May 17, 1966 R.

GARDON 3,251,671
- METHOD OF ANNEALING OF GLASS RIBBON
Filed Feb. 5, 1962 9 Sheets-Sheet

F.G.

U
S.u
o
o

2OO

IOO
ric
V O
t t2. t3
J

TIME, t, SEC.
sINGLE BREAK"ANNEALING WITH R> R2
INVENTOR.
Aeoaaar Gaaloo/V
BY

a 77OAeMay
May 17, 1966 R. GARDON 3,251,671
METHOD OF ANNEALING OF GLASS RIBBON
Filed Feb. 5 , l962 9 Sheets-Sheet 2
F. G. 2

N ANNEAung
RANGE
O SOO - N -

E.- fu
IN 2A
U
-

tu
35.
2
-

8
300
Z . 200
O2
2S
He
100
O

t
U
4 -100

-200 O SO 1OO 90

TIME, t,sec.
singLE BREAK ANNEALING WITH R < R2

INVENTOR.
3OAA767 Ga?cAO/y
BY

94thw a 77 oe/Way
May 17, 1966 R, GARDON 3,251,671
METHOD OF ANNEALING OF GLASS RIBBON
Filed Feb. 5, 1962 9 Sheets-Sheet 3
F.G. 3A

180

60

O2 O.4 O,6 0.8 2. 3 4 5

RATIO OF COOLING RATES, R/R2 INVENTOR.
aAAa7 Gaa AOM
BY

a77Oaway
May 17, 1966 R. GARDON 3,251,671
METHOD OF ANNEALING OF GLASS RIBBON
Filed Feb. 5, 1962 9 Sheets-Sheet 4

F. G. 4

4A

SN
3.
TIME, t, sec.

INVENTOR.
Maoaaa/ Caadow
BY

a 77aa/May
May 17, 1966 R. GARDON 3,251,671
METHOD OF ANNEALING OF GLASS RIBBON
Filed Feb. 5, 1962 9 Sheets-Sheet 5

do M.L.
007 Nye INVENTOR.

BY Moaaay Gaé202/V

477 Oaway
May 17, 1966 R. GARDON 3,251,671
METHOD OF ANNEALING OF GLASS RIBBON
Filed Feb. 5, 1962 9 Sheets-Sheet 6

F. G. 8
O TEMPOC
O7
9 27
.8
7 S38
.6
5
4. 5A9

09
56 | | | |
07O OIIHI
20 30 A0 50 60 7O 80 90 OO
TME, (SEC)
GENERALZED STRESS RELAXATION CURVES
FO2. SHEET GLASS USED N LLUSTRATIVE
CALCULATON

INVENTOR.
M32a3A%27 GaAOMy

177 Oaway
May 17, 1966 R. GARDON 3,251,671
METHOD OF ANNEALING OF GLASS RIBBON
Filed Feb. 5, 1962 9 Sheets-Sheet 7

its?2

INVENTOR.
-47 eaazer Gaédéw
BY

a 77aa/WAY
May 17, 1966 R, GARDON 3,251,671
METHOD OF ANNEALING OF GLASS RIBBON
Filed Feb. 5, 1962 9. Sheets-Sheet 9

INVENTOR.
F. G. 12 BY aOaaa/ Gaadow

a 77OAWay
United States Patent Office 3,251,671
Patented May 17, 1966
2
temperature of the ribbon from the forming temperature
3,251,671 to the upper limit of the annealing range rapidly and uni
METHOD OF ANNEALNG OF GLASS RBBON formly, and then holding the temperature constant
Robert Gardon, Farmington, Mich., assignor to Pittsburgh throughout the area being annealed for the proper length
Plate Glass Company, Pittsburgh, Pa., a corporation
of Pennsylvania of time. Because of these difficulties in accurately foll
Filed Feb. 5, 1962, Ser. No. 170,989 lowing the prescribed theoretical curve of Adams and Wil
9 Claims. (C. 65-118) liamson, industrial processes have generally only followed
the annealing curve of Adams and Williamson approxi
This invention relates to the annealing of glass and mately or have attempted to establish constant cooling
especially to methods for cooling and annealing a con 10 rates through the annealing range. Such schedules gen
tinuous glass ribbon in an easily controlled manner and erally involve a gradual cooling through the annealing
in a relatively short time while achieving relatively low range with a reduction in the cooling time above and
levels of resultant stress in the finished products. below the critical range where the cooling process does
The necessity for applying a controlled cooling sched not affect the residual stress.
ule to reduce the stress which appears in glass products 5 U.S. Patents Nos. 2,774,190 and 2,952,097 are examples
as they are cooled from their molten form or from a tem of the application of the theoretical annealing curve of
perature to which they were heated to relieve existing Adams and Williamson to an industrial process of glass
internal stresses is well recognized. In the production of a drawing. These patents indicate the necessity of main
continuous glass sheet or ribbon, as by a drawing process, taining a relatively high temperature of the glass for a
proper annealing is necessary to produce glass which has 20 Substantial proportion of the annealing range and for
a Substantially uniform over-all residual stress pattern accurately controlling the cooling curve through the an
and acceptable cutting characteristics. There is an opti nealing range.
mum range of residual stress in sheet glass which renders The present invention, relating to a novel annealing
the glass most suitable for cutting. High stress values tend Schedule, has eliminated the necessity for maintaining the
to increase cutting difficulties by increasing the tendency 25 glass at a high constant temperature through a part of
of a running fracture to depart from a score line made the annealing range and does not require cooling the
by a cutting tool. glass through the annealing range at progressively increas
Broadly, the process of reduced residual stress involves ing cooling rates, which are difficult to control and dupli
the cooling of glass, formed from a molten source or heat cate. At the same time, the minimum degree of anneal
ed to a temperature at which existing stresses relax, in a 30 can be achieved, as compared with the known theoretical
manner such that substantial residual stresses will not calculations of Adams and Williamson, consistent with the
reappear when the glass temperature has reached equilib original stress conditions and the time available for anneal
rium. To minimize stress, exact control of temperature ing. This has now been made possible by the realization
is important only during a narrow temperature range that a temperature gradient can be introduced in stress
the annealing range. The upper limit of the annealing 35 free glass being cooled through the annealing range of
range is capable of practical definition and determination the glass to establish actual stresses in the glass within the
for a given glass. In general, it is the lowest temperature annealing range which, by relaxation through viscous flow,
at which the relaxation of stresses is so rapid that stresses reduce the final stress appearing in the glass. For exam
cannot be detected on the time scale of the process being ple, if a glass plate just produced from a melt and still
considered. The lower limit of the annealing range is 40 at an elevated temperature is cooled and a temperature
more indefinite but is generally taken as that temperature gradient impressed upon it, no stress will be produced be
from which a piece of glass can be quickly cooled without cause the glass is too fluid. If the glass is cooled through
introducing permanent stress. It must be recognized the annealing range with the temperature distribution re
that the temperature limits of the annealing range for a maining constant, no stress will be induced and hence
given glass composition are not fixed but, rather, vary with 45 none will relax; yet, when the temperature is allowed to
the annealing time. equalize below the annealing range, the glass will be found
In accordance with this invention, it has been found to have a stress distribution proportional to the cooling
that in annealing continuous glass sheet which is produced rate. If, however, this same temperature distribution is
by the conventional drawing or forming processes where used to establish an actual stress, as by rapidly changing
short annealing times (for example, of the order of 100 50 the gradient within the annealing range, the stress estab
seconds) are available, the annealing range for a given lished is capable, by relaxation through viscous flow of
glass is somewhat higher than that heretofore regarded to the glass, of reducing the final resultant stress in the glass
be valid. at isothermal room temperature conditions. By comput
Heretofore, to obtain the lowest residual stress with the ing the retained stress during intervals of relaxation until
greatest economy of time, an annealing procedure devel 55 the temperature "breaks' again with a resulting instan
oped by Adams and Williamson has been applied. This taneous change in stress, or until the glass passes out of
procedure consists of two essential steps: (1) holding or the annealing range below which stresses do not relax
soaking the glass at a constant temperature high in the but change only as the temperature distribution changes,
annealing range for a portion of the annealing time and the final residual permanent stress in the glass may be
then (2) cooling the glass from such a high constant tem 60 calculated. This treatment of the relaxation of actual in
perature to the lower limit of the annealing range at a stantaneous stresses differs from the treatment of stress
gradually increasing rate so as to introduce a final permis relaxation of Adams and Williamson which deals with
sible stress during the remainder of the available anneal the relaxation of what might be called potential stresses,
ing time. The success of this method is dependent upon i.e., stresses existing in the cold isothermal glass. This
an accurate computation of the initial annealing tempera 65 new treatment leads to a different method of annealing
ture selected and the cooling rate, and the cycles must be glass.
followed very closely. From a practical standpoint, it is In this approach to annealing, then, which forms the
difficult to apply this procedure to a continuous ribbon of basis of the present invention, the final permanent stress
glass of Substantial width. It necessitates lowering the is considered a function of both the temperature distribu
70 tion in the glass above the annealing range and the
L. H. Adams and E. D. Williamson, J. Franklin Inst., amount of relaxation of instantaneously existing stresses,
190,597-631 and 835–868 (1920). Moreover, the relaxation of instantaneously existing
 3,251,671
3. 4.
stresses may or may not induce a final permanent Stress FIG. 2 shows a series of three graphs for a second cool
in the glass, as will be shown. This treatment of stress ing schedule of the present invention in which the tem
relaxation does not contradict the well-known principle perature (FIG. 2A), the cooling rate and the thermal
set forth by Adams and Williamson, but, rather provides gradient (FIG. 2B) and the center tension (FIG. 2C)
a new method of calculation which allows the determina are plotted on a common time scale.
tion of new and equally effective annealing schedules for F.G. 3 shows temperature-time curves which traverse
glass. the annealing range within a specified time (FIG. 3A)
The present invention is concerned with methods for and the residual center tension resulting from following
cooling glass in an easily controlled manner to produce such representative curves (FIG. 3B).
minimum residual stresses in a given annealing time. O FIG. 4 shows a series of three graphs for a preferred
This is accomplished through a novel annealing Schedule. cooling schedule of the present invention in which the
Broadly, this schedule comprises cooling the glass through temperature (FIG. 4A), the cooling rate and the thermal
the annealing range in a plurality of distinct cooling rates, gradient (FIG. 4B) and the center tension (FIG. 4C)
each of which may be. Substantially constant for its dura are plotted on a common time scale.
tion. Preferably, the glass will be cooled rapidly from 15 FiGS. 5, 6 and 7 each show a temperature vs. time
its forming temperature to a point within the annealing curve for a specific example of an annealing schedule
range, whereupon it will be cooled at a second diminished (FIGS. 5A, 6A and 7A) and a family of curves indi
substantially linear cooling rate and thereafter, still with cating the residual stresses produced by the annealing
in the annealing range, will be cooled at a third Sub schedules (FIGS. 5B, 6B and 7B).
stantially constant but increased cooling rate to at least 20 FIG. 8 is a graph showing an example of the stress re
the lower limit of the annealing range. The time during laxation curves used in the method of calculation dis
which each change to a different cooling rate takes place closed herein. These curves were derived from the data
must be extremely short relative to the time duration of for the stress relaxation of a particular glass composition,
the rates themselves. Alternative cooling schedules could as disclosed by A. F. Van Zee and H. M. Noritake,
include curves utilizing more than three distinct cooling 25
rates. The preference for three cooling steps is based “Measurement of Stress Optical Coefficient and Rate of
on practical engineering considerations. Stress Release in Commercial Soda-Lime Glasses,' J.
Advantages attendant upon this process include the Am. Ceram. Soc., 41, 164-175, 1958.
relative ease by which linear rates of cooling may be es FiG. 9 is a schematic, perspective view of a glass draw
tablished and maintained or reproduced as compared with 30
ing machine illustrating a preferred embodiment of the
nonlinear rates, the low requirements of additional heat invention.
for controlling an annealing schedule in which the glass is FIG. 10 is a chart showing preferred cooling curves
continually cooled as compared with a schedule in which for vertically drawn sheet glass wherein the surface tem
the glass must be maintained at an elevated temperature peratures of different thicknesses of glass ribbons having
for a substantial portion of the available annealing time, identical mean temperatures are compared with their ver
and the reduction of the susceptibility of the glass to tical positions in a schematically shown vertically dis
roll marking or other surface deformation by virtue of posed lehr during the upward travel of each ribbon.
the preferred rapid lowering of the ribbon temperature FIG. 11 is a chart showing a preferred cooling curve
during the initial stages of annealing. . for horizontally rolled glass wherein the temperature of
Just as industrial applications of other annealing proce 40 glass ribbon is compared with its horizontal position in
dures have resulted in annealing schedules simplified from a schematically shown horizontally disposed lehr during
more ideal curves, it is possible to further simplify the pre the travel of the ribbon.
ferred embodiment of the instant invention. It has been FIG. 12 is a chart showing a preferred cooling curve
found that an acceptable anneal can be obtained using the for reannealing glass sheets or plates, either flat or curved,
teachings of this invention by utilizing only two cooling wherein the temperature of the glass is compared with
rates within the annealing range. Where the time avail 45 its position in a schematically shown horizontally dis
able for annealing is determined, it has been found that posed lehr during the travel of the glass.
certain combinations of two separate cooling rates within An understanding of the theory upon which the present
the annealing range produce a product having lower result invention is based may be best obtained by following
ant stress than cooling at a single constant rate. The first two annealing schedules, as shown in FIGS. 1 and 2,
of the two cooling rates will most advantageously be more 50 respectively, of the drawings, which focus attention on the
rapid than the constant rate of cooling so as to reduce two principal mechanisms involved: (1) the origin of
the susceptibility of the glass to marking from the draw stresses in glass by rapid changes in temperature distribu
ing rolls during the initial stages of the drawing opera tion and (2) the relaxation of stresses by viscous flow.
tion, thereafter changing to a diminished cooling rate in In developing these theoretical concepts, the fact that the
termediate the annealing range so that the temperature of 55 corresponding intrinsic rates of stress release differ in
the glass reaches the lower limit of the annealing range various parts of a glass sheet which are at different tem
in the same time as the straight line schedule. Alterna peratures has been neglected. For example, as glass ap
tively, the first cooling rate may be more gradual than the proaches the lower limit of the annealing range, its sur
straight line rate, with a change intermediate the anneal face layers may have been cooled below this limit while
ing range to a more rapid rate. In either case, a proper 60 the central parts are still capable of viscous flow. Con
selection of the cooling rates and the temperature at which siderations of differences thus created have been neglected
the change in the cooling rates is made will result in a in this simplified treatment.
final product having less residual stress than if cooled at The two schedules indicate that the relaxation of in
a constant rate through the same temperature range in the stantaneous stresses may either help or hinder the attain
same time.
Other attendant advantages of this invention will be ment of a low final permanent stress, and that there is
more readily appreciated as the same becomes better an optimum Schedule to produce the least permanent
StreSS.
understood by reference to the following detailed descrip FIG. 1 is composed of three graphs, 1A in which the
tion when considered in connection with the accompany temperature of a glass sheet undergoing a cooling process
ing drawings in which: 70 is plotted against time, 1B in which the cooling rate
FIG. 1 shows a series of three graphs for a cooling R and the thermal gradient AT between the center and
schedule of the present invention in which the temperature the Surfaces of the glass sheet are plotted against time,
(FIG. 1A), the cooling rate and the thermal gradient and 1C in which the instantaneously existing stress in the
(FIG. 18), and the center tension (FIG. 1C) are plotted glass in terms of center tension expressed in central bire
on a common time scale. 75 fringence Ac is plotted against time. Because the time
 3,251,671
5. 6
scale of the annealing processes discussed herein is of a available for annealing is often limited in an industrial
magnitude that-except for a very short transition period process, there exists within the available time an optimum
following changes in cooling rate-the temperature dis combination of the temperature at which the cooling rate
tributions within the glass may be considered in their is changed and either the rate to which it is changed, or
quasiequilibrium state, i.e., parabolically distributed. the time interval ti-ta between the change in the cooling
The stress distributions, therefore, will also be parabolic rate and the time at which the glass reaches the lower
ally distributed and can be conveniently characterized by limit of the annealing range. The graph of FIG. 1C
a single parameter such as the instantaneously existing shows that the relaxation of the instantaneously induced
center tension. stress caused by the change in cooling rates decreases the
A cooling gradient applied to the glass sheet above the 10 center tension and thus the final stress in the glass.
annealing range of the glass cannot produce a stress in As the cooling schedule and stress curve in the graphs
the glass because any stress would, due to the low viscosity of FIG. 2 indicate, the relaxation of temporary stresses
of the glass, instantaneously relax through viscous flow; does not necessarily diminish the final permanent stress.
and that no stress will be produced as the glass is cooled FIG. 2A shows a cooling schedule in which the glass is
at a constant rate (which corresponds to a constant ther cooled from above the annealing range to a temperature
mal gradient as may be seen by correlating the curves of TB within the annealing range, and at time t is cooled at
FIGS. 1A and 1B), it will be evident that there will be an increased rate through the annealing range to room
no stress produced in the glass until the rate of cooling is temperature where the glass becomes isothermal. In this
changed at some point below the upper limit of the -an case, the change in cooling rates produces an increase in
nealing range. As shown in FIG. 1A, a glass sheet is 20 the temperature difference between the mid-plane and the
cooled at a first rate R1 from above the annealing range Surface, as shown by the graph and schematic inserts in
to a temperature TB within the annealing range. FIG. 1C FIG. 2B. The corresponding elastic response of the glass
indicates that the glass being cooled at rate R1 by the is for the center of the sheet to be put in compression, as
thermal gradient AT in FIG 1B is stress free. If the con 25 is indicated by the line AB in FIG. 2C. This stress re
stant cooling curve is changed to a second constant cool laxes as in the previous example until the temperature of
ing rate R2 at a point within the annealing range, as illus the glass passes through the lower limit of the annealing
trated at point TB in FIG. 1A, an instantaneous stress range at time to, after which the stress remains constant
proportional to the change in cooling rates is also pro until the gradual disappearance of the temperature gradi
duced. Regarding the relatively rapid change in tempera ent brings the glass to its final stress condition. In this
ture distribution at time t in FIG. 1A as instantaneous, 30 case, the relaxation of stress from time t, to time t is in
the dominant response of the glass is likely to be an elastic the positive direction, i.e., in the direction of increased
one. Thus, the sudden diminution of AT (the tempera center tension, and it therefore tends to increase the re
ture difference between the surface and mid-plane of the sidual stress. While it would appear to be desirable to
plate as indicated by the inserts in FIG. 1B diagram Suppress this stress relaxation, as by increasing the termi
matically showing the temperature distribution through nal cooling rate so as to diminish the time available for
the plate thickness) is expected to produce an instantane stress relaxation, the establishment of an increased rate
ous increase in center tension. See FIG. C. If the glass would increase the step in stress from D to E (which is
were elastic, the stress induced would remain constant as proportional to the terminal cooling rate) and would off
long as the temperature distribution in the glass did not set the gain due to the reduction of stress release between
again change. However, because the glass, when at a tem 40 B and C. Once again, for any externally imposed condi
perature within the annealing range, is not elastic, it tions there will exist an optimum combination of the
physically adjusts through viscous flow to relax the in break temperature, i.e., the temperature at which the rate
stantaneously induced stress. The relaxation occurs at a of cooling is changed, and the terminal cooling rate that
diminishing rate because the absolute level of stress and will produce the least amount of final permanent stress.
the temperature of the glass are decreasing. Relaxation A Specific example of the application of the above prin
continues, as shown in FIG. 1C, until the temperature of ciples to the annealing process used for drawing a con
the glass reaches the lower limit of the annealing range. tinuous ribbon of glass is illustrated in the graphs of
From the lower limit of the annealing range to room FIGS. 3A and 3B, as applicable to glass of 742-inch thick
temperature the glass is essentially an elastic material and ness, the annealing taking place within the short periods
the permanent stress is not changed by that portion of the of time herein contemplated. The speed of draw, which
cooling schedule between ta and ta. As the temperature is determined by factors relating to the phenomenon of
of the glass approaches room temperature, the cooling ribbon formation and desired thickness, governs the time
rate which was constant is diminished as is, therefore, the in which the glass will pass through the annealing lehr
temperature difference between the surface and mid-plane of the drawing machine. Such factors as the susceptibility
of the glass. Because the glass is now an elastic material, 55 of hot glass to marking and deformation by the drawing
this decrease in AT from time t3 results in an increase in rolls, and the necessity for lowering the temperature of
stress which is one component of the final perma the glass to about 90 degrees centigrade by the time it
nent stress produced in the glass by the thermal history. reaches a cut-off station where it must be cut and handled,
This stress is equivalent to the stress which would have make it desirable to lower the temperature of the glass
been produced by maintaining a constant cooling rate 60 through the annealing range at an early stage after its
corresponding to the terminal rate R2 through the anneal formation. The specific distance during which the glass
ing range plus the unrelaxed stress from the change in is annealed in the drawing machine is arbitrarily estab
cooling rates within this annealing range. It is also lished, consistent with the above-mentioned manufactur
equivalent to the stress which would have resulted from ing factors and the degree of anneal which can be attained
continuing the first cooling rate through the annealing 65 during the time it takes the glass to traverse any estab
range, less the amount of the stress caused by the change lished distance. It happens that, once a distance zone of
in cooling rates which relaxes. the machine is established for annealing, it is possible to
The final permanet stress could, of course, be dimin anneal within that zone all the different thicknesses of
ished by decreasing the terminal cooling rate. R2. How glass drawn by the machine. This is possible, notwith
ever, this would take more time for annealing. The final standing the different drawing speeds (and therefore dif
70 ferent times within the annealing range) required for the
permanent stress could also be diminished by increasing
the amount of stress which relaxes, as by lengthening the different thicknesses of glass, because thinner ribbons of
time interval ti-ta between the change in the cooling rate glass which are drawn more rapidly than thicker ribbons
and the time at which the temperature of the glass reaches also cool more rapidly. This characteristic is graphically
the lower limit of the annealing range. Because the time 75 illustrated by the curves in FIG. 10 of the drawings.
 3,251,671
7 8
In the description of the annealing schedules included cooling schedule passing from the upper limit to the
herein, all temperatures, unless otherwise specified, refer lower limit of the annealing range in the specified time
to the mean temperature of the glass. Because the pres of anneal can be characterized by one additional inde
ent annealing schedules are primarily applicable to what pendent variable, the initial rate of cooling R or the
is known in the industry as "flat glass,' i.e., glass sheets, terminal rate of cooling R2, or the ratio of these two.
plates and other glass products having essentially parallel There is shown in FIG. 3B a plot of the residual stresses
major surfaces and including bent or curved sheets or produced in 742-inch thick glass as a function of the
plates, the temperature at the surface or mid-plane of break temperature TB and the ratio of the initial and
the glass, as well as the mean temperature, could be used terminal cooling rates for single break annealing sched
in specifying the annealing schedules. IO ules in which the glass is cooled from 600 to 526 de
Temperature limits of annealing ranges, as specified grees centigrade in 100 seconds. All of these curves
herein, are capable of practical determination by com pass through point p, which corresponds to cooling with
paring stress in a glass sample cooled at a constant cooling out a break and a terminal stress of 145 millimicrons per
rate through a temperature range which clearly encom inch. With the break occurring at 570 degrees centi
passes the annealing range, with the stress in various sann 5 grade, it may be seen that the lowest obtainable degree
ples cooled at two cooling rates, one of which is the of stress occurs where the ratio of cooling rates is equal
same as the constant cooling rate of the first sample. to 1, i.e., the condition of no break at all. For breaks
As long as the change in cooling rates occurs above or occurring at temperatures below 570 degrees centigrade,
below the limits of the annealing range and the constant decreasing the ratio of the initial cooling rate to the ter
cooling rate equivalent to that of the first sample is main 20 minal cooling rate produces a lower residual stress, while
tained through the annealing range, the resultant stress increasing the ratio increases the residual stress. On
will remain the same. Once the change in temperature the other hand, when the break occurs at temperatures
occurs within the annealing range, the resultant stress will higher than 570 degrees centigrade, lower terminal
change because the cooling schedule through the anneal stresses are produced where the initial cooling rate is
ing range will no longer be a single constant rate. Hence, 25 greater than the terminal cooling rate. Break tempéra
if the temperature at which the cooling rates are changed tures which occur higher than 570 degrees centigrade
is varied, the limits of the annealing range will manifest have very well defined minima showing that, for any
themselves by a change in resultant stress because of the given break temperature, there is an optimum ratio of
appearance of the break temperature within that tempera initial to terminal cooling rates beyond which the ratio
ture range which affects the final stress. In general, 30 should not be increased. This optimum arises from a
the annealing range of typical soda-lime-silica window balance between a decreased contribution to the perma
and plate glass for annealing schedules of a time mag nent stress from the decay of the temperature gradient
nitude contemplated in the examples set forth herein (i.e., corresponding to a smaller terminal cooling rate R, and
on the order of 100 seconds for 7%2-inch thick glass) is an increased contribution to the final permanent stress
approximately 600 to 520 degrees centigrade. This refers from the higher instantaneous stress induced at the time
to the mean temperature of the glass, For purposes of at which the change in cooling rates is made as a result
control it is convenient to determine this mean tempera of a more drastic change of cooling rates from initial
ture from measurement of surface temperature which will to terminal, which clearly has not time to decay suffi
be invariably lower. For example, for 732-inch thick ciently before the glass is cooled below the annealing
glass, the surface temperature, while the glass is in the an 40 range.
nealing range, will extend from about 585 to 505 degrees Turning now to FIG. 4 of the drawings, the general
centigrade. For any thickness of glass, the surface tem cooling schedule of a more complicated two-break tem
perature observed while the glass is within the annealing perature curve is shown in FIG. 4A plotted against time.
range would vary as indicated by the above equation, and The accompanying changes in the thermal gradient and
remains roughly in the range of 600 to 500 degrees centi 45 center tension are shown on the same time scale in FIGS.
grade. Variations in the composition of the glass can, of 4B and 4C, respectively. The center tension created
course, raise or lower the annealing range. by the first change in cooling rates from R to R at
The determination of stress values expressed herein is time t1 relaxes in a direction which decreases the ulti
based on a comparison of the path difference formed be mate center tension. In addition, the gradual slope of
tween two plane-polarized light waves passing through the temperature curve R2 (FIG. 4A) during the second
stressed glass, as determined by a known standard pro cooling rate indicates a low thermal gradient (see FIG.
cedure.
Taking as an illustrative example, the production of 4B) which, if continued through the annealing range,
a standard 742-inch thick continuous ribbon of glass, Would result in a relatively low final permanent stress.
manufacturing considerations as discussed above have However, if this rate were continued, the lower limit of
imposed the requirement of lowering the temperature
55 the annealing range would not be reached in the re
of the glass from 600 to 520 degrees centigrade (the quired time. Therefore, a second break at time t and
annealing range) in 100 seconds. The simplest pos a third more rapid cooling rate Ra are established to
sible cooling schedule is that corresponding to cool lower the temperature of the glass below the annealing
ing at a uniform rate throughout the annealing range. range in the required time. As the curve in FIG. 4C
With the above conditions set for the present example,
60 shows, this change causes an instantaneous stress in the
straight-line cooling permits no variations and corre direction of increased center compression. Any relaxa
sponds to cooling the glass at a uniform rate of 0.8 tion of this stress is therefore harmful because it decreases
degrees centigrade per second. The corresponding stress the center compression (i.e., increases center tension).
(center tension) produced in %2-inch thick sheet glass is However, because the temperature of the glass at this
about 145 millimicrons per inch. point is near the lower limit of the annealing range,
A system having a single break in the cooling schedule, there is very little actual relaxation and the glass does
as shown by the dotted lines in FIG. 3A, permits some not, to any great degree, physically adjust through viscous
variation in the annealing process. This may range from flow to the third cooling rate. Therefore, the final perma
a high initial cooling rate and a low terminal rate, as nent stress when the third gradient is removed will not
indicated by curve ABC, through a constant cooling
70 be proportional to this high gradient. Rather, it will
be proportional to the stress which would result from
rate AC, to a low initial rate and high terminal rate, the removal of the initial cooling gradient, less the por
as indicated by curve ADC. The break may occur tion of the instantaneous stress produced by the first
at any temperature TB within the annealing range. In change in cooling rates that relaxes during the second
addition to the break temperature TB a single break 75 cooling rate, plus the portion of the instantaneous stress
 3,251,671
9 10
produced by the second change in cooling rates that re shown) several floors above the bath. The molten glass
laxes during the third cooling rate. forming the bath 17 is maintained at a temperature of
FIGS. 5, 6 and 7 illustrate specific examples of the about 1000 degrees centigrade, which is suitable for the
more complicated cooling schedules entailing two breaks, drawing process. As the ribbon is formed and drawn
still conforming to the requirement of cooling the through the drawing chamber and into the annealing
%2-inch thick sheet of glass from 600 to 520 degrees lehr 30, which encloses that portion of the ribbon that
centigrade in 100 seconds. m is within the temperature limits of the annealing range, it
FIG. 5A illustrates an annealing schedule in which is first cooled from the high temperature of the bath
the temperature of the glass is lowered from above the by coolers 22, the ambient air within the drawing cham
annealing range at an initial cooling rate R1 to a tem O ber 18, and the cooled catch pans 24. In this manner,
perature level T within the annealing range and held the glass is cooled to a temperature of about 650 de
constant (R=0) between times t and t2, after which grees centigrade by the time it leaves the drawing cham
the glass is cooled at terminal cooling rate Rs. Residual ber. This much of the cooling cycle takes place above
center tension is plotted in FIG. 5B as a function of the upper temperature limit of the annealing range of
the three independent variables R1, R3, and T. The 5 the glass and is common to known processes and the
family of curves illustrated shows the variation in re process of the present invention.
sidual center tension for various terminal cooling rates FIG. 10 schematically indicates the surface tempera
and different temperature levels of the holding period tures of ribbons of different thicknesses and having iden
when the initial cooling rate R1 is equal to 2 degrees tical mean temperatures relative to their positions in
centigrade per second. These curves would, of course, 20 the drawing machine. The dotted-line curve indicates
vary with different initial cooling rates. It may be seen the temperature distribution of a glass ribbon drawn in
that the residual stresses depend strongly on the tem the known manner wherein the ribbon cools at a natural
perature level T, and to a somewhat lesser extent on rate characteristic of the drawing machine. A glass rib
the terminal cooling rate Rs. For all but the slowest bon allowed to cool in such a substantially uncontrolled
cooling rates, for which the holding period is neces 25 manner has been found to have an unacceptable level
sarily very short, the least degree of residual center ten of residual permanent stress and, as a result, the glass is
sion is produced when the holding temperature is about difficult to cut. Any loss of glass through breakage
568 degrees centigrade. The lowest minimum produced in cutting operations is, of course, undesirable, and for
under this annealing schedule, 98 millimicrons per inch, this reason it is particularly important that an effective
corresponds to the most critical cooling schedule, as in 30 cooling schedule be established that is compatible with
dicated by the steepness of the graph on either side of existing manufacturing processes.
the minimum. Similar families of curves (not shown) Easily controlled cooling curves of the type shown in
obtained for other values of the initial cooling rate R FIG. 10 by the solid lines, based upon the principles ex
also exhibit minima in the region of 100 to 120 milli
35
plained above, have resulted in a reduction of residual
microns per inch for holding temperatures between 565 stress in drawn glass to a level heretofore thought pos
and 575 degrees centigrade. sible only by following a theoretical annealing curve
FIGS. 6A and 7A illustrate examples of linear cool of the type proposed by Adams and Williamson. This
ing schedules which entail two breaks at two different has been accomplished by establishing a large initial
temperatures, TB and TB, thus eliminating the constant 40
thermal gradient in the glass ribbon as it exists from the
holding temperature of the previous case. Cooling sched drawing chamber so as to rapidly cool the ribbon along
ules of this type are characterized by four independent the initial cooling rate R1 greater than the natural rate
variables, such as the two break temperatures and either of cooling from a temperature above the upper limit of
the corresponding break times t1 and to or two of the the annealing range to a temperature within the anneal
three cooling rates, R, R2, R3. FIGS. 6B and 7B are ing range. This may be accomplished in the manner
plots of residual stress as a function of the break tem 45 shown in FIGS. 9 and 10 by positioning heat exchang
peratures TB and TB for initial cooling rates of 2 de ers, such as Water coolers 32, on each side of the rib
grees centigrade per second and terminal cooling rates bon at the entrance of the annealing lehr. These cool
of 1.3 and 2 degrees centigrade per second, respectively. ers are, in part, Substantially parallel to and extend the
Several of these schedules, for example those correspond Width of the ribbon transversely of the direction of draw.
ing to an initial cooling rate R2 of 2 degrees centigrade 50 They extend in the direction of the draw a sufficient
per second and a first break temperature of TB between distance to lower the temperature of the glass to the
570 and 580 degrees centigrade, give terminal stresses proper break temperature at the proper time, as op
in the range of 94 millimicrons per inch, which are even timized for the annealing time of the particular process.
lower than those considered optimum in schedules that This distance, while substantially the same for differ
entail holding at a constant temperature during the an 55 ent glass thicknesses drawn on the particular machine,
nealing schedule, as in FIG. 5A. Practically speaking, as indicated by the temperature curves of the different
the decrease in calculated residual stresses from about ribbon thicknesses in FIG. 10, will vary with different
98 to 94 millimicrons per inch is less significant than the types of drawing machines and with different arbitrarily
fact that these curves have more flat minima than those Selected annealing distances in the machines. Prefer.
of FIG. 5B. This suggests that operation of the latter 60 ably, the terminal portion 33 of the coolers are in
heat treating schedules at their optimum would be less clined toward the ribbon to prevent the adjacent heating
critical than operation of the best schedule entailing elements 34 from influencing those portions of the rib.
holding at a constant temperature. bon still adjacent the coolers.
Reference is now made to FIG. 9 which shows an ap When the ribbon passes the coolers 32, its rate of
65 cooling is influenced by the heating elements 34 and the
paratus for drawing a continuous ribbon of glass 6
from a molten bath 17. As in conventional drawing enclosure 30 which retard the normal cooling rate (shown
apparatus, a drawing chamber 18 formed in part by in dotted lines in FIG. 10) to the diminished, substan
L-blocks 19, ventilator water coolers 20, and cooled tially constant rate R2. Of course, from a manufactur
catch-pans 24 encloses an area above the bath and pro ing standpoint, a linear rate R is advantageous as it is
vides a suitable environment for the formation of the
70 more easily established and reproduced than non-linear
ribbon. Heat exchangers, such as coolers 22, hasten cooling rates. However, slight variations in the linearity
the solidification or "setting” of the glass as the rib of this rate due to the varying locations of machine.
elements or other causes, such as errors or inaccuracy
bon is drawn from the bath 17 by a series of pairs of in control, will generally not adversely affect the re
drawing rolls 26 and conveyed to a cutting station (not 75 Sultant anneal to an unacceptable extend.
 3,251,671
2
Positioned at the upper or exit portion of the lehr the next approximately 20 seconds, the cooling rate of
in FIG. 9 are coolers 36 on each side of the ribbon. the ribbon is increased above the natural cooling rate
These coolers are substantially parallel to the ribbon and by opening the sides of the machine so as to lower the
extend the width thereof transversely of the direction temperature of the ribbon to approximately 515 degrees
of draw. In this manner, the desirable accelerated final 5 centigrade measured surface temperature. It should be
rate of cooling of the glass may be established to lower pointed out at this time that the annealing range of the
the temperature from a point within the annealing range glass, in terms of mean temperature, is, in this example,
to a temperature below the annealing range. Alterna taken to be about 600 to 520 degrees centigrade. This
tively, as shown in FIG. 10, doors 37 in the sides of corresponds to a range in terms of the surface tempera
the machine may be opened at this point to obtain an in ture of the glass of about 585 to 515 degrees centigrade.
crease in the cooling rate. While these coolers or open The ribbon is cooled through the annealing range in ap
ings are used to establish the preferred curves of the proximately 100 seconds. The ribbon is then cooled
type shown in the graph of FIG. 10, an acceptable anneal at a natural rate and is at a temperature of approxi
may be established by omitting the cooling at this point mately 90 degrees centigrade when it reaches the cut
and establishing only a single break in the annealing 5 off floor where the ribbon is cut into discrete sheets.
CW3. The central birefrigence of a 742-inch thick glass sheet
The following is an example, by way of illustration annealed in this manner, as indicated by the path dif
only, of a preferred mode of manufacturing sheet glass ference formed between two plane-polarized light waves
utilizing the present invention to obtain a low resultant passing through the glass measured in the manner previ
stress in the final product: ously explained, is approximately 110 to 120 millimi
A ribbon of glass approximately 100 inches wide and crons per inch. As would be expected, calculated stress
%2-inch thick is continuously formed by being mechani minima will generally only be approached rather than
cally drawn from a partially enclosed bath of molten reached in actual operations. By way of comparison,
glass maintained at a temperature of approximately 1000 similar ribbons of glass drawn through the same ma
degrees centigrade and consisting of: 25 chine but allowed to cool naturally (i.e., without the
Percent by use of the water coolers and heaters above the draw
weight ing chamber) have resulting stresses in the range of 220
SiO2 -------------------------------------- 71.38 to 250 millimicrons per inch.
Na2O ------------------------------------- 12.79 Under various operating conditions, of course, the ob
CaO -------------------------------------- 9.67 30 Served annealing range and break temperatures might well
MgO ------------------------------------- 4.33 vary from the temperatures of the examples, used only
Na2SO4 -- - - a m - was arm a or - - - - - - - - - - -- - 0.75 for purposes of illustration, due both to machine and op
NaCl ------------------------------------- O. 12 erating factors as well as to variations in the time of
Fe2O3 or no a ran was m ms - --- - - - - - , . .. . 0.15 anneal. The primary applicability of the disclosed an
Al2O3 ------------------------------------ 0.81 nealing schedules, from a commercial standpoint, is to
The ribbon, formed and drawn upwardly by the traction processes where. the time available for annealing is less
of pairs of opposed drawing rolls acting upon the already than ten minutes and usually between one-half to five
formed portion of the ribbon, first passes between a pair minutes. With respect to glass drawing machines, it
of Vertically disposed heat exchangers within the draw must be emphasized that no two machines are identical,
ing chamber which remove heat from the ribbon and and the slight differences in surroundings, machine po
40

drawing chamber at a rate of approximately 8640 Brit sition relative to the melting tank, etc., have a pronounced
effect upon the operation of each machine.
ish thermal units per minute. In addition, the ventilator As a general rule the annealing herein contemplated
coolers, in part forming the drawing chamber, remove is achieved by rapidly cooling the glass from the top of
heat from the ribbon and drawing chamber at a rate of the annealing range to a temperature within 20 to 70
1970 British thermal units per minute. As the ribbon percent of this range (measured from the top thereof and
travels upwardly and the portion thereof being consid considering the bottom of the range to be 100 percent
ered leaves the drawing chamber, catch-pan coolers on thereof) and thereafter cooling the glass at a distinctly
each side of the ribbon at the exit end of the drawing slower net average rate to the bottom of the range.
chamber remove heat from the ribbon and drawing cham 50 Normally, the first rate of cooling is at least about 10
ber at the rate of 12,600 British thermal units perminute. percent faster than the linear average cooling rate for
In this manner, the temperature of the glass ribbon is the total available annealing time.
reduced to approximately 635 degrees centigradte sur The net average rate for any portion of the annealing
face temperature as it leaves the drawing chamber. An Schedule and the linear average cooling rate for the en
other pair of coolers, one cooler on each side of the rib tire schedule is determined by dividing the temperature
bon, located just above the drawing chamber and within
the enclosed portion of the drawing machine, withdraw whichthrough
drop
the
which the glass is cooled by the time during
cooling takes place. For example, if the sur.
heat from the ribbon at a rate of 1085 British thermal
units per minute to lower the temperature of the rib face temperature of the glass is lowered 75 degrees centi
grade during the entire annealing schedule and the avail
bon at a substantially constant and increased rate over. 60 able
the natural cooling rate from 635 to 560 degrees centi coolingannealing time is 100 seconds, the average linear
rate for the entire schedule, in terms of surface
grade measured surface temperature in approximately temperature, is 5400 or 0.75 degree centigrade per second.
45 seconds from the time the ribbon enters the enclosed Alternatively, this rate may be expressed in degrees centi
portion of the machine above the drawing chamber. For
the next approximately 65 seconds, the natural cooling 65 grade per percent of the total available annealing time.
For most purposes the first and fasternet average linear
rate of the ribbon is retarded by insulation and heating
elements positioned within the enclosed portion of the rate of cooling should be at least 0.9° C. per one percent
of the total available annealing time. -
drawing machine, and in this time the ribbon is cooled The lower rate may take place as a linear cooling rate
to a temperature of approximately 545 degrees centi
grade measured surface temperature. For this purpose, 70 from the point of change in cooling rate to the bottom
of the annealing range or may occur as two or more
the heating elements are divided into three vertical zones, cooling rates, for example, a slower rate and then a
as indicated in FIG. 10 of the drawings. Electrical faster rate. In the latter case the second change in cool
power is distributed in the amounts of 400 watts to the ing rate may take place in a range of 30 to 55 percent
first Zone, 300 watts to the second, and 200 watts to of the annealing range measured from the top of the
the third to retard the cooling rate of the ribbon. In range, as discussed above. In the former case the change
 8,251,671
3 14
in cooling rate usually takes place within the range of or other conveying means as the glass passes through the
about 20 to 45 percent of the annealing range measured annealing stage of the operation.
as discussed above. With reference to the specific example previously dis
In general, for best reduction of residual permanent closed in connection with FIG. 10, where a glass ribbon
stress when a two-break annealing schedule is being ap is first cooled from 585 degrees centigrade to 560 degrees
plied to a soda-lime-silica glass of the thickness and the centigrade (surface temperature) in approximately 15
type mentioned in the example and at the times therein seconds of a 100-second annealing schedule (i.e., in 15
contemplated, the first change in the cooling rate within percent of the total annealing time), the first net average
the annealing range should occur between mean tempera rate of cooling is:
tures of 585 and 550 degrees centigrade (or between 0. 259 C.
surface temperatures of 575 and 540 degrees centigrade) 15% of total annealing time = 1.67 C. per one percent
and the second change should occur at a lower tempera
ture than the first and between mean temperatures of total annealing time
570 and 525 degrees centigrade (or between surface tem The glass is then cooled at a decreased rate from 560
peratures of 565 and 520 degrees centigrade). Where 15 degrees centigrade to 545 degrees centigrade in approxi
a single break annealing schedule is being followed, in mately 65 seconds (i.e., 65 percent of the total annealing
which the first rate is more rapid than the second, the time) and the net average rate of cooling for the second
change in cooling rates should occur between mean step is:
temperatures of about 585 and 570 degrees centigrade
(or between surface temperatures of about 575 and 555 20 65% of total15°annealing
...
C.
al-Hall- o
time 0.23° C. per one percent
degrees centigrade). The ratio of the first net average
cooling rate to the second, diminished, net average cool total annealing time
ing rate over the balance of the annealing time available, The third cooling step alone lowers the temperature of
whether as a single- or double-break. or other multiple the glass from 545 degrees centigrade to 515 degrees
schedule, normally is not more than 10, and preferably 25 centigrade in 20 seconds (i.e., 20 percent of the total
is less than 8, but, of course, must be greater than unity. annealing time) and the net average cooling rate for this
In the preferred annealing schedules, where the tem step is:
perature of the glass ribbon is first lowered into the an
nealing range at a cooling rate more rapid than the next 30° C.
subsequent cooling rate, the change from the first rate 30 20% of toiannealing
run-ru-------"mum-
time -15° d
C. per one percent
to the second, diminished rate should occur before the
lapse of about 40 percent of the total available annealing total annealing time
time and in most cases is after the lapse of about 10 per
cent of such available annealing time. The first net Because the second and third cooling steps together lower
average rate of cooling is at least about 0.9 (rarely being the temperature of the glass from 560 degrees centigrade
over about 5 to 10) degrees centigrade per one percent of to 515 degrees centigrade in 85 seconds (i.e. 85 percent of
total available annealing time. the total annealing time), the net average cooling rate
To economically achieve the desired reduction in ulti of the glass from the first break in the cooling schedule
mate stress in the glass, it is important that cooling pro to the lower limit of the annealing range is:
gress substantially continuously. Hence, the net average 40
rate of cooling over any interval of 5 percent of the 45° C.
total annealing time should not be less than 0.2 degrees 85% of total annealing time = 0.53° C. per one percent
centigrade per one percent of total annealing time avail total annealing time
able.
As previously stated, the second cooling may take place 45 FIG. 12 schematically illustrates an annealing lehr 60
in two stages, for example, first at 0.3 degree and later and a temperature curve established for annealing re
at 0.6 or 1.5 degrees (or even more) per one percent heated glass sheets or plates (such as sheets which are
of total annealing time. In such a case the net average heated above the annealing range to an essentially stress
cooling rate for the second cooling may be somewhere free condition for the purpose of bending) in accordance
between the rates of the two stages. The net average 50 with this invention. Heating elements 62 within the lehr
cooling rate for lowering the temperature from the range on each side of the glass sheets or plates 64 raise the
of 575-540 degrees centigrade to the bottom of the an temperature of the glass to a temperature above the
nealing range is not above 0.65, usually being not over annealing range as it is conveyed through the lehr on
0.6 degree centigrade per one percent of total available conveying means such as rollers 66. After the glass has
annealing time (time available for cooling from the top attained a temperature sufficiently above the annealing
of the annealing range to the bottom thereof). range to allow the internal stresses to quickly relax, it
In an embodiment of this invention utilizing the cool passes by coolers 68 extending across the inside of the
ing schedule shown by curve ADC in FIG. 3A, the glass lehr on each side of the glass. These coolers lower the
is cooled from about 590 degrees centigrade surface tem temperature of the glass at a rapid and substantially con
perature to a temperature below about 560 degrees centi 60 stant rate to a temperature within the annealing range.
grade over a period which constitutes the major portion Heating means, such as electrical elements 70 within the
of the total available time, normally 55 percent or more lehr on each side of the glass, then established a reduced,
of such time. Such first-stage cooling normally takes substantially constant cooling rate until the temperature
place either at a uniform rate of cooling or a plurality of the glass has been lowered to a point near the lower
of rates, but in any event the net average rate of cooling 65 limit of the annealing range, after which coolers 72 ex
is not less than about 0.2 degree centigrade nor more tending across the width of the lehr on each side and at
than about 0.7 degree centigrade per one percent of total the exit end thereof rapidly lower the temperature of
available annealing time. Following cooling to a level the glass below the annealing range.
below 555 degrees but above about 540 degrees centi The annealing of glass in the manner disclosed herein
grade surface temperature, the glass is then cooled at a 70 will result in a final anneal as low as the minimum anneal
more rapid rate to complete the annealing in the avail obtainable by following the calculated curve of Adams
able time. While this process is effective, it has the and Williamson but without the expense of maintaining
disadvantage that the temperature of the glass, remain the glass at an elevated temperature over a substantial
ing high during a major portion of the anneal, may result portion of the cooling cycle and without the difficulty
in undesirable marking of the glass by the drawing rolls 75 inherent in following a continuously increasing rate of
 3,251,671
5 6
cooling. It should be understood that the foregoing dis is not less than about 500 C. in terms of the Surface
closure is intended to cover all changes and modifications temperature of the glass, and wherein a glass ribbon is
of the examples of the invention herein chosen for the formed from a pool of molten glass and the ribbon is
purposes of the disclosure which do not constitute de cooled, the improvement which comprises annealing the
partures from the spirit and scope of the invention as 5 ribbon during said cooling over an annealing time of less
set forth in the appended claims. than 10 minutes duration by continuously cooling the rib
claim: bon from the upper limit of the annealing range to a sur
3. In the method of producing sheet glass wherein a face temperature below 575 degrees centigrade but above
glass ribbon is formed from a pool of molten glass and 555 degrees centigrade at a first net average rate of at
the ribbon is cooled, the improvement which comprises O least 0.9 degree centigrade per one percent of total avail
annealing the ribbon during said cooling over an anneal able annealing time, then further continuously cooling the
ing time of less than 10 minutes duration by continuously glass through the lower limit of the annealing range at a
cooling the ribbon from the top of the annealing range to net average rate of not less than 0.2 degree centigrade nor
a temperature range within 20 to 70 percent of the anneal more than 0.6 degree centigrade per one percent of said
ing range measured from the top of said range at a rate of total annealing time.
at least 10 percent faster than the average linear cooling 7. In the method of producing sheet glass wherein a
rate, then further continuously cooling the glass to the glass ribbon is formed from a pool of molten glass and the
bottom of said range at a net average rate which is slow ribbon is cooled, the improvement which comprises an
er than said first rate. nealing the ribbon during said cooling over an annealing
2. In the method of producing sheet glass wherein a 20 time of less than 10 minutes duration by cooling the rib
glass ribbon is formed from a pool of molten glass and bon from a surface temperature of 585 degrees centigrade
the ribbon is cooled, the improvement which comprises an to a surface temperature below 575 degrees centigrade
nealing the ribbon during said cooling over an annealing but above 555 degrees centigrade at a rate of at least 0.9
time of less than 10 minutes duration by continuously degree centigrade per one percent of total available time,
cooling the ribbon from the top of the annealing range to then further continuously cooling the glass to a tempera
a temperature range within 20 to 70 percent of the an ture below 515 degrees centigrade at a net average rate
nealing range measured from the top of said range at a of not less than 0.2 degree centigrade nor more than 0.6
rate of at least 0.9 degree centigrade per one percent degree centigrade per one percent of said total annealing
of total available annealing time, then further continuous 30 time.
ly cooling the glass to the bottom of said range at a net 8. The process of claim 7 wherein a net average rate of
average rate which is slower than said first rate. at least 0.2 degree centigrade per one percent of said total
3. The process of claim 1 wherein the top of the an annealing time is maintained over any 5 percent time in
nealing range is between about 585 C. and 590 C., and terval of the total available annealing time.
the bottom is between about 505 C. and 515° C. in terms
of surface temperature and wherein the glass is rapidly 9. The method of claim wherein the center of the
cooled from the surface temperature of 590 C. to a range sheet remains under tension throughout substantially all
of 575 to 540 C. and then is cooled at a net average of the annealing range.
slower rate not less than 0.2 degree centigrade per one
percent of total available annealing time. 40 References Cited by the Examiner
4. The process of claim 3 wherein the glass is cooled UNITED STATES PATENTS
at said net average slower rate in at least two stages, the 1,585,542 5/1926 Henry-------------- 6.5-117
first of which is relatively slow, the second of which being 1,837,311 12/1931 Amsler -------------- 65-119
relatively fast.
S. The process of claim wherein the net average cool 45
2,774,190 12/i956. Atkeson -------------- 65-95
ing rate of said further cooling is at least 0.2 degree centi OTHER REFERENCES
grade for a 5 percent time interval of the total available
annealing time. Textbook of Glass Technology, by Hodkin and Cousen,
6. In the method of producing soda-line-silica glass 1925, pp. 47 to 50.
having an annealing range, the upper limit of which is not 50
higher than about 600° C. and the lower limit of which DONALL H. SYLVESTER, Primary Examiner.
 UNITED STATES PATENT OFFICE
CERTIFICATE OF CORRECTION
Patent No. 3, 25 l, 671 May l7, 1966
Robert Gardon
It is hereby certified that error appears in the above numbered pat
ent requiring correction and that the said Letters Patent should read as
corrected below.
Column l, line 28, for 'reduced' read - - reducing - - ;
column 4, line 39, before 'glass ribbon' insert -- the - - ;
column 8, line 10, for '526' read - - 520 - -; column 9, line
50, for "R" read - - R --; column 10, line 39, for 'exists'
read - - exits - - ; line 41, for 'the initial' read - - an
initial - - ; line 75, for 'extend' read - - extent - - ; column ll,
line 52, for 'centigrate' read - - cent grade - -.
Signed and sealed this 22nd day of October 1968.
(SEAL)
Attest:
Edward M. Fletcher, Jr. EDWARD J. BRENNER
Attesting Officer Commissioner of Patents