose miscibility or opacification problems (Scott, 2012). Scott combined agar¹ with organic solvents during the cooling phase of gel preparation when the gel was still in liquid phase, and noted the tendency of lower polarity solvents to ‘destabilize the colloid mixture’. This was mitigated by adding ethanol, a high-polarity solvent. Tangelin prepared 2% agar solvent-gels using 8:2 volumetric proportion of water to solvent (Tengelin, 2017).² While ethanol was successfully introduced in the gel, lower polarity solvent like ethyl acetate (EA) could not be integrated into the gel despite extensive stirring.

In their characterization of alcogels prepared with low-acyl (LA) gellan gum and ethanol or propanol, Cassanelli observed that the addition of alcohol to the hydrogel hardens its texture (Cassanelli et al., 2017). This could be detrimental to establishing good contact with the slightly irregular surface of some papers.

Adjusting the hardness of rigid gels

Combining rigid and paste-like gelling material to obtain gels with desirable qualities of both types can help improve contact between a rigid gel and the treated surface, while avoiding residues. The rigid texture of gellan gum LA can be softened by the addition of a small proportion of softer high-acyl (HA) gellan gum (Richard, 2023). Commercially available Nanorestore gels are prepared with different types and proportions of polymers to adjust the texture, elasticity, and retentiveness of the gels (CSGI Nanorestore gel® commercial website 2023).

Water-soluble poly (vinyl alcohol) (PVA) combined with a solution of borate ion forms a highly viscous solution of poly (vinyl alcohol)-borax, or PVA-B gel: this gel flows to adopt the irregular shape of artwork s’ surfaces and sticks to itself to remove potential gelling residues after application. PVA-B’s affinity for organic solvents increases as the degree of substitution of hydroxyl groups (–OH) by lower polarity acetate groups (–OCOCH₃) increases on the polymer’s backbone (Angelova, 2017).

El Eman used agarose as an additive to paste-like PVA-Borax gels in order to adjust the texture to lessen risks of gel residues when cleaning antique murals (Al-Emam et al., 2020). Their approach inspired the addition of a small proportion of PVA-B to agar gels in order to increase the conformability of the solvent-gel to textured paper, while avoiding the risk of gel residues on treated surfaces.

Emulsions as an alternative to pure solvents in solvent-gels

Using oil-in-water emulsions instead of pure solvents overcomes the difficult integration of low-polarity solvents in high-polarity hydrogels. The affinity of the gel matrix for water facilitates the penetration of dispersed micro-droplets of organic solvent in the continuous water phase. Gorel used low-polarity petroleum ether dispersed in water, using a surfactant and pentanol to obtain a stable and transparent oil in water micro-emulsion. They were able to make agar gels containing up to 1.5% weight of low polarity solvent petroleum ether (Gorel, 2010).

Along the same lines, commercially available Nanorestore Cleaning® (CSGI Nanorestore Cleaning® commercial website 2023) microemulsions and nanostructured fluids contain water, alcohol, surfactant and low-polarity solvents. Case-studies of application on paper artefacts demonstrate the efficiency of these emulsions for the removal of PST adhesives (Mirabile et al., 2020). The solvent-in-water emulsion (or nanostructure fluid) can be introduced in rigid gels commercially available or in studio-made hydrogels (Al-Emam et al., 2019).

The presence of water in solvent-gels remains a drawback for the treatment of water-sensitive surfaces. On paper, local application of water can cause visible tidelines upon ageing, by the accumulation of water-soluble products at the wet-dry interface (Jeong et al., 2012). Water contained in gels poses similar risks (Barbisan, 2018). Additionally, emulsions can leave residual surfactants, which may locally increase the affinity of the paper for water.

Water-free or water-poor solvent-gels

Organogels

The use of water-free gels, or organogels, would alleviate the risk of water exposure on paper. Mirabile used Di-Ethyl Carbonate in a water-free poly (ethyl methacrylate gel (PEMA-DEC)) (Mirabile et al., 2020). However, the preparation of PEMA organogel requires materials and equipment beyond the reach of most paper conservators operating in private practice. The use of ready-made organogels from Nanorestore® is reported (Udina, 2018), however, these crossed-linked vegetal oil-based gels are not yet commercially available (CSGI Nanorestore gel® commercial website 2023).

Solvent-rich hydrogels

The method developed in this study to prepare water-poor, solvent-rich, agar gels using the limited equipment of a conservation studio was inspired by Cassanelli’s study on the effects of ethanol on LA (low-acyl) gellan gum gel structure (Cassanelli et al., 2017). They compared the shrinkage of LA gellan gum hydrogels after water in the gel was replaced by ethanol or propanol. The gradual introduction of alcoholic solvent in the gel by soaking in increasingly concentrated solvent and water solutions (25%, 50%, 80% and finally pure alcohol) leads to moderate shrinkage (∼13%), while hydrogels soaked directly 100% alcohol shrank by ∼50%. The volume and the shape of the hydrogel were retained with the gradual alcohol treatment. Gels structured similarly are likely to display similar results: the gradual increase of alcohol to replace water in the gel was successfully applied to agar-based gels in the present study. Once treated with alcohol, gels were soaked in solvents selected to solubilize PST adhesive.

Solvents for PST adhesive removal

Graphic representation of solvent’s characteristics

In practice, solvents are selected for their ability to dissolve material to remove after solubility tests are carried out. In the case of PST adhesive removal, the type of adhesive and, in the case of rubber-based adhesive, the oxidation stage, would influence the choice of solvent.

The Paper Conservation Catalog (O’Loughin & Stiber, 1992) discusses choosing solvents using their location on the Teas Triangular Solubility Chart, a simplified graphic representation of solvent’s cohesive energy forces.

The selection of solvents based on their physical properties is the underlying idea of an alternative graphic representation that classifies solvents according to their polarity and vapour pressure (Zumbühl, 2019). Solvents classified as belonging to the same group regarding their polarity will be considered ‘slow’ or ‘fast’ at dissolving a substrate according to their vapour pressure.

Practical application for conservation treatment

Consequently, solvents could be selected according to the duration of application for a given treatment. If solvents are applied through the paper onto a suction table, they are in contact with the material to dissolve for few seconds at most: choosing solvents with high dissolving rate (high vapour pressure) is adequate. Conversely, applying solvents by means of a gel lasts longer, several minutes to half an hour or more: fast or slow action solvents can be selected. In practice, if solvents from the dipolar group (IV) are selected for PST adhesive removal treatment, methyl ethyl ketone (MEK) might be applied with a solvent-gel, while acetone would be more suitable for suction table application.

Silicone-based solvents used as a barrier layer

Benefits in practice

The Paper Conservation Catalog advocated the use of slow evaporating solvents such as mineral spirits (US) or white spirits (UK) to use as a barrier in the area surrounding the treatment area to ‘contain lateral movement and tidelines caused by solvents with higher polarity’ (O’Loughin & Stiber, 1992). The current use of low-polarity, low-surface tension silicone-based solvents derives from the same principle. Cyclomethicones D4 and D5³ were introduced in the 2010s as an ingredient in microemulsions used in painting conservation (Stavroudis, 2012), before being used as a barrier solvent or temporary masking agent in paper conservation (Sullivan et al., 2014; Köhler et al., 2021). Silicone solvents such as D5 are seen as chemically inert and selected for treatment precisely because they are not expected to react with or solubilize any materials on the surface of paper. When used as a barrier solvent, Köhler observed that D5 did not produce any tideline under visible light or UVA radiation, and could significantly diminish the formation of tidelines caused by the application of droplets of acetone and ethanol–water mixture on various papers after artificial ageing.

Potential effect on health and the environment

How safe are cyclomethicones for the user and the environment? In 2016, the Committee for Risk Assessment of the European Union (EU) classified both D4 and D5 as very persistent and very bio-accumulative. Moreover, D4 was also classified as toxic, with a risk of reprotoxic effects (Scientific Committee on Consumer Safety, 2015). Health risk combined with the fact that D5 has a longer evaporation time than D4 geared conservators in Europe towards the use of D5. Growing concerns regarding the environmental impact of human activities on the waterways have led the EU legislator to ban the use of cyclomethicones in cosmetic wash-off products since 2020 (Official Journal of the European Union, 2018). Although conservation’s environmental impact is at a much smaller scale than the cosmetic industry, applying D5 as a masking agent on an artwork before immersion in water could result in releasing minute amounts in the waterways. When used as a barrier solvent to lessen the risk of visible side-effect of local solvent application, D5 dissipates in the air, potentially contributing to atmospheric pollution. Multiple cycles of oxidation are necessary to significantly degrade highly volatile cyclomethicones in atmospheric conditions (Alton & Browne, 2022). The environmental impact of these persistent chemicals and their oxidation by-products in the air is not yet addressed by EU laws, however, legislation currently in preparation aims to limit their access, use and release in the environment in the coming years. Derogations would be granted for specific fields including art conservation⁴ (Annex to the Commission Regulation, 2023). Although the prospective EU regulation would allow art conservators to use D5 in the future, environmental concerns should prompt the profession to moderate their use.

Materials

Materials and equipment to make gels

Agar, PVA Mowiol 88, borax decahydrate (99.9% pure mineral borax salts), ethanol (99.8%), methyl ethyl ketone MEK (100%), ethyl acetate EA (>99.5%) and cyclomethicone D5 (decamethylcyclopentasiloxane 90–100%), as well as glassware were supplied by Labshop (Labshop commercial website 2023).

Stock solutions of Mowiol 10% (w/v) and 6% borax (w/v) in deionized water were prepared to make Agar-PVA-B gels.

Microwave MN203S was supplied by Coolblue (Coolblue commercial website 2023).

A supporting video available online shows how to make and use agar-based solvent gels: https://youtu.be/cU8ThkAbCh0.

Paper samples

Pure cellulose Whatman paper was supplied by Sigma Aldrich (Sigma Aldrich commercial website 2023).

Machine-made paper, 80 gsm (77% mechanical pulp 23% chemical pulp, alum-rosin sizing with starch surface sizing), and machine-made paper, 78 gsm (100% sulphite pulp, no additive) were supplied by the Cultural Heritage Agency of the Netherlands (RCE)

Archival materials from the 1930–40s with unknown fibre and sizing content, were donated by the Instituut voor Oorlogs-, Holocaust- and Genocidestudies (NIOD, Amsterdam, the Netherlands). A newspaper, two types of office copy paper, and a manuscript letter were selected.

100% Kozo, 31 gsm Japanese paper was donated by the Conservation Studio of the University of Leiden Libraries (UBL, Leiden, the Netherlands).

Instrumental

Paper samples were artificially aged in a Weiss ClimEvent chamber using one-hour fluctuations of relative humidity from 50% to 90% at a constant temperature of 80°C for a week.

A Video Spectral Comparator VSC8000/HS Foster + Freeman illuminant was used for digital photography under UV 254 and 365 nm.

Images of samples and treatments were obtained using a Canon Eos 5D Mark III camera equipped with UV filters and an iPhone 12.

Gels and paper samples were weighed using an analytical balance with a precision of 0.0001 g (RCE)

ATR-FTIR spectra were measured on a Perkin-Elmer Frontier spectrometer equipped with a Pike Technologies diamond GladiATR module. Spectra were collected as an average of 4 scans at 4 cm⁻¹ resolution. For measuring liquids, a cap was placed immediately over the solvent droplet on the ATR-crystal to minimize evaporation.

Scanning Macro X-Ray fluorescence imaging was carried out using a Bruker M6 Jetstream. The Rh anode X-Ray tube was operated at 7.5 W (50 kV, 199 µA) and a dwell time of 120 ms/pixel and a pixel size of 100 µm. The data was processed making use of PyMCA and Datamuncher.

Methods

Making agar-based solvent-gels

A three-step gel-making procedure was developed in order to replace water initially present in hydrogels with organic solvents (Figure 1). A concentration of 5% w/v gelling material was selected because gels made in this proportion are easy to prepare, cut and handle.⁵

• Step 1: gelling material (1 g agar) was dissolved in deionized water (20 mL) using a microwave oven.

• Step 2: after the gel was cast and cooled to its rigid form in a glass petri dish, strips of gel were cut and soaked into successive ethanol/water baths with increasing concentration in ethanol (25%, 50%, 75%, and 100%).

• Step 3: the gel was then transferred to other types of organic liquids (MEK or EA).

Figure 1. Three-steps process to prepare agar-based solvent-gels. The gradual addition of ethanol (step 2) as a co-solvent facilitates the ingress of methyl ethyl ketone MEK or ethyl acetate EA during Step 3.

Three frames of text arranged horizontally with arrows in between summarize each of the three steps to prepare the gels, starting on the left with step 1, heat up the gelling material and water.

A modified approach consisted in starting to introduce ethanol during the cooling phase.

• Step 1: 1 g of agar was dissolved in 15 mL water using a microwave oven.

• Step 2: 5 mL ethanol was introduced and stirred in vigorously during the cooling phase, before casting, in order to obtain a gel containing 75% water and 25% ethanol as a liquid portion. Strips of gel were cut and soaked into successive ethanol/water baths with increasing concentration in ethanol (50%, then 75% and 100%) to progressively replace water by ethanol.

• Step 3: the gel was then transferred to other types of organic liquids (MEK or EA).

Another modification consisted of replacing a third of the agar by PVA-B in order to obtain agar/PVA-B gels made with water and ethanol.

• Step 1: 0.7 g agar, 0.3 g PVA (3 mL of a 10% solution of PVA in water), and 0.036 g borax (0.6 mL a 6% solution of borax salts in water) were dissolved in 11.4 mL water using a microwave oven.

• Step 2: 5 mL ethanol was introduced and stirred vigorously during the cooling phase, before casting, in order to obtain an agar/PVA-B gel containing 75% water and 25% ethanol as liquid portion. Strips of gel were cut and soaked into successive ethanol/water baths with increasing concentration in ethanol (50%, then 75% and 100%) to progressively replace water by ethanol.

• Step 3: the gel was then transferred to other types of organic liquids (MEK or EA).

Optimisation of the preparation of slabs of gels

In order to decide how much ethanol to introduce during the cooling phase without negatively affecting the appearance and mechanical characteristics of the gels, homogeneity and flexibility of the gels were assessed. The homogeneous appearance of the gels was observed directly by placing freshly made gels in front of a black-and-white background. The mechanical soundness and flexibility of the gels were assessed by placing strips of gel across the edge of a table.

Proposed solvent soaking sequence

Soaking slabs of gels in solvent promotes liquid exchange: water migrates from the gel to the soaking solution while organic solvents ingress the soaking slabs of gel. The three-step process to prepare solvent-gels proposes soaking pieces of gel solutions of ethanol and water with a gradually increasing proportion of ethanol to progressively raise their alcohol content before soaking them in 100% ethanol and finally in EA or MEK (Figure 2).

Figure 2. Proposed organic solvent treatment for slabs of gels made of pure agar or agar/PVA-B, combined with pure water or a mixture of water and ethanol. Preparing the gels with water and ethanol was determined to be applicable to both pure agar and agar/PVA-B gels. Using 100% water was only applicable to pure agar gels. The soaking time indicated was applied to pieces of gel ∼3 mm thick. Thicker gels would need to soak for a longer time period.

Slabs of gels containing water and ethanol or pure water, and a series of five jars containing, respectively, 25%, 50%, 75%, 100% ethanol, and lastly MEK or EA, are aligned horizontally and linked by arrows going from one jar to the next, indicating the proposed solvent-gels preparation protocol in successive solvent baths.

Before starting the soaking sequence, slabs of gel should be cut in convenient shape, such as strips matching the width of the PST adhesive to remove, and short enough to fit at the bottom of the solvent-soaking glass jars. Solvent should largely cover the pieces of gel, which should not be stacked together. Giving the tightly closed containers an occasional gentle stir keeps the slabs of gel apart.

As an indicative preparation process, soaking 3 gel strips measuring 10 × 30 × 3 mm in 40 mL liquid at the bottom of 5 cm diameter glass jars would allow all surface area of the gels to be exposed to the liquid. In terms of duration, 2.5 h in each solvent bath is sufficient for a 3 mm thick gel. This time can be extended to overnight or a couple of days.

Characterization of agar-based solvent-gels

Variation of the solvent-soaking sequences

An experimental protocol was designed to test the role of ethanol used as a co-solvent to ease the penetration of EA or MEK in the gels.

A series of six solvent-gel preparation methods were tested, assessed and compared: the methods considered best were the ones causing the least distortion or shrinkage of the gel, and which have the highest concentration of organic solvent in their liquid portion. The shrinking rate of gels was measured by weight ratio. The measurement of the proportion of different solvents in the gel was carried out using ATR-FTIR spectroscopy.

Of the six preparation methods, only three methods starting from gels made with 75% water and 25% ethanol were applied to agar/PVA-B gels. For each method described Table 1, pieces of gels measuring 1–2 cm² were soaked for 2.5 h in each solvent bath, following a specific solvent soaking sequence. All sequences ultimately included MEK or EA in order to assess the role of polarity on the solvent ingress in the ethanol-loaded gels.

Table 1. Agar and agar/PVA-B (in bold) solvent-gel preparation methods and nomenclature.

Type of gel	Numbering	Initial liquid content	Solvent soaking sequence
Agar	EA_1 or MEK_1	100% water	1 = 100% ethanol; EA or MEK
Agar	EA_2 or MEK_2	75% water, 25% ethanol	2 = EA or MEK
Agar/PVA-B	EA_2 or MEK_2	75% water, 25% ethanol	2 = EA or MEK
Agar	EA_3 or MEK_3	75% water, 25% ethanol	3 = 100% ethanol; EA or MEK
Agar/PVA-B	EA_3 or MEK_3	75% water, 25% ethanol	3 = 100% ethanol; EA or MEK
Agar	EA_4 or MEK_4	100% water	4 = 25%, 50% ethanol; EA or MEK
Agar	EA_5 or MEK_5	100% water	5 = 25%, 50%, 75%, 100% ethanol; EA or MEK
Agar	EA_6 or MEK_6	75% water, 25% ethanol	6 = 50%, 75%, 100% ethanol; EA or MEK
Agar/PVA-B	EA_6 or MEK_6	75% water, 25% ethanol	6 = 50%, 75%, 100% ethanol; EA or MEK

Notes: Solvent-gels tested in this study are named using the type of gel (agar or agar/PVA-B), type of solvent ultimately loaded in the gel (EA or MEK), and solvent soaking sequence (1–6).

ATR-FTIR analysis

After the sequence of solvent soaking, water, ethanol, and EA or MEK were present in unknown proportions in the solvent-gels. The proportion of water and organic solvents contained in the liquid portion of the gels was assessed using ATR-FTIR spectroscopy. All data processing was carried out using custom scripts in Wolfram Mathematica software. Details of the data analysis method and example spectra can be found in Appendix.

Weight loss ratio

Changes to the shape of the gels were directly observed by comparison of changes in thickness, corresponding to the shrinkage of the gel. Shrinkage was evaluated by proxy by measuring the weight loss ratio: $\mathrm{Weight}\mathrm{loss}\mathrm{ratio}={\displaystyle \frac{\mathrm{Weight}\mathrm{of}\mathrm{gel}\mathrm{after}\mathrm{solvent}\mathrm{treatment}}{\mathrm{Weight}\mathrm{of}\mathrm{gel}\mathrm{before}\mathrm{solvent}\mathrm{treatment}}}$Each value was calculated as an average for three pieces of gels treated simultaneously with the same solvent-soaking sequence. The weight difference between before and after solvent treatment was due to two factors:

• The actual structural shrinkage affecting the shape: a shrunk gel holds less liquid than an intact gel, hence weighs less than before shrinkage.

• The replacement of water and ethanol by lighter solvents. In the absence of structural shrinkage, the mere replacement of liquid initially contained in the gel by less dense solvents causes a loss of weight. This expected weight loss was calculated using the experimental ATR-FTIR solvent fractions measurement and solvent density data.⁶

Application of solvent-gels to paper using cyclomethicone D5 as a pre-treatment

The proposed solvent-gel application included the pre-treatment of the treated paper surface with cyclomethicone D5. The combined application of D5 and a solvent-gel to remove PST adhesive from the paper promotes the migration of liquids to the gel rather than across or through the paper porous network (Figure 3).

Figure 3. Influence of pre-treatment using D5 on liquid migration during gel application on paper. In A, direct application of solvent-gel on the porous paper network results in partial cleaning and migration of residual adhesive in the paper. In B, the pre-treatment with D5 to saturate the paper’s porous network favours migration of solubilized matters towards the gel, resulting in better cleaning of the paper surface.

The two possible application set-ups of solvent-gels on paper (A or B), are schematically represented with light blue rectangles of gels applied on the adhesive stain embedded in a piece of paper underneath. The gels are covered with polyester film. Liquid migrations during treatment are represented with three colors of thin, linear arrows: blue (evaporation of liquid from the gel), dark red (migration of solubilized adhesive to the gel), and light red (migration of solubilized adhesive to the paper). The two settings differ by the number of light red arrows: in the presence of D5, fewer red arrows indicate less migration of the solubilized adhesive towards the paper than in the absence of D5.

Assessing the efficiency of D5 as a treatment aid to limit the migration of liquids during solvent-gel application

The efficiency of D5 to limit the side-effects of local gel treatment on paper was assessed by comparing tidelines due to the application of rigid gels directly, or after pre-treatment with D5.

Filter papers were prepared according to the description in Figure 4 to compare the effects on the support of different gels applied directly or after locally applying D5 with a brush on filter paper. The 5% agar gels were prepared with pure water, or water and organic solvents. The solvent-gels were made either by adding organic solvents during the cooling phase or by soaking slabs of hydrogels in solvent baths to gradually replace most of the water. Ultimately the amount of water in the gels ranged from 100% to 70% and 60% when organic solvents were added during the cooling phase, to less than 10% for the gel prepared following the three-step solvent-gel protocol. The four different gels applied on the filter papers contained the following liquid portions:

• Pure water

• 70% water and 30% ethanol (in volume) stirred in the warm, gelling solution before it cools down to form a rigid gel

• 60% water, 20% ethanol and 20% benzyl alcohol (in volume). The alcohols were mixed together before being stirred in the gelling solution during the cooling phase.

• Over 70% MEK, 20–25% ethanol, and less than 10% water. The gel was prepared following three-step solvent gel protocol to gradually replace water by ethanol, then ethanol by MEK.

Figure 4. Preparation template of the filter paper samples, divided in quadrants to delineate the application area of 5% agar gels with different liquid portion. The liquid portion of each gel is detailed next to the quadrant on which it is applied. The water content marked in bold varies from 100% (bottom right), to 70% (top right), 60% (bottom left) and finally less than 10% (top left).

A plain line circle represents a piece of filter paper. Vertical and horizontal dotted lines run through the center, dividing the paper in four quadrants. A square designates the area of gel application in each quadrant.

A piece of transparent polyester film was used to cover the gels during 5 min of application time. After gels were removed, the filter papers were left to air dry completely on a piece of polyester film. One filter paper sample from each series (direct application or use of D5) was artificially aged. Aged and unaged samples were photographed under visible light, and two wavelengths of UV radiations (365 and 254 nm) using a Video Spectral Comparator. Images obtained were compared visually to assess the role of D5 to limit the stains induced by gel application.

Short and long-term effects of D5 on naturally and artificially aged paper

The risk of short or long-term side-effects of the application of D5 on paper, namely visible stains and modification of the affinity of paper for water, was assessed by observing a variety of paper samples on which D5 was applied. Observations were carried out before and after artificial ageing.

D5 was applied with a brush on seven different paper samples:

• Four archival documents from the 1930s and 1940s

• Two late twentieth-century machine-made papers of known composition

• Early twenty-first-century Japanese paper (100% kozo, 31 gsm).

On all samples, D5 was applied alone, or as a pre-treatment before application of ethanol or water in free, liquid form, water held in a hydrogel, and solvent-gel loaded with MEK. All samples were photographed under visible light and UV 365 nm, before and after artificial ageing. Specific areas of interest were photographed under UV 254 nm. Additionally, paper samples previously treated with D5 alone were immersed in water to assess the effect of D5 on paper’s affinity with water.

Results

Influence of ethanol on homogeneity and flexibility of the gels

Introducing ethanol in agar-based gels affected the homogeneity and flexibility of the gels. The agar/PVA-B gel prepared with pure water lacked homogeneity: the addition of a small amount of ethanol during the cooling phase overcame this effect (Figure 5). Increasing the proportion of ethanol tended to increase the gels’ flexibility, however adding too much decreased homogeneity, resulting in a tendency for the gels to break under their own weight (Figure 6).

Figure 5. The effects of introducing increasing amounts of ethanol to agar and agar/PVA-B gels during the cooling phase was visualized by placing petri dishes with freshly casted gels on a black and white background. Ethanol opacified both types of gels. With pure water, agar/PVA-B gel displayed white specks that decreased (gel B) and disappeared (C, D, E) with addition of ethanol. Surface bubbles visible on gel C did not compromise the homogeneity of the bulk of the gel. Higher proportions of ethanol (D and E) caused the agar/PVA-B gels to set faster and rigidify before covering the entire surface of the petri dish.

Two series of five petri dishes (A to E) containing 5% gel are aligned from left to right on a black and white background. The top row contains pure agar and the bottom row agar/PVA-B. The proportion of ethanol increases from 0% (gel A) to 12.5% (B), 25% (C), 37.5% (D) and 50% (E).

Figure 6. Gels were prepared with an increasing proportion of ethanol (from A to E: 0, 12.5%, 25%, 37.5% and 50% v/v). Strips of gels were cut, handled, and placed across the edge of a table to assess their mechanical properties. The further gels bent, the more flexible or softer they were.

Strips of gels A to E are placed across the edge of a table are viewed from above. Pure agar gels A, B, and C and agar/PVA-B gels B and C present good strength and flexibility.

Based on the observation and handling tests results, agar gels A (100% water), agar gel C and agar/PVA-B gel C (75% water, 25% ethanol) were selected as starting points from which to pursue with solvent-gel preparation by soaking in solvent solutions.

Assessing the accuracy of the ATR-FTIR measurements to determine solvent fractions

The gels were composed of 95% (w/w) liquid, thus the ATR-FTIR spectra were dominated by solvent signals, with only very weak bands corresponding to the gel matrix. Therefore, each gel spectrum could be approximated as a weighted sum of ATR-FTIR spectra corresponding to each of the pure solvents (water, EtOH, EA or MEK), obtained by least-squares fitting. The weight factors of each spectral contribution could then be used to obtain solvent volume fractions. This approach assumes a constant measurement volume and ignores possible shifts or intensity variations in vibration bands due to interactions between solvent molecules. However, these effects tended to be prominent only in the OH stretching region above 3100 cm⁻¹, and had only a minor effect on the outcome of the fit. Details of the data analysis approach and example spectra can be found in Appendix.

The accuracy of this approach was assessed by measuring a series of tertiary liquid solvent mixtures with known composition, and calculating the error margin. Nearly three-quarters of the 39 measurements were within 0–5% error margin, the rest was between 5% and 10%, except for one measure that had a 12% error. This method of measuring solvent fractions was therefore considered sufficiently accurate to compare the solvent compositions in different gels. The results of ATR-FTIR analysis are presented in Table 2. Due to the accuracy limitations, small differences of less than 5 units in the solvent proportion of the gels should not be considered as significant. The results of the assessment of the accuracy of the method are presented in Appendix (Figure A1, A2, and Table A1).

Table 2. Values and weight loss ratios for solvent-gels loaded with EA (top) and MEK (bottom).

	FTIR			Weight loss ratio
	etOH	EA	Water	Expected	Measured
agar_EA_1	25	68	7	0.88	0.82
agar_EA_2	17	1	83	0.97	0.73
agar-PVAB_EA_2	15	3	82	0.97	0.72
agar_EA_3	26	71	3	0.88	0.90
agar-PVAB_EA_3	29	67	4	0.87	0.89
agar_EA_4	25	1	73	0.94	0.45
agar_EA_5	25	72	3	0.88	0.85
agar_EA_6	27	71	2	0.87	0.89
agar-PVAB_EA_6	20	77	3	0.88	0.88
	FTIR			Weight loss ratio
	etOH	MEK	Water	Expected	Measured
agar_MEK_1	20	75	5	0.81	0.76
agar_MEK_2	19	56	25	0.85	0.83
agar-PVAB_MEK_2	12	67	20	0.83	0.78
agar_MEK_3	28	69	3	0.81	0.83
agar-PVAB_MEK_3	25	72	3	0.81	0.82
agar_MEK_4	21	60	20	0.85	0.79
agar_MEK_5	24	74	3	0.82	0.79
agar_MEK_6	25	73	2	0.80	0.82
agar-PVAB_MEK_6	23	76	1	0.80	0.82

Notes: ATR-FTIR values are the average of two measures per gel; measured weight loss ratios are the average of three measures per gel (complete set of data presented in Appendix Table A2).

Effect of solvent-soaking sequence on the structure and solvent-content of the gels

The values of expected and measured gel weight loss presented in Table 2 is colour coded in green when the gap between both values is 2% or less, yellow between 3% and 10%, and red for 10% and above, corresponding to heavily distorted and structurally shrunk gels. The low shrinkage of EA-loaded gels prepared using protocols number 3 and 6 could be attributed to solvent density rather than structural shrinkage affecting the gel matrix. In contrast, Protocols 2 and 4 led to structural shrinkage of the gels. This result underlined the importance of 100% ethanol as a preparation step to obtain low-shrinkage agar-based gels with high EA content. The shrinkage rate measured for MEK-loaded gels indicated little structural shrinkage following protocols 3 and 6, however, none of the tested protocols resulted in heavily shrunk gels.

Figures 7 and 8 show the combined results of weight loss ratios and solvent portion measurements using ATR-FTIR. The highest columns correspond to the smaller weight loss. For each solvent-gel, the measured solvent portions are represented in blue (water), green (ethanol) and dark red (EA or MEK). The solvent-gels corresponding to the highest columns with the most EA or MEK and the least water are considered more desirable.

Figure 7. Weight loss ratios and solvent portion of gels prepared with EA according to solvent soaking sequences. With protocols 3 and 6, gels displayed smaller weight loss, combined with low amounts of water and high amounts of EA. 25% alcohol gel soaked directly in EA (protocol 2) contained mostly water. Gels soaked in 50% alcohol before soaking in EA (protocol 4) displayed a bigger weight loss and contained mostly water. In both protocols 3 and 6, gels were prepared with 75% water and 25% ethanol. Gels were soaked in pure ethanol before EA for protocol 3, and in progressively increasing concentration of ethanol (50%, 75%, then 100%) before the last bath in EA for protocol 6.

Each EA-loaded gel listed along the horizontal axis corresponds to a stacked bar. The total height indicates the weight loss ratio. Within each bar, the solvent portion is colour-coded.

Figure 8. All solvent soaking sequences tested allow the ingress of MEK in the gels, however, the water portion remained relatively high (20–25%) for the gels prepared following protocols 2 and 4. As for EA-gels, MEK-gels made following protocols 3 and 6 displayed the smaller weight loss combined to high proportion of MEK and little water.

Each MEK-loaded gel listed along the horizontal axis corresponds to a stacked bar. The total height indicates the weight loss ratio. Within each bar, the solvent portion is colour-coded.

Efficiency of D5 to limit the migration of liquids during local rigid gel application

D5’s efficiency to limit the formation of stains due to local gel application on paper was tested on pure cellulose filter paper, selected in order to limit the potential sources of water or solvent-soluble components coming from the substrate. After application of the gels, under visible light, none of the unaged filter paper samples showed tidemark stains. Upon ageing, a stain is very likely due to the migration of cellulose degradation products⁷ formed in the area of contact between paper and 100% water with and without D5. The 70% water–30% ethanol gel caused a stain only when applied directly.

Under both UV radiations, direct gel application resulted in areas of fluorescence expanding beyond the area of contact of paper and gel. The bright outer lines corresponding to the wet-dry interface are likely to become visible with time.

The MEK-loaded gel (top left quadrant) left a very faint fluorescence perceptible under UV 254 nm when applied directly. With D5 there was no detectable trace of MEK-loaded gel application, even under UV radiations after ageing. Similarly, D5 lessened the UV-induced fluorescence caused by the gels containing water or a mixture of water and alcohol (Figure 9).

Figure 9. Stains left by 5% agar gels containing different proportions of water viewed under visible light (bottom row), UV 365 nm (middle row) and 254 nm (upper row). To the left gels were applied directly, and to the right D5 was applied with a brush as a pre-treatment before gel application. For each series, unaged and aged samples are presented side to side. The stars adjacent to samples viewed under visible light indicate quadrants with gel-induced light brown stains visible on the sample but hardly visible on the printed images. White dotted circles indicate a stain that is not significant for the study (artefact of artificial ageing).

Photographs under different lighting conditions of non-aged and aged filter papers on which 4 different gels were applied. UV lighting reveals stains that are not visible before aging, and hardly visible after aging, under visible light.

These results illustrate the efficiency of D5 in limiting lateral liquid migration and the intensity of UV fluorescence caused by local gel treatment. The intensity of the fluorescence under UV 365 or 254 nm seemed related to the amount of water in the gel: water seems to be the main factor of long-term risk of stain after local gel treatment on paper.

Possible side-effects of D5 on paper

When used as a pre-treatment to solvent-gel application on paper containing solvent-soluble sizing or media (i.e.: alum-rosin size, printing ink), D5 did not prevent local solubilization of these solvent-sensitive compounds. Stains due to their migration were visible before artificial ageing. D5 attenuated but did not prevent the effect of water on water-sensitive media or paper additive (i.e.: starch surface sizing). These observations were coherent with a previous study on the limits of D5 used as a fixative for soluble media during wet treatment (Köhler et al., 2021).

Once evaporated from the paper surface, D5 did not leave any detectable trace under visible light, before or after ageing, on any of the seven tested papers. However, in one of the mid-twentieth-century office papers, the area of D5 application displayed fluorescence under UV 254 nm (Figure 10). After artificial ageing, the overall increased fluorescence of the paper sample rendered the local UV fluorescence induced by D5 imperceptible.

Figure 10. Template of the test implemented on the seven paper samples (left); one of the NIOD donated mid-twentieth-century office paper sample under visible light (left), and under two UV wavelengths: 365 nm (middle) and 254 nm (closeup, right). Under UV 365 nm, darker traces left by water or solvent gels were visible. Under UV 254 nm, additionally to these dark marks the outer fluorescent D5 evaporation line was visible.

Three side by side photographs show an overall view of one of the paper sample under visible light and UV 365 nm. A close-up view of area of interest is viewed under UV 254 nm.

Kozo fibres Japanese paper samples treated with D5 were left to air dry for 48 h before immersion in water. While the papers absorbed water readily, the D5 evaporation line remained dry, denoting a change of wettability in that area. After 16 months of natural ageing, repeated immersion resulted in the same effect (Figure 10). A similar effect was observed on Japanese paper samples artificially aged after the application of D5. The local hydrophobicity of the paper along the evaporation line of D5 was only observed on the Japanese paper samples. The four papers from mid-twentieth-century archives and the two machine-made papers displayed homogeneous wettability upon immersion in water.

Mapping XRF analysis of paper samples on which D5 had been applied did not detect any silicone residues on any of the samples, in the detection limits of the equipment used for such a light element. The hydrophobicity of the Japanese paper visible upon immersion in water is therefore unlikely to be due to remaining by-products of D5. It could be due to the migration and local accumulation of Japanese paper degradation products. The potential causes of this side-effect were not explored further.

Analysis

On the preparation of agar-based solvent-gels

The nature of the gelling matrix, pure agar or agar/PVA-B, did not significantly influence the solvent content: similar solvent-soaking sequences lead to similar solvent contents for protocols 2, 3 and 6 that were applied to both types of gels. Soaking gels in low-polarity solvent EA only resulted in solvent ingress if the gels were first gradually soaked in 100% ethanol. The higher polarity solvent MEK penetrated gels containing 25–50% ethanol, however, the water content remained quite high (20–25%) in these formulations. Higher MEK and lower water contents were obtained after gradual exposure of the gels to ethanol before soaking in MEK. Gels made with 25% ethanol could be soaked in 100% ethanol, or step by step: 50%, 75%, then 100% ethanol, without significant difference between both. The water initially present in the gel was replaced with ethanol, which was then partially replaced by the chosen tape removal solvent (EA or MEK). The resulting solvent-gels contained a relatively large portion of ethanol: the liquid portion of gels made using protocols 3 or 6, considered more desirable with regards to shrinking ratio and solvent content, was composed of 20–29% ethanol.

During the course of this research, concerns were raised regarding the classification of ethanol as carcinogenic by the European Chemicals Agency (ECHA) which might result in a difficult supply for conservation studios (Weller & Phan Tan Luu, 2022). Consequently, weight loss ratio tests were repeated using isopropanol (IPA) as a substitute for ethanol (Table A3 in Appendix). Similar ratios were measured when implementing a progressive increase of alcohol content as described in protocols 3 or 6. Protocol 1 (direct immersion in alcohol of the gel made with 100% water) resulted in further weight loss when using IPA instead of ethanol.

On the efficiency and side-effects of cyclomethicone D5

The comparison of the effects of gels applied on pure cellulose filter paper either directly or after pre-treatment with D5 illustrated the benefit of D5 to lessen or prevent the formation of tidelines. The presence of water in the gels increased risks of tidelines perceptible as a fluorescence under UV 365 nm before artificial ageing, which can become visible under visible light after artificial ageing. The filter paper selected purposely for this experiment is a very porous substrate on which the application of D5 as a treatment-aid visibly limits the formation of stains due to local gel application. Less porous papers would be less prone to the formation of stains in the absence of D5. Moreover, the filter paper samples were air-dried after gel application, and no attempt was made to mitigate the potential side-effects of the treatment. The conservation treatment of a paper object might include extra steps, such as drying the treated area on suction table in order to quickly remove the liquids, and limit the risk of local stain on the paper substrate.

The potential changes induced by D5 were the fluorescence discerned under UV 254 nm on one historical, mid-twentieth-century machine-made paper sample (Figure 10), and the local modification of paper’s affinity for water on kozo Japanese paper (Figure 11). After artificial ageing, there was no visible change due to D5 on any of the samples. Nevertheless, these observations pose the question of long-term risks induced by the application of D5 on paper. The outcome of long-term natural ageing could differ from observations made after artificial ageing. The formation of a local hydrophobic line around D5 application area could compromise the outcome of future wet treatment, for instance, an overall wash by immersion could potentially leave that line unwashed. Although no visible trace was detected on any of the paper samples over the course of this study, the risks of D5 side-effects becoming visible in the long-term cannot be assumed. However, the benefit of D5 application as a pre-treatment prior to local gel application might outweigh these risks.

Figure 11. Japanese paper samples immersed in water after local application of D5 with a brush displayed a dry ring around the application area, still visible after 16 months of natural ageing. The dates indicate the days pictures were taken.

The same samples of Japanese paper on which D5 had been applied are photographed immersed in water on 25 June 2021 and on 30 October 2022.

Investigating mechanisms leading to the observed changes is beyond the reach of the present study. Conservators need to decide on a case-by-case basis if the beneficial effects of D5 to reduce the migration of liquids through and across paper during gel application overweigh the risks described above.

Case study: removal of aged, PST adhesive from the spine area of a children’s book

A paper-bound children’s book previously repaired with PST around the spine area was selected as a case-study to demonstrate the efficiency of using agar-based solvent-gels for the removal of aged, PST adhesive residues. This object was chosen because its typology limits the choice of possible solvent application methods to remove or lessen the adhesive. The application of solvents through the paper using a suction disc is not possible, and local treatment can only be carried out through from the front. Besides solvent-gel, other possible application methods would be free solvent (for instance with a cotton roll) or poultices (for instance using clay powder as poulticing material). However, free liquid solvent would potentially cause the adhesive stain to migrate across or deeper in the paper, and clay poultices may leave residues of powder which would need to be removed as an extra treatment step. Applying solvent-gel directly, without application of D5 beforehand, could be possible but presents the risk of migration of the stain beyond its current shape, and through the substrate. Given the porosity of the paper and board underneath, the application of D5 onto the substrate to locally and temporarily saturate the porous network lessens this risk.

The discolouration of the adhesive stain, the fact that it had penetrated the substrate, and the relative ease with which the carrier layer could be lifted led to assess that the adhesive was probably rubber-based in the advanced oxidation stage. After mechanically lifting the carrier layer of the tape, solubility tests showed that the residual adhesive layer was soluble in both MEK and EA. Some of the inks were readily soluble in MEK but displayed very little solubility in EA. EA was therefore preferred as an active ingredient in the gels.

5% agar gels (w/v) were prepared following protocol agar EA-6 described in the methods section (see Table 1).

After the application of D5 with a brush, the 20 min application of squares of EA-agar solvent-gel along the spine was repeated 2–3 times to remove the old PST adhesive. Gels were covered with a piece of polyester film to slow down solvent evaporation during treatment (Figure 12, top middle row). After treatment, there was no planar distortion, no lifting or cockling of the layered cardboard and paper support thanks to the quasi-absence of water in the system. Upon application with a brush, D5 formed a temporary oily-looking stain that dissipated completely after few hours. D5 successfully promoted the migration of dissolved substances to the porous agar solvent-gel, leaving no tidelines behind (Figure 12, right column, photos under visible and UV 365 nm taken before and after solvent gel application).

Figure 12. Experimental treatment of ‘Vadertje Langbeen’ (children’s book) showing successful removal of old PST adhesive with minimal disruption of the printed coloured inks. Agar-EA gels prepared according to protocol 6 were applied for a total duration of 40 or 60 min after pre-treatment with D5.

To the left, the overall book’s front cover is shown with a dotted-line rectangle indicating the treated area. The two center images show the treatment in progress with squares of solvent-gels during application, covered with a polyester film (top), and the gels’ aspect just after application (bottom). Four images on the right show an area of detail before and after solvent-gel treatment under visible light and UV 365 nm.

Conclusion

This study resulted in the development of a three-step protocol to make rigid agar-based solvent-gels using material and equipment easily available to conservators. The collaborative approach between conservator and conservation scientists lead to the formulation of solvent-rich, water-poor agar and agar/PVA-B gels, and their characterization in terms of liquid content.

The versatility of the method was tested by modifying the preparation steps: the progressive increase in ethanol is a crucial step before loading the gels with MEK and EA. Using acetone and DEC as alternative solvents to MEK and EA is possible but the resulting solvent portion was not measured.

A case-study illustrated the possible use of three-step solvent-gels in paper conservation. Using D5 as a treatment aid helps to obtain visually satisfying results. While changes to the paper due to D5 were occasionally noticed under UV 254 nm and upon immersion in water of paper samples previously treated with D5, the effects remained invisible after artificial ageing. It was therefore considered that the benefits largely balanced the potential drawbacks.

After preparation, MEK or EA-loaded agar-based gels contain around 20–29% ethanol. In this study, no attempt was made to lessen this proportion. The principle of progressively soaking gels in ethanol-EA or ethanol-MEK mixtures containing increasing amounts of EA or MEK could be applied for the third step of the gel-making procedure as an attempt to reduce the concentration of ethanol in the gels.

Acknowledgements

We would like to thank Leila Sauvage (UvA, Rijksmuseum) for carrying out MA-XRF analysis, and Butcher Walsh for the realization of the video on making solvent-gels.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was financially supported by Metamorfoze, the Netherlands’ national programme for the preservation of paper heritage, hosted by the Koninklijke Bibliotheek (National Library of the Netherlands). The Cultural Heritage Agency of the Netherlands (RCE) kindly hosted the research.

Notes on contributors

Françoise Richard

Françoise Richard After graduating in 2005 with a MA in Conservation from Paris 1 Panthéon-Sorbonne University, Paris, France, Françoise Richard worked for private and institutional Book and Paper conservation studios in the US, Europe and the UK. In 2017, she moved to Amsterdam, Netherlands, where she joined the Rijksmuseum conservation team. In 2019, she started her own private practice. Two years later she joined the department of Conservation and Restoration at the University of Amsterdam with a part-time lectureship while continuing to provide conservation services to Dutch cultural institutions.

Joen J. Hermans

Joen J. Hermans Having graduated with a BSc and MSc degree in chemistry at Utrecht University, Joen Hermans obtained his PhD in 2017 at University of Amsterdam studying the mechanisms of metal soap formation in oil paintings. After five years of post-doctoral research at the University of Amsterdam focusing on infrared spectroscopy methods and the behaviour of water in oil paints, he now holds a position as assistant professor in Conservation Science at the Conservation & Restoration program and at the chemistry institute of the same university. Since 2017, he also works as a researcher at the Conservation & Science department of the Rijksmuseum.

Lora Angelova

Lora Angelova is a heritage science and conservation researcher based in Berlin, Germany. She is currently working with Scientific Analysis of Fine Art. Lora was previously Head of Conservation Research at The National Archives, UK, a conservation science researcher at Tate, and a Newton Fellow at the Material Studies Laboratory in University College London. She holds a PhD in chemistry from Georgetown University in collaboration with the National Gallery of Art, Washington, DC.

Notes

1 Agar was selected over gellan gum because during the cooling phase, agar gels become rigid at a lower temperature than gellan gels. The addition of organic solvents at room temperature during the cooling phase speeds up the system’s temperature drop, potentially compromising the gel homogeneity if the gel starts to set before the solvent becomes fully integrated in the liquid phase. A gel with lower setting temperature is therefore preferable.

2 2 g agar powder in 80 mL deionized cold water with later addition of 20 mL solvent (Tengelin, 2017)

3 At room temperature cyclomethicone D5 has a surface tension of 18 mN/m to be compared with water: 72 mN/m.

4 At the time of writing, prospective EU legislation would restrict the market of Cyclic Volatile Methyl Siloxanes. Derogations would however be granted for ‘D5 for professional use in the cleaning or restoration of art and antiques’.

5 Characteristics of agar from other suppliers may differ so conservators might find that other proportions are more suitable. Testing what works best by making agar hydrogels in different proportion is recommended.

6 The density of organic solvents and water at 20°C is available on Wikipedia (https://en.wikipedia.org). The following values in g/mL were used: ethanol 0.789; EA 0.902; MEK 0.805; water 1.

7 To confirm that the stain was not a result of the solvents, tests were carried out using neat solvent on the filter paper. These samples did not show stains. There is a secondary possibility, that the stains are caused by gel components migrating with the solvent front. This was not explored further in the current study but is worth further investigation.

Appendix. Estimating solvent ratios in gels using ATR-FTIR spectroscopy

Solvent ratios in the gels were estimated using ATR-FTIR spectroscopy. A set of pure solvent spectra (MEK, EA, EtOH and water) were collected to serve as reference spectra, as well as spectra of solvent mixtures with known composition for validation. All data analysis was carried out using custom Wolfram|Mathematica scripts. While there were only very minor variations in background signal for these solvent and gel samples, all ATR-FTIR spectra were baseline-corrected using a statistics-sensitive nonlinear iterative peak clipping (SNIP) method. Examples of corrected spectra of the pure solvents are given in Figure A1. These spectra show that there are several unique spectral features for each solvent, which facilitates reliable and automized estimation of solvent fractions.

Figure A1. Baseline-corrected ATR-FTIR spectra of four solvents.

Each spectrum of a solvent mixture or gel was approximated as a linear combination of the four pure solvent spectra, using a non-negative least-squares (NNLS) algorithm. Modeling the spectra with this method assumes that the measurement volume remains approximately constant between samples, and ignores potential shifts or intensity variations in vibration bands due to interactions between solvent molecules or between solvent and gel molecules. An example of a modeled ATR-FTIR spectrum is shown in Figure A2.

Figure A2. Example of a fit of an ATR-FTIR spectrum of a loaded solvent gel, and the fitted model based on three solvents.

The example in Figure A2 demonstrates that the method yields a reasonable match between modeled and true spectra. The main region where discrepancies occur is the O–H stretch vibration region (3100–3600 cm⁻¹). This broad band, in this sample caused by the present of EtOH, is known to be quite sensitive to the concentration of hydrogen-bonding molecules in solution, and therefore the band is shifted towards higher wavenumbers in solutions than contain relatively minor EtOH or water concentrations. However, the exclusion of this spectral region from the analysis had no significant effect on the resulting solvent ratios, which indicates that there are enough other features in the spectra to yield a reliable fit (as demonstrated by the results on solvent mixtures with known composition).

In our method, the weight factors for each solvent spectrum in the linear combination model correspond to volume fractions for each solvent. Finally, to correct for small variations in contact area between samples and ATR crystal, the solvent volume fractions for each sample were divided by the sum of solvent volume fractions.

Table A1. Tertiary solvent mixtures were prepared by measuring known quantities of solvent by weight.

Notes: Each mixture (A to P) was analyzed using ATR-FTIR. The table to the left presents the experimental results of solvent portion measurement, using the weighed sum of FTIR spectra. The central table presents the expected proportion of solvent from weighing components of each mixture during the preparation of the solvent mixtures in the lab. The table to the right presents the difference between both set of data, marked in green for a difference comprised between 0 and 5, orange between 5 and 10, and red above 10. Mixture P was excluded because of a mistake when measuring the solvent weights. Mixture I and J were excluded because of a mistake in the treatment of the FTIR data. 13 sets of data were considered, with 3 results per set, so 39 in total.

Table A2. Weight measurement and rate of weight loss for gel samples before and after solvent soaking, following preparation protocols 1–6.

Measurements November–December 2022
Pure agar	Agar_EA_1			Agar_EA_2			Agar_EA_3
Weight before solvent soaking	0.4269	0.772	0.82118	0.4839	0.79202	0.6656	0.25097	0.7046	0.53164
Weight after solvent soaking	0.341	0.6324	0.699	0.3548	0.5182	0.5271	lost	0.6364	0.4757
Shrinking rate	0.80	0.82	0.85	0.73	0.65	0.79		0.90	0.89
Average	0.82			0.73			0.90
Error bar +	0.03			0.07			0.00
Error bar −	0.02			0.07			0.00
Pure agar	Agar_EA_4			Agar_EA_5			Agar_EA_6
Weight before solvent soaking	0.42914	0.91009	0.78532	0.40322	0.77824	0.62518	0.41864	0.7718	0.86928
Weight after solvent soaking	0.17645	0.3476	0.4024	0.3442	0.6612	0.5293	0.3748	0.6967	0.7602
Shrinking rate	0.41	0.38	0.51	0.85	0.85	0.85	0.90	0.90	0.87
Average	0.45			0.85			0.89
Error bar +	0.06			0.00			0.01
Error bar −	0.07			0.00			0.02
Agar/PVA-B	AgarPVAB_EA_2			AgarPVAB_EA_3			AgarPVAB_EA_6
Weight before solvent soaking	0.37442	1.0944	0.70854	0.36886	0.72035	0.69318	0.3198	0.63239	0.5298
Weight after solvent soaking	0.2789	0.8618	0.4382	0.3249	0.6444	0.61847	0.2835	0.5612	0.46409
Shrinking rate	0.74	0.79	0.62	0.88	0.89	0.89	0.89	0.89	0.88
Average	0.72			0.89			0.88
Error bar +	0.07			0.00			0.00
Error bar −	0.10			0.01			0.01
Measurements December 2022
Pure agar	Agar_MEK_1			Agar_MEK_2			Agar_MEK_3
Weight before solvent soaking	0.48745	0.79845	0.72666	0.58415	0.8977	0.82677	0.3634	0.81438	0.73075
Weight after soaking	0.3707	0.6117	0.5573	0.4696	0.7778	0.685	0.2992	0.6782	0.6082
Shrinking rate	0.76	0.77	0.77	0.80	0.87	0.83	0.82	0.83	0.83
Average	0.76			0.83			0.83
Error +	0.00			0.03			0.00
Error −	0.00			0.03			0.01
Pure agar	Agar_MEK_4			Agar_MEK_5			Agar_MEK_6
Weight before solvent soaking	0.4567	0.8663	0.7973	0.4403	0.75095	0.73626	0.3483	0.89994	0.63593
Weight after soaking	0.3647	0.6967	0.6197	0.353	0.595	0.57856	0.2837	0.73625	0.522
Shrinking rate	0.80	0.80	0.78	0.80	0.79	0.79	0.81	0.82	0.82
Average	0.79			0.79			0.82
Error+	0.01			0.01			0.00
Error −	0.02			0.01			0.00
Agar/PVA-B	AgarPVAB_MEK_2			AgarPVAB_MEK_3			AgarPVAB_MEK_6
Weight before solvent soaking	0.46462	0.81105	0.9257	0.3897	0.82679	0.80046	0.39334	0.8091	0.7449
Weight after soaking	0.3609	0.6602	0.693	0.3177	0.6831	0.6596	0.3181	0.6647	0.6121
Shrinking rate	0.78	0.81	0.75	0.82	0.83	0.82	0.81	0.82	0.82
Average	0.78			0.82			0.82
Error +	0.03			0.00			0.00
Error −	0.03			0.01			0.01

Notes: For each preparation protocol (numbered 1–6), 3 samples of gel were used in order to calculate an average weight loss, indicative of shrinking rate due to solvent. Pure agar and agar/PVA-B gels were tested and ultimately soaked in ethyl acetate (EA) or methyl ethyl ketone (MEK).

Table A3. Ethanol used throughout the study was replaced with isopropanol IPA in order to compare the effects of these two alcohols on the shrinking rate of solvent-gels prepared according to protocols 1, 3 or 6.

Protocol	Alcohol treatment	EtOH + MEK	EtOH + EA	IPA + MEK	IPA + EA
Agar_1	Direct	0.80	0.82	0.78	0.75
Agar_3	Progressive	0.81	0.86	0.81	0.87
Agar_6	Progressive	0.82	0.86	0.80	0.87

Note: All figures presented in the table are the average three measurements.