When the surface of a ceramic object lifts in flakes or islands, it is easy to think of a consolidant first. That is the wrong starting point: consolidating and re-adhering are two different operations, which respond to two different failures and are carried out with different materials or, when the material is the same, at very different concentrations.
Costa, Leal, Mimoso and Pereira put it in writing for glazed tiles: "Mass consolidation of a ceramic body requires the use of different solutions and products from those used as adhesives to repair the glaze-biscuit connection" (Materiales de Construcción, 2017).
This guide is based on the conservation-restoration literature and on manufacturers' technical data sheets. It is no substitute for seeing the object: everything is tested first on a hidden area.
Two different failures, two operations
| Consolidating | Re-adhering (reattaching) | |
|---|---|---|
| What has failed | Cohesion: the material is disaggregating, powdering, losing grains | Adhesion: the layer is intact, but detached from what lies beneath it |
| What it looks like | Powder when touched, a surface that crumbles | Lifted edges, a shadow under the layer in raking light, islands that move |
| What is needed | A low-viscosity product that penetrates the pores and restores cohesion | An adhesive introduced under the edge, at the interface, that joins layer and substrate |
| Where it acts | Inside the material | Between the layer and its substrate |
The general difference between loss of cohesion and loss of adhesion is covered in our guide Consolidants for Restoration: Which to Choose and When Not To. On glazed ceramics it has a particular feature: both can happen at once. Mimoso and Esteves, studying glazed tiles, distinguish the flake that separates cleanly at the interface from the one that carries a thin layer of the ceramic body stuck to its back (Conservar Património, 2016). In the second case the body underneath is weakened too, and both operations are needed. That is why the back of a fallen flake, examined under a magnifier, is one of the first things worth keeping.
Why a consolidant does not re-adhere a flake
Ethyl silicate
Ethyl silicate (tetraethoxysilane, TEOS) has a documented use for consolidating porous, disaggregated stone and ceramic bodies: it penetrates and deposits silica inside the pores. For a lifted flake it has a physical limit: its gel does not bridge gaps larger than about 50 µm (micrometers), according to the Getty Conservation Institute (GCI) monograph on alkoxysilanes (Wheeler 2005, citing Wendler 1999). The same work documents flakes that, after treatment with ethyl silicate, remained solid but not well adhered to the stone.
In addition, the technical data sheets of the commercial ethyl silicates reviewed[1][2][3] indicate it for absorbent mineral substrates, and silicon sol-gel treatments are not reversible (Chapman and Mason, Journal of the American Institute for Conservation, 2003).
A nuance we will not hide: in laboratory specimens of terracotta with slip and glaze, fired at around 950 °C, impregnation with an elasticized ethyl silicate increased the adhesion between glaze and slip; the author concludes that every solution tested, acrylics included, improves it (D'ham, HAWK Hildesheim, 2003). That is a weakened interface in a porous substrate, not a flake already detached with a gap beneath it.
Colloidal silica
It is a different product, although it gets confused with the previous one: a water-based dispersion of silica nanoparticles. The technical data sheet of one of the commercial products (30% silica, pH between 9.8 and 10.4) states that its penetration capacity is reduced compared with ethyl silicate. In a laboratory test on archaeological ceramics, applied undiluted it left a transparent residual layer with some cracking and darkened the surface, with a color difference (ΔE) of 9.2 (Braco, master's thesis, Universitat Politècnica de València, 2019).
Very dilute acrylic resin
Here the material works; what fails is the concentration. Chapman and Mason put it bluntly: "too low a concentration, and the resin lacks adhesive strength". A solution designed for impregnation is not an adhesive.
Paraloid B-72 concentration: for consolidating and for re-adhering
Paraloid B-72, an ethyl methacrylate and methyl acrylate copolymer, appears in many of the published ceramic cases, and in several of them for both operations at once, at different concentrations. These are the published figures, with their source:
| Operation | B-72 concentration | Solvent | Source |
|---|---|---|---|
| Consolidating glazed brick: first the friable areas, then the whole brick | 1.5-3%, then 3-5% | Acetone:ethanol 1:1 (not stated for the 3-5%) | Whyte, Muros and Barack 2004 |
| Consolidating delamination in tiles, by micropipette until the opening is saturated | 5% | Acetone:ethanol 70:30 | Laanbroek, Manshanden and Overhoff 2012 |
| Sealing fine cracks by capillary action | ~10% or less | — | Preservation Equipment usage sheet |
| Priming the edges of porous ceramics before bonding | 5-10% w/v | Acetone | Koob 2018 |
| Re-adhering glaze flakes | 20% | Acetone | Laanbroek, Manshanden and Overhoff 2012 |
| Re-adhering fragments of the tile body | 50% | Acetone | Laanbroek, Manshanden and Overhoff 2012 |
| Tube adhesive prepared in the studio | ~48% w/v (for glass) and ~55.5% w/v (for porous earthenware) | Acetone | Koob 2018 |
w/v: grams of resin per 100 ml of solution.
Three points from Koob (AIC Objects Specialty Group, Postprints 25, 2018) worth keeping at hand:
- On porous ceramics, prime first. Earthenware, terracotta and low-fired pottery need their edges sealed or "primed" before the adhesive is applied, so that the adhesive retains its solvent long enough to set. Glass and high-fired china need no special preparation.
- Injection that is drawn in by capillary action ("wicking in") only works with non-porous materials, such as glass and porcelain.
- If ethanol is added to the acetone to slow its evaporation (Koob proposes this for hot conditions), between 5 and 15% by volume, no more: above that, ethanol, which is highly hygroscopic, draws water into the mixture.
And a warning about the solvent under a flake, where the evaporation surface is very small relative to the volume: with slow solvents, the solvent is retained and setting takes longer (De Henau 1996, cited in Chapman and Mason 2003).
The general preparation of B-72 is covered in our Paraloid B-72 guide.
How it has been solved in practice: museum case studies
- On a tile panel at the Nederlands Tegelmuseum (Otterlo), in 2012, the panel was desalinated for two weeks and the delamination was then consolidated with 5% B-72 in acetone and ethanol, applied several times by micropipette; the glaze flakes were re-adhered with B-72 at 20% in acetone, and the body fragments at 50%. The treatment was completed; one tile that had already got wet before its fragile glaze was consolidated was desalinated separately, glaze side up, to keep damage to a minimum (Laanbroek, Manshanden and Overhoff 2012).
- On the Assyrian glazed bricks at the Oriental Institute in Chicago, from 2002 onwards, the glaze flakes were "tacked" in place with B-72 from a tube and the friable areas were consolidated separately with dilute B-72; the bricks were then consolidated several times and placed in polyethylene bags after each application to slow evaporation. The façade was reassembled (Whyte, Muros and Barack, AIC OSG Postprints 11, 2004).
- On the glazed stoneware mural by Joan Miró and Josep Llorens Artigas at Barcelona airport, in 2005, outdoors, the glaze flakes were consolidated with Paraloid B-44 at 20% in acetone, a peeling polyurethane varnish was removed and the decision was taken not to revarnish the stoneware. Color and surface were recovered; no follow-up has been published (Gall Ortlik, UNICUM, 2006).
- On the glazed stoneware sculpture Textured Dog at Silpakorn University (Bangkok), fired at around 1100 °C and outdoors for more than ten years, the lifting glaze was re-adhered with dilute B-72 in acetone, applied with a fine brush or microsyringe at the edge so that it ran in under the flake, in two coats. The surface was stable at the end of treatment; it is a master's thesis and there is no follow-up (Mukherjee 2025).
- On Luca della Robbia's lunette in Urbino, the Opificio delle Pietre Dure (OPD) in Florence re-adhered glaze lifted by salts with a silica sol-gel. Adhesion was not measured, and under the microscope the silica film covers the original glaze. The substrate is porous terracotta (Afra et al., AIC OSG Postprints 29, 2022).
- On terracotta garden sphinxes with slip at the Amiens museum, three materials were used for three problems: ethyl silicate in the powdering areas, an acrylic dispersion injected by catheter into the gaps between laminae, and an acrylic resin bulked with microballoons in the large voids. The author notes that the consolidation did not manage to fill the voids between laminae. The pieces did not go back outdoors (Vétillard, CeROArt, 2012).
None of these cases publishes a medium-term follow-up of the bond.
Before choosing a material: five questions about the object
- Is there a gap under the layer, or is the layer crumbling? Raking light and a magnifier tell them apart: shadow and a lifted edge, or powder and loose grains. That is what decides whether you re-adhere, consolidate or do both.
- What is the layer? Glaze, slip or a finish applied cold after firing: unfired finishes on ceramics do exist (National Park Service, Museum Handbook, Appendix P). An organic finish can soften or dissolve in acetone, so any solvent is tested first on a hidden spot.
- Are there salts? In glazed ceramics, salts can be deposited between the glaze and the body and, if they are not removed, the glaze may flake off (Hamilton, Texas A&M University, 1999). If the piece cannot withstand washing, whatever is moving is secured before wetting and desalination comes afterwards. To check for them, the conductivity of deionized water in contact with the surface is a spot test (Frame, Segalman and White 2007); commercial chloride strips do not read below 500 ppm (parts per million), and a negative result does not prove absence (Skipper and Rubinstein 2018).
- Indoors or outdoors? The glass transition temperature (Tg) of B-72 is 37-41 °C (Podany et al., JAIC, 2001), and Koob 2018 does not recommend it outdoors or above 50 °C. The resin is chosen according to the temperature the surface will have to withstand; in the outdoor cases reviewed, the choice was a resin with a higher Tg or moving the object under cover.
- Has it been treated before? B-72 dissolves in acetone and in ethanol and is insoluble in white spirit (Preservation Equipment usage sheet). One swab with odorless petroleum essence and another with acetone, rolled without rubbing on a hidden area, help to find out what is on the surface before adding anything.
Protecting the surface while you work
Where there are flakes, applying a consolidant over the whole surface can dislodge fragments. The documented practice is to secure the flakes first with locally injected adhesive, often with a protective facing (Wheeler 2005, citing Nonfarmale 1976). For that facing, a lightweight Japanese paper is used, such as Tengujo Kashmir 9 g or Bib Tengujo 12 g.
In summary
| What you see | Operation | What was used in the cases |
|---|---|---|
| Intact layer, detached, with a gap beneath | Re-adhere | B-72 at 20% in acetone, or tube adhesive (~48-55.5% w/v) |
| Fine delamination, fragile layer but still in place | Consolidate the area | B-72 at 5%, several applications by micropipette |
| Porous body disaggregating under the layer | Mass consolidation | Ethyl silicate; dilute B-72 |
| Flake carrying body material on its back | Both: secure the flake first | Adhesive under the flake and consolidation of the body |
| Salts present | Secure what is moving, desalinate, then treat | As in the steps above |
Frequently asked questions
Can ethyl silicate be used to re-adhere lifting glaze?
For a flake with a gap beneath it, that is not what it does: its gel does not bridge gaps larger than about 50 µm (Wheeler 2005, citing Wendler 1999). Ethyl silicate consolidates porous, disaggregated bodies. If, besides the loose glaze, the body underneath is crumbling, those are two operations and they are done separately.
What concentration of Paraloid B-72 is used to re-adhere glaze flakes?
In the published cases, at 20% in acetone (Nederlands Tegelmuseum, 2012), or as a tube adhesive prepared in the studio, at around 48% w/v for glass and 55.5% w/v for porous earthenware (Koob 2018). Concentrations of 1.5 to 5% appear for consolidating, not for re-adhering.
Are colloidal silica and ethyl silicate the same thing?
No. Ethyl silicate is a tetraethoxysilane in solvent that forms silica inside the pore; colloidal silica is a water-based dispersion of silica particles that are already formed. The technical data sheet of one of the commercial products states that it penetrates less than ethyl silicate.
Should you desalinate before re-adhering?
It depends on how fragile the piece is. If it cannot withstand washing, whatever is moving is consolidated first and the piece is desalinated afterwards (Hamilton 1999). At the Nederlands Tegelmuseum the tiles were desalinated first and then consolidated and re-adhered; the tile that had already got wet before its fragile glaze was consolidated was desalinated separately, glaze side up, to keep damage to a minimum (Laanbroek, Manshanden and Overhoff 2012).
What this guide does not cover
- Structural joins of broken fragments, which have their own technique.
- Diagnosing the cause of the defect (firing defects, frost, glaze fit).
- Stone consolidation. A different family of problems.
- A specific object. None of the above replaces seeing the piece.
What you need
- Paraloid B-72, in 1 kg.
- Tengujo Japanese paper for the protective facing: Kashmir 9 g, Bib 12 g and 18 g.
- MIR odorless petroleum essence for the solvent test.
Everything is in Adhesives and Consolidants and in Ceramics and Glass.
If you have a specific case
Tell us what the object is, which layer is lifting and where it has been. We will tell you what to look at before you buy anything.
Álvaro