R.F. Schumacher, R.J. Workman, and T.B. Edwards
Westinghouse Savannah River Company
Aiken, SC 29808

This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

This report has been reproduced directly from the best available copy.

Available for sale to the public, in paper, from:  U.S. Department of Commerce, National Technical Information Service, 5285 Port Royal Road, Springfield, VA 22161,  phone: (800) 553-6847,  fax: (703) 605-6900,  email:  orders@ntis.fedworld.gov   online ordering:  http://www.ntis.gov/support/ordering.htm

Available electronically at  http://www.osti.gov/bridge/

Available for a processing fee to U.S. Department of Energy and its contractors, in paper, from: U.S. Department of Energy, Office of Scientific and Technical Information, P.O. Box 62, Oak Ridge, TN 37831-0062,  phone: (865 ) 576-8401,  fax: (865) 576-5728,  email:  reports@adonis.osti.gov

The Harrop, High-Temperature Viscometer has been in operation at the Savannah River Technology Center (SRTC) for several years and has proven itself to be reasonably accurate and repeatable. This is particularly notable when taking into consideration the small amount of glass required to make the viscosity determination [1,2]. The volume of glass required is only 2.60 cc or about 6 to 7 grams of glass depending on the glass density. This may be compared to the more traditional viscosity determinations, which generally require between 100 to 1000 grams of glass. Before starting the present investigation, the unit was re-aligned and the furnace thermal gradients measured. The viscometer was again calibrated with available NIST Standard Reference Material glasses (717a and 710a) and a spindle constant equation was determined. Standard DWPF Waste Compliance Glasses (Purex, HM, and Batch 1) were used to provide additional verification for the determinations at low temperature. The Harrop, High-Temperature Viscometer was then used to determine the viscosity of three random samples of ground and blended DWPF, Black, Start -Up Frit, which were obtained from Pacific Northwest National Laboratory (PNNL). The glasses were in powder form and required melting prior to the viscosity determination. The results from this evaluation will be compared to "Round Robin" measurements from other DOE laboratories [3] and a number of commercial laboratories. The results obtained in this investigation, over a three month time period, indicated excellent consistency (< 3% relative standard deviation) and good agreement (2.5%) with viscosity measurements obtained during a 1987 characterization of the Frit [4] by the commercial laboratory, Sharp-Schurtz. These results also agreed to within 6% of a round robin determination on the same ground samples at Corning Engineering Laboratories (CELS)[5].

Keywords: DWPF, Glass, Viscosity, Start-Up Frit Measurement, Calibration, Round Robin

The DWPF Start-Up Frit was selected for this Round Robin evaluation because there is no available standard reference material (SRM) from the National Institute for Standards and Testing (NIST) for the determination of glass viscosity in the temperature region of interest for radioactive waste vitrification. The Start-Up Frit is an alkali-alumino-borosilicate glass that is roughly representative of glasses to be produced at the U.S. high-level vitrification sites (West Valley, Savannah River, Idaho National Engineering and Environmental Laboratory, and at Hanford).

Three 1000 gram samples of DWPF Start-Up Frit (Sample No’s 57, 58, and 59) were obtained from John Vienna at Pacific Northwest National Laboratory. The glass was originally obtained by PNNL from the DWPF Start-Up Frit material [4], ground to a fine powder, and then blended and split into lots of 1000 grams. From the three 1000 gram samples received at SRTC approximately 50 to 60 grams were removed. The 50 to 60 gram samples were melted at 1150°C for two hours in covered platinum crucibles. The density was determined twice for each of the three samples and then the viscosity was determined at least three times for each sample. The remaining amounts of the three 1000 gram samples were shipped to CELS for their measurement of glass viscosity and other physical properties. This measurement also included the re-melting of the three powder samples at 1150°C for two hours in closed platinum crucibles. CELS carried out three independent determinations of the glass density, thermal expansion, and glass viscosity. Table 1 presents a summary of the calculated viscosity at 1150°C for CELS and Sharp Schurtz and the calculated viscosity measurement for the eleven SRTC measurements. The SRTC results were somewhat lower (~3-6%) than either of the commercial laboratories at 1150°C. Very recent preliminary data from PNNL provides calculated viscosities which were closer to the SRTC numbers.

A graph of the SRTC measurements is presented with a comparison of the Sharp Schurtz data (10/27/87) and CELS measurements as Figure 1. The diamond markings are the Sharp Schurtz calculated viscosity for the middle sample (10/27/87)[4]. The CELS markings or triangles, are calculated from the viscosity equation for the middle sample #58 [5]. The small horizontal bars are the nine SRTC determinations. The agreement is excellent at the high temperature end and the SRTC measured viscosity values only slightly higher at the low temperature end. It should be noted that the liquidus temperature for this glass is between 1050 to 1075°C and this may account for some of the increased variability observed on the low temperature measurements.

The Brookfield Viscometer was a Brookfield Model RVDVII+ with serial number RT49625. The Harrop High-Temperature Viscometer unit, as shown in Figure 2 attached at the end of this report, was assembled and aligned with the spindle as close to the furnace "center line" as possible. The platinum spindles were straightened at the glass shop using a glass lathe. The platinum spindles rotated evenly within the empty platinum crucible, but at times the spindle made contact with the crucible. It is now believed that when the glass is present this "wobble" contact is minimized by the presence of the glass and perfect alignment is not required. Originally two crucible / spindles were employed (A and B), but as time progressed two additional crucibles (C and D) were obtained and used. All the crucibles and spindles were fabricated to the same dimensions and there did not appear to be a correlation of viscosity with crucible identity. The spindles A and B continued to be employed through out this investigation.

A calibration thermocouple and read-out (GT 1423 and GT1101) were used to measure the thermal gradients within and above the platinum crucible. The estimated uncertainty of this measurement was ± 2.5°C as determined by the Standards Laboratory. During this evaluation, the furnace was closed off at the bottom and partially closed at the top, similar to actual operation. While the platinum crucible was positioned in the alumina holder, the spindle and Brookfield Viscometer were removed from the instrument. The furnace was raised to the normal operating position (0.0 inches on the scale) and the thermocouple lowered into the crucible from above. The gradient was determined from the inside-bottom of the crucible and up to four inches above the crucible bottom along the furnace center-line. Furnace set point, furnace crucible digital read out, and calibrated thermocouple temperatures were recorded and plotted. At the end of this report, see Figure 3 for a plot of the measured thermal gradient and location in respect to the crucible location. The furnace temperature was held at the set points of 800, 1000, 1200, and 1400°C. The crucible digital temperature was noted slightly above the X axis and the measured (calibrated) temperature with respect to the bottom of the crucible was presented on the Y axis.

Examination of Figure 3 indicates that the actual temperature in the crucible was higher than the digital furnace read-out at low temperatures and lower at high temperatures. In the initial set-up of the viscometer, when the NIST standards were used to determine the spindle constant (K), it was found that K could best be described as a linear equation as a function of temperature. The use of a "variable constant" will be shown in this report to be a very convenient means to correct for the observed temperature bias and excludes complicated calibration adjustments. The explanation for the "variable constant" was previously unknown. It can now be shown that the cause of the slope with temperature was the thermal gradient bias. Since the glass is contained in the one inch zone inside the crucible, the glass melt would be at a higher temperature than the digital reading around 800°C and lower than the digital reading at 1400°C. When the temperature bias is compensated, the value of K approaches a constant (ca. 110). See Figure 4 for a plot of data obtained with only the 717a NIST standard. The inclusion of a temperature correction nearly straightens the slope to a flat line. However, it was much easier to use the sloped equation for K than it would be to adjust all the measured digital temperatures to a correct value. The use of NIST standards to obtain the equation for the spindle constant compensates for any error due to temperature bias in the sample crucible and provides a viscosity / temperature relationship which should be correct and directly traceable to the NIST Standards.

The amount of glass required for a viscosity determination was selected as 2.60 cubic centimeters of glass [1]. In order to calculate the weight of glass required for this volume of glass, it was necessary to carefully measure the glass density. This was accomplished using fractured samples of the NIST and the DWPF glasses with a GTOP Procedure [6]. This procedure measured the buoyancy in distilled water in order to determine the glass volume, while the dry weight was measured directly with the same balance. NIST glass density standards were used to control the density measurements. The density for the DWPF type glasses was also calculated using an algorithm (Equation 1) based on the calculated molecular density correlated to a measured annealed density [7]. This algorithm was developed under the DWPF Batch 1 variability study and is best suited for Batch 1 glasses. These values were used to compare to the measured values.

Calculated Density = 1.124 + 0.598 x Molecular Density   (Equation 1)

Compositions for the glasses were obtained from published works [4,8]. Error in the density determination can raise or lower the melt level in the crucible and cause error in the viscosity determination. Most of the density determination error is caused by irregularities in the fractured glass surfaces. This may have caused some of the inconsistencies observed for the density determinations of the DWPF - HM type glasses. The estimated, calculated and measured densities are provided in Table 2 along with other measured densities for comparison. It appeared that the calculated density value provided the best approximation for the Batch 1 glass as was expected.

The furnace heating and measurement schedule for the viscosity determinations of the two high temperature NIST standards (717a & 710a) was programmed into the furnace as follows:

Room Temperature to 1450°C – Time 2 Hours
Hold at 1450°C – 1 Hour (Determine % Torque)
1450 to 1250°C – ½ Hour
Hold at 1250°C – ½ Hour (Determine % Torque)
1250 to 1150°C – ¼ Hour
Hold at 1150°C – ½ Hour (Determine % Torque)
1150 to 1050°C – ¼ Hour
Hold at 1050°C – ½ Hour (Determine % Torque)
1150 to 1250°C – ½ Hour
Hold at 1250°C – ½ Hour (Determine % Torque)
Total Time = 6.5 Hours

The furnace schedule for the viscosity determinations of the SRM 711 and all the DWPF type glasses was programmed as follows:

Room Temperature to 1200°C – Time 2 Hours
Hold at 1200°C – 1 Hour (Determine % Torque)
1200 to 1150°C – ¼ Hour
Hold at 1150°C – ½ Hour (Determine % Torque)
1150 to 1100°C – ¼ Hour
Hold at 1100°C – ½ Hour (Determine % Torque)
1100 to 1050°C – ¼ Hour
Hold at 1050°C – ½ Hour (Determine % Torque)
1050 to 1000°C – ¼ Hour
Hold at 1000°C – ½ Hour (Determine % Torque)
1000 to 950°C – ¼ Hour
Hold at 950°C – ½ Hour (Determine % Torque)
Total Time = 6 ¾ Hours

After each Hold period, the Brookfield RVDV-II+ was used to measure the viscosity. The measurement was repeated until stability was obtained (usually three times). At the completion of the measurements, the crucible was cooled to room temperature, placed in another furnace to drain and then cleaned in hydrofluoric acid. By using multiple crucibles and spindles, one glass could usually be measured per day.

Three National Institute of Standards and Technology (NIST) Standard Reference Materials (SRM) were available for the development of the spindle constant equation. These glasses were the Standard Reference Material (SRM) 710a (Soda-Lime-Silica Glass) and SRM 717a (Borosilicate Glass) and SRM 711 (Lead Silica Glass). A very limited amount of the more appropriate SRM 711 was obtained from Pacific Northwest National Laboratory and only one determination could be made. The SRM’s have the accepted Fulcher equations as follows:

SRM 717a     Log10 [Viscosity (Poise)] = -1.560 + 4852.2 / (T - 192.462)   (Equation 2)
SRM 710a     Log10 [Viscosity (Poise)] = -1.729 + 4560 / (T – 240.8)     (Equation 3)
SRM 711       Log10 [Viscosity (Poise)] = -1.621 + 4254.649/ (T – 152.1)    (Equation 4)

The symbol T is introduced into the equation as the glass temperature in degrees Centigrade. These Fulcher equations allow the calculation of the NIST accepted viscosity for a particular temperature. The spindle constant can then be calculated from the Brookfield Viscometer values and Equation 5:

K = (NIST Viscosity) x (Rotational Speed) / (% Torque) Equation 5

The measurements for the NIST SRM’s are presented in attached Tables 3 and 4. It should be noted that the spindle constant varied as a function of temperature as previously explained. The temperature and spindle constant data for the all the NIST glasses was plotted using the EXCEL Ô Chart Tool to obtain an average linear relationship for the spindle constant with temperature. The plot of this data is presented as attached Figure 5 and the derived linear equation for K as a function of T was:

K(T) = 175.19 - 0.0493 (T) Equation 6

Where K is the spindle constant, T is the temperature in ° C. All of the data for the NIST SRM determinations appeared to lie in a band of approximately K +/- 10. Since K is generally greater than 100 this calculates to less than a 10% variation.

Small amounts of the DWPF "Waste Compliance Glasses" Purex, HM, and Batch 1 were obtained and the viscosities determined. The resulting viscosities were calculated using Equation 6 to obtain the appropriate spindle constant K and Equation 7 to calculate the viscosity.

Viscosity = K (T) x (% Torque) / ( Rotation Speed) Equation 7

The viscosities were calculated and compared to the viscosity determined by a commercial laboratory (Sharp Schurtz) and reported previously [4]. See Table 5 for the results. In general the viscosity measured with the Harrop High Temperature Viscometer was about 10% higher than the commercial measurement.

A procurement of services contract was obtained with Corning Engineering Laboratory Services (CELS) to measure the viscosity of the three frit samples (#57, #58, and #59) from PNNL. The Purchase Order Number 6C9846 was provided for this procurement. The viscosity determination was made on the three samples after measurement of the physical properties required to control the viscosity determination.

These physical properties are presented in Table 6 for glass melted at 1150°C for two hours from the supplied powder samples.

The viscosity results, reported in Table 7, will be included in the round robin evaluation and were discussed in some detail at the beginning of this report.

Each set of the viscosity measurements generated by SRTC for the DWPF, Start-up Frit was fit to a Fulcher equation. The viscosities (in Poise) of each of these glasses at various temperatures (including 1150°C) are to be estimated from the corresponding Fulcher equation. The functional form of the (three-parameter) Fulcher equation (expressed in Poise) used to fit these data is given by

Equation 8

In this equation, represents the natural logarithm of the estimated viscosity (in Poise), , and A, B, and C represent the parameters of the model that were determined from the measurements provided in Table 7. The model fitting was conducted using JMP® Version 4 [11], and the results for each set of data are summarized in Appendix D.

The fitted models of Appendix D were then used to predict viscosities of the given glass at various temperatures and at 1150°C. The results are provided in the following Table 7. It should be noted that the relative standard deviations at all temperatures between 950 to 1200°C were less than 3%. Again the variability was larger at lower temperatures. The average of the viscosities at 1150°C was near 42 Poise.

Table 7. Calculated Viscosities from Fulcher Equation Fits to SRTC Measured Viscosities

All of the viscosities determined at SRTC with the Harrop High Temperature Viscometer were very repeatable and agreed within 10 % of the values obtained from CELS and Sharp Schurtz over the temperature region of interest to waste vitrification (1000 to 1200°C). For the DWPF Start-Up Frit the relative standard deviation between measurements at 1150°C was a little over 2%. The relative standard deviation was less than 3% over the temperature range between 1000 to 1250°C. This is a very low variability considering this technique required a very small amount of glass, approximately 6 to 7 grams, per determination, and these measurements were obtained over a three month period. This technique should be very advantageous to the viscosity measurement of radioactive glasses where the amount of glass should be minimized to limit the radiation dose.

The use of a spindle constant equation as a function of temperature was explained and was found to be due to temperature gradients in the vicinity of the sample crucible. However, it was found that the use of this equation to calculate a spindle constant for each measurement temperature was a simple means of linking the measurement directly to the NIST viscosity standards.

The measurement of the glass density may be an important factor in the accurate determination of the viscosity and may account for some of the larger variability observed between the SRTC values and the commercial determinations for the waste compliance glasses.

After the Round Robin is completed, it would be appropriate to re-calibrate the Harrop High-Temperature Viscometer with the DWPF Start-Up Frit. If the NIST SRM 711 finally becomes available, this glass should be used as well.

Preparations to operate the second Harrop High Temperature Viscometer for use with radioactive glasses should be initiated.

Just a few words of appreciation to Phyllis Workman who performed all the viscosity determinations reported here. The care and effort required to make these measurements is obvious, especially to anyone who has measured glass viscoity. What is not obvious is the cheerful cooperation and understanding she provided when it was often necessary to make additional measurements to clarify these findings.

1. R.F. Schumacher & D.K. Peeler, "Establishment of Harrop, High-Temperature Viscometer-(U)," WSRC-RP-98-00737, Rev. 0, September 30, 1998.
2. T.B. Edwards, J.R. Harbour, R.F. Schumacher, & R.J. Workman, "Measurement of DWPF Glass Viscosity – Final Report," WSRC-RP-99-01053, Rev. 0, November 11, 1999.
3. R.F. Schumacher, "Task Technical & QA Plan," WSRC-TR-2000-0097, Rev. 0, April 14, 2000.
4. C.M. Jantzen, "Characterization of the Defense Waste Processing Facility (DWPF) Startup Frit (U)," WSRC-RP-89-18, April 18, 1989.
5. Corning Engineering Laboratory Services Report 11988-150, March 30 1999.
6. D.K. Peeler, "Glass Density Using the Mettler AT 400 Balance," GTOP 3-105, Rev. 0, November 16, 1998.
7. R.F. Schumacher, Basic Data Report: Simulated New Batch 1 Glasses-(U)," WSRC-RP-95-0539, Rev. 0, July 15, 1998.
8. C.A. Cicero, S.L. Marra, & M.K. Andrews, "Phase Stability Determinations of DWPF Waste Glasses (U)," WSRC-TR-93-227, Rev.0, May 1993.
9. Personnel Communication: John Vienna , Pacific Northwest National Laboratory, Density Data, August 29, 2000.
10. R.F. Schumacher, "Determination of Glass Viscosity," GTOP 3-111, Rev. 0, L13.1 May 17, 1999.
11. SAS Institute, JMP® Statistics and Graphics Guide: JMP Version 4, SAS Institute, Inc., Cary NC, 2000.

Figure 2. Schematic of Harrop High-Temperature Viscometer

Figure 3. Measured Furnace Gradients for Harrop Viscometer at Various Set Point Temperatures

Figure 5. Spindle Constants for All NIST Glasses (710a, 717a and 711) Compared to Temperature

Table 3. Spindle Constant for SRM 717a Determinations

NIST Standard Reference Material 717a(Borosilicate)

APPENDIX C
Date	12/15/00		Crucible C & Spindle B
Glass	#59		Crucible w/glass		291.263
Density	2.6879		Empty crucible		284.364
Goal	6.9885		Actual amount		6.899
TIME	TIME	SET RATE	FURNACE	SAMPLE	SPEED	TORQUE
MIL	LEFT	°C	°C	°C	RPM	%
1000	0.1	1200	1200	1203	60	14.8
1005	0.05	1200	1200	1203	60	14.8
1010	0	1200	1200	1203	60	14.8
1045	0.1	1150	1149	1147	60	21.8
1050	0.05	1150	1149	1147	60	21.75
1055	0	1150	1150	1147	60	21.75
1130	0.1	1100	1100	1094	60	33.4
1135	0.05	1100	1100	1094	60	33.4
1140	0	1100	1100	1094	60	33.4
1215	0.1	1050	1049	1040.5	60	53.75
1220	0.05	1050	1050	1040.5	60	53.75
1225	0	1050	1050	1040.5	60	53.75
1300	0.1	1000	999	987.5	30	46.3
1305	0.05	1000	1000	987.5	30	46.3
1310	0	1000	1000	987.5	30	46.3
1345	hold	950	950	934	12	34.3
1350	hold	950	950	934	12	34.3
1355	hold	950	950	934	12	34.3

Date	12/13/00		Crucible C & Spindle B
Glass	#58		Crucible w/glass		291.363
Density	2.6899		Empty crucible		284.368
Goal	6.9937		Actual amount		6.995
TIME	TIME	SET RATE	FURNACE	SAMPLE	SPEED	TORQUE
MIL	LEFT	°C	°C	°C	RPM	%
1010	0.10	1200	1200	1202.5	60	14.7
1015	0.05	1200	1200	1202.5	60	14.7
1020	0.00	1200	1200	1202.5	60	14.7
1055	0.10	1150	1149	1146.5	60	21.55
1100	0.05	1150	1150	1146.5	60	21.55
1105	0.00	1150	1150	1146.5	60	21.55
1140	0.10	1100	1099	1093	60	33
1145	0.05	1100	1100	1093	60	33
1150	0.00	1100	1100	1093	60	33
1225	0.10	1050	1049	1040	60	53.25
1230	0.05	1050	1049	1040	60	53.25
1235	0.00	1050	1050	1040	60	53.25
1310	0.10	1000	999	987	30	45.8
1315	0.05	1000	1000	987	30	45.7
1320	0.00	1000	1000	987	30	45.75
1355	hold	950	950	933	12	33.85
1400	hold	950	950	933	12	33.85
1405	hold	950	950	933	12	33.85