Restoration of paper artworks with microemulsionsconfined in hydrogels for safe and efficientremoval of adhesive tapesNicole Bonellia,b,1, Costanza Montisa,b,1, Antonio Mirabilec, Debora Bertia,b, and Piero Baglionia,b,2

aDepartment of Chemistry, University of Florence, 50019 Florence, Italy; bConsorzio Interuniversitario per lo Sviluppo dei Sistemi a Grande Interfase (CSGI),University of Florence, 50019 Florence, Italy; and cPaper Conservator, 75009 Paris, France

Edited by Joan Selverstone Valentine, University of California, Los Angeles, CA, and approved April 25, 2018 (received for review February 7, 2018)

The presence of pressure-sensitive tapes (PSTs) on paper artworks,either fortuitous or specifically applied for conservation purposes,is one of the most frequent and difficult issues encountered duringrestoration. Aged PSTs can damage or disfigure artworks, com-promising structural integrity, readability, and enjoyment. Currentprocedures are often inherently hazardous for artistic media andpaper support. Challenged by the necessity to remove PSTs from acontemporary and an ancient drawing (20th century, by artists daSilva and Hayter, and a 16th-century drawing of one figure fromthe Sistine Chapel by Michelangelo), we addressed this issue froma physicochemical perspective, leveraging colloid and interfacescience. After a characterization of the specific PSTs present on theartifact, we selected a highly water-retentive hydrogel as the hostof 23% wt/wt of “green” organic solvents uniformly dispersedwithin the gel in the form of nanosized droplets. The double con-finement of the organic solvent in the nanodroplets and into thegel network promotes a tailored, controlled removal of PSTs ofdifferent natures, with virtually no interaction with the solvent-sensitive artwork. This noninvasive procedure allows complete re-trieval of artwork readability. For instance, in the ancient drawing,the PST totally concealed the inscription, “di mano di Michelan-gelo” (“from Michelangelo’s hand”), a possibly false attributionhidden by a collector, which is now perfectly visible and whoseorigin is currently under investigation. Remarkably, the samemethodology was successful for the removal of aged PST adhesivepenetrated inside paper fibers of a drawing from the celebratedartist Lucio Fontana.

pressure-sensitive tapes | paper artwork cleaning | cultural heritageconservation | hydrogel | microemulsions

The invention of pressure-sensitive tape (PST) is attributed toHorace Day, a surgeon who made, in 1845, the first basic

surgical sticking plaster tape by combining India rubber, pinegum, turpentine, litharge, and turpentine extract of cayennepepper and applied that mixture to strips of fabric (1). It tookalmost 75 y, from Day’s first PST until the early 1920s, for thefirst industrial tape application to appear (1). Since then, PSTshave been used in several applications, including to mend pre-cious mementoes, repair torn book pages, and hold togetherparts of lacerated documents, hence reflecting the social andcultural “fast at any cost” aspects of modern society. PSTs mostfrequently encountered (2) in works of art have rubber or acrylic-based adhesive pastes. Depending on their composition, PSTadhesives undergo several degradation processes, which lead, attheir final stages, to materials that become dark and oily and canentirely penetrate the medium; i.e., they are quite different incolor, form, and function from the original. Time and experiencehave shown that PST(s) on paper can be disfiguring, damag-ing, and difficult to remove. Conservators are familiar with avariety of tape removal methods, including mechanical, immer-sion, poultice, rolling, and suction table (2). However, eachmethod has some associated risks/disadvantages, which may re-sult in undesirable changes such as skinning of the medium with

mechanical removal, tidelines and media bleeding when suctiontechniques and poultices are used, and media bleeding andpenetration of the adhesive into the cellulose fibers when usingimmersion treatments. For artistic, ethical, and practical pur-poses, it is important to find a way that ensures selective removalof the PSTs without affecting the artifact or leaving residues.Over recent years, we have designed several aqueous-basedcolloidal systems, where organic solvents can be confined withstructural and dynamic control at the nanoscale, termed“nanostructured fluids” (NSFs), tailored to clean surfaces ofartworks. These complex fluid media overcome many of the dis-advantages of neat organic solvents, such as toxicity and solventspreading (3–5). In addition, the loading of water-based NSFswithin hydrogel scaffolds with high water retentiveness guaranteeseffective yet safe cleaning of water-sensitive artifacts (6, 7). Thesemethods have been applied to remove soil, wax, polymeric coat-ings, and varnishes from a variety of important artistic surfaces (8,9). Recently, we were confronted with the removal of PSTs fromtwo completely different artworks: a recto-verso contemporarydrawing, unusually realized by two artists at different times (therecto was realized by the Portuguese−French abstractionistMaria Helena Vieira da Silva, while Helen Phillips Hayter realizedthe drawing visible on the verso) and an ancient 16th-centurydrawing representing one of the figures of the Ascesa dei Beati, a

Significance

From Dead Sea Scrolls to Federico Fellini and Lucio Fontanadrawings, pressure-sensitive tapes (PSTs) have been used asadhesive fasteners or as part of temporary conservationtreatments that frequently became permanent. Their safe andefficient removal poses ethical and aesthetic questions: Adhe-sive tape residues damage the paper substrate and, due todiscoloring, prejudice the artwork enjoyment and conserva-tion. Selective removal without affecting the underlying sup-port is challenging and often impossible. We tackled this issuefrom a physicochemical and a colloidal perspective, by designinga system where nanosized solvent droplets are confined withina hydrogel. This method has the potential to revolutionize theapproaches used so far in the removal of PSTs and coatings froma plurality of materials.

Author contributions: A.M. and P.B. designed research; N.B., C.M., A.M., and D.B. per-formed research; N.B. and C.M. analyzed data; C.M., D.B., and P.B. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Published under the PNAS license.1N.B. and C.M. contributed equally to this work.2To whom correspondence should be addressed. Email: baglioni@csgi.unifi.it.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1801962115/-/DCSupplemental.

Published online May 21, 2018.

5932–5937 | PNAS | June 5, 2018 | vol. 115 | no. 23 www.pnas.org/cgi/doi/10.1073/pnas.1801962115

Dow

nloa

ded

by g

uest

on

Mar

ch 2

8, 2

020

scene of theGiudizio Universale realized byMichelangelo Buonarrotiin the Sistine Chapel.We decided to tackle this complex issue using a hydrogel, close

to complete water retentiveness (6), as a scaffold to confine ananostructured fluid composed by a surfactant that stabilizesdroplets of organic solvents in water as continuous phase (NSF),tailored to remove specific PSTs. This approach allows a verysophisticated control of the cleaning system: (i) The confinementof the organic solvent—needed for PST softening—in dropletswithin a continuous water phase controls the penetration of thesolvent in the PST, and prevents spreading in the underlyingsupport; (ii) the confinement of the NSF within the highly re-tentive hydrogel network allows NSF−PST contact with no lat-eral migration of the liquid phase. To fine-tune the hydrogel/NSF system, we investigated the interaction with the mostcommon PSTs, whose multilayered structure had been charac-terized through SEM, attenuated total reflectance (ATR)-FTIR,and thermal analysis. The NSF ability to interact with the PSTsand diffuse through the backing of the different PSTs types wasstudied by thermogravimetric analysis and steady-state fluores-cence measurements, respectively. We then challenged the dif-ferent PSTs with the hydrogel/NSF system and monitored theNSF interaction with confocal microscopy, to determine themechanism of action and predict the outcome of PST removal inreal cases. The selected system was applied to remove the PSTsfrom the two-mentioned artworks, leading, in one case, to theastonishing discovery of the inscription “di mano di Michel-angelo” (“from Michelangelo’s hand”). Moreover, we extendedthe same method to the removal of strongly defacing PST resi-dues from a 20th-century Lucio Fontana drawing, demonstratingthe potentiality of the proposed methodology to fully controlevery step of the restoration process.

Results and DiscussionPST Models. Fig. 1 A and B displays an ancient 16th-centurydrawing representing one of the figures of the Ascesa dei Beati,a scene of the Giudizio Universale realized by MichelangeloBuonarroti in the Sistine Chapel, with an aged PST, at a firstinspection characterized by a backing made of cellulosic material(paper), attached on the bottom left side of the recto. Fig. 1 Cand D displays a recto-verso contemporary drawing (the recto bythe Portuguese−French abstractionist Maria Helena Vieira daSilva, the verso by Helen Phillips Hayter), with a PST charac-terized by a transparent backing on the verso side (highlighted byan arrow in Fig. 1D). These two extremely different artworksclearly demonstrate the ubiquitous nature of PSTs’ applicationand the complexity of their removal. In fact, several classes ofPSTs can be found on both ancient and contemporary artworks,with different structural and physicochemical features. To tacklethis issue with a general approach and to set up a procedureenabling the safe and efficient removal of different PSTs fromworks of art, we selected the most common PSTs, to investigatetheir physicochemical properties and address their removal.PSTs present a multilayered structure composed of a pressure-sensitive adhesive (PSA) and its carrier (backing). Minor com-ponents include a release coat, ensuring an easy unrolling of thetape, and a primer, that enhances adhesion between backing andthe adhesive mass. Backing materials may include paper, fabric,cellophane, cellulose acetate, and oriented polypropylene, whilePSAs include natural and synthetic rubbers, acrylic copolymers,and silicones (2, 10). Both the adhesive and the backing layerdetermine the fate of the PSTs, in terms of aging and degrada-tion effects.We selected three types of PSTs as the most common repre-

sentative models of PSTs in artworks: Filmoplast P (FPP) is aPST with a cellulose-based backing, specifically designed forpaper conservation; MagicTape (MT) is the popular “matte-finish tape” (which makes it invisible on paper); and ordinarytape, OT, is a very common PST for domestic usages. In Fig. 2A,SEM images of the three backings of PSTs are displayed. FPP(i.e., paper) exhibits the typical morphology of compressed cel-lulose fibers; MT displays a rough surface, indicating that themacroscopic matte effect is obtained through treatment of thesurface; and the surface of OT appears smooth and compact atthis length scale.Fig. 2 shows the ATR-FTIR characterization of the adhesive

(Fig. 2B) and backing (Fig. 2C) of the PST models. In all of thestudied PSTs, the typical infrared pattern of acrylic adhesives,widely used in PSAs formulations, is present (Fig. 2B) (11).Conversely, meaningful differences in the composition of the

Fig. 1. (A) Sixteenth-century drawing from Ascesa dei Beati, a scene of theGiudizio Universale of Michelangelo Buonarroti, Sistine Chapel, (B) detail ofan aged PST on the drawing; and (C and D) 20th-century drawing (C) rectoby Maria Helena Vieira da Silva and (D) verso by Helen Philips Hayter.

Fig. 2. (A) SEM images of PSTs. (B and C) ATR-FTIR spectra of the (B) ad-hesives and (C) backings.

Bonelli et al. PNAS | June 5, 2018 | vol. 115 | no. 23 | 5933

SOCIALSC

IENCE

SCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Mar

ch 2

8, 2

020

PSTs are observed in the backings (Fig. 2C): cellulose for FPP,cellulose acetate for MT, and polypropylene for OT (full ATR-FTIR analysis is reported in SI Appendix, Fig. S2). As alreadydiscussed, PSTs are currently removed with procedures that candamage the artifacts, as rough mechanical processes are oftenextremely dangerous for the integrity of works of art. The se-lection of an appropriate solvent for PST removal using theclassic immersion procedure is by no means trivial. The solventshould efficiently penetrate the backing and soften the adhesive.However, since the supports of artworks are generally very po-rous, a complete solubilization of the adhesive could lead to thepenetration of the solubilized adhesive within the support,leading to uncontrolled spreading and penetration of greatquantities of solvent that can be extremely dangerous for theartistic media (e.g., inks), with the risk of irreversible damage oreven the loss of the artistic message.

Choice of the NSF. With these prerequisites in mind, we chose amultifunctional system able to efficiently interact with differentbackings, without causing damage to the support, i.e., withoutaffecting the painted layer or the underlying paper. Our researchgroup designed several nanostructured aqueous-based formula-tions as cleaning tools for works of art, with high efficiency forthe removal of hydrophobic polymer coatings from wall paint-ings, unwanted varnishes from easel paintings, or grime andother contaminants from modern and contemporary arti-facts (12). Nanostructured fluids present several advantages fordealing with restoration issues: The confinement of organicsolvent in a water-based medium limits or completely avoids theadverse effects on the support that are due to the uncontrolledspreading of the organic solvent in the porous matrix of theworks of art. The use of small amounts of organic solvents,typical of NSF, improves the system in terms of environmentalimpact and health compatibility. Finally, the intrinsic high in-terfacial area of the NSFs leads to very high efficiency and se-lective removal of defined components from artifacts (13). Oneof these NSFs, termed EAPC, is composed of water, SDS, 1-pentanol, ethyl acetate (EA), and propylene carbonate (PC).This system has been fully characterized from a structural pointof view using small-angle neutron scattering with contrast vari-ation (4), determining the localization of each component in theNSF. Nanosized (major axis 12.8 nm) (4) ellipsoidal dropletsmainly composed of EA are stabilized in a mixed continuousphase (water and about 20% PC) by a film of SDS and 1-pen-tanol. PC, mainly located in the continuous phase, is also parti-tioned at the micelle interface, conferring enhanced cleaningcapacity due to its high dipole constant (a cartoon sketching thestructure of EAPC is displayed in SI Appendix, Fig. S1). EAPCdisplayed high efficiency in removing acrylic coatings from ar-tistic surfaces (5, 14). Finally, both EA and PC are classified asenvironmental friendly solvents. Based on chemical affinity, wecan expect this NSF to efficiently penetrate the backing of FPP(thanks to the hydrophilicity and porosity of the backing), andto efficiently interact with MT backing (due to the partial solu-bility of cellulose acetate in ethyl acetate and propylene carbon-ate), while scarce interaction is expected for the OT backing.Finally, since NSFs can dewet thick acrylic coatings (15–18 μm),we can predict some swelling of the acrylic adhesives. We willshow how NSF opportunely confined into hydrogels allowsthe removal of PSTs from artistic surfaces with unprecedentedperformances.

Interaction of PSTs with NSFs. Before uploading the NSF inside thehydrogel matrix for application to the removal of PSTs fromworks of art, we investigated, in detail, its interactions with themost common PSTs used in the conservation practice.The thermal degradation profiles of chemical components

are robust tools for analytical characterization. We exploited

differential thermogravimetric analysis (DTG) for determiningthe chemical modifications of PST components constituting theadhesive and backing layers, upon interaction with EAPC NSF.In SI Appendix, Fig. S4, the DTG curves of the three completePSTs (adhesive and backing) and bare backings, before and afterimmersion in NSF for 48 h, are displayed. The DTG results(whose complete discussion is reported in SI Appendix) highlight(i) the absence of chemical modification of the acrylic compo-nents of PST adhesives upon interaction with the NSF, for thethree PSTs; and (ii) different effects of the NSF, depending onthe backings: FPP and OT show no significant variations inthermal behavior, suggesting that the NSF does not alter thechemical composition or structure of the backing. On the otherhand, after interaction of MT backing with the EAPC NSF, anadditive [probably a plasticizer, diethyl phthalate, commonlyused in the production of cellulose acetate-based PSTs (19, 20)]is removed from the cellulose acetate film, possibly due to its highsolubility in ethyl acetate, one of the components of EAPC NSF.In summary, DTG suggests that the different backings steer

the interaction pathway of the EAPC with the PSTs. To furtheraddress this point, we performed a fluorescence assay, to mon-itor the ability of EAPC NSF to penetrate PST backings. Thethree backings were layered on a quartz cuvette filled with NileRed (NR)-labeled EAPC NSF in contact with a second cuvettefilled with unlabeled EAPC NSF; NR fluorescence inside theoriginally unlabeled EAPC solution was then monitored for aperiod of 96 h (Fig. 3). The fluorescence increase due to NRdiffusion across the MT backing was compared with a referencecurve (SI Appendix, Fig. S6) to obtain an NR concentrationprofile (Ct). Fig. 3B compares, for the three backings, the con-centration profiles normalized for the theoretical final concen-tration, Cf (Ct/Cinf). Consistently with DTG results, no penetrationis observed across OT backing, while NR crosses MT and FPPbackings, with a higher permeation efficiency in the first case.The dye diffusion to the second cuvette implies (i) interaction ofthe NSF with the backing and (ii) release of the dye from thebacking to the unlabeled solution, due to concentration gradient.The data were analyzed with a model for the release of activeprinciples from a thin film, according a power law Ct/Cinf = ktn,with k dependent on the structure and geometry of the systemand n related to the diffusion mode (21–24). For pure Fickiandiffusion, n = 0.5 is expected, while a linear dependence overtime (n = 1) is related to Case II transport, connected to swellingor relaxation of the support. The data, yielding n values of0.47 for MT and 1.0 for FPP, can be correlated with the DTGresults. EAPC closely interacts with MT backing, eventuallyleading to the solubilization of some of the additives (Fig. 3A),and the dye diffusion across this barrier is not hampered,resulting in a Fickian diffusion. Conversely, the possibly strong

Fig. 3. (A) Fluorescence spectra of NR-labeled EAPC diffusing across MTbacking over 96 h; (B) picture of cuvettes containing originally unlabeledEAPC, after 96 h contact with NR-labeled EAPC through FPP, MT, OT back-ings; and (C) NR diffusion kinetics through FPP, MT, and OT backings(markers); curve fit of experimental data according to Ct/Cinf = ktn (contin-uous lines) equation; curve fit of FPP with n = 0.5 fixed (dashed line).

5934 | www.pnas.org/cgi/doi/10.1073/pnas.1801962115 Bonelli et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

8, 2

020

affinity for FPP cellulose backing might lead to adsorption−desorption of NR on the cellulose fibers, resulting in a non-Fickian diffusion. If we “force” the model to a pure Fickiandiffusion for both backings (n = 0.5), we obtain k values of0.05 and 0.025 for MT and FPP, respectively. The diffusion ki-netics of NR across MT is considerably faster than for FPP.Despite its simplicity compared to more complex approaches(25), this model nicely captures some essential details, account-ing for the different affinity of EAPC for the backings, andcomplement DTG results. In particular, for FPP, the interactionwith cellulose probably determines a partial disruption of NSF, anonnegligible physical adsorption of the NR on fibers, withoutany chemical modification of the backing (as highlighted fromDTG). This leads to the overall slowing down of NR diffusionacross the backing (decrease in k) and to a change of the dif-fusion mode (variation of n), due to a strong “matrix effect.”In summary, the NSF is able, to a different extent and with

different mechanisms, to cross the backings of FPP and MT overtheir entire thickness, while no penetration is observed for OT.This very simple method allows ex situ selection of the NSF toachieve a slow and controlled interaction with the backings andcan be applied generally to address PST removal without in-volving the artifact.

Interaction of PSTs with Gel-Confined NSFs. The NSF’s confine-ment, key for restoration interventions, ensures control over theremoval process. A partial control is guaranteed by dispersingthe organic solvent as nanodroplets in the NSF, but a lateralconfinement is necessary to limit spreading outside the applicationarea. For this reason, the NSF was inserted in a semiinterpenetratingpolymer network composed of poly(hydoxyethyl metacrylate)

[p(HEMA)] and polyvinylpyrrolidone (PVP). The hydrogel prep-aration and NSF upload are described in SI Appendix, HydrogelPreparation and Loading. Unlike the physical gels most com-monly used in art conservation (e.g., agar, gellan, solvent gels)(26, 27), this system is a chemical gel, whose network is estab-lished by formation of covalent bonds. This difference ensuresthe absence of gel residues on support after the treatment.Depending on composition and viscoelastic properties, chemicalhydrogels can be easily shaped to perfectly match the PSTfootprint on the artwork. In addition, the high retentiveness canfurther improve the confinement of the NSF, determining acontrolled release of the NSF to the PST, avoiding the lateralspreading over the application area of the NSF to the very sur-face covered by the PST.The interaction of p(HEMA)/PVP gels loaded with EAPC

NSF with the different adhesive tapes was monitored with con-focal laser scanning microscopy (CLSM). PST samples werefluorescently labeled by immersion in a 10−3 mM aqueous so-lution of rhodamine B isothiocyanate (RhBITC) and then airdried for 24 h. RhBITC physically adsorbs within the acrylicadhesive layer without altering the structural and chemical fea-tures of the PSTs. EAPC dispersions, fluorescently labeled withRh110 10−2 mM, were uploaded in the gel by soaking the net-work in the Rh110-labeled fluid overnight. For CLSM (Fig. 4),the RhBITC-labeled PSTs (red) were attached on a coverglasswith the gels loaded with Rh110-labeled EAPC NSF (green)layered on top.In Fig. 4 A–C, some representative 3D reconstructions of FPP

PST (red) interacting with p(HEMA)/PVP gel containing theRh110-labeled EAPC (green) are displayed, after 5 min (Fig.1A), 10 min (Fig. 1B), and 20 min and (Fig. 1C). The FPP paperbacking is opaque, and, at the beginning, only the fluorescentlylabeled adhesive (red) is visible (Fig. 4A). With increasing in-cubation times (Fig. 4 B and C), the NSF progressively pene-trates the backing, reaching the backing−adhesive interface after20 min (Fig. 4C). Concerning MT (Fig. 4 D and E), PST iscompletely transparent and allows visualizing both the labeledPST (red) and the gel. In this case, some Rh110 is able to deeply

Fig. 4. Confocal microscopy of PSTs with RhBITC-labeled adhesive (red)interacting with p(HEMA)/PVP gel loaded with Rh110-labeled EAPC NSF(green). (A–C) A 3D reconstruction of FPP PST after (A) 5 min, (B) 10 min, and(C) 20 min of interaction with the gel; due to the opacity of FPP backing, thegreen-labeled hydrogel layered on the PST is not visible in CLSM 3D re-construction; after 20 min, the NSF penetrates the backing, which is thenhomogeneously fluorescently labeled (C). (D) A 3D reconstruction and (E) a2D horizontal section of MT after 20 min of interaction, where, due to thetransparency of the PST, both the initially unlabeled backing and thehydrogel are visible; after 20 min, the NSF penetrates the backing fromthe hydrogel (the pale green region between the red adhesive and thebright green gel). (E) A 2D section of MT PST adhesive after 20 min of in-teraction with the NSF: The red and the green emissions are separatelydisplayed with the transmission (grayscale); the overlay of these images(with colocalization of the probes appearing as yellow) highlights the suc-cessful penetration of the NSF in the upper parts of the MT adhesive. (F) A3D reconstruction and (G) a 2D vertical section of OT PST after 20 min ofinteraction. Due to the transparency of the PST, both the unlabeled backingand the hydrogel are visible; after 20 min, the NSF from the hydrogel isseparated from the adhesive layer by the backing, which, unlike for MT andFPP, remains unlabeled. (Scale bars, 50 μm.)

Fig. 5. Confocal microscopy of unlabeled PSTs interacting with p(HEMA)/PVPloaded with Rh110-labeled (green) and NR-labeled (red) EAPC NSF. (A and B)A 3D reconstruction of FPP PST after (A) 10 min and (B) 20 min of interaction;after 20 min, the unlabeled backing is green, due to the penetration of thehydrophilic components of the NSF, while the adhesive remains unlabeled.(C) A 3D reconstruction and (D) a 2D vertical section of MT PST after 20 min ofinteraction, with the channels displayed both separately and overlaid (withemission colocalization in yellow): The backing transparency allows visualizingthe adhesive, backing, and gel, showing that the doubly labeled NSF migratesin the backing, reaching the first layer of the adhesive enriched with the NSForganic components [consistent with the higher intensity of NR fluorescence(red) compared to the hydrophilic Rh 110 fluorescence (green)]. (E) A 2Dvertical section of OT PST upon 20 min of interaction, with the channels bothdisplayed separately and overlaid (the colocalization of Rh110 and NR fluo-rescence is highlighted in yellow): The transparency of the backing allows vi-sualizing adhesive, backing, and gel, showing the confinement of the NSF inthe hydrogel, with no significant penetration in the backing and adhesive,which remain unlabeled. (Scale bars, 50 μm.)

Bonelli et al. PNAS | June 5, 2018 | vol. 115 | no. 23 | 5935

SOCIALSC

IENCE

SCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Mar

ch 2

8, 2

020

penetrate the backing (Fig. 4D), eventually reaching the lowerpart of the adhesive. As a matter of fact, a horizontal section ofthe same sample, acquired after 20 min of incubation (Fig. 4E),highlights that the red and the green fluorescence are colocalizedinside the PST. Fig. 4F displays a representative 3D recon-struction of OT PST (red) upon 20 min of interaction with theNSF (red) loaded in the p(HEMA)/PVP gel (green). In this case,the probes’ emissions are well separated, as confirmed also bythe vertical section displayed in Fig. 4G.In summary, a different penetration of the NSF loaded in p

(HEMA)/PVP gels is observed in the three PST models: (i)EAPC efficiently penetrates the FPP paper structure, poorlyinteracting with the adhesive underlying layer; (ii) it efficientlyinteracts with MT, completely penetrating the backing andeventually reaching the lower part of the labeled PST (the ad-hesive layer); (iii) conversely, the polypropylene backing of OTstrongly opposes the penetration the NSF.From these experiments, the composition of the fluid phase

diffusing inside the PSTs remains unclear. To address this issue,we designed a complementary experiment to determine the axialpenetration of the components with different polarity of thecomplex fluid. Unlabeled PSTs were attached to coverglassesand covered with the gel loaded with a doubly labeled NSF,containing 10−2 mM Rh 110 (green) and 10−2 mM NR (red).The two dyes are characterized by extremely different affinitiesfor water and organic solvents: Rh 110 is hydrophilic, while NRhas a marked hydrophobic nature. The penetration degree of thetwo dyes can be considered a reliable representation of thepenetration extents of the components of the NSF with differentpolarities. Fig. 5 A and B shows some representative 3D recon-structions of NR−Rh110 doubly labeled NSF/gel after 20 min ofinteraction with unlabeled FPP. The whole paper backing issoaked with the NSF, with homogeneous fluorescence of thedyes visible as a continuous layer overlying the adhesive (un-labeled and transparent, and thus not visible). Interestinglyenough, a different distribution of the two dyes is observed: Thehydrophilic green dye is more evenly distributed compared to thehydrophobic red dye. This is consistent with a higher affinity ofthe hydrophilic components of the NSF for the cellulosic struc-ture of FPP. Concerning MT (Fig. 5 C and D), after 20 min ofinteraction, the hydrophobic red NR penetrates more deeply thePST compared to the hydrophilic green dye. Therefore, the NSForganic components have a higher affinity for the MT backing

than the hydrophilic ones. Finally, Fig. 5E reports vertical sec-tions of the NSF/gel after 20 min of interaction with unlabeledOT. The perfect colocalization of green and red, represented inyellow, highlights that the NSF components remain confined inthe gel, without penetration in the backing, nor in the underlyingadhesive layer.Overall, these experiments provide fundamental knowledge on

the interaction mechanism of the NSF with the different PSTs.For FPP, the NSF does not chemically modify the backing and

adhesive but can efficiently penetrate across the backing. Whendiffusing within the PST, the NSF structure is partially disrupted,and a water-enriched fluid reaches the lowest part of the backing,in contact with the adhesive. This evidence can possibly explainthe non-Fickian diffusion of NR across the backing.Concerning MT, the NSF produces chemical modifications in

the backing, solubilizing some of its components (diethyl phthalate),then efficiently crosses the backing with a purely diffusive mech-anism to reach the farthest region of the backing, with final ad-sorption inside the adhesive. In this case, the organic componentof the NSF is able to penetrate at higher depths compared to theaqueous components.In the case of OT, the NSF does not modify the backing and

the adhesive and is not able to penetrate the entire thickness ofthe backing.The gel-confined NSF can then remove FPP-type and MT-

type PSTs from artworks, while NSFs with a higher amount oforganic solvents have to be designed to tackle the OT removal.

Removal of PSTs from Artworks. The removal of aged PSTs wasperformed with the same procedure for both drawings: The NSF-loaded hydrogel was applied on the PST after being shaped tomatch its exact profile and size. After 5 min, the softening of thePST was tested with a scalpel. Detaching was performed by avery gentle mechanical action, without risk of abrasion of theunderlying paper support. Fig. 6A displays the entire procedureleading to the removal of the PST from the 16th-century draw-ing, while Fig. 6 B and C shows the artworks after removal. Theconfinement of the EAPC NSF inside the hydrogel permitted asafe application, avoiding lateral migration of the liquid. Thanksto the facile handling, which permits removal and reapplication ofthe loaded hydrogel if needed, the minimum time of interactionbetween the NSF-loaded hydrogel and the PST is readilydetermined.

Fig. 6. (A) Removal of a PST from the bottom of the 16th-century drawing.The detail shown in the red square highlights the EAPC-loaded hydrogelshaped to precisely match the PST to be removed to avoid contact betweenthe cleaning system and the artwork. (B) Detail of the drawing after removalof the PST, where the inscription “di mano di Michelangelo” appears,probably a false attribution which was concealed by the collector. (C) Con-temporary drawing by Helen Phillips Hayter after removal of the PST. Insetshows the detail of PST before removal.

Fig. 7. Lucio Fontana, Untitled, 1956 (ballpoint pen and tempera on paper)(A) before and (B) after removal of PSTs adhesive residues upon applicationof the chemical hydrogel.

5936 | www.pnas.org/cgi/doi/10.1073/pnas.1801962115 Bonelli et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

8, 2

020

After the successful removal of the two PSTs from both theancient and the contemporary drawing, an FTIR ATR analysiswas performed (SI Appendix, Fig. S10). The PST removed fromthe Helen Phillips Hayter drawing is made of a cellulose acetate-based backing with an acrylic adhesive, which makes it consistentwith the composition of the MT sample. On the other hand, thePST removed from the 16th-century drawing consists of acellulose backing with a rubber-based adhesive. Due to thebacking, we expect a mechanism of diffusion similar to FPP.Since this adhesive is different from the three selected modelsamples, this evidence shows that the proposed system can beapplied to a large variety of PSTs.In the discussed examples, the intervention on aged PSTs aims

at softening the adhesive to facilitate the backing removal,minimizing the mechanical action needed. In some cases, thePST adhesive residues can be found even inside the paper fibers,due to complex aging processes or inappropriate solvent appli-cations by previous restoration interventions.Fig. 7A displays a ballpoint pen and tempera drawing by the

20th-century celebrated Italian artist Lucio Fontana, with defacingdiscolorations, due to aged adhesives deeply penetrated inside thepaper fibers, in the absence of the backing.The ink found on this drawing is a black ballpoint pen ink from

the 1950s. Ballpoint pen inks consist of organic dyes dispersed ina mixture of appropriate solvents containing various additives.These inks are very sensitive and may be altered by either organicsolvents or water, depending on the specific composition (28). Forthese reasons, and due to the obvious difficulty of assessing theexact composition of each ink on original artworks, the treatmentof drawings made using ballpoint pens as artistic media is par-ticularly challenging to conservators, making the wet cleaning ofareas containing ballpoint pen strokes almost impossible.The use of the system proposed in this work enables the

control of penetration and lateral spreading of the liquid phase,minimizing the contact with sensitive components of the artworkand limiting the possible movements of the inks. The applicationof the previously described hydrogel successfully allowed thesoftening of the penetrated adhesive and its removal (Fig. 7B).

These case studies demonstrate that a highly versatile tool isavailable to conservators, guaranteeing complete control of theremoving fluids during all of the steps required for PST removalfrom paper artworks.This confinement-based methodology ensures the achieve-

ment of unprecedented safe and efficient removal of PSTs frompaper artworks, thus restoring their full readability for publicenjoyment.

Materials and MethodsPST Models. OT was purchased from Tesa (product code 56100), MT is from3M, and FPP is from Neschen; analysis of backing was carried out after re-moval of the adhesive with isopropanol.

Microemulsion Preparation. EAPC system is an oil-in-water microemulsionprepared by dissolving the SDS surfactant in water and propylene carbonate.Ethyl acetate as dispersed organic phase and the cosurfactant 1-pentanolwere added drop-wise to the aqueous surfactant solution.

Hydrogel Preparation. Semiinterpenetrating p(HEMA)/PVP hydrogel net-works are obtained through free radical polymerization of HEMA usingazoisobutyronitrile (AIBN) as an initiator and N,N-methylene-bis(acrylamide)(MBA) as cross-linker. Hydrogels were loaded with the nanostructuredcleaning fluid (i.e., EAPC) through immersion for at least 12 h.

PST Characterization. PST characterization was performed through SEM, ATR-FTIR, and thermogravimetry.

EAPC Nanofluid Interaction with the PSTs. EAPC nanofluid interaction with thePSTs was evaluated through thermogravimetric analysis and a designedsteady-state fluorescence kinetic experiment.

p(HEMA)/PVP EAPC Interaction with PSTs. p(HEMA)/PVP EAPC interaction withPSTs was investigated through CLSM. Full methods and materials are avail-able in SI Appendix.

ACKNOWLEDGMENTS. Consorzio Interuniversitario per lo Sviluppo dei Sistemia Grande Interfase (CSGI) and the European Union [NANORESTART project(nanomaterials for the restoration of works of art), Horizon 2020 researchand innovation program, Grant H2020-NMP-21-2014/646063] are acknowl-edged for financial support.

1. Skeist I (2007) Handbook of Adhesives (Springer, New York).2. Smith MA, Jones NMM, Page SL, Dirda MP (1984) Pressure-sensitive tape and tech-

niques for its removal from paper. J Am Inst Conserv 23:101–113.3. Giorgi R, Baglioni M, Berti D, Baglioni P (2010) New methodologies for the conser-

vation of cultural heritage: Micellar solutions, microemulsions, and hydroxide nano-particles. Acc Chem Res 43:695–704.

4. Baglioni M, Berti D, Teixeira J, Giorgi R, Baglioni P (2012) Nanostructured surfactant-based systems for the removal of polymers from wall paintings: A SANS study.Langmuir 28:15193–15202.

5. Baglioni M, et al. (2010) Removal of acrylic coatings from works of art by means ofnanofluids: Understanding the mechanism at the nanoscale. Nanoscale 2:1723–1732.

6. Domingues J, Bonelli N, Giorgi R, Fratini E, Baglioni P (2013) Innovative method forthe cleaning of water- sensitive artifacts: Synthesis and application of highly retentivechemical hydrogels. Int J Conserv Sci 4:715–722.

7. Domingues JAL, et al. (2013) Innovative hydrogels based on semi-interpenetratingp(HEMA)/PVP networks for the cleaning of water-sensitive cultural heritage artifacts.Langmuir 29:2746–2755.

8. Mastrangelo R, Montis C, Bonelli N, Tempesti P, Baglioni P (2017) Surface cleaning ofartworks: Structure and dynamics of nanostructured fluids confined in polymerichydrogel networks. Phys Chem Chem Phys 19:23762–23772.

9. Chelazzi D, Giorgi R, Baglioni P (2017) Microemulsions, micelles and functional gels.How colloid and soft matter preserve works of art. Angew Chem Int Ed, 10.1002/anie.201710711.

10. Satas DE (1989) Handbook of Pressure Sensitive Adhesive Technology (Van NostrandReinhold, New York).

11. Gorassini A, Adami G, Calvini P, Giacomello A (2016) ATR-FTIR characterization of oldpressure sensitive adhesive tapes in historic papers. J Cult Herit 21:775–785.

12. Baglioni P, Carretti E, Chelazzi D (2015) Nanomaterials in art conservation. NatNanotechnol 10:287–290.

13. Baglioni M, Giorgi R, Berti D, Baglioni P (2012) Smart cleaning of cultural heritage: Anew challenge for soft nanoscience. Nanoscale 4:42–53.

14. Giorgi R, Baglioni M, Baglioni P (2017) Nanofluids and chemical highly retentive hy-drogels for controlled and selective removal of overpaintings and undesired graffitifrom street art. Anal Bioanal Chem 409:3707–3712.

15. Baglioni M, et al. (2017) Dewetting acrylic polymer films with water/propylene car-bonate/surfactant mixtures–Implications for cultural heritage conservation. PhysChem Chem Phys 19:23723–23732.

16. Raudino M, et al. (2015) Polymer films removed from solid surfaces by nanostructuredfluids: Microscopic mechanism and implications for the conservation of cultural her-itage. ACS Appl Mater Interfaces 7:6244–6253.

17. Baglioni M, et al. (2017) Polymer film dewetting by water/surfactant/good solventmixtures: A mechanistic insight and its implications for the conservation of culturalheritage. Angew Chem Int Ed Engl, 10.1002/ange.201710930.

18. Raudino M, et al. (2017) Probing the cleaning of polymeric coatings by nano-structured fluids: A QCM-D study. Langmuir 33:5675–5684.

19. Wypych G (2012) Handbook of Plasticizers (Elsevier, New York), 2nd Ed, pp 431–520.20. Phuong VT, et al. (2014) Cellulose acetate blends–Effect of plasticizers on properties

and biodegradability. J Renew Mater 2:35–41.21. Ritger PL, Peppas NA (1987) A simple equation for description of solute release I.

Fickian and non-fickian release from non-swellable devices in the form of slabs,spheres, cylinders or discs. J Control Release 5:23–36.

22. Boyd BJ, Whittaker DV, Khoo S-M, Davey G (2006) Lyotropic liquid crystalline phasesformed from glycerate surfactants as sustained release drug delivery systems. Int JPharm 309:218–226.

23. Negrini R, Mezzenga R (2011) pH-responsive lyotropic liquid crystals for controlleddrug delivery. Langmuir 27:5296–5303.

24. Negrini R, Mezzenga R (2012) Diffusion, molecular separation, and drug delivery fromlipid mesophases with tunable water channels. Langmuir 28:16455–16462.

25. Siepmann J, Peppas NA (2011) Higuchi equation: Derivation, applications, use andmisuse. Int J Pharm 418:6–12.

26. Gulotta D, et al. (2014) Setup of a sustainable indoor cleaning methodology for thesculpted stone surfaces of the Duomo of Milan. Herit Sci 2:1–13.

27. Wolbers R, Stavroudis C (2012) Aqueous methods for the cleaning of paintings.Conservation of Easel Paintings (Routledge, Abingdon, UK), pp 500–523.

28. Bügler JH, Buchner H, Dallmayer A (2008) Age determination of ballpoint pen ink bythermal desorption and gas chromatography-mass spectrometry. J Forensic Sci 53:982–988.

Bonelli et al. PNAS | June 5, 2018 | vol. 115 | no. 23 | 5937

SOCIALSC

IENCE

SCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Mar

ch 2

8, 2

020