DENSITY OF MOLTEN GLASS 
 
 BY 
 
 SHEO-HEN El 
 
 THESIS 
 
 FOR THE 
 
 DEGREE OF BACHELOR OF SCIENCE 
 
 IN 
 
 CHEMICAL ENGINEERING 
 
 COLLECE LIBERAL ARTS AND SCIENCES 
 
 UNIVERSITY OF ILLINOIS 
 
 1922 
 

 
 
 
 
 
 ■ 
 
 ■ 
 
 
 
 
 
 
 
 ' 
 
 
 
 
 i 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ACKN OWLEDGMEUTS 
 ******* 
 
 The writer v?ishes to take this opportunity to express 
 his appreciation of the guidance and help that Professor E.W. Wash- 
 bum has so kindly given. 
 
 He also wishes to express his appreciation for the help 
 given by Dr .E. IT. Bunting in the preparation of this thesis. 
 
 ************* 
 
Digitized by the Internet Archive 
 in 2015 
 
 https://archive.org/details/densityofmoltengOOIish 
 
TABLE 05’ 
 
 CONTENTS 
 
 Page 
 
 I. INTRODUCTION 1 
 
 II . EXPERIMENTAL PART. 2 
 
 III. CALIBRATION OF THE SPRING 5 
 
 IV. CALCULATIONS... 7 
 
 V. RESULTS FROM EXPERIMENTS 9 
 
 VI. DISCUSSION 11 
 
 1 . Suggestions 13 
 
 VII . SUMMARY • 14 
 

 
- 1 - 
 
 DENSITY OF MOLTEN GLASS 
 
 ■X-W-W X‘ k X' Ar X : 'k k '{-if -jf k )’; vif X* 
 
 I. INTRODUCTION 
 
 The importance of tne determination of the density of 
 molten glass can be discussed from two aspects. First, from the 
 
 purely scientific point of view, and secondly, from the practical 
 (point of view. 
 
 The art of glass making has been known since ancient times. 
 But since then there has not Been very much advancement until rather 
 modern times. People now-a-days know how to make glass, but the 
 reactions in various steps during its course of manufacture are not 
 completely known. Only little work has been done on the various 
 properties of glass at high temperatures. No work has been done 
 on the density of molten glasses. Thus although the art of glass 
 manufacture is quite perfect, yet the knowledge of its underlying 
 ! facts is still incomplete. Work of this kind will, of course, be 
 
 of some value for the literature. 
 
 From the practical point of view, density has a close 
 1 relation with viscosity, surface tension, and other physical constant 
 By knowing the densities at high temperatures, we can figure out 
 the coefficient of expansion at different temperatures. This work 
 has been carried on to 600°C by C.G. Peters and C.H.Cragoe.* 
 
 * 
 
 J.Opt. Soc.Am.4, 105-44, 1920 
 

- 2 - 
 
 The density of molten glass is an important factor in the process 
 of feeding the molten glass to the molds or machines for making 
 different articles. 
 
 
 
-3- 
 
 II. EXPERIMENTAL PART. 
 
 The method used for the determination of the density of 
 molten glass was to immerse a platinum sphere of known volume in tie 
 fluid , and the loss of weight was measured . The density was then 
 
 obtained by a simple process of division with a few necessary 
 corrections. The sphere was suspended on a Jolly spring balance 
 iwith a fine platinum wire. In order to diminish the effect of 
 jsurface tension, which was one source of error, as much as possible, 
 and at the same time to have the wire strong enough to nold the 
 weight at high temperatures, wire of 0.3 mm. in diameter was used. 
 The furnace was heated by electricity by passing current through 
 a platinum resistance coil. As shown on the diagram, the furnace 
 consists of a long porcelain cylinder w r ith platinum heating 
 element wound outside. The glass pot in which the glass was melted 
 was of similar construction, but smaller in size so that it would 
 just fit into the furnace. One more porcelain pot a little 
 larger than the heating pot was used to protect the latter and 
 facilitate the task of mending in case any accident happened to the 
 furnace. ho w the whole apparatus was put in a large fire clay 
 
 cylinder containing a lot of insulating materials, such as silocel 
 and calcined kaolin. In order to prevent radiation loss on the 
 
 top, a well fitted porcelain cover was used. The cover consisted 
 of two equal semi-circular discs with a hole in the center so that 
 we could see what was going on during the run. The temperature 
 was measured by a plat inum- rhodium thermocouple inserted between 
 
 ^ ^ ~ 1 — ^ ■— — —— — — i 
 
the glass pot and. the heating cylinder. how the glass in small 
 pieces was charged into the pot with the thermocouple fitted in 
 position, and the current turned on. After the glass 136031116 
 quite fluid the platinum hall was inserted. Care was taken 
 
 not to immerse too much of the wire beneath the surface, but just 
 enough so that the ball was completely immersed. The temperature 
 of the glass was raised to a little above 1400°C and icept at this 
 temperature for a few moments in order to secure equilibrium. 
 
 Then readings were ready to be taken. The distance or stretch 
 between two definite points, A and B ( see the diagram) on the 
 spring was read by means of a cathetometer to 0*02 mm. Headings 
 were taken at 1200°, 1300°, and 1400° respectively. A reading 
 was first made at 1400°, and then the furnace was coded down very 
 slowly to 1300°C, maintained at this temperature, when another 
 reading was taken. A similar reading was made a„t 1200°C. After 
 
 this, the temperature was again raised and another set of readings 
 obtained in order to check the first set of readings. 
 
? A 
 
 Spring 
 
 i 
 
 porce i ion Insu /at'ors 
 
 P/ohnam boll 
 
 Fire brick 
 
 
 PI a bin um 
 suspension wire 
 
 glass pot 
 
 , Platinum 
 
 heating e/errjenh 
 
 O l 
 
 • \ 
 • \ 
 • « 
 
 t t 
 
 ^ ; > 
 ' * 
 
 i ; 
 
 mil 
 
 
 m 
 
 IBjTY 1 . vviV/ 
 
 'U 
 
 Molten g/ass 
 Si/oce/ insulation 
 
 X 
 
 Fire c/ay cy/inoter 
 
 FURNACE AND SPRING FOR THE 
 DETERMINATION OF THE density 
 OF MOLTEN GLASS 
 

 
 
 
 
 
 V* v • . ■ 
 
 
 ^\»r\v\ ,\' . 
 
 V ' :■> <\f. \v,\ y.,. hxrt 
 
 V v: f. . t>\c , 
 rA'WW. \ 
 
 
 S> 1 ? 
 
 Ws V Vo ^ ^ ', iV-A. - ' 
 
 V wi . A'Av,' -\ 
 
 . 
 
 v z<'\^ *\VV»$V\ 
 
 
 v • -V ’•- <l.v; . $ ' • 
 
 
 • ,A<\ , . \j\V. 3 v\ 
 
 
 Av,'\^i <svr\ 
 
 
 
 
 
 
 wys vm ^vv\ c m p\i ,i '.u- 
 
 7' , i 3 . ' : -7 7.7 ■ 7VFv7 t. 
 
 
 
 
-5- 
 
 III. CALIBRATION OR THE SPRING. 
 
 The spring was calibrated by applying to it successively 
 a series of known weights, and at each time after the spring came 
 to rest, the amount of stretch between the points A and B on the 
 spring was read, a,nd the amount of elongation by one gram additional 
 weight was obtained by averaging all the readings. 
 
 Table I shows a calibration of the spring used. 
 
 Table I 
 
 Weight 
 on spring 
 
 i 
 
 10 gm. 
 
 10.5 
 
 11 
 
 11.5 
 
 12 
 
 12.5 
 
 13. 
 
 Upper 
 
 reading 
 
 104.082 
 
 104.082 
 
 104.082 
 
 104.082 
 
 104.082 
 
 104. 082 
 
 104.082 
 
 Lower 
 
 reading 
 
 80.764 
 
 80.320 
 
 79.880 
 
 79.435 
 
 78.986 
 
 75.543 
 
 77.095 
 
 Total 
 
 Stretch 
 
 25.318 
 
 23.762 
 
 24.202 
 
 24.647 
 
 25.096 
 
 25.537 
 
 25.987 
 
 Elonga- 
 
 tion 
 
 
 0.444 
 
 0.440 
 
 0.445 
 
 0 .449 
 
 0.441 
 
 0.437 
 
 The wt . of the pan (0.5956 gm.) used for holding the weights 
 was not included on the table because what we wanted was the elonga- 
 tion of one gram weight. The weight of the pan w as measured 
 because it had to be included for the calculation of the check 
 weight. 
 
 Nov/ immediately after finishing each determination, the 
 weight of the platinum ball in the molten glass was calculated by 
 means of this table and a metal button of any kind having its 
 weight equal to the calculated value wss weighed out. Without 
 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
- 6 - 
 
 letting the spring go back, the platinum ball with its suspension 
 wire was removed from the spring and the check weight put on it. 
 After the spring came to rest, the reading of the stretch between 
 the points A and B on the spring was again made. 
 
 By comparing this amount of stretch with that of the actual run, 
 the difference was interpolated and added or subtracted from the 
 known weight, as the case may be. In this way any source of 
 
 error euch as that due to fatigue of the spring, or to change of 
 temperature of the spring tending to alter the amount of stretch 
 from its normal value, was eliminated 
 

 
 
 
 
 
 
 
 
 
 . 
 
 < 
 
 < 
 
 . 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
-7- 
 
 IV. CALCULATIONS 
 
 Tne volume of the platinum ball was obtained by Archimede' s 
 water displacement method. If the water in which the platinum 
 ball was weighed, were at 4°C, then loss of weight in water would 
 give us its volume directly. But if it were not at 4°C, the loss 
 was divided by the density of water at the corresponding temperature. 
 Necessary corrections were applied to the calculation. The volume 
 of the platinum ball at high temperatures was obtained by 
 applying the following equation: 
 
 V » ( 1 + at) V 0 
 
 where V = the volume at temperature = t 
 
 t 
 
 V = Volume of platinum ball at 0° 
 
 o 
 
 CX = the cubical coefficient of expansion 
 The volume of the platinum ball calculated from the above equation 
 at temperatures 1200°, 1300° and 1400°C was 0.6056, 0.6064, and 
 0.6030 cc, respectively. After securing the readings at various 
 temperatures, the weights corresponding to the stretches were 
 calculated and a check weight of metal having the same weight was 
 applied to the spring. from this we obtained the true weight 
 of the platinum ball in glass with one more correction, i.e. the 
 Surface tension effect of the molten glass, which was exerted on 
 the suspension wire and helped to pull down the spring. The sur- 
 face tension of the glasses was determined by Mr .E.E.Libman in 
 his doctorate thesis, and it varies from 140 to 170 dynes per 
 
 Since this was not a big amount the average value 
 
 centimeter. 
 
- 8 - 
 
 of 155 dynes per centimeter was used in this case. The following 
 
 formula was used in calculating the results: 
 
 Density = Wt . in air -(Wt.in gl ass-Surf ace tension corr.) 
 
 Volume of tn*e pt / FalV ’ 
 
 The density of thirteen samples of glass having the following- 
 composition was determined. 
 
 TABLE II. Composition of the Glasses 
 
 No. 
 
 SiO 
 
 2 
 
 Na 0 
 2 
 
 CaO 
 
 2 
 
 82.3 
 
 17.7 
 
 
 3 
 
 70.0 
 
 30.0 
 
 
 4 
 
 60 • 0 
 
 40.0 
 
 
 5 
 
 62.7 
 
 14.7 
 
 22.6 
 
 6 
 
 60.5 
 
 20.0 
 
 19.5 
 
 7 
 
 60 
 
 30 
 
 10 
 
 8 
 
 70 
 
 20 
 
 10 
 
 9 
 
 52.5 
 
 40.0 
 
 7.5 
 
 10 
 
 70.0 
 
 10 
 
 20 
 
 12 
 
 72 
 
 14.5 
 
 13.5 
 
 13 
 
 73.5 
 
 16.5 
 
 10.0 
 
 15 
 
 67.5 
 
 15.5 
 
 17.0 
 
 16 
 
 65 
 
 20 
 
 15.0 
 
The results from the experiments are tabulated as follows: 
 
-10 
 
 In order to make the results easily understood, two other 
 methods as given below were used, (a) by the temperature density 
 curves and (b) by the composition density model. 
 
 (a) The curves were all drawn to the same scale so as to 
 give a better comparison. One extra curve was drawn for glass 
 No.16. In this case a duplicate was run and more frequent 
 density readirgs were taken at smaller temperature intervals. 
 
 This curve shows a maximum density at 1380°C, 
 
 (b) The composition density model was built for the density 
 
 of the different glasses at 1420°C. The composition of the 
 
 glass is represented on the triangular coordinate paper. The 
 vertical distance of the model at any point represents the density 
 at 1420° of the glass, having its composition represented by the 
 point. Each centimeter vertical distance of the model repre- 
 
 sents an amount of 0.1 for the density. As shown on the photo- 
 graph attached below, the first white horizontal line represents 
 
 a density of 2.15; the second 2.2, etc., The base of the model 
 represents a density of two. 
 
200 400 600 800 /OOO / 200 /40O 
 
o(H>/ oo?/ 000/ 00‘S 009 00^/ __ oo? 
 
. 
 
200 ~ 400 goo goo /ooo /200 /4oo 
 
1200 1250 BOO /J50 !*oo /450 
 

-11 - 
 
 VI. DISCUSSION. 
 
 Accuracy of the Res ult.- The results do not show much 
 uniformity in the amount of expansion from room temperature to 1400°C 
 nor do they show any indication of the relation "between the 
 composition of the glass and the amount of expansion. The dupli- 
 cates run f or a few samples only check to the second place after 
 the decimal. small errors are unavoidable in making readings 
 
 of temperature and elongation. Dissolved gases in glass have 
 considerable effect upon the experiment. This difficulty, however, 
 could be overcome by putting the molten glass under vacuum so as to 
 lift the gas out of the mass. But another source of gas buble 
 may arise from the decomposition or volatilization of certain 
 components of the glass. This is shown by an experiment with 
 
 lead flint glass. The density of the lead flint glass was 
 
 determined. When the temperature reached 1300 °C , gas bubbles 
 began to appear, and as the temperature went up more gas bubbles 
 appeared. At the same time the platinum ball was raised up 
 
 considerably. This probably was due to the gas bubbles adhering 
 to the ball and tending to rise up to the surface. As the 
 
 temperature coded down again, the ga.s bubbles disappeared gradually 
 until at 1200°C no more bubbles were visible. This phenomenon 
 perhaps was due to the vaporization of arsenic compounds at high 
 temperature. Difficulties of this kind may not be a-le to be 
 overcome without changing the composition of the glass. Still 
 another source of gas bubbles may be due to the excluded gases 
 
- 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 . 
 
 
 
- 12 - 
 
 in platinum escaping at high temperature. In high silica glasses 
 the melting range is so high that the glass is still very viscous 
 at 1200°C or 1300°C* and it is no doubt difficult to obtain reliable 
 measurements under these circumstances. For these reasons it is 
 not expected that the results are very accurate. Since the 
 duplicates of a few samples check to the second place after the 
 decimal, the third place after the decimal shown in the table is not 
 of much value. 
 
 Glasses No. 4 and No. 16 show a peculiar phenomenon, i.e., 
 as the temperature goes up the density goes up also. A duplicate 
 was run on glass No. 16 and the readings check fairly well. 
 
 The temperature density curve of glass No. 16 shows a maximum at 
 1380®C. This may be due to certain errors, but since the two 
 runs check it is not likely that any errors made are merely coin- 
 cident. Whether or not glasses do have this property needs 
 
 further investigation. 
 
 Jaeger in Holland used a balance suspended over the fur- 
 nace in determining the density of molten salts. The platinum ball 
 was suspended f r an the bottom of the pan, the other arrangements 
 being similar to those described above. This method may take 
 longer time to get equilibrium. The spring suspension method has 
 the advantage of simplicity and ease of manipulation. The sources 
 of error due to any effect tending to alter the amount of elongation 
 from the normal value can be eliminated by immediate checking after 
 each run. Tne disadvantage of this method is that if the two 
 telescopes are not set quite parallel, e. small angle will maxe 
 considerable error in the result. 
 

 
 
 
 
 
 
 
 
 
 
 
13 - 
 
 Suggestions ; - Evacuate every glass before each experi- 
 ment. Keep the molten glass at a certain temperature long enough 
 in order to secure equilibrium before a reading is taken. 
 
 Run the temperature up and down through the range at least twice in 
 order to see if the readings check. 
 
 Lastly take good care in leveling the two telescopes so that they 
 are exactly parallel. In this way readings checking to the third 
 place after the decimal might be obtained. 
 

 
 
 . 
 
 
 
 
 
 
 
- 14 - 
 
 VII. SUMMARY 
 
 (a) The range of expansion of different soda-lime glass 
 from room temperature to 1400°C is from 2.49$ to 14.8$. 
 
 (b) The errors in the experiment are mostly due to the 
 behavior of the glass at high temperature and can be removed by 
 evacuating the glass and by prolonged heating. 
 
 (c) Whether or not the molten density has a maximum point 
 of density needs further investigation.