- Email: [email protected]
Nuclear microprobe and optical investigation of sparkling wine bottles
Nuclear Instruments and Methods in Physics Research B 158 (1999) 695±700
www.elsevier.nl/locate/nimb
Nuclear microprobe and optical investigation of sparkling wine bottles J. Padayachee *, V.M. Prozesky, C.A. Pineda Van de Graa Group, National Accelerator Centre, P.O. Box 72, Faure 7131, South Africa
Abstract Glass bottles, used for sparkling wine, are treated with freon during manufacturing to harden the inside surface. Although this type of treatment normally improves the properties of the glass, in this case the occurrence of ``egg'' formations (egg-shaped rough areas) on distinct areas of bottles, as well as yeast sticking to the insides of bottles at speci®c areas pointed to the possibility of dierent areas showing dierent properties in the same bottle. The question was whether the correct gas was used for the treatment, and secondly, whether the process was controlled well enough to obtain the correct properties for the inside of the glass. We present results of an optical microscopy and nuclear microprobe (NMP) investigation. Ó 1999 Elsevier Science B.V. All rights reserved.
1. Introduction Glass bottles used for sparkling wine were treated with 134-A freon during manufacturing to harden the surface, and to prevent blooming (or weathering). The aim of the treatment process is to deplete the surface region of Na and Ca in order to obtain nearly pure silica on the surface, which gives the surface the desired qualities. The treatment has been shown to improve the properties of the glass, although in all previous studies [1,2] other freon gases, such as 152-A have been used and the question was, ®rstly, whether the correct gas was being used for treatment, and secondly, whether the process was controlled well enough to
* Corresponding author. Tel.: +27-21-843-3820; fax: +27-21843-3543; e-mail: [email protected]
obtain the correct properties for the inside of the glass. The occurrence of ``egg'' formations on distinct areas of bottles, as well as yeast sticking to the insides of bottles at speci®c areas pointed to the possibility of dierent areas showing dierent properties in the same bottle. Such dierences were the focus of this investigation, using optical microscopy and the nuclear microprobe (NMP) [3] of the National Accelerator Centre (NAC), Faure, South Africa.
2. Sample preparation Glass bottles were supplied, representing different problems that are being encountered. Some bottles showed areas of ``egg'' formation, some where the sticking of yeast was a problem and some bottles with no problems. The bottles sup-
0168-583X/99/$ - see front matter Ó 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 5 8 3 X ( 9 9 ) 0 0 3 3 5 - 3
696
J. Padayachee et al. / Nucl. Instr. and Meth. in Phys. Res. B 158 (1999) 695±700
plied for analysis, described in Table 1, were all ®lled, except the TREATED-NEW and UNTREATED-NEW. Bottles were emptied, broken and glass fragments, collected from the regions of interest, were washed in nano-pure water, dried and mounted for analysis. 3. Optical microscopy and the nuclear microprobe at NAC Optical microscopy was performed using a stereoscopic NIKON SMZ-2T microscope with maximum magni®cation of 125, and a NIKON LABOPHOT microscope with maximum magni®cation of 1000. In both cases re¯ective microscopy was used, together with polarisation ®lters in the cases of high magni®cation, to improve contrast. The NAC NMP set-up has been described previously [3]. A custom-made lid that allows for stepper motor control of the target ladder in the X, Y and Z-directions, has been recently added. The stepper motors are computer-controlled and a permanent set of standards has been mounted on the target ladder. The positions of these standards are stored in computer memory to simplify the calibration process.
4. Results 4.1. Optical microscopy Fig. 1(a) and (b) show photographs of EGG and EGG-REST areas, respectively, taken with the stereoscopic microscope at a magni®cation of 68. There is a clear indication of surface spots that represent a form of surface roughness. The density of these spots is much higher in the EGG area, compared to the EGG-REST area. This was tested for dierent pieces of glass containing EGG and EGG-REST areas, and was found to be consistent. Similar analyses were carried out for the remaining samples. Results can be summarised as follows: There were small irregularities in the YEAST-REST area and these were fairly densely spaced whilst the YEAST area showed large protrusions, which were likely to be the areas where yeast got stuck to the sides. Some of the other samples showed similar trends to the EGG-REST area, these were TREATED-NEW, UNTREATED-NEW (only a few very large protrusions were visible) and TREATED-OK. Similarly the areas that can be compared to the EGG result are
Table 1 Description of bottles analysed and abbreviations used. Bottles indicated as `treated' were treated with 134-A freon Description
Abbreviation
Aected area of bottles with areas of ``egg'' formation (treated) Unaected area of bottles with areas of ``egg'' formation (treated) Bottles with areas where yeast stuck on the sides (treated) Bottles showing no problem during usage (treated) Bottles not yet used (treated) Bottles not yet used (untreated) Bottles with no pressure after fermentation (treated)
EGG EGG-REST YEAST TREATED-OK TREATED-NEW UNTREATED-NEW TREATED-JUNK
Fig. 1. (a) and (b) Photographs of EGG and EGG-REST areas at a magni®cation of 68. The bright angular area in Fig. 1(b) shows the result of a subsequent beam scan of the NMP, resulting in damage to the surface.
J. Padayachee et al. / Nucl. Instr. and Meth. in Phys. Res. B 158 (1999) 695±700
TREATED-JUNK and UNTREATED-NEW. These results, with the exception of the UNTREATED-NEW are consistent with these irregularities being responsible for the problems experienced with either egg formation or yeast sticking to the glass. The NMP analysis was conducted on the basis of these results. 4.2. Proton induced X-ray emission Proton induced X-ray emission (PIXE), using 3 MeV protons and a Si(Li) detector with an 80 lm Al ®lter, was used to analyse one piece of glass, which had a transition area from EGG to EGGREST. Even though no direct depth information can be obtained from PIXE measurements, one would expect to detect the depletion of Ca at the surface from the reduced yields of Ca X-rays. X-ray measurements were made on a line from an EGG area to an adjacent EGG-REST area to see whether there were dierences consistent with the expected treatment eects. Even though Na can be measured with the NMP using PIXE, in this case the strong X-ray yield from Si had to be ®ltered out, and therefore the low energy X-rays from Na were strongly attenuated. Therefore, only results from Ca are presented and this is shown in Fig. 2. It is clear that the Ca content decreases (this was con®rmed to be surface-related with subsequent Rutherford Backscattering Spectrometry (RBS) measurements) as the beam is moved from the EGG to the EGG-REST area. This is consistent with the expected eect of the surface treat-
Fig. 2. Ca X-ray yields as the beam is moved from an EGG to an EGG-REST area.
697
ment on the EGG-REST area, with less or no treatment in the EGG area, which might be an indication of the reason of the EGG formation on the eected area. As this area is also the area with the high density of ÔspotsÕ compared to the EGGREST area, this suggests a relation between nonoptimal treatment, a problem during manufacturing and the appearance of the ÔspotsÕ. The elements that were measured by PIXE are shown in Table 2 together with the concentrations of a typical analysis, error bars based on two standard deviations and the minimum detection limits (MDL). Except for the Ca data, other elements did not show any correlation, although concentrations varied signi®cantly between dierent points separated by a few tens of micrometers. The F content was measured simultaneously during the PIXE measurement, using the 19 F(p,a)16 O reaction, and although low concentrations were measured, there were no correlations between dierent areas. This is to be expected since the F-compounds that form during treatment would be soluble, and would dissolve as the bottles are ®lled. 4.3. Rutherford Backscattering Spectrometry All bottles supplied were analysed using 2 MeV alpha particles and both the backscattered partiTable 2 Typical results of a PIXE measurement, showing the detection of all elements heavier than K. Results are expressed in terms of ppm by weight unless otherwise indicated Element
Concentration
MDL (ppm)
K Ca Ti V Cr Mn Fe Cu Zn As Kr Rb Sr Zr
4460 190 8.3% 0.016% 190 7 30 11 1600 46 118 19 2100 91 60 25 85 18 170 36 190 75 190 64 600 130 400 100
30 30 14 15 15 19 18 40 46 77 150 170 200 250
698
J. Padayachee et al. / Nucl. Instr. and Meth. in Phys. Res. B 158 (1999) 695±700
cles and X-rays were collected for analysis. The annular surface barrier detector (used for the RBS measurement) is at an angle of 176° to the incoming beam and subtends a solid angle of 13 msr. A typical RBS spectrum of an EGG area is shown in Fig. 3, together with a theoretical spectrum (dotted line) generated using RUMP [4]. The energy calibration was performed with a thin Ir layer on a Si substrate, which correlated well with O and Si. As Fe and Cr cannot be separated during these measurements, the Fe edge represents both Fe and Cr. There are signi®cant dierences between the theoretical spectrum and the measured data in two regions of the spectrum. The ®rst dierence of interest to this investigation is found at the surface edge of the Ca signal, where the measured Ca signal is actually measured below the surface of the sample, indicating depletion of Ca in the surface layer. This represents an indication that there has been at least partial treatment at this area. It is, however, not accompanied by a similar depletion of Na in the surface layer, a phenomenon to be expected in the case of successful treatment. The second signi®cant difference of interest, between the theoretical spectrum and the measured data, is at the Na that is deep in the sample, where there is an indication of a slightly higher yield for the measurement, com-
pared with the simulation. This represents an enrichment of Na at higher depths, possibly not related to the treatment of the glass. Table 3 summarises the bulk compositional results of the RBS measurements, indicating the average concentration measured by RBS. As RBS can only measure major and minor elements, trace elements are not indicated. As the signals from Fe and Cr cannot be separated, the sum of their concentrations is given. The comparison of EGG and EGG-REST areas is shown in Fig. 4. The only signi®cant dierence between the EGG and EGG-REST area can be found just beyond the surface signals of Ca, with the EGG-REST signals showing the depletion of Ca just beyond the surface, when compared with the EGG area. This suggests slightly better treatment results at the EGG-REST area. There is also a dierence between the Fe/Cr signals at large depths, but this should not have any in¯uence on the eect studied. A similar comparison was made for the TREATED-OK and TREATED-JUNK areas, and the comparison is shown in Fig. 5. In this comparison there is a clear dierence between the surface edges of Ca, and to some extent Na, and again the interpretation can be made that the treatment received in the case of TREATED-OK was superior than that received by the TREATED-JUNK area. The Na content of the TREATED-JUNK area is also higher than that of the TREATED-OK area. These areas were from different bottles, however, and may indicate variations of the dierent bottles. The comparison for the TREATED-OK and UNTREATED-NEW areas is shown in Fig. 6. In this case signi®cant reduction of the surface concentration of Ca was found in the TREATED-OK Table 3 Average composition of the glass, based on RBS measurements
Fig. 3. A typical RBS spectrum of glass, with the experimental data shown together with a theoretical spectrum (dotted line).
Element
Concentration (wt%)
Si O Na Ca Fe + Cr
24.0 64.0 7.3 4.4 0.3
J. Padayachee et al. / Nucl. Instr. and Meth. in Phys. Res. B 158 (1999) 695±700
Fig. 4. Comparison of RBS spectra from an EGG vs. EGGREST area showing the full RBS spectra. The inset shows a zoomed picture of the Ca edge area.
699
Fig. 6. Comparison of RBS spectra from a TREATED-OK vs. UNTREATED-NEW area.
5. Summary and conclusions
Fig. 5. Comparison of RBS spectra from a TREATED-OK vs. TREATED-JUNK area.
sample, with subsequent higher concentration in the Si concentration at the surface. This shows that the treatment was essentially successful, however, the signi®cant dierence between the surface concentrations of Na as represented by the spectra, seems to be opposite to that expected from the treatment, i.e. the Na composition at the surface is enriched for the TREATED-OK sample.
Large protrusions were found on the surface of the glass in all the aected bottles. This might have given the necessary roughness to the surface for EGG and YEAST formation. On all surfaces it was found that the Ca concentration at the surface was depleted, the amount of depletion varying from aected areas (EGG and YEAST) to unaffected areas. There is therefore a correlation between the depletion of the Ca concentration at the surface, and the formation of EGG and YEAST areas on the surface. These correlations are linked to inecient production and/or surface treatment processes and can be linked to manufacturing or treatment of the glass. The treatment given to the bottles yielded results which were consistent with that expected from the literature. There were clear indications that the surfaces that were treated did show depletion of Ca at the surface, although the expected simultaneous decrease of Na at the surface was not found in general. This is dicult to explain on the basis of ionic exchange, as the Na ion is more prone to ionic exchange during treatment than the Ca. The treatment was, however, not applied in a consistent manner, based on the variability of the
700
J. Padayachee et al. / Nucl. Instr. and Meth. in Phys. Res. B 158 (1999) 695±700
results obtained. As indicated by the ®gures presented, there are clear and signi®cant dierences and these dierences could also be related to the satisfactory behaviour in cases where treatment seemed adequate, and formation of EGG and YEAST indicated areas where treatment was not adequate. No correlation could be found between the results of the optical microscopy, showing a high density of ÔspotsÕ in areas where problems occurred, and elemental analysis, which showed no signi®cant dierences of composition at these spots. There should be some dependence, however, based on the results obtained in this study and that
of the problems encountered in the bottles with optical problems due to spots on the surface.
References [1] R.K. Brow, W.C. LaCourse, J. Amer. Ceram. Soc. 66 (1983) C123. [2] C.R. Hense, J. Mecha, H.A. Schaeer, Glastech. Ber. 63 (1990) 127. [3] V.M. Prozesky, W.J. Przybylowicz, E. van Achterbergh, C.L. Churms, C.A. Pineda, K.A. Springhorn, J.V. Pilcher, C.G. Ryan, J. Kritzinger, H. Schmitt, T. Swart, Nucl. Instr. and Meth. B 104 (1995) 36. [4] L.R. Doolittle, Nucl. Instr. and Meth. B 15 (1986) 227.
www.elsevier.nl/locate/nimb
Nuclear microprobe and optical investigation of sparkling wine bottles J. Padayachee *, V.M. Prozesky, C.A. Pineda Van de Graa Group, National Accelerator Centre, P.O. Box 72, Faure 7131, South Africa
Abstract Glass bottles, used for sparkling wine, are treated with freon during manufacturing to harden the inside surface. Although this type of treatment normally improves the properties of the glass, in this case the occurrence of ``egg'' formations (egg-shaped rough areas) on distinct areas of bottles, as well as yeast sticking to the insides of bottles at speci®c areas pointed to the possibility of dierent areas showing dierent properties in the same bottle. The question was whether the correct gas was used for the treatment, and secondly, whether the process was controlled well enough to obtain the correct properties for the inside of the glass. We present results of an optical microscopy and nuclear microprobe (NMP) investigation. Ó 1999 Elsevier Science B.V. All rights reserved.
1. Introduction Glass bottles used for sparkling wine were treated with 134-A freon during manufacturing to harden the surface, and to prevent blooming (or weathering). The aim of the treatment process is to deplete the surface region of Na and Ca in order to obtain nearly pure silica on the surface, which gives the surface the desired qualities. The treatment has been shown to improve the properties of the glass, although in all previous studies [1,2] other freon gases, such as 152-A have been used and the question was, ®rstly, whether the correct gas was being used for treatment, and secondly, whether the process was controlled well enough to
* Corresponding author. Tel.: +27-21-843-3820; fax: +27-21843-3543; e-mail: [email protected]
obtain the correct properties for the inside of the glass. The occurrence of ``egg'' formations on distinct areas of bottles, as well as yeast sticking to the insides of bottles at speci®c areas pointed to the possibility of dierent areas showing dierent properties in the same bottle. Such dierences were the focus of this investigation, using optical microscopy and the nuclear microprobe (NMP) [3] of the National Accelerator Centre (NAC), Faure, South Africa.
2. Sample preparation Glass bottles were supplied, representing different problems that are being encountered. Some bottles showed areas of ``egg'' formation, some where the sticking of yeast was a problem and some bottles with no problems. The bottles sup-
0168-583X/99/$ - see front matter Ó 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 5 8 3 X ( 9 9 ) 0 0 3 3 5 - 3
696
J. Padayachee et al. / Nucl. Instr. and Meth. in Phys. Res. B 158 (1999) 695±700
plied for analysis, described in Table 1, were all ®lled, except the TREATED-NEW and UNTREATED-NEW. Bottles were emptied, broken and glass fragments, collected from the regions of interest, were washed in nano-pure water, dried and mounted for analysis. 3. Optical microscopy and the nuclear microprobe at NAC Optical microscopy was performed using a stereoscopic NIKON SMZ-2T microscope with maximum magni®cation of 125, and a NIKON LABOPHOT microscope with maximum magni®cation of 1000. In both cases re¯ective microscopy was used, together with polarisation ®lters in the cases of high magni®cation, to improve contrast. The NAC NMP set-up has been described previously [3]. A custom-made lid that allows for stepper motor control of the target ladder in the X, Y and Z-directions, has been recently added. The stepper motors are computer-controlled and a permanent set of standards has been mounted on the target ladder. The positions of these standards are stored in computer memory to simplify the calibration process.
4. Results 4.1. Optical microscopy Fig. 1(a) and (b) show photographs of EGG and EGG-REST areas, respectively, taken with the stereoscopic microscope at a magni®cation of 68. There is a clear indication of surface spots that represent a form of surface roughness. The density of these spots is much higher in the EGG area, compared to the EGG-REST area. This was tested for dierent pieces of glass containing EGG and EGG-REST areas, and was found to be consistent. Similar analyses were carried out for the remaining samples. Results can be summarised as follows: There were small irregularities in the YEAST-REST area and these were fairly densely spaced whilst the YEAST area showed large protrusions, which were likely to be the areas where yeast got stuck to the sides. Some of the other samples showed similar trends to the EGG-REST area, these were TREATED-NEW, UNTREATED-NEW (only a few very large protrusions were visible) and TREATED-OK. Similarly the areas that can be compared to the EGG result are
Table 1 Description of bottles analysed and abbreviations used. Bottles indicated as `treated' were treated with 134-A freon Description
Abbreviation
Aected area of bottles with areas of ``egg'' formation (treated) Unaected area of bottles with areas of ``egg'' formation (treated) Bottles with areas where yeast stuck on the sides (treated) Bottles showing no problem during usage (treated) Bottles not yet used (treated) Bottles not yet used (untreated) Bottles with no pressure after fermentation (treated)
EGG EGG-REST YEAST TREATED-OK TREATED-NEW UNTREATED-NEW TREATED-JUNK
Fig. 1. (a) and (b) Photographs of EGG and EGG-REST areas at a magni®cation of 68. The bright angular area in Fig. 1(b) shows the result of a subsequent beam scan of the NMP, resulting in damage to the surface.
J. Padayachee et al. / Nucl. Instr. and Meth. in Phys. Res. B 158 (1999) 695±700
TREATED-JUNK and UNTREATED-NEW. These results, with the exception of the UNTREATED-NEW are consistent with these irregularities being responsible for the problems experienced with either egg formation or yeast sticking to the glass. The NMP analysis was conducted on the basis of these results. 4.2. Proton induced X-ray emission Proton induced X-ray emission (PIXE), using 3 MeV protons and a Si(Li) detector with an 80 lm Al ®lter, was used to analyse one piece of glass, which had a transition area from EGG to EGGREST. Even though no direct depth information can be obtained from PIXE measurements, one would expect to detect the depletion of Ca at the surface from the reduced yields of Ca X-rays. X-ray measurements were made on a line from an EGG area to an adjacent EGG-REST area to see whether there were dierences consistent with the expected treatment eects. Even though Na can be measured with the NMP using PIXE, in this case the strong X-ray yield from Si had to be ®ltered out, and therefore the low energy X-rays from Na were strongly attenuated. Therefore, only results from Ca are presented and this is shown in Fig. 2. It is clear that the Ca content decreases (this was con®rmed to be surface-related with subsequent Rutherford Backscattering Spectrometry (RBS) measurements) as the beam is moved from the EGG to the EGG-REST area. This is consistent with the expected eect of the surface treat-
Fig. 2. Ca X-ray yields as the beam is moved from an EGG to an EGG-REST area.
697
ment on the EGG-REST area, with less or no treatment in the EGG area, which might be an indication of the reason of the EGG formation on the eected area. As this area is also the area with the high density of ÔspotsÕ compared to the EGGREST area, this suggests a relation between nonoptimal treatment, a problem during manufacturing and the appearance of the ÔspotsÕ. The elements that were measured by PIXE are shown in Table 2 together with the concentrations of a typical analysis, error bars based on two standard deviations and the minimum detection limits (MDL). Except for the Ca data, other elements did not show any correlation, although concentrations varied signi®cantly between dierent points separated by a few tens of micrometers. The F content was measured simultaneously during the PIXE measurement, using the 19 F(p,a)16 O reaction, and although low concentrations were measured, there were no correlations between dierent areas. This is to be expected since the F-compounds that form during treatment would be soluble, and would dissolve as the bottles are ®lled. 4.3. Rutherford Backscattering Spectrometry All bottles supplied were analysed using 2 MeV alpha particles and both the backscattered partiTable 2 Typical results of a PIXE measurement, showing the detection of all elements heavier than K. Results are expressed in terms of ppm by weight unless otherwise indicated Element
Concentration
MDL (ppm)
K Ca Ti V Cr Mn Fe Cu Zn As Kr Rb Sr Zr
4460 190 8.3% 0.016% 190 7 30 11 1600 46 118 19 2100 91 60 25 85 18 170 36 190 75 190 64 600 130 400 100
30 30 14 15 15 19 18 40 46 77 150 170 200 250
698
J. Padayachee et al. / Nucl. Instr. and Meth. in Phys. Res. B 158 (1999) 695±700
cles and X-rays were collected for analysis. The annular surface barrier detector (used for the RBS measurement) is at an angle of 176° to the incoming beam and subtends a solid angle of 13 msr. A typical RBS spectrum of an EGG area is shown in Fig. 3, together with a theoretical spectrum (dotted line) generated using RUMP [4]. The energy calibration was performed with a thin Ir layer on a Si substrate, which correlated well with O and Si. As Fe and Cr cannot be separated during these measurements, the Fe edge represents both Fe and Cr. There are signi®cant dierences between the theoretical spectrum and the measured data in two regions of the spectrum. The ®rst dierence of interest to this investigation is found at the surface edge of the Ca signal, where the measured Ca signal is actually measured below the surface of the sample, indicating depletion of Ca in the surface layer. This represents an indication that there has been at least partial treatment at this area. It is, however, not accompanied by a similar depletion of Na in the surface layer, a phenomenon to be expected in the case of successful treatment. The second signi®cant difference of interest, between the theoretical spectrum and the measured data, is at the Na that is deep in the sample, where there is an indication of a slightly higher yield for the measurement, com-
pared with the simulation. This represents an enrichment of Na at higher depths, possibly not related to the treatment of the glass. Table 3 summarises the bulk compositional results of the RBS measurements, indicating the average concentration measured by RBS. As RBS can only measure major and minor elements, trace elements are not indicated. As the signals from Fe and Cr cannot be separated, the sum of their concentrations is given. The comparison of EGG and EGG-REST areas is shown in Fig. 4. The only signi®cant dierence between the EGG and EGG-REST area can be found just beyond the surface signals of Ca, with the EGG-REST signals showing the depletion of Ca just beyond the surface, when compared with the EGG area. This suggests slightly better treatment results at the EGG-REST area. There is also a dierence between the Fe/Cr signals at large depths, but this should not have any in¯uence on the eect studied. A similar comparison was made for the TREATED-OK and TREATED-JUNK areas, and the comparison is shown in Fig. 5. In this comparison there is a clear dierence between the surface edges of Ca, and to some extent Na, and again the interpretation can be made that the treatment received in the case of TREATED-OK was superior than that received by the TREATED-JUNK area. The Na content of the TREATED-JUNK area is also higher than that of the TREATED-OK area. These areas were from different bottles, however, and may indicate variations of the dierent bottles. The comparison for the TREATED-OK and UNTREATED-NEW areas is shown in Fig. 6. In this case signi®cant reduction of the surface concentration of Ca was found in the TREATED-OK Table 3 Average composition of the glass, based on RBS measurements
Fig. 3. A typical RBS spectrum of glass, with the experimental data shown together with a theoretical spectrum (dotted line).
Element
Concentration (wt%)
Si O Na Ca Fe + Cr
24.0 64.0 7.3 4.4 0.3
J. Padayachee et al. / Nucl. Instr. and Meth. in Phys. Res. B 158 (1999) 695±700
Fig. 4. Comparison of RBS spectra from an EGG vs. EGGREST area showing the full RBS spectra. The inset shows a zoomed picture of the Ca edge area.
699
Fig. 6. Comparison of RBS spectra from a TREATED-OK vs. UNTREATED-NEW area.
5. Summary and conclusions
Fig. 5. Comparison of RBS spectra from a TREATED-OK vs. TREATED-JUNK area.
sample, with subsequent higher concentration in the Si concentration at the surface. This shows that the treatment was essentially successful, however, the signi®cant dierence between the surface concentrations of Na as represented by the spectra, seems to be opposite to that expected from the treatment, i.e. the Na composition at the surface is enriched for the TREATED-OK sample.
Large protrusions were found on the surface of the glass in all the aected bottles. This might have given the necessary roughness to the surface for EGG and YEAST formation. On all surfaces it was found that the Ca concentration at the surface was depleted, the amount of depletion varying from aected areas (EGG and YEAST) to unaffected areas. There is therefore a correlation between the depletion of the Ca concentration at the surface, and the formation of EGG and YEAST areas on the surface. These correlations are linked to inecient production and/or surface treatment processes and can be linked to manufacturing or treatment of the glass. The treatment given to the bottles yielded results which were consistent with that expected from the literature. There were clear indications that the surfaces that were treated did show depletion of Ca at the surface, although the expected simultaneous decrease of Na at the surface was not found in general. This is dicult to explain on the basis of ionic exchange, as the Na ion is more prone to ionic exchange during treatment than the Ca. The treatment was, however, not applied in a consistent manner, based on the variability of the
700
J. Padayachee et al. / Nucl. Instr. and Meth. in Phys. Res. B 158 (1999) 695±700
results obtained. As indicated by the ®gures presented, there are clear and signi®cant dierences and these dierences could also be related to the satisfactory behaviour in cases where treatment seemed adequate, and formation of EGG and YEAST indicated areas where treatment was not adequate. No correlation could be found between the results of the optical microscopy, showing a high density of ÔspotsÕ in areas where problems occurred, and elemental analysis, which showed no signi®cant dierences of composition at these spots. There should be some dependence, however, based on the results obtained in this study and that
of the problems encountered in the bottles with optical problems due to spots on the surface.
References [1] R.K. Brow, W.C. LaCourse, J. Amer. Ceram. Soc. 66 (1983) C123. [2] C.R. Hense, J. Mecha, H.A. Schaeer, Glastech. Ber. 63 (1990) 127. [3] V.M. Prozesky, W.J. Przybylowicz, E. van Achterbergh, C.L. Churms, C.A. Pineda, K.A. Springhorn, J.V. Pilcher, C.G. Ryan, J. Kritzinger, H. Schmitt, T. Swart, Nucl. Instr. and Meth. B 104 (1995) 36. [4] L.R. Doolittle, Nucl. Instr. and Meth. B 15 (1986) 227.