TECH REPORT: Devitrification of glass

Research Glassblower
Colorado State University
Department of Chemistry
Ft. Collins CO 80523-1872
(970) 491-5229 (voice)
(970) 491-1801 (FAX; attn. M Olsen)

Q: What is devitrification, how does it happen, and how can it be prevented?

A glass is by definition a liquid at all temperatures. Compositionally a typical glass consists of the 'glass (or network) formers' (SiO2 and B2O3), fluxes (also called alkaline metals, K2O and Na2O), and the network (or matrix) modifying alkaline earths (CaO, BaO, PbO, ZnO).

Other oxides capable of glass network formation include GeO2, P2O5, As2O5, As2O3, Sb2O3, and to a limited degree V2O5, ZrO2 and Bi2O3. The oxides of Ti, Zn, Pb, Al, Th and Be can be included in varying concentrations, but will not on their own, yield a glass. These, and other oxides that will not form a glass (including Sc, La, Y, Sn, Ga, In, Mg, Li, Sr, Cd, Rb, Hg, and Cs) are used as network modifiers.

Within the glassy matrix there are bridging oxygen ions and non-bridging oxygen ions (which ionically bond to the cations Li+, Na+, K+ or Ca++ in a typical glass). Technically, glass is a metal oxide polymer with SiO2 being the principal copolymer of all commercial glasses. In three dimensions, this polymer can be depicted as a random 'cobweb' of silicon ions each bonded to either three or four oxygen atoms. Progressing along any branch of the polymer chain we see Si-O-Si-O-Si-O-Si-O with an occasional matrix modifier ion (as Pb, lead) substituting for a Si. Branching from each Si will be another similar chain of O-Si-O-Si-etc. Occasionally, a chain will terminate at a non-bridging oxygen, which has a strong negative charge. This oxygen will be protruding into a microscopic cavity in the glassy matrix within which resides a counterbalancing, positively charged alkaline metal flux ion (as Na+, sodium).

A pure SiO2 glass is called fused quartz. It can be manufactured from rock quartz crystals, and has no flux nor matrix modifiers within it. Because of the uninterrupted strength of the Si-O bonds it yields a very high temperature softening ('hard') glass. Occasional interruptions of the Si-O matrix with modifying ions (with weaker metal-O bonds) will lower the viscosity profile of (or, 'soften') the glass. The more alkaline earth oxides added to the glass, the more the viscosity profile is depressed.

This is glass when everything 'goes right'. When glass devitrifies, it doesn't necessarily revert to its former, solid, crystalline constituents. In many instances a devitrified glass will not be chemically altered, but rather it will crystallize into its ceramic analogue.

Q: Is it true that devitrification can happen after prolonged stressing and exposure to contaminants and this in turn will result in a degradation of physical properties?

First, the jargonal term for devitrification is 'devit'. The term however describes more than one phenomenon.

Yes, both stress and surface contamination both result in what is termed devit, however I have never read nor heard a discussion of the former, and I honestly don't remember where I 'learned' of the decompositional (reductional or oxygen liberating) role I describe in a document discussing the role of oxygen in glass - I don't think I have anything in print that I can reference, so what follows is largely conjecture.

I have a theory about stress induced devit, and I'd appreciate any input outlining prior experiments conducted to test it.

First, it is commonly observed that a piece of glass, when heated to the point where it is somewhat plastic but not yet at its 'working point' (these glassworking points - including strain and annealing - are all best defined by specific viscosities) will exhibit a surface 'defect' termed 'devit' when a compressive or flexing force is applied. This devit commonly appears as a milky haze on only (a hint!) the outer surface of the workpiece, however it may also appear as a surface crazing.

In the experience of most glassblowers, the milky haze devit is reversible by 'burning' it back into the bulk - e.g. focused heating. It is 'common knowledge' among glassblowers that this burning procedure 're-melts' the devit. As discussed below, I do not agree with this.

The craze-type devit is a little more troublesome, as the repair procedure is the same however the fix is not always successful as there may be some 'wrinkles' in the glass that remain no matter what is done once they somehow become 'fixed' in the glass. The only thing that can then be done is to cut out, grind off, or (with a torch) 'pick' these defects from the surface of the workpiece.

As a result of 20+ years experience, I have observed that certain repair procedures for this 'stress-induced devit' are more successful than others, and I have now developed techniques which nearly always avoid inducing the phenomenon in the first place, and a repair procedure to reverse the phenomenon that is nearly 100% successful. But first, here are a couple of additional points that I think relate to the surface-devit phenomenon:

I do not recall ever seeing the milky haze devit in quartz (not to be confused with 'bloom').

I have induced 'wrinkles' in under-heated quartz, but I do not recall seeing quite the surface crazing that 'softer' (alkaline flux-containing) glasses commonly suffer from.

It is my impression that softer glasses are more prone to irreversible stress devit than the harder glasses. This needs to be tested. If true, it would strongly support my hypothesis of what underlies the phenomenon.

'Flame cutting' a piece of glass will often produce a quartzlet if the freshly 'torn' edge is exposed directly to a sharp (very hot and O2-rich) flame.

All of these phenomena can be diminished or avoided entirely if the effected glass is simply not exposed to a sharp flame! The 'devit' will simply flow back into the bulk.

MY HYPOTHESIS: Stress-induced devit is only indirectly induced by applied force and actually has little to do with either the force (stress) or deformation (strain). The minute surface geometric irregularities which are formed by applied force (wrinkles in regions of compression and crazing in regions of tension - neither of which affect the bulk structure of the glass) merely increase the surface to volume ratios of the outer few microns of the glass which allows these so-called 'devit' micro-structures to become super-heated by the flame, allowing the higher vapor pressure fluxes to volatilize, thus creating a harder, more quartz-like glass. Note specifically that what I am asserting is that stress-induced devit is not a form of devit at all but a form of chemical decomposition at the surface which results in a different glass!

If my hypothesis is true, it should be easily testable.

Return to Home

This page 2004 by Michael Olsen. All rights reserved. Permission is granted to link to this page, but you may not copy, modify, publish, or display any of the content of this page, in whole or in part, without the express written permission of Michael Olsen.

Last edited 02-18-03