Fundamentals of Inorganic Glasses Arun K. Varshneya New York State College of Ceramics Alfred University Alfred, New York @ ACADEMIC PRESS, INC. Harcourt Brace & Company, Publishers Boston San Diego New York London Sydney Tokyo Toronto“This hook is printed on acihtce pape. @ Copyright © 194 by Asidemie Pres, le AD igs eeserved. No part ofthis publication may be reproduced or transmitted in tng form or b) ay means electronic oF rechanicalineuding photacopy, recording, or sny ilormnation storage ard retrieel System: without pesmission in writing for the publisher. Academic Pes, Ine 1250 Sixth Avenue, San Diego, CA 92101-4311 United Kingom Eaton published by Academie Prse Limited 24-28 Oval Road, London NWI 7DX Library of Congress Catalogin-i-Publiaton Data Vorshneya, Arun K. Fundementals of inorganic glases / Arun K, Varshney Includes bibliographical reerences end index ISBN 012-714970.8 (acide paper) 1 Glass, 1 Tile TPRSTS? 1993 620.44 —8e20 5.16391 cP Printed in the United States of Americn 9394 959697 BBY #7654321aA 7 Fundamentals of organic Glasses foal vests 70} 2 & non of temperature oft) smd (eile laser (ater Litton and ‘Sow of he Aiernn Ceram Sec reduced time. It therefore relaxes to the areas: description of far-from-cquitirium relaxation function measured using one as much of the relaxation data for smal re are consistent with the assumption of hat TRS could not deseribe the enthalpy S fibers. Likewise, Moynihan et al. (23) ‘Jaxation function may be property-type- Tao Te 020 th time at $30°C for sodaclime silica glass 500°C {0} (Data of Haraand Suctoshi [27] sensitive, For instance, the fictive temperatu diferent from that for enthalpy relaxation |. ues Tanaion Range Behavior 27 138. Development of Permanent Stresses in Glass by Cooling through the Glass Transition Range | SRE weadg:R.Gardon and ©. & Neayanevamy. J. Am. Ceram | Soc. $3(7), 380 (1970), | B Siardon Glass Sci & Tech. Vol. $ (D, R. Uhlmann and Ni. 4. Kreidl, | 30) pp 146-217, Academic Press New Var, So [08.Neraanaswans, am. Cham, Soe Boy 6-189 97m i i dercing Bt 8 cooled through the glass transition range, permanent stresses develop in diferent regions. In the later chapter on the can srl be emphasized that glass invariably Breaks in tence regions of ath of glass, it Hence, those ss that develop permanent compression upon cooling are srcnathened. and those where tension develops are weakened Since com- & Products fos sim 80 hand in hand, the forming technologs op lass Frat {or low scrap rate during process and 1 sale predece vn service | demands attention in relation to the magnitude and localion ane developed stresses in glass | _itatss were pert lati cooling fom the outside wou generate Danita sey cui 8d compression on the inside (Section tog ne | Sent that aoc, StU cooling generates a parabolic temperate se renpereenngilts in a parabolic stress gradient. Upon the slate ie igless wore Redd a FOO temperature, al sreses Would alse daapoees H2lss were Aud any stresses that developed would be soc rapidly even tat secon of epee sradiens Is inte ae tansiies eo te tal Feateangement occurring at diferent rates av ania et permaseas co ancompletestess relaxation, and hence tothe soho se | Remanent sires, Qualitatively, there [stress development: the or permanen ecurtet clasied depending upon which of te wesc ens {permanent structural heterogeneities cause it Trex are discussed next | [HE Miscottasric conteisuTion. This mechanism ig schematically shown £&f 2 alass plate in Fig. 13-35. In (a) the glass plowe been high uniform _feumerature and hence no stress a all. Upon cooling a temperature gradient éevelops, cooler on the outside and warmer om the Inside, as shown in (b) 'n a perfectly clastic body. such temperature paces Would cause theFigure 18-85. Simplified concept of permanent sires production in glass due to “frozen temperature gradient.” (a) Glass with no temperature gradient well above the transition region has no stress (6) Temperature gradient develops upon cooling; however, no stress develops beceuse of rapid relaxation while the glass remains well shove Ty {@) Cooling to below T, while maintaining the same “frozen temperature tradient™ produces no stress yet. (@) Finel emoval of the temperature gragient produces stresses that are now permanent, development of tension on the outside and compression on the inside, However, since glass is still well above T, into the supercooled liquid region, these stresses relax rapidly, In (o}, the’ plate has been brought down in temperature with the same “frozen” temperature gradient — however, in such a manner that the outside has reached room temperature. Ithe effective thermal contraction coefficient of the body is assumed to be independent of temperature, then since each layer of glass cooled through exactly the same temperature interval, the various layers of the plate in situation (e) still do not have any dimensional incompatibility. Hence, the plate does not develop any stresses, In the next step, we superimpose a reverse temperature gradient (as in (d}) to make the temperature uniform (room temperature) everywhere. Such a gradient where the outside is warmer than the inside would cause the generation of a stress gradient with compression on the outside and tension on the inside. We may recognize that the final stress distributio the negative of the stress that relaxed viscoclastically. It really does not matter how many steps we take fo arrive at the final state. We are allowed to perform these idealized operations conceptually, hecause of linearity and the Boltzmann superposition principlepermanent stress production in glass due 102 ‘ass with no temperature gradient well above Temperature gradient develops upon cooling; f rapid relaxation while the glass remains wel ‘le maintaining the same “frozen temperature 1 Final removal of the temperature gradient outside and compression on the inside rove T, into the supercooled liquid region, 3, the’ plate has been brought dowa in sn” temperature gradient — however, in reached room temperature. I'the effective the body is assumed 10 be independent of ‘of glass cooled through exactly the same layers of the plate in situation () still do uibility, Henee, the plate does not develop sperimpose a reverse temperature gradient + uniform (room temperature) everywhere is warmer than the inside would cause nit with compression on the outside and sognize that the final stress distribution is Yaxed viscoclastically. It really does not 6 arrive at the final state. We are allowed ‘ons eonceptusily, because of linearity and ciple | | | i | | | | | | t ‘lass Transition Range Behavior 299 [STRUCTURAL CONTRIAUTION DUE TO TRANSIENT HETEROGENESTY. In the preced~ ing discussion, we ignored the rather large volume change a glass-forming liquid undergoes during and above the glass transition range. For simplicity the structure, and hence the volume, may be characterized by the fictive temperature, 7;, which is essentially the physical temperature in the liquid range and which lags gradually in the transition range until “frozen” at some value Tj, in the “appareatly solid” state. When a glass with a temperature gradient is being cooled uniformly through the transition range, the various layers also possess a gradient in T;, ie., a structural heterogeneity, whieh vanishes rapidly as the glass approaches Jj, just below the transition range. We noted earlier that the removal of the physical temperature gradient AT in the solid state brings about permanent stresses. Likewise, we may expect the development of permanent stresses due to the removal of the fictive temperature gradient, There is one difference, however, which is best under- stood with the help of Fig, 13-36. The figure shows the temperature, time and stress development interrelationship. Since the fictive temperature gra- ) (After Gardon, i ono. 7th dnt Cong Glass, paper no, 79, Inst. Nat ci Verre, Chariot, Belgium, 1965), Figure 19-41. Degree of temper as a jon of the different parts of a mercury thermometer homogeneity of expar ould cause the thermometer to deform ‘Quneaf the simpler procedures employed for annealing isto hold the glass axsround the snnealing temperature, for 15 min, and then eBol Sow 1 Sratb forthe buildup of stresses to well below the strain point, after which ihe ware is cooled a fapidly as the thermal endurance OT tbe e glass If we assume a KWW relaxation law (Eq. (13.76)}, an isothermal hold for is min at 10"? Pars viscosity (assured shear modulus G = 3 x 10%° Pa) gives S,a/Sya(0) = expt —{900/(1072/3 x 10'4)}°*] = 0.0055. In other words, essentially all the inital stress relaxes. At strain point {hiscosity ~ 10°? Pas),a similar release would require roughly eight hours Fcnane glasses in general do not all conform to a KWW relation, and tocanse the value of G is not a universal constant, itis best to acknowledge the annealing point as the temperature where stress release occurs jn 2 matter Shminotes and the strain point as the temperature where the stress releastsn teavent #2 are) | eel ion ce cone icereaste teed se) Temoarotire To (°C) function of initial glass temperature and beat sg cooling, Glass thickness 0.61 em, To convert We atPay, nultiply by 0.039. Experiments results wgted by Barienev’ theory fn). (After Gardon, ‘inst Nat du Verre,Charoi, Belgium, 1965) te different parts of a mercury thermometer + deform. ‘cemployedt for annealing is to hol {ature for 15 min, and thes cool lowly fo 5) to well below the strain point, ater which She thermal endirance of the ass will allow: Givlaw (Eg, (13:76), an isothermal hold for Samed shear modulus G = 3 x 50!” Pa) si the glass = 0,0058. {909,(10"2/3 x 10°) I the initial stress relaxes. At strain point Jar release would reguire roughly eight hours sot all conform to a KWW relation, nd is best to acknowledge yours in a matter Ine stress releas® «universal constant, iti erature where stress release 0% at as the temperature where ¢ ‘loss Transltion Range Behavior 307 Annealing point _ paint Figure 13-42. Suggested schedules for commercial annealing of soda-lime si igassware. (Courtesy of Corning, Ine) eee takes 2 few hours. common practice to perform annealing is to follow a sel in Fig, 13-42, The various regions of the schedule are given b; tale shown A (heating rate) = 839/[zd") ‘min B (holding time) yd min C (slow coo!) = 42.6/fxd*)°C/min X(C below strain point) = $°C for 0.3 em thick plate = 10°C for 06cm thick plate PC for 1.3 em thick plate (fast coo!) = 2 times the slow coo! E (final cool) = no greater than 10 times the slow coot (13.95) here « is the thermal expansion coefficient of gl " ransion coefficient of glass in 10°7/°C units, and {iste thicknes of the las (em), when cooled from one side oly, and the illthickness (em), when the glass is cooled from both sides. The re ‘commended values of A-E for three different values of x and three different plate thicknesses are listed in Table 13-4Fundamentole of Inorganic Glassas ov, 3045 om wo Buy00>, anpoid Se ais ‘ret agen ax os™ g 130 3x 107 0 % 0 260 % 88 ss 5010-7 ass Transiton Range Behavior 308 43.410. Summary of Glass Transition Range Behavior Upon cooling a liquid through the supercooled region, the viscosity rapidly rises. Glass transition occurs when the viscosity is of the order of 10"? Pas, ‘The intersection of the extrapolated liquid line and the glassy state line on the ¥-T diagram is called the “fictive temperature 7,” of the glass. In approximate language it may be stated that the structure of the glass corresponds to the structure of the equilibrium liquid et ‘Jj at constant pressure, Usually, the rapid rise in viscosity implies & rapid reduction in free volume However, there are exceptions, particularly vitreous silica, where the volume ‘ofa corresponding crystal is higher than that of the glass In such cases, the meaning of the free volume is not entirely clear. A more appealing description is the Adam-Gibbs concept, where the increase in viscosity with falling temperatures is ascribed to rapid loss of configurational entropy In the transition range, both the action of 2 mechanical impulse and a temperature impulse cause the system to be displaced from equilibrium. As a resoll, glass undergoes structural rearrangement, The physical properties, therefore, also vary. In the impulse at constant temperature, the change in physical properties due to structural rearrange- ment is small when the applied deformations are small {ie., for small displacement out of the equilibrium), For such a case, the assumption of linear viscoctasticty is generally applicable, Glass may behave as a combina~ tion of Hookean springs and Newtonian dashpots, offen represented by 4 Maxwell element, a Voigt-Kelvin element, or a Burger element. The former two are based upon a single exponential relaxation retardation law. The latter two simulate the delayed elasticity observed commonly. in lass The change of temperature, on the other hand, brings about a large change in viscosity, causing the approach to equilibrium to become nonlinear. However, such nonlinearity, being due to the change in viscosity in part due to the changing temperature and due to the changing fictive temperature (structure), can still be converted to linearity by compressing or expanding the time scale to that at a reference temperature (again, as long as the deviations from the equilibrium are small) using the assumption of the thermorheological simplicity. At any point during the cooling of a glass, the structural change through a small step of physical temperature change can be found to be equivalent to a small change in the fictive temperature on the equilibrium liquid line, Because of nonlinearity, however, upon cooling through the glass transition range, glass retains a particular fictive history lor a memory) characteristic of the cooling path. ‘The fictive history de ‘ermines the value of the physical properties such as volume, specific heat of a mech: