Ceramic Engineering and Science Proceedings
`Charles H. Drummond Ill
`copyright @The American Ceramic Society, 1987
`
`Ceram. Eng. Sci. Proc., 8 (3-41 pp. 200-207 (1987)
`The Effect of Amber Cullet Additions on Amber Glass
`Transmission
`
`STEVEN M. WEISER
`Brockway, Inc.
`McCullough Ave.
`Brockway. PA 15824
`
`Results of experimental melts are giuen which indicate that amber cullet additions
`to an amber glass ofsimilar composition and 550 nm transmission cause the 550
`nm transmission of the final glass to increase by approximately 0.15% for euery
`percent addition ofarnber cullet abooe 10% total cullet. It is also shown that amber
`glasses containing amber cullet tend to exhibit lower retained sulfur leuels (expressed
`as SO,) for a giuen level of carbocite/ton of glass from batch than do similar no-
`cullet batches. Transmission measurements also indicate that carbocite adjustments
`are not as efjectiue in an amber glass when higher leuels of amber cullet are pres-
`ent. A set of cullet-related carbocite transmission factors are deueloped which show
`that a 0.45 kg or one pound (per ton of glass from batch) addition of carbocite in
`a 0.28% Fe,O, amber glass reduces the 550 nm transmission by22.1% at 0% cullet,
`whereas at 50% cullet, the change is only 5.2%.
`
`Introduction
`Within the last several years, the demand for glass recycling has increased. To
`keep in step with this demand for recycling. the amount of redemption cullct
`used in our batches has been steadily increasing. The best example of this is at
`our Plant No. I where cullet levels are being maintained at between 50% and 70%
`in our amber and emerald green compositions. Company wide, we have increased
`our redemption cullet utilization from 20 700 tonne (23 000 t) in 1982 to 99 000
`tonne ( 1 10 000 t) in 1985. 1986 year-to-date levels equate to an annual consump-
`tion of 104 400 tonne ( I 16 000 t).
`As a result of this increased cullet usage. problems associated with cullet con-
`tamination and with cullet compositions which were dramatically different than
`our own glasses have had to be addressed. An unknown factor, which until now
`had not been studied. was the effect of amber cullet additions on amber glass
`transmission. It had been assumed that as long as the cullet had a 550 nm transmis-
`sion that was similar to that of the glass from batch. the transmission of the final
`glass would not be affected. It had also been assumed that the method of calculating
`carbon adjustments to compensate for transmission changes would also remain the
`same. This method based carbon adjustments on the weight of glass from batch,
`excluding cullet. It was the intent of this study to determine if these assumptions
`were correct. Carbocite, an anthracite coal product, is used to supply the carbon
`reducing agent in the amber batches.
`
`Experimental Procedure
`Melt Preparation
`To prepare the experimental glasses for this study. a rotary hearth furnace'
`
`200
`
`O-I Glass, Inc.
`Exhibit 1027
`Page 001
`
`

`

`was used. This furnace has six gas-fired burners located around its circuinfercnce
`to provide uniform heat distribution. Located inside the furnace is a rotating table
`on which six crucibles are placed, each having a glass capacity of approximately
`7.2 kg (16 Ib). The rotation of this table during a melt further ensures a uniform
`thermal history for the experimental glasses. A hole is located in the crown of the
`furnace. through which the premixed batches can be charged into the pots while
`the furnace is in operation.
`The batches were melted at 1482°C (2700°F) for four h. After cooling, samples
`of glass were cut from the center of each pot. These samples were analyzed for
`total Fe20, and sulfur (expressed as SO3) by X-ray fluorescence spectrometry. The
`550 nm and 1000 nni transmissions of each sample were then measured using a
`dual-beam spcctrophotometer. All of the transmission data that is reported has been
`corrected to 2 mm thickness.
`The experimental amber glasses that were prepared in this study were targeted
`to a theoretical Fe,Ol level of 0.28%. A combination of barytes and gypsum was
`used to supply the sulfur in the batch and a processed mill scale was used to supply
`the Fe,O,. The barytes, gypsum and mill scale weights per 900 kg (2000 Ib) of
`glass (from batch) were 7.3 kg (16.2 Ib), 4.9 kg (10.8 Ib), and 2 kg (4.4 Ib).
`respectively.
`Transmission Factor Definition
`To quantify the effect of various batch constituents on amber glass transniis-
`sion. a value known as a transmission factor is calculated for each material that
`influences anibcr color.
`A transmission factor is defined as being a number which indicates the
`magnitude and direction of the change in the 550 nm transmission which results
`from a change in the quantity of a particular material used in the batch. These have
`historically been determined from experimental melts in which a single batch com-
`ponent has been varied and its effect on light transmission measured. These melts
`are normally prepared using no cullet.
`Since a plot of the 550 nm amber transmission versus the carbocite level is
`not a linear function. as indicated in Fig. 1, it is appropriate to determine the slope
`ofthc curve at a particular transmission level. This slope, or the transmission fac-
`tor. is typically calculated at a standard 550 nm transmission of 34%. This is ac-
`complished by fitting the data to a second degree polynomial. This yields an equa-
`tion of the form
`y = bo + b,X+ b2X2
`( 1 )
`where bo, b,, and bz are constants. From this equation, the value of X (carbocite
`weight) which corresponds to a 34% 550 nm transmission can be calculated. The
`slopc of the curve at that point can then be found by taking the derivative of Eq. I .
`( 2 )
`d, = ( b , +2b2 X)dx
`(3 1
`The value of dyidx in Eq. 3 is equivalent to the percent change in transmission
`per pound of material per 2000 pounds of glass from batch at 34% 550 nm
`trammission.
`Results and Discussion
`In the initial experimental melt, six glasses were prepared to determine the
`effect of 50% amber cullet on the amber glass transmission at several carbocite
`
`dyldx=b, +2bz X
`
`20 1
`
`O-I Glass, Inc.
`Exhibit 1027
`Page 002
`
`

`

`levels. The amber cullet used in this melt had a 40.5% 550 nni transmission and
`it had a composition that was similar to that of the experimental batches.
`The first three pots of glass were prepared with all batch, while the second
`three pots contained 50% batch and 50% amber cullet. The carbocite levels were
`such that the theoretical transmission of glass No. 1 would be approximately 7 %
`lighter than that of glass No. 2. Glass No. 2 was the standard glass targeted to
`31% transmission. Glass No. 3 was targeted to be approximately 15% darker than
`glass No. 2. These same theoretical variations applied to glasses 4, 5. and 6 .
`As indicated in Table I, the measured transmission differences between glasses
`Nos. 1 , 2, and 3, as well as their respective transmissions, were quite similar to
`the targeted values. In contrast, the transmissions of glasses Nos. 4, 5, and 6 (Table
`11) were all higher than anticipated and they did not show the same changes in
`transmission as seen between glasses Nos. I , 2, and 3. There are several reasons
`which might be used to explain this behavior.
`First, the simple dilution of the all-batch glass with cullet of higher transmis-
`sion would result in a lighter final glass. The effect of this dilution, as seen by
`comparing the transmissions of glasses Nos. 1 and 4, 2 and 5, and 3 and 6 , became
`more evident as the transmission of the glass from batch was made increasingly
`lower than the transmission of the cullet.
`Second, since the carbociteh of glass from batch adjustments were to have
`yielded the same glass transmission changes at the 0 and 50% cullet levels, the
`measured transmission data indicated that the reducing power of the carbocite was
`less at 50% cullet. However, had the carbocite adjustments been calculated using
`the total glass weight (batch and cullet), the calculated transmission changes at 50%
`cullet would have been only half as large. Thus at 50% cullet, the disagreement
`between the measured and theoretical transmissions was partially a result of the
`method of calculation
`Finally, the measured transmissions of glasses Nos. 4 and 5 were both higher
`than those of the corresponding all-batch glasses or the cullet alone. This would
`indicate that even above the amount needed to compensate for the cullet color dilu-
`tion, additional reducing agent is needed when operating at high cullet levels. A
`decrease in either the ferric iron or sulfide sulfur concentrations of the glass2
`associated with the use of cullet in place of the batch would explain the increased
`transmission.
`To further quantify the effect of amber cullet on amber glass transmission,
`a second melt was prepared in which the cullet was varied from 0 to 50%, in 10%
`increments. In this melt, the weight of carbocite pert of glass from batch was held
`constant at 2 kg (4.4 Ib). In terms of the total glass weight, the carbocite was de-
`creased from 2.2 kg/tonne (4.4 Ib/t) of glass at zero cullet to 1.1 kg (2.2 Ib) at
`50% cullet. As measured prior to remelting, the cullet had a 550 nm transmission
`of 30.0%.
`On the basis of the constant weight of carbocitekon of glass from batch and
`the 30% transmission of the cullet, it was assumed that the transmissions of the
`finished glasses would be relatively close to the 34% standard. However, the data
`in Table 111 shows that there was a steady increase in the 550 nm transmission
`with each cullet increase. While this increase was not dramatic in comparison to
`the size of the cullet increases, it does show that steady cullet additions to an amber
`glass batch require reducing agent adjustments, even if the cullet transmission is
`comparable to that of the furnace glass.
`As indicated in Fig. 2, cullet increases above 10% increased the amber glass
`550 nm transmission by 0.15% for each percent of cullet increase.
`
`202
`
`O-I Glass, Inc.
`Exhibit 1027
`Page 003
`
`

`

`Upon review of the X-ray analyses from Melt No. 2 (Table 111), it would ap-
`pear that the increase in the 550 nm transmissions resulted from a decrease in the
`retained sulfide sulfur concentrations (expressed as SO,). The best example of this
`decreased sulfur retentior! w’a~ g!ass No. 6 (50% cullet) in which the retained SO,
`was lower than the levels analyzed in either glass No. 1 (all batch) or the cullet
`itself. As with the transmission, the total iron and SO, analyses of the cullet were
`made prior to remelting. The noted SO, content of the cullet is typical of our 0.28 %
`Fe,O, production amber glass.
`As estimated from the 1000 nni transmissions3 approximately 77% f 3 % of
`the total iron in glasses Nos. I through 6 was present as ferrous iron. Since there
`were essentially no differences detected in the ferrous iron or total iron concentra-
`tions of these glasses, this meant that no differences existed between the ferric iron
`concentrations, either. Therefore, the transmission changes had to result from
`changes in the sulfide sulfur concentrations.
`It was interesting to note that the transmissions of all of the glasses which con-
`tained cullet were higher than either the all-batch melt or the cullet. Possible ex-
`planations for this behavior could be that the sulfide sulfur present in the amber
`cullet oxidized during melting and then volatilized from the glass or that the car-
`bocite was simply less effective. The effectiveness of the carbocite may be tied
`to the fact that as the cullet is increased, lesser amounts of carbocite and total sulfur
`are added in comparison to the amount of finished glass.
`Regardless of the exact mechanism, the results from the first two experimen-
`tal melts indicated that additions of amber cullet to an amber glass increased the
`550 nm transmission and that the existing carbocite transmission factors were not
`adequate for calculating routine carbocite adjustments at high cullet levels.
`Revised Carbocite Transmission Factors
`To generate revised transmission factors, several experimental melts were
`prepared in which carbocite was varied while maintaining a constant cullet percen-
`tage. The 550 nm transmission results from these melts are shown graphically in
`Fig. 3.
`Using the procedure described earlier, this transmission data was used to
`generate revised carbocite transmission factors for each of the cullet levels that
`were investigated. These new factors, expressed as the % transmission change per
`pound of carbocite per ton of glass from batch, were - 16.0, - 10.8, and -5.2
`for 10%. 30% and 50% cullet levels, respectively (Table IV).
`
`New Transmission Factor Evaluation
`To evaluate these new transmission factors, carbocite adjustments that were
`recently made in one of our amber furnaces were evaluated.
`The first change, a decrease of about 140 g 5 ounces of carbocite per ton of
`glass from batch, was made while operating at 30% cullet (25% amber, 5% green).
`This carhocite reduction should have produced a 7% increase in the 550 nm
`transmission based o n the existing, no-cullet factor. The actual change was only
`about + 2 % . Using the - 10.8%/lb/t of glass from batch (@ 30% cullet) carbocite
`transmission factor from this current study, the predicted change was approximately
`+ 3 % .
`The next two carbocite adjustments were made while operating at a 50% amber
`cullet level. The first of these, an addition of 388 g/tonne (12 9’2 ozh) of glass from
`batch, should have produced a 17% decrease in transmission based on the existing
`factor. The resultant change was only about -5%. Using the -5.2%/lb factor
`
`203
`
`O-I Glass, Inc.
`Exhibit 1027
`Page 004
`
`

`

`from the current study, the change was predicted to be -4%.
`The final change to be reviewed was an addition of 194 gltonne ( 6 % oz/t)
`of glass from batch. This adjustment should have produced a 9% decrease in
`transmission based on the existing factor. The resultant change was only about
`-2.5%. The calculated change, based on the -5.2%/lb. carbocite factor, was
`- 2 % .
`As indicated by each of the above examples, the transmission changes that
`were calculated using the newly developed carbocite transmission factors were far
`superior than those that were calculated when cullet was not considered a factor.
`As shown in Fig. 4, a plot of these revised transmission factors can be used
`to estimate transmission factors for carbocite for use at any cullet level.
`Summary and Conclusions
`In summary, the results from this investigation have shown that additions of
`amber cullet to an amber glass caused the 550 nm transmission of the final glass
`to increase. However, these transmission changes were relatively small in com-
`parison to the size of the cullet additions. It is important to note that in a produc-
`tion situation similar results would be expected, but only as long as the cullet had
`a 550 nm transmission that was similar to that of the furnace glass and that there
`were no changes in other variables which might affect the transmission. One such
`variable is the level of contamination of the cullet by carbonaceous or aluminous
`materials.
`It was also shown that to compensate for transmission variations which might
`occur when using high cullet levels, the newly derived carbocite transmission fac-
`tors must be used. Because these factors indicated the need for larger than “nor-
`mal” carbocite adjustments when fairly large, say 5 % and above, transmission
`changes were desired, care was taken to determine if such carbocite adjustments
`would effect the amber glass redox stability. No production problems have been
`observed.
`It appears that in the presence of high cullet levels, the reason for the diminished
`effectiveness of carbocite adjustments is not related to any changes in the iron redox
`or the total iron concentration, but to a decline in the sulfide sulfur content of the
`glass. This may be associated either with oxidation of sulfide sulfur in the cullet
`during melting or just with simple dilution of the sulfate and carbon input concen-
`trations. A definite conclusion pertaining to these possible mechanisms could not
`be reached based on the results of this investigation.
`References
`‘J. P. Poole, “An Experimental Glass Melting Furnace,” J. Am. Cerutn. Sor. 32 [7] 233-36 (1949).
`‘F. L. Harding and R. J. Ryder. “Amber Color in Comniercial Silicate Glasses.“ J. Can. Cerum.
`Sot,. 39 59-63 (1970).
`‘P. Close. H. M. Shepard. and C. H. Drummond. “Determination of Several Valence States of
`Iron. Arsenic. Antimony. and Selenium in Glass.” J. Am. Cemm. SOC. 41 455-60 (1958).
`
`204
`
`O-I Glass, Inc.
`Exhibit 1027
`Page 005
`
`

`

`Table I. Melt N o . I . 550 nni Transmission Data. 0% Cullet
`Measured
`G I ; l \ \ N o ,
`'Iheore[ical
`_
`_
`
`I
`38.0%
`41,074
`34.0
`32.7
`?
`19.0
`21.2
`3
`
`- --
`
`k:.From glazs 2
`
`+ 5.3%
`
`-11.5
`
`~
`
`~
`
`~~~~~~
`
`Table 11. Melt No. I . 550 nm Transmission Data. 50% Cullet
`
`it
`5
`6
`
`45.7%
`41 .O%
`41 .5
`34.0
`34.2
`19.0
`Cullet-40.5 5% Transmission
`
`+5.3%
`
`-7.3
`
`Table 111. Melt N o . 2 , Incremental Changes in Cullet Level
`% Fe,O,
`'.; C'ullet
`%SO,
`GILl5\ No.
`
`1
`2
`3
`3
`5
`6
`Cullet
`
`0
`10
`20
`30
`40
`5 0
`
`,267
`,376
`.28 1
`.28 I
`,280
`,274
`.277
`
`,085
`,083
`,073
`,085
`,068
`.067
`,095
`
`Y 550
`Tram.
`27.9
`32.7
`33.9
`36. I
`37.7
`38.4
`30.0
`
`___
`
`Table 1V. Carbocite Transmission Factors at Various Amber Cullet Levels
`Tmnsmihsion factor"
`c ; Cullet
`-22.1
`0
`10
`- 16.0
`- 10.8
`30
`50
`- 5.2
`'k% per Ib per ton of glass from batch
`
`205
`
`O-I Glass, Inc.
`Exhibit 1027
`Page 006
`
`

`

`Fig. 1. Effect of carbocite on amber transmission.
`
`55-
`
`5 0,
`
`0
`
`MELT 2
`Yz.152 (XI + 31.2 ABOVE 10%
`
`A MELT I
`
`GLASSES
`
`1 - 4
`
`GLASSES
`
`2 - 5
`
`GLASSES
`
`3 - 6
`
`15
`0
`
`1
`
`10
`
`1
`
`20
`
`30
`40
`AMBER CULLET
`Fig. 2. Effect of cullet on the 550 nm transmission
`
`1
`
`I
`
`1
`
`50
`
`I
`
`60
`
`7
`
`206
`
`O-I Glass, Inc.
`Exhibit 1027
`Page 007
`
`

`

`30.
`
`A 0 Yo CULLET
`
`E
`
`0
`In
`8 20.
`In
`
`IOYoCULLET
`0 30YoCULLET
`X 50Y.CULLET
`
`51 5
`4: 5
`315
`~.~
`LBSJTON OF GLASS FROM BATCH
`
`6
`
`Effect of carbocite o n amber transmission at various amber cullet
`
`BATCH
`
`I
`
`10
`
`1
`
`30
`2Io
`O/o AMBER CULLET
`
`40
`
`30
`
`6
`
`10.-
`2.5
`
`Fig. 3 .
`levels
`
`
`
`5. 5.
`
`
`
`0. 0.
`
`-3Y
`
`354
`0
`
`Fig. 4. Carbocite transmission factors as a function of amber cullet level.
`
`207
`
`O-I Glass, Inc.
`Exhibit 1027
`Page 008
`
`