R . W . C A R L S O N AN D L . R . F O R R R IC II, M a s s a c h u s e t ts I n s t i t u t e o f T e c h n o lo g y , C a m b rid g e , M a s s.
M
EASUREM ENTS of heat of hydration of cement have been made extensively only during the past few years, beginning a t a time when little was known about either the characteristics of heat liberation or the thermal properties of the materials involved. I t is not surprising, then, th a t discrepancies between results from different calorimeters were encountered. The present paper is an attem pt to correlate Massachusetts Institute of Technology during the past three years. in the concrete if no heat were lost. A plot of the equivalent temperature rise against time is called an adiabatic time- temperature curve, which is of interest to engineers.
For the purpose of comparing cements, on the other hand, by the specific heat of the concrete. The specific heat varies with the type of aggregate and with temperature.
Four factors th a t affect the conversion of heat of hydration of cement into adiabatic tem perature rise of concrete are (1) the immediate heat liberation, (2) the variation with temperature of the specific heat of concrete, (3) the water- cement ratio, and (4) the curing temperature. Some of these factors were overlooked when heat of hydration studies first began to be made only a few years ago. Because they are of general interest, they will be discussed before the various calorimeters are described.
I m m e d ia t e H e a t L ib e r a tio n o f C e m e n t When cement is mixed (in a Dewar jar) with water of equal temperature, an appreciable temperature rise occurs almost immediately and continues a t diminishing rate for a half hour or more. This tem perature rise is believed to be due mainly to the solution of free oxides and impurities, and only a small am ount is believed to be due either to the hydra
tion of prim ary compounds or to the wetting of the cement grains. The am ount of immediate heat liberation, com
puted from tem perature rise of cement paste, is shown in Table I for a number of different cements. The am ount of heat liberated up to 30 minutes varies, for the cements tested, from 1.5 to 6.3 calories pfer gram. The average 30-minute heat liberation is about 5 per cent of the average potential
heat of hydration (not shown) for the cements listed in Table I.
The immediate heat liberation is included in the deter
m ination of heat of hydration by the heat-of-solution method, b u t is ap t to be missed in determinations by other "methods. tem perature changes within the range commonly encountered by concrete structures. Likewise, the specific heat of water is almost constant. B ut the specific heat of cement paste increases with tem perature to an extraordinary extent. At ordinary temperatures, the specific heat of the hydrated paste is lower than the sum of the heat capacities of the amounts of water and cement contained in a gram of the paste, as would be expected if the water and cement were chemically combined. In other words, hydration reduces the specific heat of cement paste a t ordinary temperatures.
Because of the increase with temperature, however, the spe
cific heat of hydrated paste may be greater than th a t of cor
responding unhydrated paste a t temperatures of about 65.56° C. (150° F.) and above. special adiabatic calorimeters designed and constructed for this express purpose.
JULY 15, 1938 ANALYTICAL EDITION 383 hydrated cement and th a t the aggregate is not responsible;
in other words, the variation is in the cement paste. In Table II, the observed specific heats of a particular concrete a t various temperatures are compared with corresponding computed values on the assumption th a t the concrete is composed of cement paste having the measured variation in specific heat, and of aggregate having a constant specific heat. The agreement is as good as the test data, and the discrepancy th a t exists is in the right direction to be explained by a small increase in specific heat of the aggregate. No measurements were made on the aggregate, which was siliceous, but aggregates in general exhibit a slight increase in specific heat with rising temperature. The specific heat of limestone, for example, shows an increase of 5 per cent due to a change in temperature from 0° to 100° C. temperature rise of concrete cannot be translated accurately into heat of hydration of cement, or vice versa, without taking the variations into account.
Before leaving the subject of specific heat, mention should be made of one phase th a t remains indefinite. The specific heat of fresh concrete is the weighted-average specific heat of the ingredients. After hydration, a considerably different value prevails. Considering only the cement paste, which
Interpolation between the known hydrated and unhydrated values, in accordance with the degree of hydration, seems to offer the safest means of ensuring a reasonable degree of accuracy where specific heat is involved.
W a t e r - C e m e n t R a t io
H eat of hydration of cement is usually determined on a neat-cement paste of relatively low water-cement ratio.
Because only a p art of the w ater in concrete was believed to combine with the cement, it was first thought th a t the dif
ference in water-cement ratio between neat paste and con
crete was not im portant. The heat of hydration increases
appreciably with increasing water-cement ratio, however, even in the range of ratios encountered in concrete. higher water-cement ratios were prepared by slowly rotating vials of the paste during setting to prevent separation of water. The results were not consistent with those for lower water-cement ratios, where test methods were straightfor
ward, so they were not included in the table. Auxiliary the rate of heat liberation of cement. The heat liberation at higher tem perature is generally greater a t early ages but tests reported by Hornibrook and associates (S).
T a b l e IV. E f f e c t o f C u r i n g T e m p e r a t u r e o n H e a t o f be duplicated readily in the laboratory. Thus, it is necessary usually to convert heats of hydration obtained under con the cement is first mixed with water, an appreciable amount of heat is liberated immediately, as described above. After set, b u t usually occurs later; final set is an arbitrary hardness th a t does not require as much hydration and heat liberation continued heat liberation accumulates to an appreciable amount over a period of days or weeks.
384 INDUSTRIAL AND ENGINEERING CHEMISTRY The most obvious method of determining the heat of hydra
tion of cement is to measure the tem perature rise of an in
sulated specimen in a room of constant temperature. B ut even after carefully sealing the specimen against moisture loss and making correction for heat losses, reliable results can be obtained by this method only up to about 3 days.
Later results are usually lacking in accuracy because the cor
rections are larger than the quantity being measured. Under typical conditions, one might find th a t a specimen of 181.44 kg. (400 pounds) weight, insulated all around with 15 cm.
(6 inches) of kapok, would reach a maximum tem perature a t about 3 days and thereafter would decline in temperature despite its heat generation. This example illustrates the difficulty of measuring heat of hydration over a long period of time and shows th a t the early method, consisting of measur
ing temperature rise of insulated concrete,, could not give satisfactory long-time results.
F i g u r e 1. A d i a b a t i c C a l o r i m e t e r i n W h i c h C o n c r e t e
Is S i m u l a t e d b y N e a t C e m e n t a n d W a t e r
A d i a b a t i c C a l o r i m e t e r . The adiabatic calorimeter avoids heat losses from a specimen of concrete by keeping the room exactly as warm as the specimen. In the most satisfactory type of adiabatic calorimeter, an automatic con
troller actuates heaters th a t maintain zero tem perature dif
ference between a thermometer in the specimen and another in the room. Temperatures are either registered by a sepa
rate recorder or are observed from time to time to provide the time-temperature curve.
An economical type of adiabatic calorimeter has recently been developed a t the Massachusetts Institute of Technology.
I t takes advantage of the fact th a t the variation in specific heat of concrete is almost solely in the cement paste, and employs a concentrated neat-cement specimen to replace the bulky concrete specimen usually employed. A cross-sec
tional drawing of such a calorimeter is shown in Figure 1. The neat-cement specimen, weighing about 1.36 kg. (3 pounds), is surrounded by a jacket containing a measured
am ount of water, sealed in metal, and thus the thermal prop
erties of concrete are simulated. A few inches from the specimen and its water jacket is a copper container th a t is maintained a t the same tem perature as the water jacket, so th a t no heat can escape from the specimen.
Advantages of adiabatic calorimeters in general are:
(1) the adiabatic time-temperature curve of concrete is ob
tained directly, (2) any type of cement can be tested, (3) the hydration tem perature simulates w hat would develop in a heavy mass, and (4) they provide the mass-curing condition for simultaneously testing other specimens for other prop
erties.
T a b l e V. E f f e c t o f C a r b o n a t i o n o f T e s t S a m p l e s o n H e a t o f H y d r a t i o n D e t e r m i n e d b y H e a t - o f - S o l u t i o n
M e t h o d
Ce W ater- H eat of H ydration
m ent Cem ent Corrected
N o. R atio Age Condition CO2 Observed for COj
D ays % Calories per gram
1 0 .4 0 60 Protected 0 .9 6 5 .1 5 9 .9
N o t protected 2 .2 75 .1 6 2 .3
2 0 .6 0 75 Protected 2 .0 106 .9 9 5 .3
N o t protected 4 .5 12 0 .2 94 .1
3 0 .4 0 90 Protected 0 .8 107.1 10 2 .5
N o t protected 3 .9 12 6 .3 103.7
Disadvantages of adiabatic calorimeters are: (1) they are expensive and require close tem perature control, (2) large specimens are generally required, and (3) they are not ac
curate for the first hour after mixing the concrete for the specimen.
H e a t - o f - S o l u t i o n C a l o r i m e t e r . Woods and his co
workers (6), realizing early th a t continuous observations on specimens for determining the heat of hydration would be exacting and expensive, applied the heat-of-solution method to the problem. In this method, it is necessary only to deter
mine the difference in heat of solution of corresponding sam
ples of cement a t two ages of hydration to have the am ount of heat liberated between those ages. If one of the ages is zero—in other words, if one sample is dry cement—and the age of the corresponding sample is 28 days, the difference in heat of solution of the two samples represents the total amount of heat evolved up to 28 days. A description of the heat-of- solution method in simplified form has been given by Lerch (5), and no detailed account need be given here.
Heats of hydration determined by the heat-of-solution method in most American laboratories have been too high to indicate the correct temperature rise of concrete. Also in England, tests reported by Lea (4) show the heat-of-solution method to give considerably higher values than the adiabatic method.
Sources of error not commonly considered in the heat-of- solution method are carbonation and drying of hydrated samples during preparation for testing. Of these two sources of error, carbonation is the one th a t makes heat-of-hydration results too high.
Carbonation of a cement sample before it is dissolved in acid reduces its heat of solution, because the heat of solution of calcium carbonate is less than th a t of calcium hydroxide.
In the particular acid solution employed, the heat of solution of calcium carbonate (ignited basis) was found to be only 102 calories per gram as compared with 557 for calcium hydroxide. As each per cent of carbon dioxide corresponds to 1.27 per cent of transformed calcium hydroxide (ignited basis), each per cent of carbon dioxide would be expected to cause an error of (5.57—1.02) X 1.27 or 5.8 calories per gram.
Each per cent of absorbed carbon dioxide would then be ex
pected to reduce the heat of solution of the hydrated sample by 5.8 calories per gram. The difference between values for dry and hydrated samples would then be greater, and the heat of hydration as determined by the heat-of-solution
JULY 15, 1938 ANALYTICAL EDITION 385 method would be too great by 5.8 calories per gram for each
per cent of carbon dioxide absorbed by a hydrated sample.
An investigation revealed th a t it was common for hydrated samples to absorb more than 0.5 per cent of carbon dioxide during grinding. D ry samples were relatively unaffected by carbonation. The computed effect of carbonation was checked by testing samples carbonated to different extents.
In one series of tests, one sample was protected so as to mini
mize carbonation and the corresponding sample was purposely carbonated more than usual during grinding. Results of such tests on three cements are shown in Table V. After applying the correction of 5.8 calories per gram for each per cent of carbon dioxide, the results are in fair agreement.
The accuracy of the calorimeter was about 2 calories per gram.
I t should not be concluded th a t the effect of carbonation is as simple as merely changing free calcium hydroxide to the carbonate. Actually, the carbonation seems to affect mainly the lime contained in the gel of the hydrated cement, and to affect bu t little the crystals of calcium hydroxide. Therefore, the carbonation involves another step, th a t of separating the lime from the gel, and this was not considered in deducing the correction value of 5.8 calories per gram for each per cent of carbon dioxide. I t is indicated th a t the separation of a small amount of lime from the gel requires little energy and th a t the correction value of 5.8 calories is therefore approximately correct where small amounts of carbon dioxide are involved.
Turning to the effect of drying of hydrated samples during preparation for test, an error in the opposite direction is encountered. Any extensive dlying of a hydrated sample would be expected to increase its heat of solution and hence the heat of hydration obtained by this method would be too small. If this fact is realized, samples can readily be prepared without drying to the point of introducing appreciable error.
Drying a t 50° C., for example, was found to increase the heat of solution by 3 calories per gram, while drying a t 110° C.
caused an increase of about 16 calories per gram. Ordinarily, drying to the equivalent of 50° C. is not encountered, although this amount of drying can be obtained a t room temperature when the humidity is low.
The factors to be borne in mind in determining equivalent tem perature rise of concrete from heat-of-solution results are as follows:
1. Heat of hydration values are often reported on ignited basis (making values too high).
2. Carbonation of hydrated samples may have occurred (making values too high).
3. Drying of hydrated samples may have occurred (making values too low).
4. Lower water-cement ratios are generally used for heat-of- solution specimens than for concrete (making values too low).
5. Immediate heat of hydration is included in heat of solu
tion.
6. Specific heat of concrete varies with temperature.
7. Temperature of curing test specimens is usually lower than that of concrete.
The advantages of the heat-of-solution calorimeter are th a t small neat-cement specimens may be used and th a t the speci
mens require no attention other than tem perature control between tests. Disadvantages are th a t a high degree of accuracy is necessary in the measurement of heats of solution to get fair accuracy in heat of hydration, and th a t some ce
ments, particularly Portland pozzuolana cements, do not dissolve quickly enough in acids to be tested by the heat-of- solution method. The seven factors listed above are not con
sidered to be disadvantages of the method, because they can be eliminated or corrected.
V a n e a n d C o n d u c t i o n C a l o r i m e t e r s . Reliable results cannot be obtained on highly insulated neat-cement
speci-Fi g u r e 2 . Co n d u c t i o n Ca l o r i m e t e r
mens because abnormally high temperatures develop th a t not only affect the heat of hydration but make even high insula
tion inadequate. The vane calorimeter (2) employs neat- cement specimens b u t avoids high temperatures by con
ducting the heat away through metal vanes almost as fast as it is liberated. The rate a t which the heat is conducted away is determined by measuring accurately the small temperature difference th a t develops between the specimen and the outer edge of the vanes. When a continuous record of the rate of heat removal from the specimen is obtained, the total amount of heat removed up to any age can be computed. Because the specimen varies so little in tem perature th a t only a small am ount of heat is stored in the specimen, the total amount of heat removed is the heat of hydration.
A modification of the vane calorimeter is the “conduction”
calorimeter, exactly the same in principle but with a metal tube replacing the vanes. Figure 2 presents a cross section of the conduction calorimeter, showing how heat is removed from the specimen by a tapered copper rod and how sub
stantially all heat is caused to flow in the direction of the metal tube by reason of a surrounding Dewar jar. Thermometers, not shown in the figure, are a t either end of the conducting tube. The advantages of the conduction over the vane calorimeter are (1) greater accuracy, (2) use of smaller specimens, and (3) a more faithful response to changes in rate of heat liberation.
The advantages of vane and conduction calorimeters as a type are: (1) early heat liberation can be studied in detail, (2) any cement can be tested, and (3) continuous results up to about 7 days can be obtained a t low cost. Disadvantages are th a t results are lacking in accuracy after about 7 days and th a t curing conditions are practically limited to a substan
tially constant temperature.
386 INDUSTRIAL AND ENGINEERING CHEMISTRY C o m p a r iso n o f R e s u lt s fr o m D iffe r e n t
C a lo r im e te r s
Comparable tests, employing the three types of calorimeters described above, indicated th a t almost identical results could be obtained from all three calorimeters when due regard was paid to possible sources of error. Results of heat-of-solution tests made on neat specimens cured on the time-temperature curve of corresponding concrete, when converted into tem
perature rise of concrete, closely represented the time-tem
perature curve obtained from an adiabatic calorimeter. Like
wise, results of heat-of-solution tests on neat specimens cured a t 21.11° C. (70° F.) checked very well with results on similar neat specimens tested in vane calorimeters a t 21.11 ° C.
(70° F.). Results were not in agreement, however, until proper account was taken of (1) immediate heat of hydration, (2) carbonation of heat-of-solution specimens, (3)
water-cement ratio, and (4) variations in specific heat of concrete with temperature. The effects of the other possible sources of error discussed above either were not involved in the com
parisons or were too small to be subject to proof of validity distillation of the steam-volatile monocarboxylic acid (6, 6, 7).
Estimations of pyrethrin I by these methods are usually lower than those obtained by difference after estim ation of total pyrethrins by the Gnadinger-Corl (1) reduction method and of pyrethrin I I by the Haller-Acree (4) methoxyl method.
T a b e e I. Loss o f C h r y s a n t h e m u m M o n o c a r b o x y l i c A c i d o n
Graham (2) has found th a t steam distillation of perfumed oil extracts to remove the essential oil (as directed in the Seil
A sample of pure chrysanthemum monocarboxylic acid (b. p., 140-2° a t 9 mm.) was subjected to repeated steam distillations— the acid after extraction from the distillate and
titration with standardized base was reacidified and steam- distilled again. All steam distillations were carried out as directed by Seil. After eleven steam distillations (Table I) 69 per cent of the original acid present was destroyed. The average loss for each distillation was over 1 0 per cent.
A series of samples containing various amounts of the chrysanthemum monocarboxylic acid was prepared both by direct weighing of the pure acid and by measuring off aliquots from a standardized alkaline solution of the acid. Each sample was steam-distilled and the acid reëstimated as in the
A series of samples containing various amounts of the chrysanthemum monocarboxylic acid was prepared both by direct weighing of the pure acid and by measuring off aliquots from a standardized alkaline solution of the acid. Each sample was steam-distilled and the acid reëstimated as in the