The Decolorization of Amber Container Glass

Melissann Ashton-Patton

Advisor: Dr. James Shelby

 

Final Report to the

 

Center for Environmental and Energy Research at Alfred University

 

Summer Undergraduate Research Fellowships 2003

 

September 31, 2003

Summary

           

            Increased consumption of packaged items has caused a large increase in the amount of trash produced in the developed nations of the world. An easy, efficient way to recycle mixed cullet (amber, green, and flint) glass will soon be a necessity in the container industry. Members of the European council will be required to use 70% cullet in all containers by 2006.

            This CEER project explored possibilities for chemically removing the amber color of container glasses.  Two mechanisms for decolorization were considered.  First, the amber chromophore, which is a FeS complex, can be eliminated by eliminated by repairing Fe with some other element.  The second method involved oxidation of sulfur to eliminate the chromophore.

            Oxidizing the glass did not prove very effective in removing the amber color.  NaNO3 had very little effect on the final color of the glass melt.  NaF was more successful in removing the color.  The removal of the color was inconsistent, with amber and clear areas. ZnO was the most successful at removing the amber color.  The glass progressively turned from amber to yellow green to blue green, with increasing concentration of ZnO, and was homogenous in color.   This effect is believed to be due to the destruction of the FeS complex, converting it to ZnS, which does not produce the amber color.  Further research is continuing to see if other oxides will produce similar results to the ZnO addition. 

Introduction

 

As the population of the world increases exponentially, waste is also increasing.  Areas of the world such as Europe, where space is a premium, the European Parliament has raised the standards for recycling.  By the year 2006, member states will be required to use 70% recycled glass (cullet) in their container glass batches. Mixing different color cullet causes two problems.  Firstly, mixed cullet glasses melt in a non-homogeneous manner resulting in an undesirable color.  Secondly, due to the differences in melting conditions (mainly oxidized or reduced) melt foaming occurs when all three colors of cullet are put together [1].   Melt foaming was not explored in this experiment, but will need to be studied before mixed color recycling can be successful.  Glass recycling can be done with ease as long as there is a similar ratio of the color of recycled glass to the amount of that color produced in the area.  Glass is heavy, therefore it is expensive to ship, and the last thing the container industry wants is an increase in costs due to the need to ship in recyclable cullet.  Issues are rising in places such as the United Kingdom where there is a large amount of recycled glass but in the wrong color.  The majority of the UKÕs glass production is flint and amber glass, the greatest amount of recycled glass is green [2].  Problems also arise in the United States where communities are not required to sort their glass into green, amber and flint.  Robots have been designed to sort broken pieces of cullet by color; which is inefficient also a waste of resources.

The study explores decolorizing amber glass by oxidation of the melt by using known oxidizing agents NaNO3 and NaF, and by addition of network modifiers, ZnO and Sn2O.  The following properties were measured to see how dopants affected the glass: glass transition temperature (Tg), density, and refractive index.

 

Methods

 

The amber glass cullet used in this study came from Labatt Blue Ice amber beer bottles.  The amber glass was cleaned, heated and quenched, then broken into smaller pieces.  The glass was ball milled with zirconia media and sifted with a 10-mesh screen.  All batches were 15 grams of cullet, and mixed with the appropriate dopant.  Batches were melted in an air atmosphere, preheated electric furnace, in an open Pt-Ag crucible for 30 minutes at 1500¡C.  The glass was then cooled to room temperature and annealed at the glass transition temperature where it was held for 30 minutes and then cooled at a rate of 3K/min to room temperature.  The annealed glasses were then cut into plates 0.70 mm to 0.95 mm thick and were polished for optical measurements.  Large samples were appropriately cut to obtain thermal expansion data, and were used to obtain density data.   The plates ranged in size from 13.85mm - 19.95 mm long and 2mm to 3mm wide. 

The optically polished samples were used to measure spectra and the refractive index.  Ultraviolet and visible spectra were measured using a Perkin-Elmer Lambda 40 spectrometer.  Infrared spectra were measured using a Thermo Nicolet Avatar FTIR spectrometer.  An Abbe refractometer was used to measure the refractive indexes of the glasses with a resolution of ±0.0001.

The glass transition temperature of the glass was determined using a differential scanning calorimeter (DSC) with a heating rate of 20 K/min.  A single push-rod vitreous silica dilatometer was also used to determine the Tg of the glass along with the TEC (over the range of 100 to 400 ¡C) and dilatometric softening point (Td).  The dilatometer has a heating rate of 4 K/min. The TEC is reproducible to ±0.3 ppm/K.   Densities were measured using ArchimedesÕs Method with kerosene as the immersion fluid.  The results are reproducible to ±0.002 g/cm3. 

 

Results

 

Prior study by Melissa Dorsey determined that the decrease in the amber chromophore intensity caused by remelting in air is minimal and has little effect on the final amber intensity.  Since her used 5-gram batches remelted at 1400 ¡C for 30 minutes, those were the melting conditions for the first melt.  The numerous properties which needed measured required a batch increase from 5 grams to 15 grams.  The increase of batch resulted in fine bubbles through the bulk of the melt, so the remelt temperature was increased to 1500 ¡C to aide fining, thirty minutes remained the same.  The increased temperature resulted in no bubbles through out the bulk of the glass, so all of the following melts were melted at 1500 ¡C. 

  

Spectral Results:      

 

Zinc Oxide

 

Since ZnO was known to reduce the amber chromophore concentration, it was where the experimentation began.  Five batches were made starting with one weight percent addition of ZnO to five weight percent addition, each batch increasing one weight percent ZnO.  The molar weight addition can be seen in Table I.  The over all color of the melt was consistent through out and there were bubbles on the bottom of the crucible, which increased with increasing concentration of ZnO. 

 

Table I.  Molar weight addition of the dopants, calculated for the decolorizing agents.

 

ZnO

NaF

1 wt. %

0.0018430

0.0035724

2 wt. %

0.0036860

0.0071449

3 wt. %

0.0055290

0.0107173

4 wt. %

0.0073720

0.0142897

5 wt. %

0.0092150

0.0178622

 

ZnO additions cause gradual decrease in the amber color.  The glasses change from a dark amber color to a yellow-green, to a blue-green.  Uv-vis analysis shows a decrease in the absorption band at 415 nm, and a slight increase around 700 nm (the blue end of the spectrum), the results of which are shown in Figure 1.  It can be concluded that the addition of one and two weight percents ZnO decreases the absorption band which coincides with the amber color.  Additions beyond two weight percent have little effect on the absorbance at 415 nm as shown in Figure 2.  The infrared spectrum shows a significant hydroxyl band. Minimal changes of were observed in the infrared end of the spectrum as seen in Figure 3.

Figure 1.   Uv-vis spectra of glasses containing the indicated amounts of ZnO.

 

Figure 2. Effect of ZnO on absorbance at 415 nm.

 

Figure 3.  Effect of ZnO on infrared spectra. 

 

Sodium Fluoride

 

Five batches were made containing one to five weight percents NaF.  The addition of NaF decolorized the glass, but in a different manner than the ZnO.  While the glasses made with ZnO were homogonous in color, the glasses containing NaF were inconsistent in color with swirls of amber and flint color, with the amber concentration higher in the center of the glasses. 

The uv-vis spectrum indicates that there was indeed a higher absorbance in the center of the glass, while the clear edges had no absorption at 415 nm.  The large variation of color in the glass means that the uv-vis results shown in Figure 4 are misleading.  The spectra plotted were taken from the clearest sections of the glass possible.   Unlike the results for ZnO, the FTIR spectrum was affect by the addition of NaF to the glass.  Figure 5 shows the results.  One weight percent addition of NaF caused a large increase in the hydroxyl.  Each addition of NaF beyond one weight percent decreased the hydroxyl.

 

Figure 4.   Uv-vis spectra of glasses containing the indicated amounts of NaF.

 

Figure 5.  Effect of NaF on infrared spectra. 

 

SnO2 and NaNO3

            The SnO2 did not completely dissolve in the glass at one weight percent, after fifty minuets of melt time; therefore it is not useful for decolorizing.  There was a decrease in the amber chromophore concentration.  NaNO3 did not decrease the amber color.  No properties were measures for these unsuccessful dopants.

 

 

Examination of Properties:

            The refractive index was measured for all of the glasses.  The remelted Òas receivedÓ glass had an index of 1.5231.  The ranges in index are linear for both NaF and ZnO additions.  The addition of the sodium fluoride causes a decrease in the refractive index (1.5214-1.5193), while ZnO increases the index (1.5244 – 1.5294), as listed in Table II.  Figure 6 shows all dopants and their effect on the refractive index. 

 

Figure 6.  The effect of dopants on the refractive index. 


 

Table II.  The effects of the dopants on the glass transition temperature, density, and refractive index.

 

Tg Dilatometer

Td

Tg DSC

Refractive Index

Density

Molar weight of addition

No additions

570

600.7

565

1.5231

2.508

 

 

 

 

 

 

 

 

1% ZnO

555

605.3

570

1.5244

2.513

0.001843

2% ZnO

557

597.7

571

1.5259

2.539

0.003686

3% ZnO

559

603.3

573

1.5277

2.559

0.005529

4% ZnO

555

599.7

575

1.5282

2.569

0.007372

5% ZnO

552

601.3

570

1.5294

2.585

0.009215

 

 

 

 

 

 

 

1% NaF

530

587.8

556

1.5214

2.515

0.0035724

2% NaF

530

576.6

552

1.5208

2.518

0.0071449

3% NaF

525

585.8

542

1.5211

2.521

0.0107173

4% NaF

520

562.7

530

1.5220

2.523

0.0142897

5% NaF

515

562.1

520

1.5193

2.525

0.0178622

 

The density of the undoped glass was 2.508.  All of the dopents caused an increase in density.  The largest increase in density is due to the addition of SnO.  The increase is linear and ranged from 2.526 to 2.600 g/cm3.  NaF had the smallest effect on the density.  The densities range from 2.152 to 2.525 g/cm3.  The ZnO had a small effect in the density with the first addition of 1 wt. %, 2.5244 g/cm3.  All of the results are tabulated in Table II and are shown in Figure 7. 

Figure 7.  The densities of all the glasses increased with the addition of the dopant.

The dopant additions have various effects on the glass transition temperature.  The undoped glass has a Tg of 565 ¡C.  Addition of NaF causes a linear decrease in the Tg from 556 - 520¡C.  There is a small change in Tg due to the addition of ZnO, but it was not a linear effect.  The Tg is greatest at 4 wt. % ZnO.  All values are tabulated into Table II, and shown in Figure 8.

 

Figure 8.  Effect of dopants on the glass transition temperature.

 

Discussion

Amber color in glass is caused by a charge transfer in the FeS chromophore between sulfide and ferric ions.  Decolorization can occur through reducing the melt to remove ferric ions, oxidation of the melt to remove sulfide ions, or by changing the chemistry of the chromophore.  The uv-vis spectra do not show large increases in the ferrous iron concentration with increasing amounts of ZnO, therefore it is unlikely that reduction is the cause for decolorization.  Oxidation of the glass is one of the possibilities to explain the decrease of the absorption band at 415 nm.  Zn2+ ions increase the alkalinity of the glass melt.  Zinc oxide could be oxidizing enough of the sulfur that a decrease in the 415nm absorption band occurs.  If oxidation is occurring then some of sulfur is coming out of the glass as SO­2 gas.  However, no bubbles were found in the bulk of the melt, suggesting evolution of SO2 gas is not a primary reaction in the melt.  The most likely explanation for the decolorization is the removal of the sulfur ion from the FeS chromophore due to preferential bond between the Zn ion and the S ion.  If there is no charge transfer, or if the transfer occurs over the visual spectrum, then none of the wavelengths of the visual spectrum will be absorbed and no color will appear.  The preferential bonding between ZnS is most likely the cause for decolorization because of the small amount of ZnO required for reducing the absorption, reinforced by the fast rate at which the glass approached saturation. 

The failure of the NaF glass to decolorize homogeneously cannot be explained by oxidation or preferential bonding.  The failure for a strong oxidizing agent such as NaNO3 to remove any of the amber color suggests that the NaF is not oxidizing the glass.  NaF will not form a new chromophore like the ZnO, so preferential bonding can not be used to explain the decolorization.  The addition of NaF into the melt may have lowered the viscosity and caused more non-bridging oxygenÕs to form.  The change in structure could have caused the ferric ions to be attracted to the fluoride ions over the sulfide ions.  Favored bonding between transition metals ions and halide ions, instead of oxygen has been found in other glass systems [1].  The blotchiness of the melt could be due to the weakness of the reaction in the glass. 

Conclusions

 

            Amber color glass is produced by the FeS chromophore.  Elimination of the chromophore results in a glass with a lower absorption band at 415nm.  ZnO was the superior decolorizing agent, because of the uniform decolorization and the large decrease in absorbance.  NaF poorly removed the color due to the lack of homogeneity in color and large absorbance at the 415nm wavelength, even with five weight percent additions.   Since ZnO is so good at removing the amber color, further research should be done with different types of cullet, including different types of amber glass, green glass and flint glass, to see if decolorization of amber glass should be done prior to cullet mixing, or if ZnO can be added to the cullet as additive.  The effects of decolorization should be further studied to see how the ZnO additions and the mixed cullet effect melt foaming and the final properties of the glass.  

References

[1] Rapp, Douglas B.  Melissa A. Dorsey, Melissann M. Ashton-Patton, and James E.

            Shelby. ÒDecolorization of Amber Glass.Ó  Fusion Meeting 2003.

 

[2] Stewart, E. ÒNew Recycling Targets Likely to Challenge UK,Ó Glass.  Vol. 78, No. 7,

            2001. p. 220.

 

[3]  Harding, F. L., R. J. Ryder.  ÒAmber Colour in Commercial Silicate Glasses,Ó

            Journal of the Canadian Ceramic Society.  Vol. 39, 1970, p.59-63.

 

[4]  Douglas, R. W.  and  Zaman M. S. ÒIron Chromophore in Iron – Sulphur Amber

            Glasses.Ó  Physics and Chemistry of Glasses.  Vol. 10, No. 4, Aug. 1969, p. 125-

            132.

 

[5] Shi, X. Y. and M. P. Brungs.  ÒThe Iron redox couple and its interaction with sulphur

            in glass.Ó  Physics and Chemistry of Glasses.  Vol. 36, No 6, Dec. 1995, p.275-

            278.

 

[6] Hopkins, R. W.  and W. H. Manring.  ÒFactors Influencing control of color in Amber

            Glasses,Ó The Glass Industry.  Vol. 34, No.5. May 1953 p.251-276.

 

[7]  Manring, W. H. and R.W. Hopkins.  ÒUse of Sulphates in Glass,Ó The Glass

            Industry.  Vol. 39#3, March 1958.

 

[8] Blume, Russell and Charles Drummond III.  ÒModeling and Optimization of Solar-

Containing Glass Melts,Ó Glastech. Ber. Glass Sci. Technol.  Vol. 71, No. 6, 1998, p. 157-165.

 

[9] Jackson, Sir Herbert.  ÒSome Colouring Agents in Glasses and Glazes.Ó  Nature. 

            Aug. 20, 1927, no. 3016, Vol. 120. p. 264-304.

 

[10] Badger, A. E.  ÒDecolorization of Manganese and Sulfur in Glass with Application

            to Use in Furnace Slags,Ó

 

[11] Hood, H.P.  ÒUltra-Violet Transmitting Substance,Ó U.S. Patent No. 1,830,902 (1931).

 

[12] Rising, W. H.  ÒHeat Absorbing Borosilicate Glass and Method of Making the Same,Ó U.S.

Patent No. 1,737,685 (1929).

 

[13] Rising, W.H. ÒMethod of Producing Heat Absorbing Glasses and Batches Therefor,Ó

Patent No. 1,737,686 (1929).

 

 

[14] Dorsey, Melissa.  ÒDecolorization of commercial amber glass.Ó B.S. Thesis, Alfred University, 2001.