T H E ADHESION OF GLUE BY MAYNARD F. WITHERELL
The object of this work was to determine the factors influencing the adhesion between glue and a high-grade alundum, and to bring about the best possible adhesion between the two. The alundum used was a product of the Xorton Company, Worcester, Mass., and is an electrically fused grain, white or nearly so in color, and designated as ii99” white alundum, for polishing. The size of the grit is 36-mesh. The glue was a 145 millipoise hide glue, “100 A,” with a moisture content of 1 4 . 2 3 7 after ~ ~ 24 hours drying at IOSOC. The method of measuring the adhesion between the glue and grit is that given by C. E. Lanyon’ who was determining the adhesive strength of glue, using a standard grit. In this case a single glue is used throughout and the grit is varied. The details of the method are as follows: 23.55 gr. of the glue are placed in a I j o cc. beaker and 7 6 . 5 cc. of distilled water are added. This is allowed to soak for 30 minutes, then it is placed in boiling water and warmed for I O minutes. About three minutes before the glue is ready to remove from the hot water, 400 grs. of the grit is weighed into a small pan. Then 85 grs. of the 20% glue solution are poured onto the grit and rapidly and thoroughly mixed in with an 8 inch porcelain spatula. This glue and grit is then packed into the standard cement briquet moulds which have been slightly altered for this use. (The ordinary mould has a crosssection of one square inch and it was found that six or eight weeks drying was necessary to bring the briquet to equilibrium with the atmosphere. To shorten this time, a section 114 in. X I in. X I in. is affixed to the mould so as to put a hole of the given dimensions in the middle of the briquet. Thus the thickness a t the breaking point of the briquet is 3 , 8 of an inch. The total area there is 3 / 4 sq. inch). These briquets are found to have reached a maximum strength after I I or 1 2 days, or slightly longer depending OD the weather. Unless otherwise stated, the briquets have dried 14 days, a period which insures their maximum strength having been attained. The dried briquets are then broken in a RiehlC cement testing machine. Six briquets are made for each run, three briquets being made from each batch of grit and glue. The average breaking strength of the briquets is the socalled “briquet strength” which is taken as a measure of the adhesion between the grit and the glue. After testing a series, it is common to find one of them has tested 1570 or more under the average. It is very difficult to make a perfect set of briquets, because of the rapidity at which it has to be done to prevent setting before the grit is packed into the moulds. This results in imperfect distriInd. Eng. Chern., 19, 1191 (1927).
3584
MAYNARD F. WITHERELL
bution of the glue and an occasionally very poor briquet. In such a case, I believe it is perfectly justifiable to omit that result, and this has been done throughout the work. T h a t it is a fault of uneven distribution is plainly visible on close examination of the broken briquet, and the use of that value is merely leading t o a false conclusion. I t cannot be argued that weakness in that briquet would be made up for in strength of the others. Theoretically there would be an added glue content t o the others but it would be negligible, whereas one poor briquet would lower the average as much as 30 in many cases. The majority of the briquets tested lie within 5% of the average, although there are some that’ do not. When one considers the variations in mixing the glue, packing the moulds, and the actual breaking test, this variation seems allowable. There is a chance for error, due to drying a t different humidities, for the remaining moisture content will doubtlessly affect the strength of the glue. Inasmuch as the glue is much stronger than the bond between glue and grit, the strength of the glue is not the determining factor. It does not appear likely that the bond would be more than very slightly affected by a slight difference in moisture content. The average drying taking place over a two weeks period is quite uniform and the change taking place the last few days is small. It is true that the glue changes quite rapidly with humidit’y conditions, but in a briquet this change would involve only the surface. For best possible results the drying should be done a t a definite temperature and humidity, but for this work it was deemed sufficiently accurate to hang the briquets in a hood where there is continuous changing of air a t room temperatures. It is not advisable to lay the briquets on anything to dry, for uneven drying is thus obtained and the briquets do not dry completely or without warping. It was early noticed that on adding the water to the glue, some of the glue took up the water rapidly and swelled up in the beaker but did not form a jelly or homogeneous mass. Again the glue took up the water slowly and finally formed a jelly in the beaker. This appeared to be caused by varying sizes of the glue flakes. As it is taken from the bag, it is alternately fine and coarse throughout the mass. I t was feared that this non-uniformity would cause a difference in the tests so it was checked up as follows: X quantity of the glue was passed through a Io-mesh and a 3 j-mesh O sieve. Roughly 1:’s is retained by the coarse sieve, and about I ~ I passes the 3 j-mesh, the rest is retained by the 35-mesh. Six briquets were made from each grade, using a 36-mesh T J-treated Xorton sharp-fracture alundum. These briquets tested as follows: Finest glue 864.7 briquet strength Medium grade 874.0 ” >, Coarse grade 859.0 ” These results show that in spite of the variations in size the glues had adsorbed the water completely by the time that they w’ere mixed with the grit The and briquets made from the different grades checked within j, 1%. different grades showed the above-mentioned characteristics when swelling.
THE ADHESION O F GLUE
3585
Another purpose of this work is to determine the relation if there be any between the “capillarity” of a grit, as determined by the Yorton method, and the attraction between that grit and glue. The method mentioned is carried out thus: there is a rack which holds in an upright position a number of glass tubes, 2 5 cms. long with an internal diameter of 5-6 mms., the bottom of the tubes rests on a wire screen in a shallow well in which there is a I cm. high baffle, behind the tubes is a scale which is graduated from the top of the baffle up. Before use, the tubes are cleaned in cleaning solution and rinsed in distilled water aAd dried. The tubes are then set up in the rack and the grit whose capillarity is desired is poured in from the top without shaking in any way. Then water betiveen i o o and 80’ F. is poured into the well until it overflows the baffle; then start a stop-watch. At the end of the specified time interval for each grit number, reading is made of the height of the water rise directly on the scale behind the tubes. This height in centimeters, expressed as a percentage of the height the water would reach if the grain were of maximum capillarity, constitutes the Capillarity of the grain. For the 36 grit, the time period is 3!j minutes: and the maximum capillarity is I O cms. Capillarity attraction of abrasive grain is that property of attracting watcr so that the surfaces of the grain particles are thoroughly wetted. This is very essential to all kinds of abrasive grains that are to be used in conjunction with adhesives. Foreign matter on the grain surface that is repellant to water would result in very poor bonding between the grain particles and the adhesive; therefore the grains must be so prepared that the adhesive mill not only spread easily on the surface of the grain particles but also adhere thoroughly after the bond has set. There is however a doubt as to the worth of this means of determining whether the glue will adhere to the grit or not. Its theory as far as it goes is right, but it goes only as far as water to grain. K e have no proof that a grain which is readily wetted by water is also readily werted by glue. Throughout this work close check will be kept between the two to decide whether capillarity is a direct measure of adhesiveness or not. A few preliminary tests were made with a sharp-fracture 36 T J Korton alundum. The quenching treatments were carried out as follows: A muffle furnace was brought up to dull red heat and the grain was put into the furnace in large porcelain crucibles, each holding about 500 grs. of grain. At the end of the time interval, the crucibles mere taken out and the glowing grains poured into a liter beaker containing the solution, as fast as was possible to prevent’ excess cooling. The grit was then throughly washed free from the solute and drained and dried by centrifugal action, then put into a drying oven at I O O ~ - - I I O ~The C . boiling treatments were carried out in large evaporators, heated with a bunsen flame. Timing was started when the solution first boiled. Any boiled off water was replaced. Drying is the same as above.
YATNARD F. WITHERELL
3586
The treatments were: I . Heat in muffle furnace to dull red heat r hour. Quench in 10% sodium silicate. Wash and dry. Heat I hour in muffle furnace a t dull red heat. Quench in a 3-1 2. mixture of IO% sodium silicate and sodium hydroxide. 3. Heat I hour in muffle furnace at dull red heat. Quench in 1.7~ sodium silicate. Wash. Boil 1 5 minutes in 10% sodium hydroxide. 4. Boil 15 minutes in 10% sodium hydrbxide. Wash. Boil I j minutes in 10% sodium chromate. 5 . Heated I hour in muffle furnace. Quenched in IO% sodium hydroxide. Washed. Boiled 1 5 minutes in 10% sodium chromate. Kashed. Dried. Heated to dull red in muffle furnace. 6. Heated 3 hours in muffle furnace at dull red heat. Quenched in 10% sodium hydroxide. Washed. Boiled I j minutes in 10% sodium chromate. 7. Heated in muffle furnace I hour. Quenched in a 5% solution of glue. 8. Grit untreated. 4 drops of glycerol added to the glue solution. 9. Grit untreated. As soon as glue was mixed in, the pan was covered air tight and the whole was heated for five minutes in boiling water. Then the moulds were filled. IO. Control. TABLE 1
KO.
Ave. Briquet
Capillarity
Yo.
54%
6.
865
58%
63%
i,
6 2j
25%
3.
642 860 823
62%
8.
4.
755
42%
9.
9’9 725
5.
808
60%
Strength
I. 2.
IO.
Ave Briquet Strength
723
Capillarity
__ __ 59%
These results are given in Table I and Fig. I . From this we see at a glance that the same capillarities do not mean the same adhesion between the grain and the glue. The silicate-treated grains do check among themselves, but not with the chromate-treated. The chromate-treated do not check with any. Inasmuch as the T J is a treated grit, any additional treatment cannot be considered as a means of improving an untreated grain. I t is thought that the T J is a silicate treatment but nothing beyond that can be even guessed at. It is apparent that an alkali-silicate treatment improves it, but so does the alkali-chromate treatment. Efforts to bring the glue itself into closer contact with the grain, without using other means to clean the grain or change its surface, have little or no effect in increasing the adhesion. After the preliminary tests on the T J alundum, similar treatments were tried on the untreated (‘99”alundum. Due to the different fracture of this grit, the packing is different and 8; grams of glue solution with 400 grams of
3587
THE ADHESION O F GLUE
the grit, gives us a nearly fluid mass. Since the object is not t o fill in the air spaces, but merely to coat the grain surfaces with glue, the amount of 2 0 % glue solution was cut down to 80 grams. The results of the first series are given in Table 11. The details will be found in Table V.
TABLE I1 KO
Capillarity Briquet Strength
lh 2
27%
859 5 620 o 504 3
29%
4=7 5
53% 3 6%
A
3h 4A
No
SA. 6-4.
7x.
Capillarity Bnquet Strength 50% 806 o 40% 544 0 56% 797 5
FIG.I Briquet Strength-Capillarity
One conclusion can immediately be drawn from these results: chromate treatments do not compare with silicate treatments. Sos. zA,3 4 qh,and j A are all chromate treatments, the others are hydroxide-silicate treatments. The next series was that in which calcium hydroxide and calcium sulphate were used. Boiling as well as quenching was tried and will be more in the future to determine whether the same results can be obtained at a lower temperature. SA, 8B, 9A, and 9B are the numbers of these treatments, details also being in Table Y. The results are given in Table 111.
TABLE I11 NO.
Capillarity
8A 8B
Not det'd.
9A
9B
1,
9,
,, ,,
,,
1)
Briquet Strength 666 6
153
0
659.0 566.0
These treatments are but a trifle better than the preceding chromate treatments. It is apparent that the strong alkali and silicate treatments give the grain the surface which is best held by glue. It is then desirable that we know whether the results vary with different sodium hydroxide/sodium silicate ratios. And so a series of treatments by quenching, with varying silicatehydroxide ratios, was made. The tabulated and graphical results are on follow-
3588
MAYNARD F. WITHERELL
ing pages. To complete the data, quenching was also done in the IO% solutions of each constituent alone. The best adhesion is obtained with the grit which was quenched in sodium silicate alone. That quenched in 5-1 mixture of sodium silicate and sodium hydroxide runs a close second. The grit treated with sodium hydroxide alone is very poor, the test increasing in general as the percent of sodium silicate is increased. Treatments I I A and 1 2 i l are the same except that the former was heated 1 2 hours in the mume furnace, the latter but one hour. Results are: Capillarity 61% 48%
KO. I IA
1zA
Briquet Strength 840
Hours Treated I2 I
721
The difference here may be due to the long heating and it may be that R higher temperature was reached and the grit was hotter when quenched. It is either way a matter of more heat giving higher adhesive properties. z z h treatment is quenching in a mixture of sodium hydroxide and tannic acid. The capillarity is but I;% and the briquet strength j33. If there is a chemical reaction between the alundum and the alkali, we should get better results using a more concentrated solution. The following results bear out this theory: Capillarity
NO.
Briquet Strength
65% 952 2 7B 60% 940 The question arises, "Wll long boiling give the same results as quenching?" The answer is in Table IT. ZOX
TABLL 1s' Briquet Strength
Capillarity 15'; 58'c
929.5
1038.3
942.0 8ji.o
495 ; ' 0 j
The last series of tests include more boiling treatments which shorn up as well AS, and better than, quenching, Table V. I
I
0
IO
20
30
40
50
60
70
80
90 100 OXNO53
FIG.2 Variations in SaOH-Ka Silicate Ratio
THE ADHESION OF GLUE
3589
TABLE V Treatments with 99 Alundum 1.1.
2-4.
3d. 4.4.
SA. 6.1, ;A. 8-4.
9A.
8B. 9B. I IA.
12.4.
13-4. I4.A.
I5A.
16.4. I
;A.
18-4. I 9.4.
20-4.
22A.
Heated in muffle furnace one hour. Quenched in a 3-1 mixture of 107~ sodium silicate and sodium hydroxide. Heated one hour in muffle furnace. Quenched in 10% sodium hydroxide. Washed. Boiled 15 minutes in 10% sodium chromate. Heated three hours in muffle furnace. Quenched in 10% sodium hydroxide. Washed. Boiled 15 minutes in 10% sodium chromate. Heated three hours in muffle furnace. Quenched in IO%, sodium hydroxide. Washed. Boiled I 5 minutes in “cleaning solution.” Heated three hours in muffle furnace. Quenched in a 3-1 mixture of IO%, sodium silicate and sodium hydroxide. (Dried at 140). Heated one hour in muffle furnace. Quenched in a 3-1 mixture of 10% sodium hydroxide and sodium chromate. Heated one hour in muffle furnace. Quenched in a 1-3 mixture of 10% sodium silicate and sodium hydroxide. (Dried at 140). Heated one hour in muffle furnace. Quenched in a 10% solution of calcium hydroxide. Heated one hour in muffle furnace. Quenched in a 1-1 mixture of IO? calcium hydroxide and calcium sulphate. Boiled I j minutes in loYc calcium hydroxide. Boiled 15 minutes in a 1-1 mixture of 10% calcium hydroxide and calcium sulphate. Heated twelve hours in muffle furnace. Quenched in a 33-2 mixture of 107‘sodium silicate and sodium hydroxide. Heated one hour in muffle furnace. Quenched in a 31-2 mixture of 10% sodium silicate and sodium hydroxide. Heated one hour in muffle furnace. Quenched in a 5-1 mixture of 1 0 7 ~ sodium silicate and sodium hydroxide. Heated one hour in muffle furnace. Quenched in a 10-1 mixture of rocc sodium silicate and sodium hydroxide. Heated one hour in muffle furnace. Quenched in a 20-1 mixture of 10% sodium silicate and sodium hydroxide. Heated one hour in muffle furnace. Quenched in a 2-1 mixture of 107~ sodium silicate and sodium hydroxide. Heated one hour in muffle furnace. Quenched in a 1-1 mixture of 107~ sodium silicate and sodium hydroxide. Heated one hour in muffle furnace. Quenched in a 1-6 mixture of 10% sodium silicate and sodium hydroxide. Heated one hour in muffle furnace. Quenched in a 1 - 2 0 mixture of IO(#; sodium silicate and sodium hydroxide. Heated one hour in muffle furnace. Quenched in 20% sodium hydroxide. Washed. Boiled in 10% sodium silicate I; minutes. Heated one hour in muffle furnace. Quenched in a 4-1 mixture of 10% sodium hydroxide and tannic acid.
3590
MAYh’ARD F. WITHERELL
TABLE I’ (Continued) Treatments with 99 Alundum 23-4. Heated one hour in muffle furnace. Quenched in 107, sodium hydroxide. 2j X . Heated one hour in muffle furnace. Quenched in 107, sodium silicate. 27B. Heated one hour in muffle furnace. Quenched in 207~ sodium silicate. 30A. Boiled four hours jn 107~ sodium silicate. 31.i. Boiled four hours in a j - 1 mixture of ioYc sodium silicate and sodium hydroxide. 32h. Boiled four hours in 2 0 ~ 7sodium hydroxide, drained and boiled I j minutes in 1oYc sodium silicate. 32B. Boiled four hours in zoVC sodium hydroxide, drained and boiled 45 minutes in 107, sodium silicate. A11 these treatments were quenchings in mixtures of IO^^ sodium hydroxide and 107, sodium silicate. The ‘‘% S a O H ” is that percent of the total solution which was IoC;, KaOH solution. The results are shown in Table V I and Fig. 2 .
TABLE VI % KaOH
No. of treatment
Capillarity
Briquet Strength
7h 1 SA I4 .4 I 3-4
54%
960 lbs.
92%
849
40% 65%
776 992
5% 9% 16.
859;
2.5%
782
33.3%
IjX
53% 59% 60
I8X
34
I9
h
385%
873 6.53 7%
50% 85 77 957c
23A
337
626
2
IIA
16A
0%
js
’
1007~
Bookbinders have for many years added glycerine to glue to make it more flexible. This suggested the addition of small amounts of glycerine to the glue used in making briquets to determine whether it affects the strength or the keeping qualities of the glue. To determine the best’ amount to use, a series was made using 2 , 4, and 6 drops of glycerol to each 80 grams of glue solution. Six briquets were made of each. The results were as follows: z
4 6
drops glycerol-Average ”
,,
”
Control, no glycerol-”
,,
briquet strength 5803 lbs.
,>
”
568 573
”
”
JI I
”
”
19
,9
”
Untreated “99” alundum was used in all cases. I t apparently makes little difference which amount is used. happy medium was taken and 4 drops used for the next runs.
Hence a
T H E ADHESION O F GLUE
3591
A series of 18 briquets were made using 1-4grit. Some of the briquets were tested after two weeks and others were set aside to be tested later. The results on these tests were as follows: Tested after 14 days-Average btrength 8473 Tested after 63 days-Average strength 8453 Control, no glycerol-Average strength 8594 When used on the untreated grit, the glycerol causes a 36% to 42% increase in the breaking strength. This cannot be due to an increase in strength of the glue, for it is known that the glue will readily stand over a thousand in briquet tests. Hence we conclude that it increased the adhesion between the untreated grit and the glue. Before the glue can be adsorbed by the grit, the air film around the grit must be removed, or replaced. The glycerol probably replaces this film better than the straight glue solution and brings the glue in better contact with the grit thus increasing the adhesion between them. With the silicate treated grit there is no increase in adhesion between grit and glue; that is, the surface of the treated grit is of such a nature that it is wetted as easilyby the glue alone as it is by the glycerol. There is a possibility that the adhesion is greater, but that its affect is destroyed by a weakened glue but this seems improbable. Two months time did not affect the glyceroltreated glue so it seems probable that there is no ill effect. Other briquets of this series are to be given six months before testing. Treatment S o . 8 on page 3 j89 shows the effect of glycerol when working with T J treated grit, an increase in strength of nearly 37%. The surface of this grain is more readily wetted by the glue and glycerol than by the glue alone. The following series of runs are made in duplicate and are known as series “E” and “H.” The treatments are chosen from those that previously gave best results and are duplicates of each other. There is some question as to the duplicatability of the results obtained. To decide this, these treatments are made separately, one series being done at a time. The briquets are made up likewise and their drying covers a different period, but one of the same length. In order to cut down the variations liable to arise in the quenching procedure as heretofore carried out, the method was changed as follows: the grain was fed from a large funnel through a small orifice into an inclined iron pipe, with an eccentric under one end to insure the grains against stopping in the pipe. The pipe was heated to red heat by three blast lamps, and the hot grit passing out of the pipe drops directly into the quenching bath. The advantages of this method are believed to be more uniformity of heating, and a great saving of time in treating the grit. The temperatures reached are not as high however so the results of the tests are not as high as had been obtained. Since the chief aim is comparison, this does not matter.
3592
MAYNARD F. WITHERELL
These rules were adhered to strictly in making the briquets: I. 2.
3. 4.
5. 6. j.
8. 9. IO.
A11 treatments started and finished the same day. N o grit left in drying oven over three hours. Drying temperature not over I I O O C . All grits thoroughly washed in tap-water. Glue shall soak 2 j-30 minutes before placing in the boiling water. Water shall be boiling when glue is placed in it. Glue shall be heated exactly IO minutes in boiling water. Briquets shall be removed from mould as soon as set. Briquets shall dry in the rack 14 days. Six briquets shall be made of each grit.
Details of the treatments are: Boil 4 hours in roy0 sodium hydroxide. Wash. Boil 30 minutes in sodium silicate. 2 . Boil 4 hours in 107: sodium hydroxide. Vash. Boil 60 minutes in ioYc sodium silicate. 3 . Boil 4 hours in 2or; sodium hydroxide. Wash. Boil 30 minutes in 20'3~ sodium silicate. 4. Boil 4 hours in 2 o C c sodium hydroxide. Wash. Boil 60 minutes in zoyc sodium silicate. j. Boil hours in 10~1; sodium hydroxide. Kash. Boil 60 minutes in 2 0 % sodium silicate. 6. Quench in IO'; sodium hydroxide. Kash. Roil I j minutes in IO:.; sodium silicate. 7 . Quench in 1 0 ~ ; sodium hydroxide. M-nsh. Boil 60 minutes in IO%, sodium silicate. 8. Quench in 2 0 5 sodium hydroxide. Kash. Boii I j minutes in 1oC% sodiuin silicate. 9. Quench in 2oc; sodium hydroxide. \Tash. Boil 60 minutes in roc;: sodium silicate. IO. Quench in 2 0 ~ ' ; sodium hydroxide. IYnsh. Boil 15 minutes in 2 0 q , sodium silicate. I I . Quench in 2 0 5 ; sodium hydroxide. Kash. Boil 60 minutes in 2055 sodium silicate. 12. Quench in a 5-1 mixture of ioc'; sodium hydroxide and sodium silicate. 13. Quench in a 5-1 mixture of 2 0 5 sodium hydroxide and sodium silicate. The tabulated results for these series are given in Table VII; and the capillarity-briquet strength coniparison curves in Figs. 3 and 4. I.
THE ADHEBION O F GLUE
3593
FIG.3 Capillarity-Briquet Strength Series “E”
1
I
FIG.4 Capillarity-Briquet Strength Series “H”
TABLEVI1 Results of Series “E” and “H” Grit KO.
Capillarities “E” “H”
I
405
40%
2
58
53
3
51
4
io
5 6
54 45
53 57 40
7
46
8
35
9 IO
49 46
I1
60
I2
42
45 49
I3
55
61
55 41 59 46
Briquet Strength “E” “H”
926 lbs.
954 lbs.
I049 930 1046.6
910 98 I 820
876
997 726 822
800
7831 700
795; 7843 832.6 6733 7533
7561 80I 752
779 7 73 858
3594
MAYNARD F. WITHERELL
As has been stated, series “E” and “H” are primarily to show vhether the results obtained can be consistently duplicated or not. From this standpoint, results are not highly satisfactory. Good checks are obtained in seven cases, very poor checks in four cases. From this we do get some interesting information. The treatments are not the kind that give results regardless of how they are done, but on the contrary are very easily affected by imperfect control of all factors. Among the possible sources of error are variation in concentration of solutions, quenching temperatures, under or over-washing the grain, drying temperatures, and possibly presence of certain substances in the solutions which will effect the results obtained. These factors were all watched during the above runs: but plainly not closely enough. I t is known that imperfect washing will lead to a high capillarity and low briquet strength with some treatments, possibly so with all. Too much washing may decrease the effect of the treatment. I n comparing the relative effect of different treatments, I believe we should accept the highest value obt’ainedfor each. S o treatment can show up better than it actually is, but one slip somewhere might decrease its value considerably. Any result once obtained can be duplicated if the proper conditions are found and followed. The danger lies in underestimating the worth of any treatment, not in overestimating. In reviewing the treatments tried, we find that no high results are obtained except, from a silicate treatment, eit’her with or without sodium hydroxide. With sodium hydroxide, the ratio sodium hydroxide/sodium silicate must be kept small. I t makes little difference whether long boiling or quenching from red heat is the method used. Long boiling has the advantage of being less trouble, and the maximum results can be obtained more easily. Where both sodium hydroxide and sodium silicate are used, there seems to be little difference whether they are used separately or together; the single treatment is of course simpler. The concentration has little effect, in general the 20% solutions gave slightly better adhesion to the grit than did the IO^^, but this was not always true, nor was there much difference where there was any. The mechanism of this increased adhesion is merely a good point to theorize on until more is known concerning the surface of the treated grains. Sodium hydroxide will react slowly with aluminum oxide to form a sodium aluminate. This will in turn react with sodium silicate to form a complex sodium aluminum silicate. The treated grit shows no microscopic evidence of et,ching however, so it is not likely that there is formed a mechanical joint which strengthens the bond. This silicate film is either formed on the grain or is adsorbed by it. This film in turn has a specific attraction, if it may be called that, for the glue solution, is readily wetted by it and forms a very strong bond with it. Since the same result may be had without the use of the strong alkali, it seems quite likely that the whole matter is an adsorption of the silicate film on the surface of the treated grain. There is, as well, a cleansing action, and removal of any grease and air films which may be on the grit. Since these films are definitely replaced by another film,
THE ADHESION OF GLUE
3595
the effect of the treatment is long-lasting. If this grit is allowed to lie around in the open air for a long period of time, it will however lose its adhesive properties slowly, due to dirt and air gradually collecting on the treated surface. Merely heating the grain causes it to have a higher adhesive strength, due to cleansing of the grain. A heated or “burned” grit loses its adhesive properties quickly, since there is nothing to prevent it rapidly taking up air and dirt. The treatment does not give as high an adhesiveness as the silicate treatment, which supports the silicate film theory and its specific attraction for glue. The author takes this opportunity to thank the director of this research. The kindly suggestions and understanding of Professor Bancroft have made the work and the associations most enjoyable. Cornell Cniversitv.
