You are on page 1of 572

UNITED STATES DEPARTMENT- OF COMMERCE • Luther H. Hodges, Secretary

NATIONAL BUREAU OF STANDARDS • A. V. Astin, Director

Chemistry of Cement
Proceedings of the
Fourth International Symposium
Washington 1960

Volume II

Symposium held October 2-7, 1960, at the
National Bureau of Standards
Washington 25, D.C.

Proceedings published in two volumes*

National Bureau of Standards Monograph 43 — Volume II

Issued September 27, 1962

For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington 25, D.C.
•Prices: Volume I $5.75, Volume II $5.60. The two volumes at $11.25 a set.

irtiiJfiaf Bureau of Standards

&S0V^2 1962
/ / D O I h

Preface v^. -O

Symposia on the chemistry cements have become established through a succession of
of
More than a half century ago such meet-
international assemblies held at irregular intervals.
ings among researchers, covering various materials including cements, were sponsored by the
International Association for Testing Materials.
The first international congress dealing exclusively with the chemistry of cements and
related materials seems to have been a meeting sponsored by the Faraday Society and held
in London on January 14, 1918. Because of the priority of that meeting in this special field,
it has come to be known as the First International Symposium on the Chemistry of Cement.

The First Symposium was designated as "A General Discussion" on "The Setting of
Cements and Plasters." Ten brief papers were presented by 11 scientists from 5 countries.
Twelve discussions followed the presentation of the papers. The Proceedings consisted of 69
pages published in the Transactions of the Faraday Society, Volume XIV, 1918-19.
The Second Symposium was sponsored by the Royal Swedish Institute for Engineering
Research and the Swedish Cement Association. A 3-day meeting was held in Stockholm in
July of 1938 under the designation "Symposium on the Chemistry of Cements." Forty-seven
persons from 13 countries participated in the presentation of 13 papers and 52 discussions.
The Proceedings, consisting of 578 pages, were published in Stockholm in 1939 by the Royal
Swedish Institute for Engineering Research.
The intervention of World War II delayed consideration of the next symposium. Pre-
liminary plans for a general meeting in Washington were abandoned, but eventually arrange-
ments were made for the Third Symposium to be held in London under the sponsorship of the
Department and Industrial Research and the Cement and Concrete Association.
of Scientific
The five-day meeting was held in September of 1952, and was attended by 260 persons, 84 of
whom participated with 23 papers and 102 discussions. The Proceedings, consisting of 870
pages, were published in London in 1954 by the Cement and Concrete Association.
Preliminary conversations on the organization of the Fourth Symposium were initiated in
April 1956, at which time agreement was reached by the National Bureau of Standards and
the Portland Cement Association that the Symposium would be jointly sponsored by those
organizations. By the spring of 1959 a tentative program had been set up with the coopera-
tion of leaders in cement research of several foreign countries. Later in that year the organi-
zation of the working committees was completed. The sponsors and members of committees
are listed below.
Sponsors: National Bureau of Standards, Allen V. Astln, Director; Portland Cement Association, G. Donald
Kennedy, President.
Executive Committee: I. C. Schoonover, and A. Allan Bates, Cochairmen; R. H. Bogue, Secretary, Robert
E. Pflaumer, Douglas E. Parsons, Myron A. Swayze, W. S. Weaver, C. E. Wuerpel, and Hubert Woods.
Program Committee: T. C. Powers and R. L. Blaine, Cochairmen; H. F. McMurdie, W- C. Hansen,
W. J. McCoy, Fred Ordway, Kenneth E. Palmer, and Stephen Brunauer.
Arrangements Committee: J. H. Walker and Bruce Foster, Cochairmen; Clayton Davis and William Lerch.
Editorial Committee: E. T. Carlson and H. H. Steinour, Cochairmen; E. S. Newman and M. D. Catton.

Papers for presentation at the Symposium were received at the Secretary's office, beginning
in January 1960, and continuing to September. After review by the Program Committee
and some preliminary editing by the Editorial Committee, they were reproduced by a photo-
duplicating process, and copies were mailed, as they became available, to the Symposium
members.
The Fourth Symposium was held at Washington, D.C., October 2-7, 1960. Invitations
to membership in the Symposium were accepted by 362 persons, including 169 from the United

ii

States and 193 from 34 other countries. The number actually in attendance was 271. All
Symposium were held at the National Bureau of Standards.
sessions of the
At the first session, held on Monday morning, October 3, Dr. I. C. Schoonover, Associate
Director of the National Bureau of Standards, and Dr. A. Allan Bates, Vice President of the
Portland Cement Association, acted as cochairmen. Dr. A. T. McPherson, Associate Director
of the National Bureau of Standards, welcomed the guests on behalf of the Bureau, and Mr.
G. Donald Kennedy, President of the Portland Cement Association, indicated his pleasure
in cosponsoring this important gathering. Dr. Eberhard Spohn, President of the Heidelberg
Cement Company, responded on behalf of the members of the Symposium.
The first formal address, given by Dr. Wallace G. Brode, formerly Science Advisor to
the Secretary of State, was entitled "Some Problems Associated with the Growth of Science."
This was followed by an address, given by Dr. F. M. Lea, Director of the Budding Research
Station of the United Kingdom, entitled "Cement Research, Retrospect and Prospect."
Succeeding sessions were devoted to technical papers, grouped under seven general topics
related to the chemistry of cement.
The titles of the individual papers will not be given here, as they are listed in the table of
contents. The program included the following sessions.
Session II— Chemistry of Clinker; Monday afternoon, October Dr. Henri Lafuma and Dr. R. H. Bogue,
3;
presiding.
Session III — Chemistry ofHydration of Cement Compounds; Tuesday morning, October Dr. Fritz
4;
Keil and Mr. Charles E. Wuerpel, presiding.
Session IV— Chemistry of Hydration of Portland Cement; Tuesday afternoon, October Dr. Niko Stut-
4;
terheim and Mr. Myron A. Swayze, presiding.
Session V— Properties of Cement Paste and Concrete; Thursday morning, October Dr. Wolfgang
6;
Czernin and Mr. Hubert Woods, presiding.
Session VI — Destructive Processes Concrete; Thursday afternoon, October
in Dr. Arturo Rio and Mr.
6;
Douglas E. Parsons, presiding.
Session VII — Chemical Additions and Admixtures — was combined with Session VIII.
Session VIII — Special Cements; Friday morning, October Prof. B. G. Skramtaev and Dr. A. R. ColUns,
7;
presiding.

The technical communications were of three categories, as follows
1. Principal papers by outstanding contributors to the literature of cement and concrete
technology, who were invited to present their appraisal of the present state of
knowledge in their assigned topics
2. supplementary papers submitted by members of the Symposium, presenting new
and original research dealing with the specific topics of the Symposium, and
3. discussion of principal and supplementary papers, presented by members of the
Symposium.
Papers were presented in condensed form, but are given in full in these Proceedings.
Some of the discussions were submitted prior to the Symposium, others at the time of the
meetings or subsequent thereto.
After the Symposium, authors were invited to submit corrections to their papers, but
did not have opportunity to read printers' proof. Readers of the Proceedings are asked to
make allowance for this, as well as for the facts that some papers are the joint product of
authors and translators, and that translations had sometimes to be edited without access to
the original manuscripts.
Following the sessions in Washington, many of the members of the Symposium accepted
a general invitation to travel to Skokie, 111., to visit the laboratories of the Portland Cement
Association and to attend a series of seminars there. A two-day expense-paid bus trip from
Washington was arranged by the Portland Cement Associa-
to the vicinity of the laboratories
tion for foreign guests to enable them to see the countrj'side and to observe some typical
highway^systems, expressways, toll roads, and city streets in the United States.
The program at Skokie, October 10-12, included seminars on cement manufacture and
on basic and applied research, as well as laboratory tours and demonstrations.

iii

List of Symposium Members and Authors
(Names of authors of papers or discussions are indicated by asterisks. Daggers indicate persons who did not attend the
Symposium)

t *Aardt, J. H. P., Van t *Bozhenov, P. 1.
National Building Research In- South Africa Academy of Construction and USSR
stitute. Architecture,
Adams, A. Byron t *Braniski, A. C.
Dewey & Almy Chemical Divi- USA Polytechnic Institute of Bucha- Rumania
sion, W. R. Grace & Co.
*Ahlers, Guenter t *Bredsdorff, P.
University of California USA Danish National Institute of Denmark
t *Aitken, A. Building Research.
University of Aberdeen United Kingdom Bresler, Boris
Alexander, J. B. University of California USA
Southwestern Portland Cement USA Brink, Russell
Co. U.S. Bureau of Public Roads USA
Allen,Harold *Brode, Wallace R.
U.S. Bureau of Public Roads USA Chemical Advisor USA
Allen, Ivey, Jr. *Brown, Levi S.
Union Carbide Plastics Co USA Consulting Engineer USA
Ami, Howard T. Brownmiller, L. T.
National Bureau of Standards USA Alpha Portland Cement Co USA
Asano, T. *Brunauer, Stephen
Onoda Cement Co., Ltd Japan Portland Cement Association., USA
fAstin, Allen V. t *Budnikov, P. P.
National Bureau of Standards.. USA Academy of Sciences USSR
t *Assarsson, G. 0. *Burke,Edward
Geological Survey of Sweden Sweden The Associated Portland Ce- United Kingdom
t *Babushkin, W. I. ment Manufacturers, Ltd.
Academy of Construction and Ukr. SSR t *Buttler, F. G.
Architecture. University of Aberdeen United Kingdom
fBardin, Pablo P. Camarda, F. V.
Huincan, Compania de Cemen- Argentina The Flintkote Co USA
tos. tCampo, M. del
fBarona de la 0, Federico Laboratorio Central de Ensayo Spain
Consulting Engineer Mexico de Materiales de Construc-
Barrett, Wayne T. cion.
Washington Research Center, USA *Carlson, Elmer T.
W. R. Grace & Co. National Bureau of Standards.. USA
Bartosch, Eberhardt Catton, Miles D.
Verein der osterreichischen Austria Portland Cement Association.. USA
Zementfabrikanten. Cesareni, C.
Bates, A. Allan Cementerie del Tirreno Italy
Portland Cement Association USA Chaiken, Bernard
Bean, B. Leonard U.S. Bureau of Public Roads USA
National Bureau of Standards.. USA Chaney, D. L.
Bechtold, Ira C. Portland Cement Association USA
Consulting Engineer USA Cheron, M.
Becker, F. Cimenteries & Briqueteries Re- Belgium
Centre Technique Holderbank.. Switzerland unies.
Berger, Ernest E. t *Chopra, S. K.
Kosmos Portland Cement Co.. USA Central Building Research In- India
*Berman, Horace A. stitute.
National Bureau of Standards. USA Christiansen, G. G.
*Bessey, George E. Allis-Chalmers Manufacturing USA
Chalk, Lime, and Allied Indus- United Kingdom Co.
tries Research Assn. tCiriUi, V.
fBest, Cecil H. Politecnico di Torino Italy
University of California USA Clausen, C. F.
*Birss, F. W. Portland Cement Association USA
University of Saskatchewan Canada Collins, A. R.
tBizakis, T. A. Cement and Concrete Associa- United Kingdom
General Cement Co Greece tion.
*Blaine, R. L. Conroy, Francis A.
National Bureau of Standards. USA Louisville Cement Co USA
*Blair, L. R. fConvey, John
Department of Mines and Canada
Johns- Manville Products Corp. USA Technical Surveys.
Blank, Alton J.
*Copeland, L. E.
Cementos Atoyac, S.A Mexico USA
Portland Cement Association
fBlondiau, Leon fCoronas, Juan
Ciments d'Obourg Belgium Universidad de Barcelona Spain
Bogue, R. H. *Coutinho, A. de Sousa
Consulting Engineer USA Laboratorio Nacional de En- Portugal
Boughner, W. A. genharia Civil.
Bessemer Limestone and Ce- USA Crandall, J. R.
ment Co. National Bureau of Standards. USA
iv

Crepaz, E. Fowler, Albert C.
Universita di Padua Italy Huron Portland Cement Co USA
Cummins, Kenneth D. Fritts, Stewart S.
American Concrete Institute USA Lone Star Cement Corp USA
Cutlifife, J. L. *Funk, Herbert
Massachusetts Institute of USA Deutsche Akademie der Wis- East Germany
Technology. senschaften zu Berlin.
*Czernin, Wolfgang jFussman, Charles L.
Forschungsinstitut des Ver- Austria Monarch Cement Co USA
eines der osterreichischen *Gaskin, Arthur J.
Zementfabrikanten. Commonwealth Scientific and Australia
Dahl, Louis A. Industrial Research Organi-
Chemist USA zation.
Dalcq, P. R. Gaylard, H. G.
Solvay & Cie Belgium Pretoria Portland Cement Co., South Africa
*Danielsson, Ulf Ltd.
Cement and Concrete Institute. Sweden Ghion, Jules
fDavies, R. J. Ciments Portland Artificiels Belgium
Pretoria Portland Cement Co., South Africa Beiges d'Harmignies.
Ltd. Giertz-Hedstrom, Stig
Davis, Clayton Vibro-Verken, AB Sweden
Universal Atlas Cement Divi- USA Gilliland, J. J. Jr.
sion. United States Steel Ideal Cement Co USA
Corp. Glantz, O. J.
Davis, Geoffrey J. Penn-Dixie Cement Corp USA
The Associated Portland Ce- United Kingdom Glasser,Fred P.
meat Manufacturers, Ltd. University of Aberdeen United Kingdom
Davis, Gilbert Goossens, Joseph
Eastwoods Ltd United Kingdom Ciments Portland J. Van den Belgium
Daxelhofer, J. P. Heuvel.
Universite de Lausanne Switzerland Goossens, Pierre
Dolch, W. L. Ciments Portland J. Van den Belgium
Purdue University USA Heuvel.
Dungan, C. K. fGoria, Carlo
Premier Portland Cement Co., Rhodesia Politecnico di Torino Italy
Ltd. *Greenberg, Sidney A.
Dutron, Pierre Portland Cement Association... USA
Centre d'Information de I'ln- Belgium *Greene, Kenneth T.
dustrie Cimentiere Beige. Ideal Cement Co USA
fDutron, Raoul *Greening, Nathan
Laboratoire d'Essais et de Belgium Portland Cement Association... USA
Controle. t *Grothe, H.
*Dutz, W. Dyckerhoff Zementwerke Germaiiy
Institut fiir Steine und Erden__ West Germany *Grudemo, A.
Dyckerhoff, Walter Cement and Concrete Institute. Sweden
Consulting Chemist Argentina Gruenwald, Ernst
jEichenlaub, A. C. Lone Star Cement Corp USA
Peerless Cement Co USA Grunewald, Max E.
Eitel, Wilhelm Coplay Cement Manufacturing USA
Institute of Silicate Research, USA Co.
University of Toledo. fGrzymek, Jerzy
Eisner, Harald von Institute for Mining and Metal- Poland
Building Binders, Ltd West Germany lurgy.
Engstrom, Bengt t *Guinier, A.
Aktiebolaget Gullhogens Bruk_- Sweden Centre d' Etudes et de Recher- France
tErnst, Fred H. ches de I'lndustrie des Liants
Cimenteries & Briqueteries Belgium Hydrauliques.
Reunies Gunn, J. W.
Esenwein, P. Lone Star Cement Corp USA
Eidg. Materialpriifungs ' und Switzerland
t *Gutt, W.
Vorsuchsanstalt Building Research Station United Kingdom
*Evans, D. N.
fHaegermann, G.
National Bureau of Standards- USA West Germany
Chemical Engineer
fFessler, Erich
Zementwerke Loruns Aktienge- Austria fHaffner,Henry J.
sellschaft. Green Bag Cement Division, USA
Finlayson, Kier Pittsburgh Coke and Chemi-
Corning Glass Co USA cal Co.
t *Fletcher, K. E. t *Hagerman, Tor H.
Building Research Station United Kingdom Technical Consultant Sweden
tForan, M. R. HIkanson, Per
Nova Scotia Technical College. Canada Sk^nska Cement AB Sweden
tForbrich, L. R.
*Halstead, Philip E.
Green Bag Cement Division USA United Kingdom
Cement and Concrete Associa-
Pittsburgh Coke and Chem-
tion.
ical Co.
Forss, Bengt U. Halstead, Woodrow J.
Pargas Kalkbergs Aktiebolag Finland U.S. Bureau of Pubhc Roads... USA
Foster, Bruce tHanly, J. B.
National Bureau of Standards. _ USA Canada Cement Co., Ltd Canada

V

fHanna, Wilson C. Kaplan, M. F.
California Portland Cement USA National Building Research South Africa
Co. Institute.
*Hansen, Torben C. Kauer, J. A.
Cement and Concrete Institute^ Sweden Huron Portland Cement Co USA
*Hansen, W. C. t *Kavalerova, V. I.
Universal Atlas Cement Divi- USA Academy of Construction and USSR
sion, United States Steel Architecture.
Corp. Keil, Fritz
t *Harker, R. I. Forschungsinstitut der Zemen- West Germany
Johns-Manville Products Cor- USA tindustrie.
poration, Research Center. t *Kenington, H. W.
Harris, Philip H. University of Manchester United Kingdom
Riverside Cement Company USA Kennedy, G. Donald
Harrison, D. Portland Cement Association.. USA
Hydro-Electric Power Commis- Canada Kennedy, Thomas B.
sion of Ontario. U.S.Army Engineer Waterways USA
*Hattiangadi, R. Jl. Experiment Station.
The Associated Cement Com- India Kerman, Guy M.
panies, Ltd. Aberthaw & Bristol Channel United Kingdom
Heck, W. P. Portland Cement Co., Ltd.
Allentown Portland Cement Co_ USA Kesler, Clyde E.
*Heilmann, T. University of Illinois USA
F. L. Smidth & Co Denmark Kester, Bruce E.
tHeller, Mrs. L. Missouri Portland Cement Co__ USA
Geological Survey of Israel Israel tKeyser, W. L. De
*Helmuth, R. A. Universite Libre de Bruxelles... Belgium
Portland Cement Association. _ USA t *Kholin, I. I.
Hime, William G. Academy of Construction and USSR
Portland Cement Association.. USA Architecture,
tHowe, H. B. t *Kjaer, A.
Canada Cement Co., Ltd Canada Danish National Institute of Denmark
*Hubbard, Donald Building Research.
National Bureau of Standards.. USA Klawansky, A. L.
*Hunt, Charles M. Palestine Cement Works Israel
National Bureau of Standards.. USA Nesher, Ltd.
Hunter, H. M. Klein, Alexander
U.S. Bureau of Reclamation USA University of California USA
Hurlburt, R. W. tKlein, William H.
St.Mary's Cement Co., Ltd Canada Dragon Cement Co USA
Hutcheon,' N. B. Klieger, Paul
National Research Council Canada Portland Cement Association.. USA
*Idorn, G. M. *Kondo, R.
Aalborg Portland-Cement-Fab- Denmark Tokyo Institute of Technology. Japan
rik. fKoroneos, D.
flnoue, Giichi Technical University of Engi- Greece
Osaka Yogyo Cement Co., Ltd.. Japan neering Science.
Insley,Herbert fKourlibinis, P.
Consulting Mineralogist USA Halkis Cement Co Greece
t *Ish-Shalom, Moshe Koyanagi, K.
Portland Cement Association.. USA Tohoku Kaihatsu Co., Ltd Japan
Ivliev, Boris t *Kramer, Walter
Belgorod Industry Comm USSR Portiandzementwerke E. West Germany
Jack, Orville E. Schwenk.
Permanente Cement Co USA t *Kramer, H.
Jackson, Elwood C. Dyckerhoff Zementwerke West Germany
North American Cement Corp. USA Kreager, E. Carl
Jacobsen, B. R. Columbia Cement Corp USA
F. L. Smidth Co USA t *Kurczyk, H. G.
*Jagus, P. J. Institut fiir Gesteinshiitten- West Germany
t
The Associated Cement Com- India kunde.
panies, Ltd. Lafuma, H.
tJanssens, P. F. Centre d' Etudes et de Re- France
Societe des Ciments du Congo.. Congo cherches de I'lndustrie des
Liants Hydrauliques.
Jenour, Maynard
Aberthaw & Bristol Channel United Kingdom Lamberton, Bruce A.
Portland Cement Co., Ltd. Intrusion Prepakt, Inc USA
Joisel, A. *Laneuville, Jean
Centre d' Etudes et de Re- France St. Lawrence Cement Co Canada
cherches de I'lndustrie des *Lauer, K. R.
Liants Hydrauliques. University of Notre Dame USA
t *Jones, F. E. t *Lawrence, C. D.
Building Research Station United Kingdom Cement and Concrete Associ- United Kingdom
Kalousek, George L. ation.
Owens-Corning Fiberglas USA *Lea, F. M.
Corp. Building Research Station United Kingdom
*Kantro, David L. Le Cour Grandmaison, J. A.
Portland Cement Association.. USA Ciments Lafarge France

vi

Leffler,Del McMillan, F. R.
Halliburton Oil Well Cement- USA Civil Engineer USA
ing Co. McMurdie, H. F.
Legget, R. F. National Bureau of Standards.. USA
National Research Council Canada McPherson, A. T.
f *Lehmann, H. National Bureau of Standards.. USA
Institut fiir Steine und Erden_. West Germany tMehra, S. R.
*Lemish, John Central Road Research Insti- India
Iowa State University USA tute.
Lereh, William t *Mehra, V. S.
Portland Cement Association USA Portland Cement Association.. USA
*L'Hermite, Robert G. Mericola, F. C.
Laboratoires du B&timent et France Wyandotte Chemicals Corp USA
des Travaux Publics, Meyer, Adolf
f *Lhopitallier, P. Laboratorium des Westfali- West Germany
Ciments Lafarge France sclien Zementindustrie.
fLiatis, C. S. t *Midgley, H. G.
General Cement Co Greece Building Research Station United Kingdom
fLiebau, Friederich Mielenz, Richard C.
Deutsche Akademie der Wis- East Germany Master Builders Co USA
senschaften zu Berlin. *Mikhailov, V. V.
Lindholm, Bo Academy of Construction and USSR
Aktiebolaget Gullhogens Bruk_ Sweden Architecture.
*Locher, F. W. Minnick, John L.
Forschungsinstitut der Zement- West Germany G. & W. H. Corson, Inc USA
Industrie, Mitusch, Hans
t *Longuet, P. Metalihiittenwerke Liibeck West Germany
Centre d'Etudes et de Recher- France G.m.b.H.
ches de I'lndustrie des Liants fMorev, George W.
Hydrauliques. U.S. Geological Survey USA
Lorman, William R. tMorgan, E. C.
USN Civil Engineering Labo- USA Nazareth Cement Co USA
ratory. fMueller, Louis J.
Lowe, Robert F. Standard Lime and Cement Co. USA
Phoenix Cement Co USA tMyers, Philip B.
t *Ludwig, U. Allentown Portland Cement Co. USA
Institut fiir Gesteinshiitten- West Germany *Nagai, S.
kunde. University of Tokyo Japan
Lund, Anker *Nerenst, Poul
Cement Limited Ireland Danish Gas Concrete Inc Denmark
Lyon, Ethel V. *Neville, A. M.
Portland Cement Association. _ USA University of Manchester United Kingdom
MacKenzie, J. A. Newlon, Howard
United Steel Cos., Ltd United Kingdom Virginia Council of Highway USA
*Majumdar, Amalendu Investigation and Research.
National Bureau of Standards.- USA * Newman, E. S.
*Malquori, G. National Bureau of Standards. USA
Universita di Napoli Italy Nickelson, H. O.
*Manabe, T. Universal Atlas Cement Divi- USA
Nihon Cement Co., Ltd Japan sion. United States Steel
Mardulier, Francis J. Corp.
Dewey and Almy Chemical USA Nishitomi, Y.
Division, W. R. Grace & Co. Yawata Chemical Industrv Japan
fMariani, E. Co., Ltd.
Universita di Roma Italy t *Nurse, R. W.
Marshall, Gordon Building Research Station United Kingdom
G. & T. Earle, Ltd United Kingdom tO'Daniel, H.
Mather, Bryant Mineralogisches Institut der West Germany
U.S. Army Engineer Water- USA Universitat Frankfurt.
ways Experiment Station. OfFutt, James S.
*Mather, Katharine United States Gvpsum Co USA
U.S. Army Engineer Water- USA Ohta, Z.
ways Experiment Station. Mino Ceramics Co., Ltd Japan
fMatsushima, K. Okuno, Chiko
Osaka Yogvo Cement Co., Ltd. Japan Nihon Cement Co., Ltd Japan
Mau, K. T. *Ordway, Fred
Hawaiian Cement Corp USA National Bureau of Standards.. USA
McAdam, Francis A. t *Orsini, P. Giordano
Huron Portland Cement Co USA University of Naples Italy
McCoy, W. J. Osborn, E. F.
Lehigh Portland Cement Co USA Pennsylvania State University. USA
t *McCurdy, K. G.
Palmer, K. E.
University of Saskatchewan Canada Ideal Cement Co USA
t *McGowan, J. K.
Commonwealth Scientific and Parker, Walter B., Jr.
Industrial Research Organi- Dewey and Almy Chemical Di- L^SA
zation. Australia vision, W. R. Grace & Co.
*Mchedlov-Petrosyan, 0. P. Parker, W. E.
Academy of Construction and Ukr. SSR Hydro-Electric Power Com- Canada
Architecture. mission of Ontario.

vii

t *Pena.S. Schluter. S. Silcox. Robert Philleo. USA Hydro-Electric Power Com. California Portland Cement USA Presidium Academy Construe. E. Research Institute for Building Czechoslovakia National Bureau of Standards. Denmark Co. de Materiales de Construc. Robert E. Robert E. United Kingdom cion. Hans J. V. Sika Chemical Corp USA Polysius G. United Kingdom Dyckerhoff Zementwerke West Germany ment Manufacturers. Portland Cement Association. Y. t *Percival. Poul Lafarge Aluminous Cement United Kingdom Aalborg Portland-Cement-Fa. Anglo-Alpha Cement Co. Dyckerhoff Zementwerke West Germany Dyckerhoff Zementwerke West Germany *Plum. H.b. Schoonover. partment of the Army. t *Sasaki. L. . J. 0. Ervin National Bureau of Standards. USA Department of Agriculture Canada Schutz. USA Smolczyk. Niels M. Building Research Station United Kingdom Smith.. *Rio.-G. Ltd Canada Penn-Dixie Cement Corp USA *Robson. Netherlands Powers. Sauzier. C. Erich t *Schimmel. Polivka. T. A. Sauman.S. . . T. Ltd. USA Materials. W. E. viii . Arturo University of Naples Italy Cementisegni Italy t *Shestopyorov. *Seligmann. L.. Delia M. le Ferrociment Belgium Ciments Lafarge France Pirtz. Franco Architecture. G. Leonard Rutle. Seager. Raymond J. C. C. USSR Co. USA Trinidad Cement Ltd Trinidad turing Co. tion. Army Engineer Water. Jr. Walter Danish National Institute of Denmark Anneliese Portlandzement. W. C. Pirotte. Carl H. U. National Bureau of Standards. D. Academy of Construction and USSR Building Research Station United Kingdom Architecture. Rabot.. Dwight K. Portland Cement Association. Pesenti. S. G.. Copenhagen.H West Germany t *Schwiete. A Cement Limited Ireland University of Aberdeen United Kingdom Sadler. F. Robert G.und West Germany Building Research. I. J. Cimenterie di Merone Italy Runnels. O. Paul Reichenbach. South Africa Office. D. *Skramtaev. V. The Associated Portland Ce.. D. C. Giampiero Fuller Company USA Italcementi Italy t *Salnikova. Ross. Ryan. West Germany Pennsylvania State University. A. C. Edith Price. hoven. fSanada. tReagel. fRobinson. Skjoldborg.. Wasserkalkwerke. USA Ube Industries. fPetrovic. USA The Technical University of Denmark t *Schuit. de la Monarch Cement Co USA Laboratorio Central de Ensayo Spain Russell. Cement and Concrete Associa. Technical University Eind. West Germany Ciments Lafarge France kunde. M. Milos t *Schmitt. M. J^n Academy of Construction and USSR Institute of Inorganic Chem. Ltd. Ltd. University of California USA Dyckerhoff Zementwerke West Germany *Pollitt. Ltd Japan American. Czechoslovakia Architecture.. R. Romig. tionand Architecture. C. USA t *Schulz. Ltd Japan turing Co. P. USA t *Santarelli. R. USA fenschlacke.Canada Smith. Hugh C. *Poulsen. Roberts. J. Deane K.m. t *Rosaman. Chief of Engineers. M. Raymond Institut fiir Gesteinshiitten. Rustum Forschunginstitut fiir Hocho. Iwaki Cement Co. University of California USA Lone Star Cement Corp USA Plassmann. D. brik. Pennsylvania State University. fPflaumer. Italcementi Italy fPierson. Halliburton Oil Well Cementing USA *Roy. Parsons. Pepper. H.. Co. National Bureau of Standards. USA Academy of Construction and USSR Partesana. G. T. S.Marietta Co USA Sandler. G. is try. Marquette Cement Manufac. USA Consulting Engineer Norway ways Experiment Station. John U. Canada Cement Co. G. H. P. Fred S. Prussing. S. James L. Douglas E. Alfred M. L. Schmoelder. B. *Roy. H. Portland Cement Association. David Sawyer. t *Royak. H. De. mission of Ontario. Zdenek *Pilce. M. R. Marquette Cement Manufac. USA Dewey Portland Cement Co USA t *Sersale..

Ube Industries. Tomes. C. Hubert Universita di Roma Italy Portland Cement Association.H. Canada Cement Co. Jerzy Penn-Dixie Cement Corp USA Institute for Mining and Metal. R. Eugene J. Vellines. . S. Brazil *Woermann. G. Myron A. Reunies. Friederich Tsumura. R. Wittekindt.. Monolith Portland Cement Co. De. Ltd Japan OfRce. F. O. Steel Corp. zur Riverside Cement Company USA Dyckerhoff Zementwerke West Germany t *Vivian.S. U. W. J. P. Portland Cement Association. George Christiania Portland-Cement. Eduard University of Sao Paulo West Germany Portland-Zementwerke Heidel- t *Torroja. *Taylor. tStein. F. Sweitzer. _ USA Onoda Cement Co. D Louisville Cement Co USA National Bureau of Standards. F. Ltd. Duncan R. *Takemoto. A. Architecture. USA *Swayze. H. West Germany Wichers. Williams. Ltd Japan *Suzukawa.USA Veltman. J. Ltd Japan * Welch. Peerless Cement Co USA tStagg. search. R. t *Suhkowski.. _ USA ix . W.. Louisville Cement Co USA Miron et Freres. D. Eberhard Uriarte. *Turriziani. A. BattelleMemorial Institute USA Onoda Cement Co. sion. A. USA t *Taplin. H. A. Kenya t *Tan Tik-Ien. United Kingdom. Stanton Calaveras Cement Co USA National Ready Mixed Con. J.. Institut fiir Gesteinshiitten. USA Sutton. Taro U. _ USA fabrik. R. H. Sven Central Building Research In. Missouri State Highway De. partment of the Army. F. Grace & Co. H. ment Co. Ltd Canada Lock Joint Pipe Co USA Wechter. USA University of Aberdeen United Kingdom *Williams. Soji Gesellschaft Deutsche Chera... H. Julien Ketton Portland Cement Co. R. R. K.. Academy of Construction and USSR t *Watanabe. Central Laboratory. C. E. George Tanaka. zation. R. S. H. Lone Star Cement Corp USA Weaver. G. Ltd Japan Werner. lurgy..O Netherlands Universal Atlas Cement Divi. Norway Portland Cement Association. fWalker. *Verbeck. T.. Swenson. zation. E. Voelker. Portland Cement Association. T. Woods. Portland Cement Association. USA Dalen Portland-Cementfabrik__ Norway W. E. K. C. British Standard Portland Ce. E. National Bureau of Standards. N. Instituto Tecnico de la Con. Building Research Station United Kingdom Onoda Cement Co. Witt. India Vanderwerp. Harry L. J. Jerome *Strassen. Snyder. Ltd Canada Weise. H. H. J. Van Hove. Ltd. USA National Portland Cement Co USA partment. G. A. Niko Commonwealth Scientific and Australia South African Council for South Africa Industrial Research Organi- Scientific and Industrial Re. Edward kunde National Bureau of Standards. Japan iker. Bureau of Pubhc Roads. W. *Thorvaldson. Crete Association. East Germany Osaka Yogyo Cement Co. Harold f Thaulow. Robert J. P. Ary F. Valore. USA t *Suvorova. Sika Chemical Corp USA Central Road Research Insti. Mexico berg. Poland Walker. F. L. f *Srinivasan. Stensrud. Waugh.India Norsk Cementforening Norway stitute. USA Japan Cement Engineering As. H. de Portland-Zementwerke Heidel. _ USA tWillis. _ USA Consulting Engineer USA t *Toropov. berg. P.N. tute. WilUam R. Y. Commonwealth Scientific and Australia Cimenteries & Briqueteries Re. Belgium Industrial Research Organi. unies. John H. C. Thyrre. Thwaite. L. University of Saskatchewan Canada tWillingham.S.. S. Japan *White. tWolf.USA tSteinour. S. Spain struccion y del Cemento. t *Takagi. West Germany Cementos Veracruz. C. M. Ltd Japan *Spohn. sociation. N. . Ivar Washington Research Center. Jack t *Uchikawa. Wille. Ube Industries. T. Academy of Sciences USSR Dyckerhoff Zementwerke West Germany fTorres. *Stutterheim. N. Vinograd. Strand. G. Chief of Engineers. Cimenteries & Briqueteries Belgium Ltd.

S. which have been psi (or p. Ltd. M=MgO.h. G. H. White's South African Portland South Africa Centre d'Etudes et de Re. USA Worsdale. N. Paul *Yang. H = H20.) =relative humidity complex compounds. Dyckerhoff Zementwerke West Germany *Wyatt. NMR= nuclear magnetic resonance Where not marked. USA tZeeh. Chichibu Cement Co. Explanatory Notes Abbreviations. FM = fineness modulus Scale of Magnification. Donald 0. U. A = Al203. Portland Cement Association. the scale of mag- IR = infrared nification frequently is indicated by a horizontal line. F = Fe203. )= pounds per square inch widely adopted by cement chemists for formulating more rh (or r. Less common abbreviations of this Paper II-I is the first principal paper of session II. In micrographs. Identification Numbers of Papers. i. T.France Cement Co.. Technical Surveys. *Yamaguchi. has been assigned an identification number. Netherlands Department of Mines and Canada hoven. N = Na20. Bureau of Public Roads___ USA Onoda Cement Co. E. Julie National Bureau of Standards. Ltd Japan *Worksman. Ltd Japan Marquette Cement Manufac. Woolf. X . II-Sl is the first supplementary paper of session II. Each symposium paper S = Si02. Examples: K = K20. the numbers correspond to those assigned include the following: to the copies of the papers distributed in advance of the DTA = differential thermal analysis symposium. E. With Commonly used abbreviations of more general nature few exceptions. "Technical University Eind. J. C. turing Co. s.. G. are used interchangeably with the w/c (or W/C) = water-cement ratio respective oxide formulas throughout this book: C=CaO. R. Wuerpel. Yung-Chi Liants Hydrauliques. _ USA Johns-Manville Products Corp. _ USA Yoshii. the length represents one micron. The following symbols. *Yannaquis.. cherches de 1' Industrie des Wu. Zoldners. N. paper type are defined as they occur.

Majumdar and Deane K. zur Strassen. A. J. G. Barker 196 H. Chemistry of Clinker Stephen Brunauer and S. Smolczyk 80 Paper III-S4. Midgley. Influence of reducing atmosphere on of glasses of the system CaO-AloOs- the constitution of clinker. Schmitt 244 Paper II-Sl. Greenberg 135 Paper II-l. Fletcher 69 rate of hydration of ^-dicalcium sili- Discussion D. General Addresses aluminoferrite in an atmosphere of hydrogen. Swayze 26 D. Smolczyk 275 process. Contents Volume I Page Page Preface . Midgley and D. F.-G. Taplin 263 H. Copeland. Taylor. The precipitation of CaO A^Oa- Paper II-S5. the reaction of /3-CaoSi04 with water Renichi Kondo 107 at 25 °C-120 °C. Cement research: Retrospect and Compounds prospect. The hydration of tricalcium silicate and /3-dicalcium silicate at room tem- perature. A. Longuet 131 growth of science. H. Discussion H. Infrared absorption spectra of some N. 201 LeviS. Guinier 26 zur Strassen 194 Myron A. Paper III-S8. A. 285 Paper II-S6. Brown 29 Author's closure 202 Rustum Roy 29 Paper III-3. zur Strassen and C. Paper II-S7. Y. H. Rosaman. Percival. G. W. D. aluminate solutions at 21 °C. Reaction velocity in portland cement Paper III-S7. 259 H. stituents.-G. Lea 5 Paper III-l. Yannaquis and A. Assarsson 190 Smith 23 H. A. zur Strassen 244 the hydraulicity of the calcium sili- M. Gutt 59 trisulfoaluminate at 25 °C. Hydraulic properties and hydration Paper II-S3. Portland cement clinker. Katharine Mather 34 F. Woermann 104 quartz pastes. H. Locher 99 A. G. W. White 67 man 247 Author's closure 68 Paper III-S2. Crystal structures of clinker con. Herbert Funk 291 Portland cement clinkers. Aitken and H. The heat of formation of calcium j. F. N. H. Schimmel. Discussion nical Portland cement clinker. F. The temperature coefficient of the and K.. Hunt 297 Paper II-S8.-G. Jones 204 Author's closure 35 Discussion Paper II-2. Session II. Gunnar O. C. Erode 1 Session III. Paper I-l. Two different ways of hydration in clinker formation. Kramer and H. Rosaman. Swayze 118 Charles M. Hydrothermal reactions in the system Portland cement clinker. H. Taylor 277 Guenter Ahlers 104 Paper III-S6. Smolczyk 27 G. E. X-ray diffraction examination of Portland cement. R. and Discussion H. Roy and R. Kantro. Taylor 167 N. Author's closure 106 A. Berman and E. Author's closure 275 T. IOH2O from supersaturated calcium sia in tricalcium silicate. Buttler. F. Wyatt. Steam curing of cement and cement- E. Locher 267 Paper II-S4. M. Hydration of calcium aluminates and H. Paper III-S3. W. F. Heilmann 87 Paper III -S5. Reactions of coal ash with portland Discussion cement clinker during the burning H. and H. Some problems associated with the P. Si02. W. Sasaki 83 F. G. Suzukawa and T. A. W. Roberts (closure) 245 Paper III-Sl. Phase equilibria and constitution of Paper III-2.I. Grothe. Yannaquis and A. Woermann 119 Author's closure 305 XI . Rustum Roy 304 E. S. L. Myron A. Toropov 113 compounds in the CaO-SiOo-HoO Discussion system. Nurse 9 — — CaO Si02 HoO and the steam curing of cement and cement-silica products. Solid solution of alumina and magne. Welch and W. E. New- C. The decomposition of alitein tech. Schuit and R. 32 ferrites. Solid solutions of the minerals of . E. Discussion H. The effect of minor components on H. H. W. M. Chemistry of Hydration of Cement Paper 1-2. and Stephen Brunauer 75 "john H.ii Discussion List of Symposium members and authors iv Guenter Ahlers 128 Explanatory notes x Author's closure 129 Paper II-S9. Note on the behavior of tetracalcium Session I. Guinier 21 Discussion A. L. Wallace R. The composition of ettringite in set Paper II-S2. F. Schmitt 243 Fred Ordway 39 C.

Page Page
Paper III-S9. Crystalline solubility and zeolitic be- Paper IV-3. Chemistry of hydration of portland
havior in garnet phases in the system cement.
CaO-AloOa-SiOj-HaO. L. E. Copeland, D. L. Kantro,
Delia M. Roy and Rustum Roy.. 307 and George Verbeck 429
exchange systems Discussion
Paper III-SIO. Calcium in of
xCaOSiO,?/H20-Ca(OH)2-H20. J.H. Taplin 465
T. Thorvaldson, F. W. Birss, and H.-G. Smolczyk 467
K. G. McCurdy 315 Paper IV-Sl. The role of gypsum in the hardening
Paper III-Sll. Kinetics of reaction in the system of hydraulic cements.
CaO-SiOa-HjO. P. P. Budnikov 469
P. E. Halstead and C. D. Law- Paper IV-S2. The mineralogical examination of set
rence 321 Portland cement.
Discussion H. G. Midgley 479
Sidney A. Greenberg 325 Discussion
Paper III-S12. The influence of high pressure steam H. Dutz 490
on the process of hardening of calcium Paper IV-S3. Nuclear magnetic resonance study on
and magnesium silicates and of ce- hardened cement paste.
ments of different compositions. K. Watanabe and T. Sasaki 491
P. I. Bozhenov, V. I. Kavalerova,
Paper IV-S4. The rate of hydration of cement com-
V. S. Salnikova, and G. F.
Suvorova 327 pounds and Portland cement esti-
Paper III-S13. Concerning the hydration products of
mated by X-ray diffraction analysis.
3CaO Si02 and /3-2CaO Si02. G. Yamaguchi, K. Takemoto, H.
H. G. Kurczyk and H. E. Uchikawa, and S. Takagi 495
Schwiete 349 Paper IV-S5. Proton magnetic resonance in hy-
drated portland cements.
Session IV. Chemistry of Hydration of
R. L. Blaine 501
Portland Cement
Paper IV-S6. Infrared spectroscopy studies on the
Paper IV-1. Early hydration reactions of portland hydration of clinker minerals and
cement. cements.
Kenneth T. Greene 359 H. Lehmann and H. Dutz 513
Discussion Paper IV-S7. Heat of hydration of cement as af-
W. C. Hansen 374 fected by water-cement ratio.
E. T. Carlson 375
Ulf Danielsson 519
T. C. Powers 376 Discussion
A. J. Gaskin 377 W. C. Hansen 526
S. A. Greenberg and V. S.
Paper IV-S8. The influence of false setting on some
Mehra 378
properties of hardening cement paste
Author's closure 384
and mortar.
Paper IV-2. False set in portland cement.
Jerzy P. Sulikowski 527
W. C. Hansen 387
Paper IV-S9. The thermodynamics of the harden-
Discussion
ing processes of cement.
Toshio Manabe 404
Paul Seligmann and N. R. O. P. Mchedlov-Petrosyan and
Greening 408 W. I. Babushkin 533
Jean Laneuville 410 Paper IV-SIO. Contribution to the determination of
Renichi Kondo 411 the heat of hydration of clinker min-
R. R. Hattiangadi 412
L. SantareUi 414 H. E. Schwiete and A. Tan Tik-
Guenter Ahlers 415
Myron Swayze
A. 416
len 545
A. Rio and R. Turriziani 417 Author index forvolumes I and II 551
A. J. GaskinandJ.K. McGowan_ 420 Subject index for volumes I and II l 561
Author's closure 424 Publications program of the NBS 573

Volume II

Page Page
Session V. Properties of Cement Paste and Discussion
Concrete L. E. Copeland and Edith G.
Schulz 648
Paper V-1. Physical properties of cement paste. Levi S. Brown 655
T. C. Powers 577 Author's closuie 656
Discussion =
Paper V-3. Volume changes of concrete.
H. W. W. Pollitt 609 Robert G. L'Hermite 659
G. E. Bessey 610 Discussion
Author's closure ^ 611 E. Torroja and C. de la Pena 695
Paper V-2. The microstructure of hardened ce- R. R. Hattiangadi and P. J.
Jagus 699
ment paste. Myron A. Swayze 700
Ake Grudemo 615 T. C. Powers 702

Xll

Page Page
Paper V-Sl. Creep of aluminous cement concrete. Paper VI-S5. Durability of concrete and reinforced
A. M. Neville and H. W. Ken- concrete and its connections with the
ington 703 composition of cement, concrete, and
Paper V-S2. Strength, elasticity, and creep as re- construction methods.
lated to the internal structure of con- S. V. Shestopyorov 889
Session VII. Chemical Additions and Admixtures
Torben C. Hansen 709
Paper V-S3. A few unsolved problems of cement Paper VII-1. Some chemical additions and admix-
hydration. tures in cement paste and concrete.
Wolfgang Czernin 725 H. E. Vivian 909
Paper V-S4. The rheology of fresh portland ce- Discussion
ment pastes. K. R. Lauer 923
Moshe Ish-Shalom and S. Green- J.H. Taplin 924
berg 731 Author's closure 925
Discussion
Gaskin 744
Session VIII. Special Cements
A. J.
Authors' closure 746 Paper VIII-1. Stressing cement and the mechanism
of self-stressing concrete regulation.
Session VI. Destructive Processes in Concrete V. V. Mikhailov 927
Discussion
Paper VI-1. Chemical reactions involving aggre- Levi S. Brown 954
gate. Paper VIII-2. Blast-furnace slags and slag cements.
Per Bredsdorff, G. M. Idorn, W. Kramer 957
Alice Kjaer, Niels Munk Discussion
Plum, and Ervin Poulsen 749 Renichi Kondo 973
Discussion H.-G. Smolczyk 975
Tor H. Hagerman : 783 G. Malquori, R. Sersale, and
W. C. Hansen 784 P. Giordano Orsini 976
T. C. Powers 788 H. Wilhams and S. K. Chopra_- 979
R. G. Pike, Donald Hubbard, J. H. Welch 981
and E. S. Newman 794 Paper VIII-3. Portland-pozzolan cement.
John Lemish 796 Giovanni Malquori 983
R. R. Hattiangadi and P. J. Discussion
Jagus I 799 W. C. Hansen 1000
A. de Sousa Coutinho 800 N. R. Srinivasan 1001
Levi S. Brown 801 L. Santarelli 1003
R. Wyatt and G. C. A. Schuit._ 802 H. -G. Smolczyk 1004
G. M. Idorn 802 E. Burke 1005
Authors' closure 806 Author's closure 1005
Paper VI-2. Frost action in concrete. Paper VIII-4. Calcium aluminates and high-alumina
Poul Nerenst 807 cements.
Discussion P. Lhopitallier 1007
R. A. Helmuth 829 Discussion
Ulf Danielsson 833 M. H. Roberts 1033
Paper VI-Sl. Deterioration of cement products in Paper VIII-Sl. The risk of unsoundness due to
aggressive media. periclase in high-magnesia blast-
J. H. P. van Aardt 835 furnace slags.
Discussion N. Stutterheim 1035
T. D. Robson 848 Discussion
L. R. Blair and Julie Chi-Sun Levi S. Brown 1040
Yang 849 Author's closure 1041
Paper VI-S2. Capillary size restrictions on ice for- Paper VIII-S2. Special masonry cement having a
mation in hardened portland cement high slag content.
pastes. S. Nagai 1043
.

R. A. Helmuth 855 Paper VIII-S3. Blast-furnace cement in the USSR.
I. I. Khohn and S. M. Royak._ 1057
Paper VI-S3. Comparison of chemical resistance of
Paper VIII-S4. High chemical resistance pozzolanic
supersulfate and special purpose ce-
cements.
ments.
R. Turriziani and A. Rio 1067
D. N. Evans, R. L. Blaine, and
Paper VIII-S5. Refractory barium-aluminous cement
Paul Worksman 871
and concrete.
Discussion
Al. Braniski 1075
T. D. Robson 877
Paper VIII-S6. Researches on the hydration of trass
E. Burke 877
CGmGO ts
Julie C. Yang 879
U. Ludwig and H. E. Schwiete. 1093
Paper VI-S4. Chemical resistivities of various types Paper VIII-S7. High strength rapid hardening con-
of cements.
Renichi Kondo 881 B. G. Skramtaev 1099
Discussion
E. Spohn 886 Author index forvolumes I and II 1101
H. -G. Smolczyk 887 Subject index for volumes I and II 1111
Author's closure 888 Publications program of the NBS 1123

xiii

Proceedings of the Fourth International Symposium
on the Chemistry of Cement
SESSION V. PROPERTIES OF CEMENT PASTE AND CONCRETE
Paper V-1. Physical Properties of Cement Paste*
T. C. Powers

Synopsis
This paper deals mainly with cement paste in terms of its porosity, internal surface area,
interaction between solid substance and evaporable water and related properties. There
are 10 sections as follows: (1) introduction; (2) properties of fresh paste; (3) specific volume
of hydrated cement and porosity of paste; (4) surface area of hydrated cement, and indicated
particle size; (5) minimum porosity of hydrated paste and specific volume of cement gel;
(6) concepts of structure of cement gel and cement paste; (7) mechanical effects of adsorption
and hydrostatic tension; (8) instability of cement paste; (9) strength; (10) permeability.
Among the subtopics are the following: definition of pore and solid; permeability of
paste at all stages of hydration; measurement of surface area; specific volume of cement gel;
computation of capillary porosity; physical aspects of hydration products; adsorption and
capillary condensation; spontaneous decrease of specific surface area of hydrated cement;
irreversible deformations; gel-space ratio vs. strength; theories of permeability; viscosity
of water in cement paste; capillary continuity and discontinuity.
A glossary of terms is appended.

Resume
Get expos6 traite principalement de la pate de ciment en fonction de sa porosite, de la
surface sp6cifique interne, de Taction reciproque entre la substance solide et I'eau Evaporable,
et des proprietes qui s'y rapportent. II y a 10 sections qui sont les suivantes: (1) introduc-
tion; (2) proprietes de la pate fraiche; (3) volume specifique du ciment hydrate et porosite de
la pate; (4) surface specifique du ciment hydrate, et indication sur la taille des particules;
(5) porosite minimum de la pate hydratee et volume specifique du gel de ciment; (6) concepts
de structure du gel de ciment et de la pate de ciment; (7) effets mecaniques de Tadsorption
et de la tension hydrostatique; (8) instabilite de la pate de ciment; (9) resistance; (10) per-
meabilite.
Parmi les points secondaires, les suivants sont traites: definition de pore et de solide;
permeabilite de la pate a tous les stades de I'hydratation; mesure de la surface specifique;
volume speciiique du gel de ciment; calcul de ia porosite capillaire; aspects physiques des
produits d'hydratation; adsorption et condensation capillaire; diminution spontanee de la
surface specifique du ciment hydrate; deformations irreversibles; rapport gel-espace versus
resistance; theories de permeabilite; viscosite de I'eau dans la pate de ciment; continuite et
discontinuite capillaires.
Un lexique des termes se trouve en appendice.

Zusammenfassung
In dem Vortrag werden hauptsachlich die Porositat, die GroBe der inneren Oberflache,
Korper und verdampfbarem Wasser und die damit zusammen-
die Reaktion zwischen festem
hangenden Eigenschaften der Zementpasten beschrieben. Die zehn Kapitel haben die
folgenden Uberschriften: (1) Einleitung; (2) Eigenschaften der frischen Paste; (3) Das
spezifische des hydratisierten Zementes und die Pastenporositat; (4) Oberflache des
Volumen
hydratisierten Zementes und wahrscheinliche TeilchengroCe; (5) Die Minimumporositat der
hydratisierten Paste und das spezifische Volumen des Zementgels; (6) Der Zementgel- und
Zementpastenstrukturbegriff (7) ;
Mechanische Effekte der Adsorption und der hydro-
statischen Spannung; (8) Unbestandigkeit der Zementpasten; (9) Festigkeit; (10) Permea-
bilitat.

•Fourth International Symposium on the Chemistry of Cement, Washington, D.C., 1960. Contribution from the Research and Development Laboratories
of thePortland Cement Association, Skokie, Illinois.

577

Als sekundare Themen behandelt sind: Definition der Pore und des festen Korpers;
Pastenpermeabilitat in alien Hydratationsstufen; Oberflachenmessung; Das spezifische
Volumen des Zementgels; Berechnung der kapillaren Porositat; Physikalische Eigenschaften
der Hydratationsprodukte; Adsorption und Kapillarkondensation Abnahme der spezifischen;

Oberflache des hydratisierten Zementes; Nicht umkehrbare Formveranderung; Die Festig-
keit als eine Function des Verhaltnisses gel/space; Permeabilitatstheorien; Die Viskositat
des Wassers in der Zementpaste; Kapillare Stetigkeit und Unstetigkeit.
Eine Liste der benutzten Worte ist am Ende hinzugefiigt.

1. Introduction

The chemical origins of the substance called properties of paste are not much influenced by such
cement paste, and the physico-chemical processes differences. Any hydrated portland cement is
of its formation, are essential elements of the study predominantly colloidal and this point of similarity
of physical properties of cement paste, and the outweighs the points of difference attributable to
properties of cement paste are reflected in almost differences in chemistry.
every aspect of concrete technology. Thus, the The words used for describing hydrated cement
study of cement paste provides a bridge between are those required for describing a chemically un-
cement chemistry and concrete technology. A identified substance. In 1947 it was not possible,
subject of such broad scope involves the whole and still in 1960 it is not feasible, to use much
literature of cement and concrete, and a complete chemical nomenclature. The ratios CaO:Si02:
critical review would have required more time than H2O for substances in cement paste are not defi-
was available. Therefore, a complete review was nitely established, to say nothing of other com-
not attempted. ponents. There is reason to suspect that unique
This review turned out to be an occasion for molecular ratios are not characteristic of hydrated
revision and reassessment of our own work, and cements. So, in this 1960 review it is still neces-
this entailed introducing some material not pub- sary to refer to grams of "nonevaporable water",
lished before. Concepts of structure have been and to "hydrated cement", instead of conven-
emphasized, and special attention is paid to a few tionally designated molecular species. The result
is rather unsatisfactory to chemists and non-
points that have proved to be somewhat con-
chemists alike, since a discussion based on arbi-
troversial. Perhaps the most important of these trary special definitions often becomes confusing
is the question of internal surface area; Are if those definitions are overlooked. (A glossary
internal surfaces real, and have we measured them? of terms will be found in appendix 1.) Possibly,
Little is said about differences in the chemical after other papers of this symposium have been
compositions of cements because the physical digested, terminology will be improved.

2. Properties of Fresh Paste

The Dormant Period
Dispersion Specific
During a short period beginning at the time of Material medium surface
cm2/g
contact between cement and water, relatively
rapid chemical reactions occur. Then follows a Cement A Kerosene _.- - 1,690
longer period of low activity which has been called Cement A Water 1,705

the dormant period. [1,2].^ It is the time during Cement B _ Kerosene . _ . 2,590
which paste normally remains plastic, and, at Cement B . Water. . . 2,630

room temperature, it normally lasts 40 to 120
min, depending on the characteristics of the ce-
ment. Powers [4] found that the specific surface area
The reaction does not seem to alter the
initial calculated from the rate of bleeding of cement
size and shape of the cement grains very much. pastes by a modified Kozeny-Carman equation,
This is indicated, for example, by the following was about the same for a paste made with kerosene
data obtained with a Wagner turbidimeter by as with water. Steinour [5] measured the bleed-
Ernsberger and France [3]. ing rates of one cement in various liquids. The
values for specific surface area, calculated from the
'Figures in brackets indicate the literature references at the end of this
paper. Kozeny-Carman equation [6] were as follows
578

essential an understanding of the properties
to
Xame of liquid Surface
and behavior of fresh concrete, but also is pertinent
area, cmVg
to various aspects of mature concrete.
Dioxane 1,680
Hexyl ether 1,630 Table 1. Mixing schedules required by different cements
Butyl alcohol 1, 670 to eliminate false set
Octyl alcohol 1,680
Diethylphthalate 1, 675
Specific surface: 1,500 cm^/g (Wagner Turbidimeter)
Tests were on 1:2 (by wt.) mortars made with Elgin sand,
Average 1, 670 w/c is that required for (l±0.05)-inch slump, with 6-in cone, after a 1 min mix
Water The first number of the mixing schedule is the length of the first period o
mixing; the second is the period of rest; the last is the final mixing period.

Condition after
ASTM w/c 1-min mix Required Remolding Slump
These figures indicate that reaction with water type Lot No. by wt. mLxing effort, jigs 6-in.
schedule cone
caused a shght increase in specific sui'face area. E.E. c Slump
The constancy' of the rate of bleeding during
the dormant period shows that the new products min in
1 15753 0.36 27 1.05 2-0-0 25 1.25
formed by reactions occurring during the dormant 1 15760 •> .35 41 .85 2-2-2 33 1.10
2 15755 » .34 32 10 2-2-2 26 10
period do not effectively alter the surface contours 3 15757 » .34 33
1.
.95 2-3-2 27
1.
1.05
of the particles, or change the viscosity of the 4 15762 b .34 42 .85 2-5-2 22 1.40

liquid. Possibly, the layer of h^^dration products
» This cement showed evidence of moderate false set.
first formed, around which water flows during •> This cement showed marked false set.
bleeding, is of loose enough texture to accommo- = Remolding effort, number of jigs.
date the small quantity of material produced
during the dormant period. The Flocculent State
During the dormant period, properly prepared
Preparation of Pastes paste is a thick suspension of particles in a floccu-
The physical characteristics of a batch of lent state. Steinour [9] found that the term
fresh paste depend on how the paste is prepared. "flocculent state" is not to be construed to mean
When mixed by gentle stirring, the paste remains that the paste consists of a collection of more
stiff relative to the consistency produced by or less separate floccules. Instead, the whole
vigorous mechanical stirring [7]. Dr}- cement, body of paste constitutes a single floe, the floe
which is normally in a flocculated state, is structure being a rather uniform reticulum of
apparently not uniformly dispersed by wetting cement particles. Because of such structure,
alone. To produce a homogeneous paste of fresh paste has some cohesive strength, as shown
lowest possible stiffness, a laboratory mixer able by its Theological properties.
to produce a high rate of shear is necessary. The flocciilent state can be modified or destroyed
Length of mixing period is important. If by means of surface-active agents. Such agents
it is of the order 30 sec or less, the paste becomes have been studied extensively, particularly in the
firm soon after mixing stops [8]. Steinour called Soviet Union, but, this being the subject of
this phenomenon 'brief-mix-set" and attributed another paper of the Symposium, it wiU not be
''

it to the grains becoming stuck together by gel dealt with here.
in process of forming on grain surfaces. Even Studies of sedimentation volume indicate that
if the mLxing period is long enough to prevent cement particles in water, though flocculent, are
brief-mix-set, which is usuaUy the case, it may not normally far from the dispersed (nonflocculent)
be too short to eliminate false set. In the labora- state [10]. The sedimentation volume of cement
tory, it is advisable to allow a period of rest in dry alcohol, in which the particles are com-
between an initial and final mixing period, the pletely dispersed, is only a little smaller than the
length of the rest period having been determined sedimentation volume of cement particles in
by trial for each cement. Examples of mixing water, whereas the sedimentation volume in a
schediiles required to eliminate false set are given nonpolar hquid, benzene, for example, is nearh'
in table 1. twice that in dry alcohol. The relatively low
With the exception of matters pertaining to intensity of mterparticle attraction between
false set, time effects and mixing procedures are cement particles m
water is probably due to the
principally laboratory- problems. Under normal hydrophflic nature of cement.
field conditions, the time of mixing can hardly Although the floe structm'e of cement paste
be so short as to permit brief-mix-set, and the has some strength and rigidity, the cement parti-
roUing mass of aggregate in a concrete mixer cles are nevertheless discrete durmg the dormant
"homogenizes" the paste as effectively as the period. This was shown by measurements of
most vigorous laboratory stirrer. hydrostatic pressure [11]. The constancy of the
When a paste is properly prepared,
its properties rate of bleeding also shows that the particles
are reproducible, and are amenable to quantita- remain discrete during the dormant period. It
tive study. A
considerable amount of such study appears that where the gel-coated grains are
has been done, and it gave information about almost in contact, they are separated by a thin
the structure of fresh paste that is not only layer of water, probabh* onh* a few angstrom

579
630446 O—62 2

units thick. The constancy of the bleeding rate The lowest possible water content of the com-
signifies in particular that such thin, separating pleted sediment is higher, the higher the original
layers of water have no solidity. water content of the paste [18]. In contrast to
This last-mentioned observation seems contrary this, Steinour found that thick suspensions of
to what has been reported for some mixtures of emery would always settle to the same final vol-
clay and water [12], namely, that rigid, ice-like ume, regardless of the initial volume of the suspen-
structures develop in the small spaces between sion, provided the particles were in a nonflocculent
solid surfaces. The flocculent state in cement state [19]. Presumably, cement particles would
paste seems explicable in terms of classical theory, do the same were it not for their flocculent state.
viz., the gel-coated grains carry a "solvated The greatest possible amount of water that can
layer" and they have a positive zeta potential [13]. be lost from a paste by bleeding under the force of
The combined efi^ect of solvated layer and electro- gravity alone has been called the "bleeding ca-
static charge is such as to prevent actual contact pacity" [18]. It is a function of the initial water
between adjacent grains. But the grains are con- content, and it appears to follow a regular law.
centrated enough to experience interparticle attrac- Steinour's studies led to several approximations of
tion, at least over parts of their boundaries. The it, one of which may be stated as follows:

effects of repulsion and attraction balance at a
certain distance of separation where the potential
energy of the particles is at a minimum. The ^H'=^^[wlc-{wlc)r^\ (1)

cement grains tend to remain in "potential
troughs", which are so located as to require spaces Where AiJ' is the bleeding capacity expressed as
between the particles. settlement per unit of original height; w is the
Solvated surfaces and electrostatic charge initial weight of water; c the weight of cement;
account in part for the kind of rheological prop-
Pc is the density of the cement, and V
is the initial
erties exhibited by cement pastes. Other factors volume of the paste. The symbols K, and {w/c)ni,
are the size, shape, and concentration of particles, are empirical constants characteristic of a given
and the viscosity of the fluid. In pastes of cement, and are dependent mostly on the surface
relatively stiff consistency, effects of interparticle area of the cement; {w/c)m was interpreted as the
forces dominate, viscosity playing a minor role [14]. water cement ratio of a "base" paste in which the
At softer consistencies, w/c=0.4 and above, particles are so concentrated that bleeding cannot
cement paste is, as Reiner puts it, "... a first occur. It is evaluated by extrapolation of plotted
approximation to a Bingham body" [15]. data, and may be regarded as referring to a hypo-
thetical paste.
Sedimentation (Bleeding) The values of the constants in eq (1), for a given
cement, can be altered by any means that changes
Because cement particles remain discrete during
the state of flocculation. Increase of interparticle
the dorniant period, and because the suspending
attraction increases the base water content, and
medium is viscous, the suspension is not stable Owing, presumably, to the interparti-
vice versa.
under the pull of gravity; cement paste "bleeds".
cle water films, the base volume is larger the finer
Bleeding has been referred to as an aspect of
the cement.
coacervation [16] but experimental evidence that
it is something other than gravitational settlement
seems lacking. Permeability of Fresh Paste
The rate of bleeding remains constant
initial
for a period determined by various factors [1, 2]. Rate of bleeding is related to the coefficient of
Although the largest cement particles are a hun- permeability of the (nonsettled) paste as shown
dred or more times the size of the smallest, all in eq (2).
sizes are forced to fall at the same rate because Q=KA(pc/Pf)-l]a-e). (2)
of the flocculent state. Hence, the fluid appearing
above the solids remains clear and free from fine Q the rate of bleeding in cm/sec, or cc/cm^ sec
is
particles. If sedimentation becomes completed Ki the coefficient of permeability to a specific
is

within the dormant period, physical factors alone fluid at a specific temperature, in cm/sec; Pa and
Pf are densities of cement and fluid
determine the particle concentration in the final respectively,
sediment. When setting arrests bleeding before in g/cc; e is the volume of fluid-filled space per unit
the process is physically complete, the final sedi- volume of paste, i.e., the porosity.
ment does not have uniform concentration of The permeability of paste made with a given
cement grains. At the bottom there may be a cement depends on temperature and water con-
layer of completed "sedimentation zone", but, at tent. There is a limiting water content below
the top, w/c remains at the initial value. Thus, which the pastes aU maintain continuous floe struc-
the final composition of a specimen of cement ture during bleeding, and above which the struc-
paste may differ from the initial composition, ture becomes ruptured and "channeled bleeding"
and the specimen as a whole may become con- occurs [20]. Most of the ensuing discussion per-
siderably less homogeneous than it was at the tains to "normal" rather than "channeled"
beginning. bleeding.

580

The mean size of pores in fresh paste can be different concentrations of calcium hydro.xide
estimated from hydrauhc radius, that is, the quo- showed that as long as the floe structure was strong
tient of water content and wetted sm'face area. enough to give normal bleeding, varying the
An example is given in table 2. The figures in the degree of flocculation did not influence c [24].
final column are based on the assumption that the (The degree of flocculation does, however, strongly
section of a typical pore resembles a rectangular influence the bleeding capacity.) Steinour's data
slit [21]. At a given water content the pore size on experiments with suspensions of monodisperse
is smaller the larger the specific surface area of the emery particles showed that the change from non-
cement. flocculated to flocculated state simpl}' increased
Wi. Thus c represents the effect of flocculation
Table 2. "Pore" sizes of fresh pastes for cement having per se. Perhaps the best explanation of it is that
a surface area of 6,000 cm^/cc (1,900 cm^/g, Wagner)
it represents water in isolated pockets excluded
from the continuous floe structure, and is there-
Water Hydraulic Estimated average
cement f radius, width of pore, fore not to be considered when computing the
ratio microns microns hydraulic radius of the floe structure.
by wt.
The experimentally observed fact that the
0.25 0.44 1.25 Between 2!i and 5 modified Kozeny-Carman equation represents
.40 .56 2.12 Between 4 and 8 experimental results accurately indicates that the
.50 .61 2.60 Between 5 and 10
.60 .66 3.23 Between 6H and 13 —
product koil Wi) is a constant that can be
.70 .69 3. 70 Between TA and 15
identified with the Kozeny-Carman constant, k.
Thus
A equation for the coefficient of
theoretical k = ko{l — Wi). (6)
permeability making use of the hydraulic radius,
and based on the Kozeny-Carman concept, is Steinour's findings, expressed in terms of eq (6),
but with (l — Wi) squared since Wi was determined
by sedimentation analysis, show that for suspen-
(3) sions of chemically inert spheres, Wi is zero and
A-o(pc2)ni-W,:) (1
A;o=4.06. Such a value for spheres is in agreement
The corresponding equation for bleeding rate is
with Camian's statement to the effect that k for
fluidized spheres is equal to 4.0 [25]. If the
{e—WiY particles are irregular but chemically inert, the
iPc—Pf)g
(4) a in Wi is finite and b is zero. In this case A:<C4.06.
Obviously, if a and b are both finite, k may be
still smaller.
Symbols not already used are defined as follows:
Empirically, it was found that the bleeding
2 is the specific sm-face of the cement grains in
rates of pastes conformed to eq (7) [26].
cmVg, as determined by a suitable permeability
method; is the viscosity of the fluid in poises.
77

Theoretically, Wi is related to a fraction of the (7)
fluid that remains with the particles during flow. dviPcSy (l-e)
As shown by Steinour's work [22], the term In this equation s is the specific surface area as
(1— Wi) of eq (4) must be squared when a specific
determined according to ASTJM designation C
surface value is used that is based on sedimenta-
115-58. The constant 5 is correct only if this
tion analysis and calculated by Stokes' law in the
measure of surface area is used. The Wt factor
usual way, which does not take account of fluid
must be evaluated empirically for each different
that accompanies the particles. The Wi seems cement by methods that have been described [27].
to be composed of three elements, that is.
Steinour showed that variation in w, among differ-
ent cements was due principally to differences in
(5) specific surface and in the initial chemical reactions
l+a+62 with water [28].
where a(l — e) is a quantity of fluid held stagnant Capillary Forces in Cement Paste
on the irregular grains, and 62(1 — e) is a volume
added to the original cement grains by the initial The bleeding rate of concrete is of the same
chemical reaction. The terms a and 62 therefore order of magnitude as the natural rate of evapora-
occru" because of an augmentation of the solids, tion from an open surface. Sun and wind together
and a diminution of the fluid, so far as flow is often cause the rate of evaporation to exceed the
concerned. The c in Wi is zero for nonflocculated rate of bleeding. When this occurs, the surface
suspensions, but always appears when a thick loses its sheen, signif3Hng that the plane surface of
suspension becomes flocculated. For spherical water has become replaced by myriad curved
particles, c =
0.1, approximately [23], and a and 6 surfaces, and this signifies that capillary tension
are zero. For crushed glass, b is zero and is has been produced. Carman has shown that the
approximately 0.18. Experiments with pulver- maximum possible capillary tension in a water-
ized silica in suspensions of lime water with filled granular bed occurs just as the water level

581

drops below the upper boundary of the bed, and at the top of the bed. Since the downward force
the meniscuses take on the greatest possible on those particles due to gravitation only only
is
curvature, that being limited by the sizes and about 0.001 atm, it is clear that when the rate of
shapes of the spaces between particles at the top evaporation exceeds the normal rate of bleeding
surface [29], The approximate value of maxi- the force causing subsidence of the surface may be
mum capillary tension can be calculated from the greatly increased. The effects of such an increase
following relationship adapted from the basic have been measured under controlled conditions
relationships given by Carman. by KUeger [30].
When resistance to consolidation becomes high
_Xap^ enough to offset capillary tension, the water
(8)'
w/c surfaces retreat to the interior, and lateral con-
solidation begins. The result is the so-called
where X is the surface tension of water in dynes/cm, "plastic shrinkage", often giving rise to "plastic-
a is the specific surface area of the cement in shrinkage cracking". Swayze has expressed the
cm^/cc, and Pc is capillary pressure, a negative
view that a significant and technically important
quantity. Substitution of appropriate figures into
degree of compaction of fresh concrete can be
eq (8) shows that maximum capillary tension will
range from about ){ atm upwards depending on produced by capillary tensions that may develop
the surface area of the cement. throughout a period of several hours, and he
The reaction to capillary tension is, of course, proposed a procedure for taking advantage of
an equal downward force exerted on the particles it [3].

3. The Specific Volume of Hydrated Cement and the Porosity
of Hardened Cement Paste
Definition of Pore and Solid The porosity of a given specimen is determined
to a small but significant degree by the balance
is undisputed that porosity is a basic property
It between opposing internal forces. To introduce a
of hardened cement paste, but there may be fluid into such a system of particles releases
disagreement as to the definition of pores. In the energy and alters the balance of forces. The
studies now being reviewed, pores are defined as surfaces change their relative positions, and the
spaces that can be occupied by water that is porosity changes too; swelling occurs. (See sec-
evaporable at a constant low external humidity, tion 7). Even the specific gravity of the solid
at a given temperature.^ A change of either the material may change slightly, owing to the change
specified temperature or humidity results in a in interfacial free energy. The quantity of fluid
corresponding change in the apparent ratio of that can be accommodated in interstitial spaces
solid to void. Thus, the choice of drying condi- thus depends partly on the amount of swelling,
tions defines the pore space and the solids, and accompanying entry of the fluid. It may depend,
does not assuredly isolate, or precisely establish, too, on how the molecules of the fluid "fit" the
the true extent of the solid phases. atomic roughness of the surfaces, and on the size
of molecule relative to the size of the smallest
Problems of Mensuration interstitial spaces. Swelling produced by water
is much greater than that produced by organic
Besides the difficulty of isolating the solid
liquids or inert gases, water molecules being
phases, except on an arbitrary basis, the quantita-
relatively small and strongly attracted by the
tive, accurate measurement of solid volume also
surface.
presents difficult problems. Hydrated cement is
Fluids that affect interparticle forces are them-
mostly colloidal (see glossary) and of the total
selves affected; they are adsorbed. Possibly,
number of molecules composing the colloidal part, strongly adsorbed water molecules penetrate the
perhaps one-half to two-thirds of them are exposed transition region of the solid more than weakly
at surfaces. Seen on an atomic scale, the surfaces absorbed molecules do. It is frequently suggested
of the particles, i.e., the boundaries of the pores,
that some of the space found by sorbed water
are transition zones of vibrating atoms, and thus
molecules is inside the crystals, but this seems
the word surface as applied here does not connote Whether or
doubtful, as will be seen further on.
the conventional geometric concept; the boundary
not these phenomena occur, it is probable that the
is neither sharp, smooth, nor static. In regions average amount of space required per molecule of
of minimum porosity (see sections 5 and 6) the
water in the adsorbed state is different from that
average distance between surfaces is about 5
required in the liquid state, which is to say that
oxygen-atom diameters, and in these places the
the average density of water in the adsorbed state
rough transition region might constitute a signif- may differ from that in the liquid state. Powers
icant fraction of the interparticle space.
and Brownyard [32] estimated the specific volume
of gel water, most of which is strongly adsorbed, at
2 This definition is not intended to include entrained air bubbles and the
like, which are regarded as cavities rather than integral parts of paste. 0.90 cc/g. But this estimate was based on density
582

of hydrated cement as determined by displacement its displacement of helium, and c the cement
in helium, and therefore it may not be correct for content of the sample, grams. The mean value
water, as was acknowledged at the time. Thus, obtained for four different portland cements was
although it is easy to determine the mass of water w„ =
0.82. The individual values ranged from
required to saturate the spaces in cement paste, it 0.81 to 0.83.
is impossible at present to determine a correspond- An advantage of eqs (11) and (12) is that they
ing volume exactly. permit evaluation of Vhc from gas displacement
data on samples of paste that contain unhydrated
Experimental Results cement.
Weir, Hunt, and Blaine [35] evaluated y„
A direct approach to the problem of measuring from experiments based on eq (13).
porosity was based on the following expressions:
w„/c
= Vc+{Vn—Vc) (13)
-m + m(l+w» —
1^s

(9)
]
In this equation, Vs is the specific volume of the
solid material in the sample, cc/g, including un-
m hydrated cement, if any. Small cylinders (ji x
6=1- (10) 4 in) of cement paste were cured for various
1 + periods and then dried by the "dry-ice" method
of Copeland and Hayes [36]. Then the speci-
e is the porosity; m
is the maturity factor (fraction mens were immersed in a light petroleum distillate,
of the original cement that has become hydrated) and the distillate and the specimen immersed
in it were subjected to pressures up to 10,000 atm,
Vnc is the specific volume of hydrated cement in
cc/g; vfn is the nonevaporable water content of the primary pm-pose of the experiment being to
completely hydrated cement in grams; Wq is the determine the compressibility of the solid phases.
original water content in grams, corrected for After release of pressure, the amount of distillate
water displaced by bleeding; c is the original remaining in the specimen at atmospheric pressm^e
cement content in grams, and is the specific was determined. Also the overall volume of the
volume of the original cement in cc/g. distillate-satm-ated specimen was measured by
All the factors in eqs (9) and (10) are subject displacement. The resulting values of were
to direct measurement, but with attendant dif- plotted against (w„/c)/(l -}-w„/c), producing a
ficulties, especially for Vh^- In atteinpts to straight line having an intercept at Vs Vc. =
The
measure Vy^c, various liquids were used as displace- slope of the line and the corresponding value of
ment media with various results [33]. For Vc gives, for a portland cement, y„ =
0.87, for an
example, water, acetone, toluene, and helium =
aluminous cement, y„ 0.75. Comparing the re-
gave the following values for v^c, based on the sult for portland cement with that reported by
normal density of each fluid: 0.395, 0.408, 0.429, Powers and Brownyard, 0.82, Weir, Hunt, and
and 0.424. These results from prehminary ex- Blaine mentioned that their liigher result might
periments were followed by a considerable amount be due to the presence of a little evaporable water
of work with helium. The displacement in in their specimens, because of incomplete drjnng.
helium of samples prepared by the magnesium However, calculation shows that even if the
perchlorate method was finally reported, from specific volume of the residue of evaporable water
work by Steinour [34], as 0.41 ±0.01 cc/g for in their specimens was as high as 1.0, the amount
four different cements. Since the four cements would have had to be about 40 percent of w„, to
were chemically different, and showed similarity account for the difference in question. Therefore,
in specific volume of hydration products, it was the data seem to indicate that helium penetrates
assumed that for any cement the approximate regions not accessible to the distillate molecules.
density of hydrated cement might be estimated The problem of determining the specific volume
from the relationship given in eq (11). of hydrated cement and the specific volume of
evaporable water has been approached by con-
sidering a saturated paste to be composed of two
(11) components: (1) the solid material and (2) the
evaporable water [37]. From a normal procedure
where is the apparent specific volume of the of physical chemistry, it foUows that.
nonevaporable water, cc/g. The values of w„ for
the four cements were calculated from their v'j, = N'^cVnc + NeV, (14)
hehum displacements by means of eq (12).
in which v'p is the specific volume of the paste, cor-
rected for any residue of anlwdrous cement; N'^c
(12)
is the weight fraction of hydrated cement; A^^' is
the corresponding weight fraction of evaporable
Vs is the volume of the dry sample, indicated by water; tJ^,. andwe are the "partial specific volumes"

583

of the solid material and evaporable water respec- this assumption one would write in place of eq
tively. For example, (14),

v'p = N',,V,c + N'aVa+{N:-N'a)v^ (17)

where A^^ is the weight fraction of adsorbed water
where V is the volume of the paste and We is the that stands in constant ratio to N'^^, Va is its partial
weight fraction of evaporable water. As used in specific volume and is the specific volume of
the present case, Ve pertains to the evaporable free water. Expressing the proportionality be-
water in a saturated specimen. tween adsorbed water and the solid material as
Among the many test data, the values of A^^ N'a = bN'f,c, and thereby eliminating A^^ from eq
varied widely. When the values of specific volume we obtain
(17),
of the specimen, v'p cc/g, were plotted against corre-
sponding values of N'^, the points conformed
v'p=NWvhc+iiva-v^)] + N',Vy,. (18)
closely to the straight line represented by the
following equation. On comparing eq (18) with eq (16), we see that
z;; = 0.398iVL+0.99A^;. (16)
+
«;„=0.99 and Vi,c b{Va—Vy,) =0.398. Therefore, the
results obtained from this analysis were:
The linearity of the plot signifies either of two
Equation (16) seems to indicate that 0.398 is the conclusions: (A) All the evaporable water has a
specific volume of hydrated cement, and 0.99 is
specific volume of 0.99 cc/g (for these particular
that of the evaporable water. However, a linear specimens) whether adsorbed strongly, weakly, or
relationship such as eq (16) is characteristic of =
not at all, in which case 0.398, or (B) Some
physical mixtures of two components that do not
of the evaporable water in every specimen has a
interact either chemically or physically when
specific volume of 0.99, and the rest, an amount
they are brought together. In the present case it proportional to the amount of hydrated cement,
is known that when dry hydrated cement and might have a different specific volume, in which
water are brought together the energy content case the specific volume of the hydrated cement is
of the system decreases, as shown by evolution of
not 0.398 cc/g. Conclusion (B) is probably the
heat, the amount of heat released being over 20
correct one.
cal/g of hydrated cement [38]. In view of this, it Another analysis [37] was based on the empirical
does not seem likely that the specific volume of relationship
either component would be the same after mixing — 0.25
t;(=0.99 wjwt (19)
as it was before. The following considerations
seem to reconcile the result indicated by eq (16) where v, is the apparent specific volume of all
and the observations just mentioned. the water in saturated specimens, including w„.
It is known that at humidities up to about 45
It would appear that at Wn/wt—l.O, 'Ot=Vn =
percent, the amount of water held in the adsorbed
0.74, for,if eq (19) is valid over its entire range,
state is proportional to the amount of hydrated
the 0.25 is Vg—Vn- However, on the basis of the
cement in the sample, and is independent of the same observations and reasoning presented in
total capacity for evaporable water. The amount
connection with eqs (16), (17), and (18), (w^—
adsorbed at 45 percent humidity is a little less than
y„)w„ may be replaced by {l-{-B)(Vu,— Va)Wn,
two molecular layers on the solid surface. This is where B is the weight fraction of evaporable
the strongly adsorbed part of the evaporable water,
water that stands in constant ratio to Wn, and
as shown by its relatively high heat of adsorption.
which may have a specific volume different from
The rest comprises weakly adsorbed water and, if that of the rest of the evaporable water, and
any, water free from the effects of adsorption.
Va is the mean of the specific volumes of the
The fraction of the total evaporable water in excess nonevaporable water and the part of the evapora-
of two molecular layers is greater the higher the
ble water whose density is altered by adsorption.
porosity of the paste, and thus that part does not
Then
bear a constant ratio to the amount of hydrated = V^—(l-\-B)(Vto—Va){Wn/Wi).
V, (20)
cement.
It follows that if a specimen of saturated cement
On comparing eqs (19) and (20) it is evident
paste is to be treated as a two-component system,
0 25
it is not permissible, a priori, to identify hydrated that (l+B)(0M-Va)=0.25, or, Va=OM-~^-
cement as one of the components, and total evap-
orable water as the other, because the properties This result presents the same impasse as eq (18).
of some of the evaporable water are certainly not It cannot be solved unless the fraction, B, of
independent of the other component, hydrated evaporable water having density different from
cement, although the properties of the rest of the that of the rest is known. Indeed, even if B were
water might be. It seems therefore that the sim- known, the apparent volume of the water in the
plest permissible assumption is that saturated ce- solid phase, f„, would still be unknown unless the
ment paste is made up of three components: hy- mean density of the adsorbed water were known
drated cement, adsorbed water, free water. On also. Only if -it is arbitrarily assumed that B=0
584

Values probably are valid for that fluid. The entropy change on is the lowest possible value for adsorption of the strongly adsorbed part [50] is Table 3 is a summary of the findings in terms of such as to suggest a considerable change of state experimental and calculated results for four dif.174 . Table 3. they do not = reported under v„ 0. the state of this subject is unsatis- which is to say that the average density of ad. Surface Area of Hydrated Cement and Indicated Size of Primary Particles Surface Area From Water. is excluded from some spaces accessible to water. Specific volume of hydrated cement !'). All the others in the last three columns are on the dry ice basis.„=0. Cc/g Cement No. of cement gel to helium is 7 percent greater than There appears to be little theoretical or experi. All the rest in this column are for samples prepared by the dry ice method. 397 .319 0.8 2. the porosity to the distillate. For general purposes it seems that the sorbed water is the same as that of the aqueous helium values give the best estimate of v. indicates either that helium from the values of Vn at the heads of the columns.74 !)„=0.2 2. It seems likely that ferent cements.Vapor Adsorption Determination of surface area of hydrated ce- ment is based on water-vapor adsorption inter- From work reported in 1946 [40] it was deduced preted in terms of the Brunauer-Einmett-Teller that the specific surface area of the solids in ma. see section 5) is about 0. theory (BET) [41].4 6. and medium.7 4. Values reported under' out that the specific volume of gel-water (We = 0.386 .easurements.74. applied to different cements on the as.394 15497 60. If first line corresponds to direct determinations by there is a change. 421 15622 49. do we obtain w„=0. in helium.9 10. tillate. it comes volume of 0.74 are based = or that when water is used as a displacement on displacement in water (aqueous solution).7 .246 » 0. cc/g c CsS C2S C3A CAF CaS04 i.0 27.411 0. or that are the smallest possible. Specific volume as given in the such a change involves a change of density. mental support for an assumption that (.210 . Assuming that the that all parts of evaporable water have a specific difference is due to densification only.394 15699 33. Subsequent work calls for an upward revision of this figure for surface area.87 represent displacements in petroleum dis.87 15754 45. and thus that Va=v„. = S. entailing the assumption both factors are involved. which is 430 mVcc.8 3. it is densified by adsorption. factory.418 b 15754 45.1 11.„=0. and the same value of v„ is assumed to Calculation on that basis shows that the porosity apply to specimens prepared by either of the two methods.408 . 0. where She is the specific surface area of liydrated 585 .417 » This figure is based on samples prepared by the magnesium perchlorate method.99 cc/g. 3. and are subject to limitations mentioned in the text.82. On the whole.394 .3 7.3 6.1 .5 4.90 cc/g.311 . The rest of the water.SOO VJc Snc (21) or about 180 m7g of dry paste.0 0.7 13.0 54.. Values in the column headed ?j„ 0. from that of free liquid.e=speoific volume of hydrated cement »c= specific volume of original cement i'n=apparent specific volume of nonevaporable water V}° —7 =grams of nonevaporable water per g cement at complete hydration Computed composition— w° Vu.4 6. an increase seems probable.2 28.82 !)„=0.4 12.c- 4.0 . helium displacement of samples prepared by the in view of the openness of the structure of liquid magnesium perchlorate method. .0 27. This solution in fresh paste.374 . Although the figures for petroleum distillate sumption that Vn is the same for all.322 .227 0.82 represent displacements seem to indicate complete penetration of pores.74.9 3.310 .319 . The value of Vn obtained from helium values in the last three columns were calculated m.374 .386 .7 4. A convenient formula is ture cement paste is equivalent to that of a sphere having a diameter of 140 A.176 .0 Vm.7 13. b This figure is on the magnesium perchlorate basis.

and (U™/w„)i 0. (22) sium perchlorate method.800 corresponds to that figure. as just uncertainty as to the relations between observed discussed. With these figures. The relationship between the values of Within the range of these data. that has become hydrated.002. in terms of ratios. = 15754) experimental values were (w„/ym)2 3. and ('m. water molecule on this basis gave 11. 354 {w„) iw„) 2] . the Copeland and Hayes method. where c is the original weight of cement and Cn is ture of the experiment.= . Hunt. Or. Powers and Brownyard [42] used to the correct dr3dng condition would result in a a lower value. This value. More- and Blaine. since the cements were tion of drying.cement. is the same as that reported by The graph of these data indicates that experi.570. the difference between the observed and the theoretically correct V^. The first factor depends Theoretically is the weight of water required not only on the size of the water molecule but to cover the surface of solid material with a layer also on the structure and composition of the 1 molecule thick. the drying conditions used for the experiments of prises Avogadro's number. Actually. there is and Copeland [43] concluded that for adsorption reason to assume that the value of Vm obtained by of water on tobermorite gel the best value is 11. the most rigorous that might be devised.8 A^ [45]. Thus. 3. Kantro. Present values of shortest period giving a value of w„ about 25 Vm are correspondingly different from the old percent greater than that given by the longest.6 to 12. The proportionality between Vm and surface Discussion of Eq (21) area depends on two factors: (a) average area covered by an adsorbed water molecule. let subscript 1 denote values obtained by the magne- iVm) 1.258 ±0.2_58.084. This destroy the practical value of the data. and the area covered by a single adsorbed cally correct one.e. It thus seems clear that Vm must be values of Vm and w„ and the theoretically correct "calibrated" for a given material and drying values. and (b) BET Surface Factor. the molecular weight eq (22) are probably not very far from the theoreti- of water.904 (UJ2. The calibration involves establishing of Vm the process of adsorption should begin with a value for molecular area. However. there is no way to assess the accuracy of Tomes. Tomes. as is shown method of establishing the molecular area is not especially well by study reported by Tomes.iV^).1 days.n and w„. other than by the degree to which the influence of various degrees of drying of test computed areas contribute to the internal con- samples on the experimentally determined values sistency of various related data. Hunt.23. m of Vrn and the correct value is proportional to the difference between the observed nonevaporable This factor is the weight fraction of cement water content and the correct one. For a particular cement (No. Even if the ultimate values of and w„ for a given procedure of drying are obtained. if the arbitrary drying not exactly alike. The numerical coefficient com. and w„ is water that is a com. The first estimate of the area of the arbitrarily established drying procedure. Powers and Brownyard [46] from direct experiment.6 AVniolecule. be somewhat fortuitous. It is necessary to depend on an of afwillite. Hunt. procedure is too short to remove all the evaporable water. is that which gave the same surface area with heat to assure this initial condition cannot be by water-vapor adsorption as was obtained by followed because the solid tends to decompose nitrogen adsorption on a laboratory preparation when heated.0. i.4 A^ [43]. Recently Brunauer.4. Thus. and Blaine empirical eq (23). To obtain the theoretically correct value condition. and Vm now obtained and those formerly obtained Blaine found that may be shown as follows: Using eq (23). The value now used. inasmuch as the weight of that part which has become hydrated. The exactness of agreement may even to the ultimate values for the given condi.. by Copeland and Hayes [36] they tested samples The values of w„ formerly obtained were about of the same material after seven different periods 8 percent higher than those obtained now by the of drying ranging from 2. and 2 those obtained by the dry ice method. the arbitrarily chosen drying procedure is not The value 3. based on the Tomes. 586 . over.354 -f^(^-l> (23) (23) gives (UJi = = 0. there is some factor depends on drying conditions. value for the numerical constant. The practice of outgassing 11. it may be m = Cnlc (24) that the humidity maintained in the desiccator is too high to produce a bare surface at the tempera. the resulting uncertainty is not such as to This was later revised to 11. the difference between the observed value Maturity Factor. mental data would conform to eqs (22) and (23) 0. it may be expected that a change water molecule. eq ^^=1-0. Using the dry ice method described the results now appear very good. surface on which it is adsorbed.„)i/(w„)2= 1. m^/g. Never. a "bare" surface. The second ponent of the solid phase. On this score. Hunt. theless. and Blaine [44] investigated the the result. very much different from the correct value. of V. values.4 A^. based on a molecular area of relationship like eq (22) with a slightly different 10.

=3800r^^1^. and is (Scoi denotes surface area of the colloidal part of B-8-3 0. .335 .210 (CH) the weight of calcium hydroxide. as described by 290 . and for C2S [43]. 45-0.174 area of the hydrated cement remains practically constant at a value given by the following equa- B-8-3 0. Cement 15758. . 100 .22 600 11 yr 6 1 .355 Correction for Calcium Hydroxide Clinker 15670.1506 .200 . 439 210 296 3 Possibly there are other noncoUoidal components. and the colloidal part of cejnent gels ^m/w„ were calculated from the data in their Sic =surface area of hydrated cement.1525 0.1328 . Brunauer. Weighted average. (CH)/w„=1.36. C!. and from the assumption that in colloidal tobermorite the ratio of CaO to Si02 is exactly Computed com- 3/2.1593 347 tion: B-15 B-20-1 — . we cannot Bulletin 86 100 . 587 .294 determined by X-ray analysis.23 11 yr . In general.55 .2196 .349 5'. the = Copeland on the hydration products of C3S and Table 5.5 11. Data used for computing surface area of hydrated cement and cement gel For most cements. m^/g table 3. Cement 15763. Data obtained from four different cements are 252 m?lg. 4 0. (CH)/m)„=1.18 219 267 10 to 20 percent lower than that of pure colloidal 15756 48. which has B-15 .9 4. (CH)/!c„=0. w°/c=0. Calculated surface areas for cement gels.2 10. and having a diameter of 6/(5.36 200 240 tobermorite. 65 315-394.55 513 600. 6 1 .858. a comparison of these data with those in the c Wn Wn m2/g mVg C3S C2S C3A C4AF upper part of the table it appears that the specific surface of the colloidal part of cement gel is from 15754 45..6 2. 654 6 1 .311 1.287 Vm/Wn remain practically constant throughout the Weighted average .5 27.CH) No. and the results of applying eqs (25) and (26) are given in the first four lines of Now that more is known about the morphology of gel particles.) When the B-20-1 .413 Clinker 15498.1900 .312 290 .160 . but if so. Neverthe- less. B-20-1 290 . 45-.452 negligible surface area.) Table 4.8 13.286 Copeland and Bragg [47]. 4 0. Cement 15754.458 L 344 279 296 measure them at present. it is perhaps pointless to express table b. cmVg dry weight.3 6. The values for S^c are experimental.35 511 1 . {CH) may be B-15 B-20-1 . On this basis the colloids in cement gel correspond to a three-cell thickness. CUnker 16367.35 505 1 .6 12.35 ..227 0.(This is not the weight of the hydrated cement.276 cement gel.227 (The degree mark indicates that the quantity is B-8-3 0.1870 .1457 .6 57.. the following relationships may be used Reference Number for any intermediate stage of hydration after the No.407 .170 . The values of w°/c.1210 .355 1. Bulletin 86./c=wJwl= VJV°=AH/AH°. 0 8 .70 81-541 10 0. w'=Jc=.23 11 yr 1 .394 . Kantro.2115 0. m)°/c=0. 7 0. and for a thickness of three cells.35 .210 ..858 1.170 To obtain the surface area of the colloidal part of cement gel (see glossary) it is necessary to make B-8-3 0. {CIT)lWn.292 0. This means also that the specific siu-face CUnker 15623.317 for completely hydrated cement. and Copeland 15763 15758 — 28.25 227 193 255 235 suggested that the sheets and ribbons of colloidal Average 210 249 tobermorite may be two or three unit cells thick. (CH)/w„=1. it may : of cement gel is not affected very much by differ- ences in chemical composition of cement.23 11 yr 1 . Kantro. . From Reference position.174 . 65 98-224 8 0.2102 .25. 9 . 65 80-543 70 .292 Experimental Results The calculated specific surface of a sheet two cells Mature Pastes thick is 377 mVg.424 Weighted average .^ Eq (26) may be used.304 B-15 .311 period of hydration at values characteristic of the cement. (25) 290 .2147 0.3 6..3 60. apparently all but ASTM Type IV.1711 . Age.378 Weighted average _ .55 513-.1570 above equalities exist it follows that AH/Wn and 1 .1 X 10^) 118 A. Cement 15756.18. the results are the same as those reported previously [48] The specific surface size of particles in terms of spheres.. . w°/c=0. 7. 50-. for comparison with the earlier figm-es. % Vm (..0 25.413 .65 259-300. be noted that a specific sm*face of 210 m^/g. or The last two lines of table 5 contain data about 510 in? Ice corresponds to that of a sphere derived from the data of Brunauer.1337 . given in table 4..409 a correction for calcium hydroxide. 40.1693 0. 50. days of values averaged Wn first few hours.277 2.

Su is the same at all stages of hydra. of important. d=<j>8. cement. The same kind of calculation for the adsorption. for a considerable edifice of vapor adsorption. Effect of curing temperature on. The suggestion that gel water might be hydrate water is not easy to reconcile with the results of Table 6. some of the hydrous calcium silicates. 1956) Particle Size From Permeability to Water Temperature Surface Pressure {iu„lc)s area. However. The degree of constancy of to question the assumption that cement gel pre- sents a like situation. For a Type I cement amount of experimental evidence that lattice cured 7 days." [52] On the other maturity factor m = w„/w°. (Ludwig and Pence.162 1 300 psi 93 200 . the results shown in table 6 were penetration does not occur. gave d=95 A after correction reasoning about this material can rest on the ac. From a less pragmatic view. Effect of Curing Temperature that effects that might be attributed to lattice Ludwig and Pence [51] reported specific surface penetration are easily accounted for without pos- as measured by water vapor for pastes cured at tulating such penetration. d. and to the degree that hand.g.. these figures gave the value for sphericity factor 588 . mVg of Powers. zur Strassen [55]) and thus is naturally to Type IV compositions for which the ratio of present in fixed ratio to the amount of hydrated VJWn changes appreciably during the course of hydration. fictitious. The validity of the equation space in cement gel. Kalousek wrote as colloidal part only gives a specific-surface diameter follows: "The very large 'areas' of cement pastes of 98 A. it has been suggested VJWn has been reported [49]. the specific surface of the hydra. Miss Moore [54] and V-nlWu remains constant during hydration of a probably others.142 103 Navier law. average. Such variations boundary indicated by evaporable water does not as there are in the specific surface values of the coincide with the boundaries of the colloidal crys- successive increments of hydration products are tals.=Wnl'w°n is discussed in other terms in another have a boundary. is 92 A. That is.. For example. and that such space must m. and obtained a figure for the size of atm 66 150 . therefore. and Copeland [56] studied the dry paste flow of water through specimens of saturated °C °F paste in terms of an equation based on the Stokes- 1 atm 27 80 0. layer penetration in certain types of crystals. 5. There is a considerable elevated temperatures. even for those composi. Does Water Vapor Adsorption Give Real 5 is the diameter of a sphere having the same Surface Area? volume as an average particle in the specimen. The figure for cement 15754. The Intermediate Stages of Hydration areas computed from total amounts [of water up- As already indicated. McConneU [53]. The measured between them. surface area of permeability tests discussed below and in sec- hydrated cement tion 10. for that part of the evaporable water a sphere having the same specific surface as a that is called adsorbed has certain important typical particle in the specimen to the diameter physical effects on the system as it enters or leaves 5 of a sphere having the same volume as the whether it occupies space in layered crystals or typical particle. in relation to the reach of interparticle attraction.. while accepting the fact of inter- given cement. However. Mann._ 160 320 139 9 "volume" diameter. From one point of view. lattice penetration does not occur is so small. Thus. by water question is important. and it is difficult to assess their meaning. except with respect that gel water may be water of crystalhzation (e. there appears to be no compelling probably insignificant in connection with most need to postulate that this possibility is the actual considerations involving the internal surface area case.139 30 droxide) in terms of a sphericity factor and a</> 300 psi . perhaps as it does in zeolites or certain clays . face area as it has been measured by water vapor yard. the cement obtained by water-vapor adsorption are not true from which a great deal of reported data were areas. Also. The value for mean distance between par- ticle surfaces obtained from the assumption that of cement paste. <A'/*5 = 201 A. the take] are. The result was . the specific surface of the hydrated cement. That early report has been verified by later work. see H. for volume of calcium hydroxide. seem tion (eq (25)). Water apparently enters the lattice of obtained."specific surface diameter". obtained. Combining cepted answer. the question is not very and 4) is defined as the ratio of the diameter. as will be seen below. to the degree that clc..152 123 75 the primary particles (exclusive of calcium hy- 300 psi 127 260 . There is a great deal of experimental tions. This figure is to be There are reasons to question the reality of sur- compared with 140 A given by Powers and Brown.. but it is possible that the paper of the Symposium [50]. One tion products is practically constant so far as this may observe that evaporable water does occupy factor is concerned. all the ratios fall within 10 percent of the data of diverse kinds bearing on these questions.

and land cement. Hunt. Kantro. The corresponding Brunauer. there is also a scattering by refraction due to a very small amount of gel produced by at the surfaces of the particles that does not reaction with moisture from the au". and it is and a little less after 6 months of liydration. about having a specific surface of about 15 m^/g showed 40. This Valis found the surface area of neat cement.4 A^ and water vapor adsorption.25. c is Surface Area by Nitrogen Adsorption about 40 X 10~* cm. ( (27) revised data respectively. the length b. approxi. 6 is the angle of scatter. For example. when dry. surface of the particles composing the sample is or from the revised datum. and the thickness c.37 and 5 = 258 A. adsorption [45]. for the 3-cell thickness. effect is confined to angles less than about 5° = w/c 0. thus could indicate respectively.ch'ljh'Idh. of surface area. there are not as was pointed out by Brunauer. material represented by the first line in table 5. This agreement indicates that water specific suriace=-i— vapor gives a real measure. with 219. In various other experi- As shown in section 10. of 5 is 9. or 500 m7cc. for which the molecular / the intensity.c=€V. the permeability data ments made in this laboratory on hyclrated port- seem extraordinarily amenable to analysis. In 589 . Emmett and DeWitt [60] reported that the According to Copeland [57] a collimated beam indicated surface area of anhydrous cement is of monochromatic X-rays passing through' a about 10 m^/g by nitrogen adsorption. If the sheets are 3 cells thick. indicated by permeability test the adsorbates were taken at 16.000 molecules and being about 3 molecules the same surface area by nitrogen as by water thick. If the width is a. Surface area indicated by nitrogen Resistance to flow through a saturated specimen adsorption for the colloidal hydration products of of paste seems to be developed by the same C3S was only about 50 percent of that indicated particles which. In section 10 a discussion therefore referred to as low angle scattering.2 and 11. The specific number of molecules per particle would be 49. given by the following relationship: To illustrate the significance of the value found for sphericity factor.52 for the 4-cell thickness. that lattice penetration occurs. and enough X-ray data to make this evidence Copeland [43]. The same kind of the figures for particle volume.~cured 1 week. (t> intensity as a function of scatter-angle as shown The volume per particle indicated by the value in eq (27). for neat cement paste. 34. But when C2S and water react Perhaps the most significant aspect of the in the form of paste. The result was 197 m^g.5. adsorb water vapor. However. by water adsorption. The difference is probably due to the particles composing the sample are of colloidal molecular roughness of the surface and possibl}^ dimensions. let us consider a S.41. The indicated results were reported by Blaine and Valis [59] and particle thickness and number of molecules per by Tomes. culminating in a revision of these figures is given. The specific surface was about 220 that the hydration products consist principally of m^/g. adsorption. adsorption to be the same as that by water = and <^)= 120/233 0.6 XIO'^ A^ for the first and (4 tt/X) sin 6»)= constant//. the specific surface area of results from permeability studies is the evidence the product indicated by nitrogen was found to be of compatibility between the deductions about only 21 percent of that indicated by water vapor particle size from two independent methods.<^=0. a laboratory preparation of afwillite thin bodies containing. but if the methods. and X is volume is approximately 195 the average is the wavelength of the beam. and a fairly accurate aoc c one. left-hand side of eq (27) the particles are thin sheets such as are found in Copeland applied method to a sample of the this the hydration product of C3S. particle seem too large to support a conclusion in the latter report. to be 10 mVg after 1 day of hydration from the direction of the primary beam. (28) hypothetical gel in which the average volume per particle corresponds to 5 = 233 A. and water to react in a ball mill and found the The sphericity factor for the hypothetical sheet specific surface area of the product by nitrogen is <^)=90/233= 0. and Weise [58] caused C2S specific-surface diameters are 90 and 120 A. provided that the molecular areas of = The value <^ 0. a value granular sample is refracted by crystal lattice much greater than that found by conventional planes according to the Bragg law. the figure obtained by water vapor „ 2ab 2 adsorption. on the average. Kantro. c is about 30X10"^ cm. or if 4 cells thick.This is to be compared mately. then. The scattering due to particle surfaces varies in The revised data give = OAl and 5 = 233. Also. and assume that The is represented by h. the nitrogen surface area Specific Surface by X-ray Scattering was 22 percent of the water vapor surface area. and Blaine [44]. the surface areas indicated hy nitro- the necessary assumptions made in the analysis gen were substantially smaller than those indicated do not seem to offer enough leeway to invalidate by water vapor adsorption. and we obtain specific sur- face=667 mVg. adsorption.000. Blaine and depend on particle shape or crystallinity. w/c=0.0X10'^ or 6.000. conclusive.39. If the material were tobermorite gel only.

6. 'M. I considered addi- will eventually become hydrated. The general relationship period. But. the porosity of the paste is lower only because of the presence of un.1 had been reached [63]. groms is from a sample prepared by the magnesium perchlorate method./c=Wo/c+0. and the corresponding the cement content of a paste is below a cer- If porosity as 26 percent.2 Xv X X\X Z . must obtain from an outside source during the The first estimate of minimum porosity of the course of hydration in order to remain in the hydration products in cement paste was based on saturated state. Minimum Porosity of Hydrated Cement Paste. it is water content. -B Although these data were obtained mostly from specimens that had been water cured about 6 months. since that method produces an ambient vapor pressure water contents for pastes made with a cement for which of about li II of mercury.59/ is the nonevaporable water content as determined — 0. with the remain unhydrated. and Specific Volume of Cement Gel Minimum Porosity gave We/w„=0. however. grams original water. When the cement content is at the limit it represents is developed from for complete hydration.45//\/y x^ by the magnesium perchlorate method.6 specific volume measurements. and the difficult to say what feature of paste structure conclusion has been drawn that the gel particles controls the uptake of nitrogen by a sample. the porosity of the hydrated hand side is the amount of water that a specimen part remains the same. The second term of the right- hydrated cement. but in this paper absence of subscript has that significance. these data being on the tain limit./(y„)8=3.38 as minimum evaporable water contents. 5. the porosity of a com- pletely hydi^ated specimen is at a minimum. which method subjects the samples to an ambient vapor pressure of about 8 m of mercury. they evaporable water. a straight line is obtained. if the tional data on minimum porosity as indicated by cement content is above that limit. Hunt [61] has observed proceeds." that. a curve is obtained which gets steeper with specific surface area seems to increase with the age. Wg being the weight of evaporable water. where (Wn)» y/y /VrfP 0.4 0. where m=wjw°. Formerly..* Another estimate was based on 0. all the cement In the course of this review. as determined 0.40/ X\/x X'sf computed corresponding figures for specimens 0.75/ xfv Xvv ^ on the surface of the solid particles.92. At w. and iVm)s the weight required to form a monomolecular layer .254:mw^/c (29) any higher cement content. the nitrogen ratio. or We/Fm = 2. additional studies were carried out later by Copeland and 0 D J Z 3 4 5 .08.3 03oX/\XX /\/ and We/Vm=S. and original by the dry ice method. wllc = 0. and by nitrogen are getting smaller as hydration then ceases to increase.227. It produced at all stages of hydration are the same is clear. and capacity for evaporable water [62]. water and at normal temperature.7o// X/Yxf^.8 0. The . Interrelation of total. . and the helium porosity is about 30 percent. evaporable.6 . Using the lowest value found in a c group of three dense specimens 11 yr old. or. 590 .35//\Xx^XLr prepared by the dry ice method aye We/Wn=1. If one makes a similar plot of nitrogen the uptake of nitrogen vapor are not those that surface areas. n = total water.6o//W)/ x/y .55/ X/x as We/(Wn)s=0.74. that the features that govern size. groms % = non-evaporable vvoter.7 on samples prepared by the magnesium perchlo- rate method.9 water was approximately We/(Vni)s=4:. Therefore. . and if the paste is kept saturated with dry ice basis. it was not certain that minimum porosity . the excess will minimum evaporable water content. a subscript h '^^s used to indicate sample preparation FiGUHE 1. using a paste of high water-cement govern the uptake of water vapor. adsorption data. . apparently for an unlimited result given in figure 1.pastes of higher water-cement ratio.7 8 -9 Hayes [64].6. The subscript b is used as a convenient means of signifying that the datum * c = weight of cement. The minimum capacity for evaporable . It is as if the structural domains measured progress of hydration up to a certain point. "If one plots water-vapor surface of an All things considered (and there is much more initially dried paste against its non-evaporable to be considered than is presented here). and on this basis the minimum evaporable water content was stated 0.65/ 0.

m= closer together.482 corresponds to a porosity without access to curing water. as indicated by the points along line such as some of those represented in figure 1 are 0-B. there. accurately the capacity of the gel for evaporable water. If this is the case. contains both anhydrous cement and chemically Therefore.0. can." If the paste samples had been reduced to the same —^=0. there is a lower standpoint of surface energy.59. the c two values of wjwn would have been brought The value is given by m=2. nearly if not quite saturated after continuous that of Welwn is 0.93. Dense specimens of paste or nearly so. it is evident that the rate of hydration cement is due to the relatively high resistance to after several years of curing is so low that it is of diffusion in very dense pastes. (31) The meaning of the quantitative difference is not entirely clear.227mwt/c. At all lower values of same measure of porosity as the quantity found in these two ratios.= 0. Perhaps in the ball-mill When the cement is in excess of that given by experiment there was a similar "self -desiccation. perhaps mature specimens which should conform to eq more probable. but it is unlikely Any sample along the line O B of figure 1 — that it would destroy the structure of cement gel. After a time when all crys- the initial period of paste hydration used in the tals in a local region have grown as much as they first experiment.19. We/c=Wo/c—0. degree of desiccation before measuring We/Wn. wjwt remains constant at 0. reagents present is not theoretically impossible. Qualitatively. reducing w/c to 0.482. the unit cell relative to the size of a gel pore. The quasi-crystalline hydrates evidently have a In his second experiment. (34) that a porosity of 21 percent would eventually be reached in pastes.61. A 591 .0. cement represented in figure 1 could differ from PZ 425 enough to account for the difference. i. their water content of 28 percent.o/c + 0. but neither would have indicated 2. It is a question. After 11 days of milling. Stoppage of the hydration reactions with both After 1 day of grinding. that is. whether the stoppage of hydration is virtual doubt that evaporable water content is a correct or real. water able to maintain a relative produced in the ball mill to have the same porosity humidity of 100 percent. the two experiments gave nearly the formation of a nucleus of a new crj^stal would same result. The paste = randomly oriented particles gradually become was water cured 28 days and then ground in a ball filled with solid material. The corresponding value of wJVm is 3. the result agrees seem a higlily improbable event. there is some reason to expect the gel free water.379.^/c='m. designed to determine whether or not the apparent Whether or not hydration in dense pastes actu- cessation of hydration in the presence of excess ally stops. which seemed to be near a limit. isting particles. The mulative misfit of contiguous lattice layers.0. cement paste. w/c 0. gel pores. the apparent stoppage is not real and is to be it does so by an improbable metamorphosis of ex- explained in terms of relative rates of diffusion. Possibly the data mean wjw(=l—0. as cement was milled with 15 percent water without suggested by Bernal. The general expressions for amount of evapora. Czernin [65] carried out two experiments measure of porosity in the ball-mill experiment. further growth requires forming viable nuclei wjwn reached about 0. the ultimate value of m is less than 1. the end of the experiment. no practical interest. Had they been sealed. Wc/Wre=0. They do so down to Wtfc = water found in the ball-mill experiment is not the 0.437.437 and Wq/c= 0.4. as the gel produced in paste. the metamorphosis being such In his first experiment. Czernin used a portland that the interstices among these variously shaped. 67]. results disagree.74:6mw^/c. if the curing continued much The plotted points in figure 1 represent fully longer than 11 yr. It seems likely that ball-milling would Is Stoppage of Hydration Virtual or Realf destroy most if not all the structure defined by capillary spaces (see section 6). Thus.28 Wtjc. is that the quantity of evaporable (33) with m=1. Quantitatively. tion point [66. Czernin used a labora- naturally restricted growth. This is now considered the best would have been considerably below the satura- estimate of minimum porosity of dry paste. The value Wc/w„=0.63 Wolc. either from the with the data cited above. the ball-mill experiment giving a ble water are : lower value than any of the values indicated by the other data. It seems unlikely that the For the cement represented in figure 1. "PZ 425". paste experiments.e. or. or of the size of a limit to the ratio We/Wn. and seemed to remain in the presumably supersaturated solution in the constant thereafter up to 40 days.058m (32) there is no indication that the minimum value of We/c=Wt/c— 0. possibly due to accu- tory-made "alite cement" (90 percent C3S). but there is reason to fore.59 indicated We/c=Wt/c—mw°Jc (30) for PZ 425 corresponds to a porosity of about 21 and percent. since 'm. If hydration does continue. The ratio We/w. Another possibility. But in places as small as gel pores.227m We/Wn is influenced much by differences in the (33) composition of cement. mill with added dry cement. water storage for 11 yr. after 50 days. WelWn=l-0%..

content. 2).weight basis. A^=2. it isto be expected that not all will agree on the cal. By definition. It is necessary to deduce particles as spheres. data on the size and shape of particles. For the present. Such in and around the sites of cement grains. in harmony with electron micro- Powers and Brownyard [69] used drawings graphs by Grudemo from calcium silicate hydrates suggesting that dense masses of gel particles form [74] and from cement paste [75] (see fig.25^ enough to accomplish this. eral expression for the specific volume of cement A^. Powers and Helmuth [70] presented the required by hydrated cement. 592 . Specific As will be seen in section 6 the massed hydration Ratio of Volume of Cement Gel to Volume of products in their densest form are called cement Cement gel. Originally. A of cement may be called the gel-cement ratio. Taplin [68] recently advanced a method for obtaining the gel-cement ratio based on the For cement gel. this assumption various details that are not directly observable. we may use for all cements the value 0.were used. and that drawings give a meager outline of a concept of these masses generally do not fill all the space structure. is found to be period. eliminating Vt by means of eq (19). gen. The Wg and (w„)„ c c (36) represent. and this was mentioned in the paper referred dividing line between valid deduction and ques- to. 6. Concepts of Structure of Cement Gel and Cement Paste Evolution of Models showing the particles to be platy. with due cogni- Volume of Cement Gel zance of the uncertainty. Each quantity is expressed C Vc{Wn)u as a ratio to the original amount of cement. One may fill in details on the basis of between the original boundaries of cement grains. The solid matter of the gel together with the characteristic porosity constitutes a solid body The volume of cement gel produced by 1 cc having a characteristic specific volume. the initial amount of water and the ultimate value of Wn for a water-cement c ratio low enough to assure an excess of cement. paste. the last quantity being the total water content of saturated cement gel. and the mode of same idea in greater detail.567. Equation (37) should give a more accurate Complete data for computing the specific result than eq (38) because it is based on the total volume of cement gel are available for only one water content of a saturated specimen at the cement. is relatively small. and indicated that the particles actually were spheri. 73] drawings tionable speculation. Although the the bleeding period. but in the swollen state.567. including the w„ A^=l + (38) nonevaporable water. the total water. time of analysis rather than on the original water y^=0. representing the gel formation of cement gel. is (37) Vc-\r{Wtlc)v. and Wt/c=w°/c. (35) For the cement represented in figure 1. or ribbonlike fibers. but later [71] some electron micrographs Such deductions are more or less speculative. and for increase due to values of wl/c vary considerably (see table 5) expansion of the specimen during the curing the value of w°/c. wJc^Wnjc.18. For this cement. following relationship: drated cement and gel pores. The w„o/c represents complete hydration at a water-cement ratio high 0. PCA lot No. respectively. If the original water content is used The indications are that corresponding values it should be corrected for any decrease during for other cements are similar. In later publications [72. 15754.similar interpretation was advanced recently by relatively large where the nonevaporable water Taplin [68]. on the dry. we obtain the following expression for the specific volume of where w^-lc and Wo/{'Wn)u are for tests at dif- cement gel ferent water-cement ratios. which is composed only of hy. Then. as to shape was adopted for simplicity and con- venience. the space Later.

3. since every grain of cement in continuously encased in gel as long as it exists. and 1. The part of the gel After an initial process peculiar to conditions that is formed by inward growth is produced that can exist only a short time. each grain residue being porosity. capillary porosity 20%.38) can eventually produce a gel to those described below.2 cc is transported outward. Space Requirements the components of hydrated cement. the main part of under the greatest possible concentration of the hydration process seems to start at the grain reagents.) having a porosity of 28 percent. Simplified model of paste structure. capillary porosity 7%. the flocculent. even when the cement is not in excess. formed outside. It is reasoned that since the hydration products (Taplin's paper [68] indicates that he also made in a specimen of paste containing an excess of this observation. it is known that cement the opposite direction to the outer border of the gel requiresabout 2. and 45 percent stays inside. w„/c=0. fresh paste (see section 2) is Water diffuses inward through the gel pores while practically in contact at several points on its 593 . therefore. lower drawing represents nearly mature paste.5. Spaces like those marked C represent capillary cavities. presumably in a state of supersaturated solution. Figure 2. in the originally water-filled space. Masses of randomly oriented groups of black lines represent cement gel. that same degree of density can be. and is produced locally at Mode of Gel Formation [76] various places tliroughout the paste. and developed concepts similar cement (Wo/c<0. It seems.2 cc of space per cc of cement. Approximately 55 percent is boundaries of the cement grains. gel layer. Wo/c=0. Upper drawing represents mature paste. Also. that at least 45 boundaries. This means that 1 cc is formed inside the original or start new ones. dift'use in As shown in section 5. where they add on to existing crj^stals. and cement gel grows outward and percent of the gel in any specimen has minimum inward simultaneously.

It follows that any region where orders of magnitude larger than gel pores. Cement gel is mostly colloidal matter. reflected light is scat- where new crystals could nucleate. but. the outward. bluish-gray cast. It is not known.0. Cement gel is one component of cement paste. Its porosity is about 28 percent.e. but at normal paste porosi- of gel water is not known exactly. is about the same as that of a largest would seem to be that size just smaller water molecule.. but whatever the case may be. it about 5 times the diameter of an unbonded may be assumed that there is a range of pore oxygen atom. it is a region where pores exist that are hundred angstrom units apart. =-2. The other component is the residue of originally . from the particles formed by outward growth. in other randomly arrayed structural discontinuities several words. of the gel particles several ways oversimplified. whether the minimum porosity reached surfaces in the gel is between 14 and 28 A. it contains noncolloidal material also. capillary pores in that region will gradually become indistinguishable from gel pores. The smallest size of a single-bonded oxygen atom. it is a useful aid formed by inward growth having a different toward understanding the properties and behavior morphology. When capillary porosity is relatively high.. cement gel is a rigid substance that occupies about 2. it is 0. With the data then available. (These are the larger than the largest characteristic of cement reasons why mature paste of high quality has a gel. Powers and Brownyard [77] estimated the width which is about 5 times the diameter of a water of pores in cement paste from the ratio of pore molecule. 3.0 A. The specific volume work through the gel. On the basis ties. The porosity of cement gel is a that the average width of pore lay between 20 and natural consequence of the growth of irregular 40 A. The main features of paste structure may be summarized as follows: chiometry. dark. whereas paste of low ponents of hydrated cement that are diffusing quality appears much whiter.9. and "outside" product) the 28 percent men- is various observations show that they are generally tioned above. of cement paste and concrete.6 A. the minimum value of cavities.Sv^ A. and this growing gel should reach minimum porosity at figure gives a hydraulic radius of 7. The these points early in the process. probably closer to 20 than to 40. However. These spaces are called capillaries or capillary As shown in section 5. One sees the Although the concept illustrated in figure 2 is in possibility. Such scattering signifies the presence of existing crystals could continue to grow. the hydraulic radius. This distance is inside material. The unbonded might be of a monomolecular dimension. :^cm=2. capillary spaces are interconnected only by of the specific volume of hydrated cement as gel pores and are accordingly called capUlary cav- 594 . the rate of increase in permeability shows that ever.2 times as much space as the cement from which it was derived. it would seem that the overall Size of Capillary Spaces porosity of any region containing only pores within the size range just described (and consisting No systematic attempts to measm-e the size of of approximately equal portions of the "inside" capillary spaces have been reported. 18 A in the outside material is the same as that of the being a reasonable estimate. How. (39) water-filled space that has not become filled with gel. There is evidence that the gel particles volume to surface area. the remaining gel pores. Thus spaces will be found in section 10 which deals with the "outside gel" tends to achieve characteristic permeability to water. wJVra is about 3. the diameter. On this basis. if not probability. are in contact with each other at many points . orders of magnitude larger than gel pores. This meant cally bonded.) Other direct outward from a grain will be likely to be deposited evidence of the relative largeness of capillary in the first over-gel-size pore encountered. When the total porosity gel porosity as it grows. and therefore the hydraulic the capillaries are a continuous interconnected net- radius of the gel is 1 . corresponding average distance between solid of course. During the process of hydration. Size of Gel Pores and the average width of its pores is about 18 A.6X10-«t. or about 13^2 times the diameter sizes in the gel as a whole. it is clear that as a given local region the capillary spaces are very much larger than approaches minimum porosity. From particles in random directions from randomly dis- data given in preceding sections we have tributed starting points. the com. or already tered.surface with neighboring grains. than the smallest in which nucleation is possible. When the porosity exceeds 28 percent is also a region there are capillary spaces. This idea is expressed of cement paste is increased by reducing the in figure 2 by showing the capillary spaces to be amount of gel and increasing the capillary space. as defined here. and this idea is Summary Description of Cement Paste represented by some areas in the drawing where only slight gaps in the gel appear. chiefly calcium hy- HydrauHc radius^g^^^^ droxide. given by helium displacement. the figure obtained and that some of the points of contact are chemi- for the hydraulic radius was 10 A. i. and perhaps even a different stoi.6We:^A.

observation that water in cement paste at humidi- ties below about 40 percent is not subject to Adsorbed Water and Capillary-Condensed capillary condensation is in agreement with the Water theoretical deduction that a meniscus cannot exist below such a humidity. Quantita. it seems that evaporation occm's from 595 630446 O — 62 3 . provide a basis for discussion. chem. capillary condensation becomes impossible. theless. except at low humidities.. sive strength of water. —1200 atm at a relative humidity of Brownyard [78]. is a familiar ring within 1 yr is about —1100 atm [82]. m[{l+wVc) ec=l ^ Computation of Capillary Porosity of Paste f (41) Since the capillary porosity is a significant fac- tor. Vg is the specific could be facilitated by dealing with it quantita. I] Paste may also contain a residue of anhydrous cement. it seems justifiable to assume that part of ities contain water. This is the negative pressure giving a probability of When a small container made of a hydrophilic unity for the spontaneous nucleation within about material contains both air and water. for a concept of physical structure is hesion. to which the gel is "diluted" with capillary spaces. It happens that the relationship between the pres- ing of the interaction of the solids and evaporable sure in the liquid phase and the corresponding water. equilibrium vapor pressure gives.e. some of the water contained in the held by capillary condensation. although the fact that eous and of negligible amount were it not for forces water in cement paste is not pure injects some that hold nearly all of them in a condensed state. toward the au\ Water m a glass capillary. bounded at each end by a meniscus.ities. uncertainty as to whether the agreement ought. e. may be regarded as strong support to the theory ically freewater molecules present in cement paste of nucleation as a means of arriving at the cohe- at any humidity below 100 percent would be gas.) 7. pure water under negative pressure at room tem- perature lead to the result that the "fracture The Process of Drying strength" of water is about —1200 atm. capillary cav. and the effects of water in gel pores and capillaries are factors that determine important characteris. The properties of cement gel and the degree indicated in eqs (40) and (41). (See tively. stress-strain- time characteristics. Its condensed state is imen depended on the porosity^of the specimen. fundamental studies of strength. proper-ties of liquid water. Two diflFer.= l-^[l-m + m(l+w°/c) (40) tics of concrete. Never- cules. It was therefore concluded that at humidities In a specimen saturated with water^ and sur- below 40 percent the water was held entirely by rounded with water or a water-satm-ated atmos- adsorption forces. water is lost from the specimen during drying. water molecules are unable to cohere. and as we taken up at humidities well above that limit is have seen. \Vhen water evaporates from such such stress. Under example. but at humidities higher than 40-45 per. the boundary 1 sec of a vapor bubble by thermal fluctuation of between the water and air is a meniscus concave the water molecules in the liquid [81]. Thus. Perhaps that agreement At temperatures above the freezing point. The theo. ent theories of the nucleation of vapor bubbles in particularly the first drying. Convenient means of a computation are section 5. Mechanical Effects of Adsorption and Hydrostatic Tension This section may be regarded as an extension and since a meniscus depends upon molecular co- of section 6. All sm-faces are usually covered with water mole. which is the present subject. but those given by Powers and Barkas [83]. pores exhibits the normal properties of water. theoretically. and —1100 at 45 percent. cent is somewhat less than enough to make two i. and. a capiUary. due to the forces of adsorption. volume of cement gel in cc/g of dry gel. The theory of nucleation helps us to understand The first question to be considered is how capillary this observation and deduction [80]. sterile unless it is combined with some understand. 40 percent. there is no hydrostatic tension. the lished results. permeability. The water held at a humidity of 45 per- °C) was proportional to the internal surface area. solid phases. as computed by tive data are few. together with more recent unpub. Powers and Brownyard [79] the water held in a specunen at humidities above observed that the amount of water held at any 40-45 percent has the ordinary properties of liquid relative humidity below about 40-45 percent (25 water. retical fracture stress based on nucleation occur. the surface area of the gel particles in the complete molecular layers over the surfaces of the specimen. and that most of the water phere. to be as close as indicated. and such a film does not have the cent the amount of water taken up by a dry spec-. and dui"ability Here ec is the capillary porosity.

Therefore a bubble is becomes supersaturated. that is assume that. From this it follows vapor pressure of the water remaining in the body. (45) though it is not able to present a liquid surface from which evaporation can occur. cannot escape by evaporation until the neck has been emptied. evaporation occurs only from curved surfaces. bubbles can develop in the sufficiently supersaturated with dissolved air. On the basis of an analogy between bubble. One con. (43) humidity drops sufficiently low to cause evapora- tion from the very small gel pores of the sur- rounding gel. on the left side in terms of the product of of saturated cement paste involves evaporation pressure and volume. it Let us consider first the conditions necessary seems clear that bubbles can form only in those for static equilibrium between a bubble and its capillary spaces large enough to permit a viable surroundings. a theory of adsorption. neglecting the force of gravity. none of the water in isolated capillary spaces can evaporate until the A=Pc+^. The following "perfect a porous solid and a "bundle" of capillaries. energy content decreases. Pc=-t (44) The amount water lost from cement pastes of at relatively high humidities cannot be accounted where t is hydrostatic tension. capillary cavities. of the radius of the bubble. and having a radius greater than r*. from Kelvin's equation. the bubble is just as likely water is saturated with air. A different theory that takes into account some consequences This equation shows that at a given temperature of hydrostatic tension is required to account for the free energy content of a bubble is a function the observed facts. it is The bubble is assumed to be spherical and to understandable that the tendency of water to contain n molecules of gas. all the cavities are pressure in the bubble is also equal to the pressure so isolated. including molecules evaporate is reduced because of the negative of water vapor. the capillary water is continuous with the gel The necessary size of cavity is that which will water that permeates the surrounding structure. hydrostatic tension value r*. likely to be viable if at one instant it can form saturation being a function of the hydrostatic with a radius greater than r*. and in some ns is the number of molecules in the bubble. and for larger values of r. The discussion fol. (42) evidence concerning the structin-e of cement paste. However. fact that relatively large amounts of evaporable lows that of Bernath [81]. Substituting from for on the basis of the theory just mentioned. and hence the magnitude of hydrostatic tension bubbles evidently can form with relatively little required to produce bubbles depends on several hydrostatic tension. Let P„ represent the gas pressure pressure in the water induced by the curved in the bubble. This is indicated by the factors. and on the right side in from the meniscuses in the capillaries within the terms of the kinetic energy of the gas molecules. Indeed. it immediately initially to remain as to disappear. r being the radius. nucleus to form. but if the radius quickly sequence of hydrostatic tension is that if the becomes equal to r*. water are lost at high humidities. as is shown by the reduction of water.) If we are to exerted on the gas by its surroundings. we obtain eq (45): Water in capillary spaces surrounded by gel evidently does escape at high humidities even (^-^-t)(^^. that if a bubble having a radius smaller than r* The magnitude of the tension is limited by the should develop. and Airr^j'i the volume of the surfaces.kT. Pc is the hydrostatic pressure in the capillary desorption hysteresis is based on the notion that water and y is the surface tension. and. It is safe to say that in all cement pastes some of the capillary spaces are in the form of T is absolute temperature. When the degree of supersaturation is enon can happen only when the water becomes sufficiently high. the meniscuses. Such a phenom- tension. (dissolve) immediately. the free- develops. accommodate a nucleus having a radius slightly 596 . The second such a situation is analogous to an ink bottle: the term on the right-hand side is the capillary inside communicates with the outside only through pressure due to the spherical meniscus of the the neck and therefore the contents of the bottle bubble. The degree of supersaturation When the water is initially saturated with air. it would probably disappear relative humidity of the surroundings. as discussed below. We have already seen that the bundle-of- capillariesconcept is hardly compatible with P^i^irr'^^nJcT. perhaps most. paste. In the present case. during drying. The cement pastes. it is gas" equation gives the free energy content of the sometimes assumed that the drying of a specimen bubble. only in those cavities The bubble is surrounded by capillary water. that is.r^)=n. eqs (43) and (44) into (42). The function is such When water evaporates from the outside surface that the free energy content of the system in- of a body of cement gel enclosing one or more creases with an increase of r up to a critical water-filled capillary spaces. (See also section 10. k is Boltzman's constant (energy per molecule per degree) and cavitiesisolated by cement gel. the degree of super.

(47) 92 85 116 226 84 43 89 48 70 495 20 25 50 963 10 15 The value of Ts. Therefore. at any given humidity all the capillary cavities below a certain size (table 7) will remain full whereas each of the larger ones will contain a bubble. Thus shrinkage caused by = with t t*. An example increase over t* would permit the bubble to expand of the results obtained is shown in figure 3. and able to accommodate also a hydrostatic tension that develops as the specimen layer of adsorbed water molecules that cannot is dried to a lower humidity is. all cavities having radii greater than about 175 A would be able to accommodate the nuclei that 18 X 10"* atn could form at that humidity. (43). with P(. 4 7 (46) Table 7. where t* is the value of tension at drying at humidities above about 45 percent is a which nucleation is possible. from specimens dried in CO2- rather unstable in any case. The formation of W/c = 0 525 a meniscus is one such phenomenon. the same thickness of this layer is estimated to be about 5 A. 0 73 As shown above. but there are some. after a state of become a part of the meniscus of the bubble. at humidities below about 45 percent. assuming the cavities to be spherical. all cavities having radii greater than about 15 A should contain bubbles. The pores in cement paste range in size from 45. The equilibrium has become established. is not the same as the stable radius for cavity r*+5 static equilibrium at the hydrostatic tension t*.200 w («) tension t*. Ex. and 1 percent are plotted against the com- molecular dimensions upwards to perhaps 0. the upper limit actually being unknown. with T constant. 18. 597 . Computed inscribed diameters of capillary cavities able to contain spherical bubbles at given humidities (eq U6)) On comparing eq (46) with eq. The result is inelastic time-dependent deformation character- istics of the paste. Nevertheless the hydro- 0 2 3 4 5 6 static tension must be the same in the cavities con- 1 ' Tension in evaporoble wofer^ atmospheres x 10 taining bubbles as in those not containing bubbles. tension. Gel Concenlrolion. bubbles cannot exist because the hydrostatic tension exceeds the fracture strength of water. would nucleate and the radius would increase about 50 percent to establish equilibrium. 100 1. - AV When the humidity has dropped to the 50 percent V level. and the tension developed.^-Ct*. and all the cavities smaller than that size would remain full of water.200 atm. Shrinkage rs. » Bubbles cannot exist at tensions above the frac- ture strength of water. static equilibrium are now available. lary space had remained filled with water while ing eq (45) with respect to r. changes in volume for drying at humidities of 75. since even the slightest free air at four difi^erent humidities. If. obtained is quite possible. However. the bubble tween —1. The expression for equilibrium is Percent aim A A 98 28 346 351 2y 2y 96 57 170 175 when P. Only a few data on the shrinkage of mature ce- cept when r* is so small that thermal fluctua.the size that would be stable at 45 40 1.= —t*. and the equation is solved static compression depends upon tlie elastic and for r. and the amount of hydro- then equated to zero. the equilibrium is about 20 yrs ago. which is apparently be- lished and kept strictly constant. is not of special interest in the present connection.1 m. Table 7 gives the calculated nucleation radius for different levels of hydrostatic tension and the correspond- ing values of required cavity size.larger than r*. These data.100 and -1. and therefore phenomena arising from molecular Reference 254 . beginning with the saturated state. as noted in the table. The derivative is hydrostatic compression. the nucleation Relative Hydrostatic Nucleation radius of humidity tension radius r* spherical radius. the Figure 3. r*. we see that eq (46) is not a statement of Required static equilibrium. puted tensions for those humidities. In other words. 6-0 cohesion of water disappear. However. drying has occurred at a relative humidity of 96 percent. The to the limit fixed by the dimensions of the cavity. It only indicates that if t* is estab. for example. throughout the capillary space as if all the capil- The value of r* can be obtained by differentiat. ment paste samples dried under suitable conditions tions might disturb equilibrium.

but an abrupt transition seems strain in the bonds also. the foregoing discussion of internal 598 . The the indicated value of Ac is about 20 percent. and the nature of shrinking and released by soaking the specimen in water.5 X per unit area of surface as could be held in areas 10^ psi or 100 kg/cm^.000 atm resembles maximum possible for the temperature of the ex- one for mechanical loading. effective tension ap. pressure in areas of obstructed adsorption. When tension points are too few to establish the locus exactly. gel struc- mately proportional to stress. further break occurred at —1. The amount discussed in section 8. pressed more than can be accounted for by reduc- tional area of capillary spaces that contained bub. periment. During the last stages of removal of evaporable bility coefficient remains about the same while water. and the specimen becomes com- area from unity to l~Ac. since the distance of separation being zero. which is reasonable for the of unobstructed adsorption. but not separated stress. The value tions that prevent adsorbed water from spreading indicated in figure 3 is 18 millionths per atm. due to surface tension in the capillary cavities hydrostatic tension is effective over the whole should disappear. areas of obstructed adsorption maintain a pres- ticity of the paste as determined from the reso. If we assume that the sions above the fracture strength of water. Thus. The that figure appears to apply to all tensions up to obstructions are probably in those areas where about 900 atm. This obstructing surfaces. but this is now believed to be a minor contri- area factor approaches zero. indicate that a shrinkage-vs. where Ac is the cross-sec. arity being greater the higher the capillary poros- ity of the paste. This is not the case. After the transition point.ssibility where Vo the volume in the is -r^-jj-' is saturated state. for 1 yr of sus. at the point where all Effect of External Pressure Although no direct experimental confirmation » As defined here. and tensile stress in the bonds. and in some of the compression is inelastic. gravity of the particles and corresponding shrink- From this approach. is greatest tension of about 1. there is enough increase in interfacial energy tension increases. per unit overall area. Shrinkage is approxi. 3) but the sion in the evaporable water is nil. However. This film pressure thus is one aspect of the stabilization of paste structure causes a slight dilation. same as that indicating the existence of inter- Other data show that when shrinkage stress is particle bonds. The evidence of the ratio of stress to strain for such a paste under existence of areas of obstructed adsorption. one ni^ht suppose that the compressi. ably be seen if more points were available. The slope of a line from the ori. and evenly over the surfaces of the particles. and the specimens. tion only. at ten- bles. contiguous particles are bonded to each other. bution to the total change shown in figure 3. shrinkage is at the curve for tensions up to about 1. however. when hydrostatic tained stress would be about —1. in evaporable water appears. one might expect some sort of that hold the gel particles together. The reduction of swelling probable because when the stress in the water pressure is effective in areas of obstructed adsorp- reaches the breaking point. the swelling phenomena itself. which effective areas of obstructed adsorption must is not far from the actual capillary porosity of the become smaller as the water content is reduced. is the sustained load. Swelling discontinuity in the stress-strain diagram at a pressure. A compressibility of 18 millionths per atm enough to accommodate as many water molecules corresponds to a Young's Modulus of about 1. or swelling. it would seem that as the age. A discontinuity does when the specimen is saturated and thus when ten- seem to be indicated by these data (fig. considered together with data for various other the evaporable water has become lost.100 atm. To understand this result.100 atm. — sure a film pressure — that tends to separate the nance-frequency of vibration of a test prism. of swelling produced by film pressure is limited by From the above discussion of the fracture tensile force corresponding to stress in the bonds strength of water. proaches zero. but the area factor falls off in at the solid surfaces to cause an increase in specific such a way as to give the observed diagram. ture must be taken into account. as already shown.150 atm. tion of swelling pressure only. isotropic tension. ' » can be cited. swelling pressure is It is not clear whether the transition should be correspondingly diminished and tensile stress and abrupt or gradual. departure from line. which. specimen. but some curvature would prob. and that expansion increase of tension serves only to reduce swelling is proportional to the reduction of effective area. areas (presumably adjacent to those spots) where formation is not usually exactly proportional to the surfaces are separated. amount of expansion per unit change of stress Those water molecules that are excluded from agrees approximately with the modulus of elas. further in. This would reduce the effective area of the paste. by figure 3. shrinkage in the high-tension range emptying of the gel pores and decreasing of the is primarily caused by cohesive forces between area factor. and such de. -stress area factor has become zero. stress tension is less than the fracture strength of water. compre. but. and the specimen should expand. Thu& for the specimen represented the solid bodies of which the gel is composed. reaching zero when all the water has become evapo- >ijreases in tension are accompanied by progressive rated. Areas of Obstructed Adsorption gin to a point representing shrinkage at a given ten- sion gives the coefficient of compressibility ^ for the The observed result seems to be due to obstruc- indicated sustained.

and since the water lost from the loaded areas must be gained Freezing of Evaporable Water by the rest of the area. store equilibrium with the ambient humidity. the film tension to increase change in heat capacity during cooling. as for concrete for expansion of water. and will not be pursued further here. Powers and Brownyard [86] studied the freezing sure and. the internal humidity rises. Amount of water frozen at —20 °C {~4° F) in concrete cured 7 days before freezing the axis of the cylinder would be reduced. as measured by the is caused to decrease. there is also a time-dependent structed adsorption maintained by tensile stress inelastic response. If a part. unfrozen is proportional to the surface area of the If the specimen had been subjected to isotropic gel. 1958) would be increased. which ratio is proportional to the reduc- grounds that if large gel particles were to grow at tion in specific surface area. hydroxide and potassium hydroxide. a substantial reduction Surface Area in specific surface area and in specific free energy content of the gel would take place. a combi.41 . The last column of shrink. During the time required for this the total volumn of concrete. Typical correspondingly and the specimen is caused to results are given in table 8. To re. 8. at least in the effects of applying pressure externally. It is clear that process. areas would diffuse into the areas of obstructed Verbeck and Klieger [88] reported the amounts adsorption. Amount of ice Percent pansion Cement -20 °C(-4 and shrinking counter to the lateral extension.75 no change in weight of the specimen. the specimen swells. water molecules from the out. Molecules from the unobstructed the characteristics of the sorption isotherms. and the swelling some of the evaporated water remains unfrozen. principally sodium of the spaces in the areas of obstructed adsorption. the subject of another paper of this Symposium nation of effects of isotropic compression and [89]. the specimen gradually dilates (swells) concrete normally contains more than enough under the sustained external tension. Some a direct manifestation of. the tension. forces leads directly to conclusions pertaining to The foregoing discussion accounts. areas of obstructed adsorption. and those oriented principally parallel to the axis (Verbeck and Klieger. of water in cement paste by means of dilatometry. At temperatures between 0 and —6°. thus creating a deficiency of water molecules in the findings have been as would be expected from those areas. rather than isotropic compression. pressure falls to its original level.9 93 70 0. but there may be no appreciable 0.8 4. force on the water films maintained by the cohesive Recently.1 57 . Over the years dur- 599 . elastic response. Spaces between surfaces oriented principally normal to Table 8. A rise of internal humidity produces an increase Following the pioneer work of von Gronow [85]. but in so doing. and hence . in effect. the instability of cement of the water is displaced immediately from the paste discussed in section 8. isotropic dilation should be present.45 change in an average internal humidity. Helmuth [87] found that the departure bonds of the gel would have become diminished. Instability of Paste Structure Spontaneous Reduction of Specific the expense of small ones. cement gel is such a change would be a reduction in the ratio often referred to as a metastable substance on the Vm/iOn. finally voids to accommodate the expansion of water annulling the shrinkage induced by the applica. the amount remaining unfrozen at a given tempera- effect would have been as follows the compressive : ture is not proportional to the surface area. for the time-dependent part of elastic re- specimen of hardened paste at equilibrium with sponse to an external force which was shown the ambient humidity is subjected to an isotropic especially well by recent data published by pressure applied to the external surface. of swelling pressure. effect of frost is not due to lack of space in the If external pressure is applied uniaxially. with a further and for temperatures below — 6 °C the amount compression (shrinkage) of the specimen. to restore equilibrium with the table 8 shows the amount of expansion of water ambient humidity. expressed as a percentage of was regained. transfer of water from the compressed areas to the extended areas. They showed that at any subfreezing temperature some water eventually evaporates. if not in the solid bonds is thereby augmented. apparently related to. that must be accommodated by space in the con- side would be received until the original humidity crete during freezing. In addition to time-dependent pressive force on the water films in areas of ob.35 .72 380 8. The effect is a temporary Water ex- swelling counter to the direction of compression. and therefore the destructive tion of the external isotropic force. Evidence of Because of its high surface area. °F) vol. percent of frozen at -7 °C(20 per cent of concrete w/o lbs/yd3 Establishing a new state of equilibrium requires concrete vol. However. Glucklich [84].49 660 520 3. opposing the external pres. Frost action is example by loading the ends of a cylinder. caused by freezing. from proportionality is due to the presence of and there would have been a concomitant widening solutes in the evaporable water. the internal humidity of ice formed in concrete. In general. the com. content at °F).

indication of stabihty may be accounted for in The second adsorption gave 97 percent of the area part by the morphology of hydrated cement. 600 . the nitrogen regard to various aspects of concrete technology.300 0. and Blaine offered. These figures are to fibers. Reduction of specific surface area. if the reduc.300» 100 100 « cycles of desorption and adsorption (Tomes. 006 0. 1960) weight of evaporable water in the sample. and Blaine found that wetting and drying not only reduced the surface area as measured by water but also Although the change in surface area produced reduced the values found by nitrogen. left in clave at temperatures upwards of 100 °C. are given in table 10. primary duced at least 20 percent before the first adsorp. Hunt. much more than the water areas. and Blaine [44] that the very process of de. terminations seems to undo some of the effect of the resulting reduction in surface energy would be desorption.ing which measurements of and w„ have been Other experiments were made by the same carried on.277 94 92 Table 9. taining small amounts of evaporable water.) The samples that ment cured in the ordinary way. are in storage in sealed containers. 89. indicated water area but not the nitrogen area. report additional observations on spontaneous re- found metamorphosis. and Blaine. the nitrogen areas were reduced tion in specific surface occurs during desorption. it is substantial and significant with fluctuations between 0 and 33 percent. For the in this way is much than that produced in the less specimens that were subjected only to humidity autoclave. The maximum readily. Hunt. it appears that. Their data. 2d adsorption 75.00002 to 0. If water vapor as seems likely. with respect to nitrogen adsorption. actually penetrated the primary particles. area was reduced by about the same percentage as well as to laboratory studies.33.20 . If indicated on the first adsorption and the third cement gel is made up of thin sheets or ribbonlike adsorption gave 89 percent. particularly a strong tend. rate of spontaneous change was found in samples ometry. See table 6 about 80 percent seemed to be stable.230 87 77 .280 98 93 samples. Hunt. are given in table 9. In an auto. the the samples deliberately.3 67 63 now be introduced.9 100 Tomes.282 . Powers and Brownyard [90] found the that produced a humidity of about 50 percent in specific surface area of autoclaved hydrated ce. the story is quite different. by the sixth adsorption only half of such penetration by water should reduce the the original surface remained. and 1st adsorption . Any substantial reduction in surface In a recent paper. Hunt.00002. the adsorption being in four steps as required Ci by the BET procedure.265 .300 . the original surface area was re. 1957) The nature and cause of a reduction in specific Specific Relative Cycle Surface surface surface under the conditions described above m2/g ot dry area remains a matter for speculation.06 .2 70 to be incompatible with such an explanation have 5th adsorption 6th adsorption 58.243 88 81 . For the specimens represented of the plotted data.295 a 100 98" .295 .5 65 64 already been presented (section 4) and more may 7th adsorption 8th adsorption 56. Tomes. a reduction tion. and Blaine. Some of kind of metamorphosis mentioned above occurs the data. surface. and Blaine [91] area and free energy would require a more pro. the sealed containers.262 . they V„iw„ after storage Relative surface area period indicated after storage period carried out eight cycles of adsorption and desorp. and the desorption in one 1 month 21 months 1 month 21 months step. (The humidity was deter- ment to be only 5 percent of that of hydrated ce. ductions of surface area occurring while samples ency toward isometry along with increase in size. is sealed specimens as a function of amount of residual Hunt. From the trend as the water area. accompanied by some change in stoichi.17 . . This saturated between successive BET determinations. Thus. Tomes. After drying samples from the saturated state to a relative humidity of about 0. Tomes. Completely Under conditions other than continuous moist dry samples seemed to be stable. vapor adsorption causes a reduction of specific c. Other data that appear 3d adsorption 66.02 . in table 10. .5 74 4th adsorption 63. caused by . Wc indicated tion. Changes in specific surface area occurring in Not so widely known the discovery by Tomes. of section 4. = weight of ignited cement. obtained on granulated 0. allowing the material to swell as growth.8 84 such an explanation. very small. Ludwig and produced humidities below 1 percent and above Pence [51] obtained a similar result. tentatively. a substantial reduction of surface area and be compared with 84 percent and 74 percent in surface energy would require more than mere table 9. but those con- storage. mined by direct measurement. evaporable water termining the surface area of cement gel by water.10 . The range of humidities was from 0. Collapse of sample a layered structure naturally suggests itself. no reduction in the ratio has occurred authors in which the samples were completely for any specimen kept continuously moist. Table 10. and thus. were unstable. Estimated by extrapolation. for if a large sheet became longer and wider much as possible between surface area de- while a smaller one changed dimensions oppositely.8 57. (Hunt.0 56.

Bogue and Lerch (1934). This phenomenon is apparently relatively constant throughout the experiment. all mentioned Cn= (CftAc/^^) X Constant. in which ideas similar to those reviewed about ently the first to observe that strength of cement were developed [97]. 9. In terms of the notation used in this review. Thus it appears that the are complicated by the effects of carbonation external force produced the same kind of effect occurring during the period of drying. by Giertz-Hedstrom in his 1938 review [94]. Steinour. [95] and Powers and Brown. such as. as did the internal stresses induced by drying. it indicates also that the midity. so that shrinkage occurs when the internal humidity is repeated cycles are required to establish stability 100 or 25 percent. and Starke (1932). Strength was considered to paste and concrete should be a function of the be a function of y such that concentration of hydrated cement. a manifestation of the same instability that was Repeated loading showed marked permanent reported in terms of decrease in specific surface seton the first loading. area as discussed above. it indicates that produced by sustained compressive force. for example. Drying Shrinkage Experiments done under less complex conditions were reported recently by Glucklich [84]. although the earlier cement-space ratio of Feret (1897) cer- (48) tainly implied such a relationship. including Pickett's. Others who dealt with similar ideas were Work and Lasseter (1931). a partial fusion externally induced stress. whether the specimen is saturated or tion published in 1953. for. considering react with all components of hydrated cement. a fraction so high as to that produced by internal tension (shrinkage). (See section 7. of the surfaces occurs. The mechanics of carbonation shrinkage intact at low humidities are weak enough to be has not yet been explained. We is on similar ideas. and concentration of CO2. Data published by Pickett the internal tension was probably moderate and [79]. Carbonation bonds have various degrees of strength. the partial reversal caused by complete swelling The surface area per gram of original anhydrous between surface-area determinations.) If this ically. it reacts considerable pressure. Cn is the volume of hydrated cement (not yard [96] independently published work based volume of gel) per unit volume of sample. or that as much as half of the original surface can a bending moment. In 1947 Shinohara. ruptured by film pressure at high humidity. Strength Werner and Giertz-Hedstrom [93] were appar. Rate of carbonation depends on particles. nearly reversible. Although a small amount reversible and that subsequent cycles of wetting of leakage through the seal. If the evidence is construed to indicate such permeability of the paste. elastic deformation. and a little on the second. gel particles are brought closer together than they have ever been before. cement appears to be slightly increased by the cludes that some of the new bonds that remain reaction. most data but further repetitions showed reversible visco- on irreversible shrinkage. He It is a familiar fact that the first isothermal used sealed specimens of neat cement subjected shrinkage caused by drying is only partially to a bending moment. If this is so. one con. the irreversibility would be accounted ratio of surface area to volume of the specimen. Thus. The indicated study of strength in relation tocement hydration constant is inserted because the method of deter- 601 . Effect of Externally Applied Force The explanation just offered suggests that as the surfaces of adjacent particles are mashed Aspecimen of paste when subjected to an together by shrinkage forces. Practically no carbonation new bonds are exhausted at one time. shows behavior similar to be eliminated in this way. and it Irreversible Deformations of Cement Paste is commoi to observe the effect of the applied load at the same time that shrinkage is going on. internal relative hu- new bond formation. Experiments of this kind are usually carried out to study plastic deformation and creep. and Lea and Jones (1935). However. are typical. Also. Perhaps the most plausible explanation is Carbonation Shrinkage that during desorption. Carbon dioxide is able to under a given set of conditions. cast doubt on the explanation. of specimen. and irreversible shrinkage accompanies should create stable chemical bonds between the the reaction. As shrinkage is greatest when the internal humidity shown in table 9 not all possibilities for forming is about 50 percent. Freyssinet (1933). Woods. and self-desiccation and drying at the same humidity are very [66] no doubt resulted in some internal tension. and about the same concepts the volume of evaporable water in a unit volume were used by Taplin [68] in 1959. Eiger (1934). Dzulynsky pubhshed a not. Cement gel is unstable in the presence of and new points of contact are produced under carbon dioxide and moistm-e chem- [92]. In a disserta. and v is the air-filled space.

In terms of 602 . They carried out measure.0 29. data published by Powers and Brownyard are From experimental data obtained from tests given in table 1 1 Calculations for . On the other hand. and i?o and k are empirical constants.24 2. than a 10 kg load. a cylinder molded with the aid of a press could support more where X is the gel-space ratio. 500 proportional to the amount of gel.7 10.000 cm7g and thus the high strength . source of strength cannot be said. where Vg is a function of its chemical composition. 18 2. k being the cements indicate that cement gels low in C3A are ratio./c=/°cX" (51) tribute to at least the van der Waals source of strength. averaging about 3. (A) that a mixture of 100 g of coarse quartz and Powers later [98] used the function 20 g of water was without strength.0 7. although chemical bonding between the particles. For reasons already given. . X R'r = By\ (49) mined by the dry ice method. iVof n of /c gel-space ratio analogous in concept to the Feret eq (37) eq (51) lb/in2 cement-space ratio. M.5 2. However. 13. as discussed in section 7. and k is connection with the chemical compositions of the fc are constants for a given cement. it duction factor on /°.0 13. 53 18.0. A'^is the volume of gel produced Dzulynsky was not the same as that on which Chc by 1 cc of cement. 6. the quantity of gel in a evidence that some of the chemical species present unit volume of specimen is properly represented do not contribute to strength. Another relation. 29. if some of the chemically combined water m Vc c is removed. The following is perhaps the most one can only speculate. Where is the observed strength. simple demonstration. 18 2. the specimen becomes stronger as its evaporable water is lost.98]. there is gain in strength. 15365 45. 500 15011J 45.7 4. the values of n come out about 12 percent higher. where Vp is the volume of the paste." Such strength is presumably due to the proximity of surfaces and van der Waals forces.22 2. Whether when aggregate is present it probably includes or not all chemical species contribute to this other factors.mining fixed (nonevaporable) water used by As seen in eq (37).0 29.)-B] = M[k (wjwo) . % Cement No. By means of a by Vrn/Vp. Constants evaluated from is based. The intrinsic strength of the seems probable that there are many points of gel thus seems to be represented by f°. As to the rela- Difl^erent methods of expressing have been X tive importance of the two sources of strength. i?.08 16.3 6. When is based on Wn deter. Table 11. CsS CzS CjA C4AF CaSOi ments that gave compressive strength and factors 14930J 22. Adopting this form of em. Czernin made it clear that The space initially available to gel is proportional the physical state of the solid material may be to Wq.0 14. i_. and arrived at 15007J 48. 68 13.«».Wo (52) in fact.8 10.8 6. Czernin remarked that in ship was proposed [72] such that cement paste the surface area is not 20. since the ratio of to Wq is responsible for strength [99]. given [72. Powers and Brownyard [96] adopted the term Calculated compomid oomp. Since the cement gel produced stronger than those high in C3A.0 3.1 9. Dzulynsky concluded that on values of w„ obtained by the magnesium per- chl orate method. Since X that the strength of cement gel is due exclusively is a fraction between 0 and 1 A'" operates as a re- . Since all the major tion of the following form components of hydrated cement are found in sub- stances having high specific surface area. Compressive strength functions for mortars pirical equation had the unfortunate consequence made with live different cements that when 7=0. 000 the data very well.000. cement paste is dried in such a way as to avoid excessive stresses during drying. 86 17. gel volume attained by the cement in time is entirely plaus- ^ gel volume -f.capillary space + air voids.00 2. all con- .000 cm^g. there is no the volume of cement gel. When a specimen of convenient expression. 22 2.=M[{VJw. ible. were based X on pastes and mortars. B.. Although there is such evidence that the strength of cement gel is from a. 800 f.' = Ro instead of 0 as it actually (Data from reference 17. He demonstrated not literally proportional to the gel-space ratio.B] (50) The figures in the last column considered in where compressive strength. 6. given cement has a characteristic specific surface area proportional to VJVg. VmlWn.5 2.7 55. tables 6-1 to 6-6) is.0 28. 64 2. 2. it seems unlikely where f° and n are empirical constants. 15 2. to physical forces.000 but 2. Experimental data conformed closely to an equa. the same proportions exhibited some strength and gel volume (C) that when the quartz was pulverized so as to gel volume + capillary space have a surface area of 20. 43 18 44 000 the following empirical equations that represented 15013J _ 39.8 2. 3. (B) that when the quartz was pulverized to "cement fineness".

van der Waals forces. Pf. where ticles. age distance between surfaces in the cement gel. it was known from Carman's i? is a constant comprising the following factors: work with clay [102]. Powers and Copeland nevertheless applied the Kozeny-Carman relationship to ma- Powers and Brownyard [101] attempted to deal ture pastes having capillary porosity.330 sive strength for curing temperatures 27.46 Cement: ASTM Type I Effect of Temperature of Curing Temp 3rature Com- Ludwig and Pence cured specimens of [51] of Cliring Pressure in Surface of solids » Qel-space ratio pressive strength.142 0. Powers and Brownyard assumed shape. discussed in section 2. Vo{T)k V I mal viscosity of water at temperature T. e. and a is specific surface area of the solid par. a function of particle "immobile". 260 127 1000 psi 30 . That is. the Stokes diameter. the surface area of the solids. and 80 27 1 atm _. n=3. seemingly with the flow of water through mature cement with some success. Permeability of Saturated Paste to Water Under proper experimental conditions. Mann. and another func- expression tion of concentration f (c) introduced by Hawksley [103] which corresponds to the Kozeny-Carman Pf9 tortuosity factor. a/T((l e)le) is a correction — term applied to normal viscosity. on porosity. and the number 27 that the quantity of immobile fluid would be which comprises the numerical constant in Stokes' proportional to and obtained the following law for the free fall of spheres. In the (54) exponential term.162 . problem Powers. and 93 °C conform to eq. the test specimen must the evaporable water is mobile. that some of the fluid might be gravitational constant. 086 relationship between gel-space ratio and compres- . cipally on Steinour's adaptation of Stokes' law to (See section 2. rioiT) (1-6) embodying a "tortuosity factor" and a shape fac- tor. and that the hydraulic radius of the prin- point of analytical studies of permeability.4 . At the two highest temperatures. ki is the amount of "immobile" evap- demonstrated that the flow of water through hard. ally abandoned in favor of another based prin- as had been done previously for fresh pastes. 27 sarily equal to all the space occupied by evaporable water.139 .85X10-12) ki cm/sec.71 1. ds'.71 1. This observation of nected system of capillary channels throughout compliance with d'Arcy's law is the beginning the gel. Nominal \v/c=0. cipal conduits could therefore not be calculated from the total porosity and total internal surface Theories of Permeability area. (51) with /°= 12.) According to this approach to the concentrated suspensions. expressed as a multiple of T^^. will be seen further on. 7 days. 1956) Curing time. 103 0. this gain of strength could strength is only 30 percent of that to be expected be accounted for in terms of the reduction in aver. As ened cement paste complies with d'Arcy's law. (Ludwig and Pence. 66. (53) B The constant k is the Kozeny-Carman constant. it can be face area. g. The 150 66 1 atm 123 . The subscript. The values of the con- 603 . and 7((1 e)/€) — The numerical coefficient is the reciprocal of the was thought of as a temperature-independent square of the proportionality between T'.139 . cm/sec. (approx) psi "F °C the nonevaporable water contents. B-. there is evidence that all To obtain a correct result.76 5. the !> Cured at 140 °F. Effect of temperature of curing on compressive may be formed during the process of drying (see strength section 8).79 beMl 200 93 1000 psi 75 152 . (56) indicates that the effective porosity is not neces. e. the cement paste. Powers and Brownyard assumed that the capil- veloped in the apparatus during the test must be lary space in paste is in the form of an intercon- practically constant [100].68 3825 obtained the results shown in table 12.„ and sur. Later. cm^cc. The term qaiT) is the nor- iri= (7. cement paste at various temperatures. be completely saturated.281 320 160 1000 psi 9. measured autoclave m2/g (?). and osmotic pressure de. from the calculated gel-space ratio. but that approach was eventu- pastes in terms of the Kozeny-Carman concepts.000 and s Including unhydrated cement. orable water. and from the work on fresh the density of the fluid in the specimen. 10. There is evidence also that new chemical bonds Table 12. factor of the same kind. and Copeland finally arrived at eq (55) [56]. and compressive strengths.

Probably the correct v* = v(T.34±0.Mo) exp {Ea+EJ/RT=exp ia/T)(l-e)/e. patterns. were reported as follows: found by Steinour. 1. derived from experi.slants in eqs (55) and (56). found for flow in gives = k 4:. Copeland. and 4. the viscosity for were based on the assumption that the mean flow through paste should be given by specific volume of evaporable water is independent of porosity and equal to 0.5. and tlie mechanical effect alone. a=l.a. The least-squares analysis was re- peated during the present writing. — Solutions of eqs (62) and (64) are given in table 13. B.99 cc/g. parallel plates. adsorption produced a function different from to require the empirical function of porosity to be eq (57).07. According to the theory of Eyring [104] tion of eq (55) showed that this might have been anticipated.242±133 and 14. Actually. from eq (55) (without 7) and from data give above.a. 32. particularly in the light of Hawksley's VoiT)=Aoexp Eo/RT (59) treatment. found by Steinour from ex- periments in which adsorption effects were absent. 4. 1 e ri.250/298 = 4. a = l. cients in eqs (57) and (58) is significant. For values of e between 0. Steinour [22]. where 17* the relative viscosity when there is no is For flow through cement pastes.3. The fit seemed good except that 7 was not sig.05.m)/vo{T) = (Ar/Ao) exp {Ea+E^)IET. it is correction factor for normal viscosity such that now clear why a Kozeny-Carman type of expres- sion cannot be applied successfully to the flow of r. omitting 7.99. and willbe and Mann.+Ea+E^)/RT (60) Based on the porosity to helium. whereas the 5=(1.m)=A^ exp {E.7. but. the effects of adsorption Thus the effect of adsorption is effect. such as between smooth.78. (63) ^-|-j4-^-exp 4. £ = (1.* = exp (a/r)(l-€)/6. (62) water in hardened cement paste. Flow of water through hardened paste does not provide the conditions just stipulated. eq (57) gives ^ = 4. the indication that 7=0.) v{T. experimenting with spherical particles of tapioca Thus suspended in oil at 25 °C.500. and 0. With e=0. did not pursue the analysis further.7±0. Thus. The exponential term in eq (55) is an empirical With the success of eq (55) established. ac- The values used in the above calculations of e cording to the same principle. calculations gave where Ea and E^ denote parts of the total activa- 5=(1. we have (1— e)/e instead of (1 e). (See section 3. The agreement between the numerical coeffi- nificantly different from zero. the same value mental data. respectively.19. being at the time mostly interested in discussed further on. (^. and for V*i = exp EJRT=exp ia/T){l-e) (64) its limitations. The agreement pointed out earlier between the (57) value 4. thatrequired for the success of the Kozeny- Carman equation. accounts for the success of the Kozeny-Carman equation. cement paste indicates that when Ea=0.1)X10-'° corresponding flgures for eq (58) are 3.19(1-6). Thus. where voiT) is the viscosity of water flowing under conditions that produce unperturbed patterns of with the following results: flow.02. This constancy.19 in eq (57). for flow through pastes.18±0.085)X10-'° tion energy for flow due to adsorption (and solutes) and to mechanical interference of flow a=l.0 could only mean that Viscosity of Water in Saturated Paste the exponential term contains nothing that is in- dependent of temperature. showing the total effect of mechanical inter- . showed that the Kozeny- Carman constant k is the following function of e. k is far from constant. Thus. A review of the deriva.36±0. For r=298° K. and Ao is a temperature-independent constant.06 within ±0.099)X10-i° Eo is the normal energy of activation for such flow.3 and 0. values are closer to those based on than to (61) those based on the assumption that the specific volume of adsorbed water is 0.5. Powers. 604 .250± 20. together with the limits on e. and its magnitude far away from 7 = 0. this equation and a/r= 1.250/T=4.06.19.432±22. 0.250 (1-6)"| ference and adsorption. relative viscosity ?7* in paste is given by The two pairs of values may be regarded as upper and lower limits.

if the scale of abscissas should have been in terms of there is still a reserve of unhydrated cement.6 600 . and from fall above line A. .489 18 8. physical characteristics of cement gels produced by different cements are similar would lead one Continuous and Discontinuous Capillaries to expect this result. fully leactied fine C . This mined from bleeding rates. 280 7.200 for 6 = 0. Magnitude of Activation Energies Activation energies for flow through different pastes ranged from 8.Table 13.60 . 1800 cm^/g. . the final point being calculated and do not have much effect if the tests are made the rest experimental. Wagner).50 .8 20. shape. normal cement. Data given in pre^ The dashed lines are estimated curves for other ceeding sections showing that the quantity and water-cement ratios. and.70 .160 for e=0. and Mann [105] showing the principal factors controlling permea- bility to water at constant temperature. At some stage of hydration of a given viscosity of water in fresh paste is negligible. and by the effect of adsorption on vis.346 10 15. relatively high permeability'. feasible to combine both functions in the same If the water-cement ratio is too high.461 16 9. Production of cement gel at first con- represents the permeabilities of fresh pastes over stricts the capillaries without destroying conti- the range of water-cement ratios indicated by the nuity. . complete graph. published by Powers. .64 . At early ages. .572. Computed relative viscositij of fluid in saturated cement pastes.700 . and con. The mean activation energy for flow of water under normal circumstances is about 4.395 12 12. shape. hardening. General View of Factors Determining Perme- V A -Moture. Since the effect of adsorption on the and A. The resistance For the particular cement represented by these to flow is determined by the size. The resistance to transition. hydration will not produce enough gel to destroy Line A represents the permeabilities of pastes capillary continuity. 605 . Line E laries. points (ASTM Type I. V^q in Fresti Paste and a/T = 4. — However it was not subsequently produced points will follow line A. in calories per mole of water. Em+Ea ranges from about 3. and the terminal points containing completely hydrated cement.700 to 1.6 134 . cosity of water in paste. . but finally may divide them into segments top scale. that are interconnected only by gel pores.Solid content of specimen ability of Saturated Paste K. which all alkali had been leached. based on eq {63) and {64) with Ai Ao. 0.414 to 6. the permeabilities having been deter. centration of the particles composing hydrated capillary continuity does not disappear at full cement. flow depends on the size. Points along line C are when the different cements are at comparable the same for a different water-cement ratio.. Permeability functions for fresh. 1 — e instead of {l e)je. and on the effect on the lines for other water-cement ratios. in terms of changes of permeability. and concentration is shown clearly by curves B and C and by dashed of cement particles. along some such curve as D.7. stages of hydration [100]. Capillary particles of the initial chemical reactions (see continuity is indicated by any point between E section 2).Various ages.Coefficient of permeability cm / sec Figure 4. = Weighl Rotio of Water (o Cement. Figure 4 is a simplified version of a diagram and mature pastes. interconnected system of capil.^'OJI by wt. and the present plotting serves the purpose. paste the point will just fall on line A. The marked points along line B represent a sample of paste at various Effect of Cement Composition stages of hydration from the fresh to the fully Differences in chemical composition of cement mature state.490 cal/mole. uliro cement..5 78 » This Is the value for cement gel. maturity if WqJc is greater than 0. 600 . eq 62 A 0. Copeland. 5 2. pastes made The water-filled space in jresh paste constitutes with a slow-hardening cement will of course have a continuous.19 Factor of viscosity increase Hyaraulic WqIC c radius Mechanical Total effect effect 77 eq 64 n'. . 5 " 47.45 . C • . X . Thus. 38 0.800 l-t cal/mole for the range of e given.

the pressure was 300 psi. 2X 10~* vs. = with w/c 0. As is apparent from figure 4. The effect was greater the denser the paste. reduce the rate of flow through 140 paste by increasing the viscosity of water. Verbeck repoTted results of the Permeability of Unsaturated Paste same kind [92]. but in section 8. 6X10"^^ cm/sec. the effect per ion being greater the greater the degree of hydration of the ion. It be expected that if the temperature of is to curing high enough to increase the size of the is primary particles. Effect of temperature of curing on permeabilities of water-cured pastes {Ludwig and Pence). All specimens cured under water for 7 days at temperature>200°F. first in humid air and then in water. This seems to be the only way in which the fineness of port- land cement. the change in permeability accompanjdng the transition from w/c=0. For example. Once about 70 times what it would have been had the water vapor is in cement paste. The mo- the structural instability of cement paste discussed tive force is not the vapor pressure difference. Verbeck [106] found that a given cement paste. Effect of Cement Fineness The higher the specific surface of the cement. influences the permeability of mature paste. practically all of specimens not been given the drying and wetting it is adsorbed and the transmission occurs in the treatment.55. fresh paste is 3 million times as permeable as mature paste. and then slowly capillaries. cement ASTM Type I. Effect of Curing Figure 5. This is probably another aspect of adsorbed or capillary-condensed state. its permeability to water is of a saturated specimen and a "coefficient of thereby increased. the farther to the left curve D will be. it is applicable only to speci- were dried very gradually to equilibrium with mens containing an interconnected system of 79 percent relative humidity. per se. There are not many data on absorptivity" of a dry specimen. mature specimens tionship was stated. the gradient in fikn tension discussed in section 7 606 .46 (nominal). points representing specimens containing alkali fall below line A. Not much systematic study of movement of Effect of Drying moisture through unsaturated paste has been reported.7. except for specimens having capillary continuity. Ludwig and Pence measured the permeabilities of pastes cured under water at elevated temperatures with the result shown in figure 5 [51]. Temperature of curing -"F. as the rela- this point. When plotted in figure 4. that is. at Wo/c=0. the higher the water-cement ratio at which capil- lary continuity can be "cured out". permeability to water will thereby be increased. In one experiment. but. Speci- mens from which alkali has been leached show as 130 - much as six tunes the permeability of companion specimens containing a small amount of alkali. Effect of Alkali - Solutes in the evaporable water. particularly 150 NaOH and KOH. was nearly five times as permeable to pure water as to a salt solution containing 12 g/1 NaCl. the "water-vapor permeability" is not ordinarily a The coefficient of permeability was found to be process of transmitting vapor as such. seems that the effect here reported It is due to the presence of hydrated cations. Powers and Brownyard [101] gave a specmien becomes dry at some time before If a relationship between the coefficient of permeability a permeability test. fresh to mature states is enormous. It has been pointed out [105] that resaturated.

exclusive of air bubbles. grams. when Hydrated cement: a collective term for all the chemical prepared at room temperature its specific surface area species produced by the reactions between cement and is about 210m2/g dry weight. capillary spaces. Ind. 74. specific volume of water under reference pressure. but noncolloids. water. unpublished laboratory report No. vapor pressure 8m of mercury. p. cm^/g. become hydrated. specific volume of water under existing pressure.567 cc/g dry weight. cc/g. and has a porosity referred to also as the Copeland and Hayes method. as applied to and Brownyard ([17]. g/cc. Its overall Gel pores: the pores in cement gel. Wi = the immobile water factor. capillary space nor residual cement. or can any. Cement Assoc. Further studies of the bleeding of [4] Reference [1]. N= volume of cement gel produced from 1 cc of = nonevaporable water water retained by a cement. cm^/cc. the pore space and physical properties are influenced to a significant in a specimen of paste in excess of about 28 percent of degree by the surface energy of the substance. (1941). Portland mortar and concrete. 249-336). g/cc. if any. density of fluid. occupied by evaporable water. In cement gel calcium hydroxide. of about 28 percent. Pc= pressure in capillary water (usually negative). adsorbed layer on a sample dried by the dry ice h = relative humidity = p/ps where p is the existing method of Copeland and Hayes [36]. {Wn)i= nonevaporable water = water retained by a Vc = specific volume of cement. = weight of water required for a monomolecular weight of cement after being ignited. Steinour. Ch-. These spaces are regarded colloidal substance may be amorphous or crystalline. cc/g. 607 . Pc = density of cement. it is the rigid body produced by cement now largely supplanted by the dry ice method. g/cc. specimen prepared by the dry ice method of P= pressure. cc/g saturated saturated specimen and the specific volume of the weight dried solid as determined by displacement of specific volume of total water in saturated speci. ditto for a specimen dried by the magnesium ln = logarithm to the base e. perchlorate method. perchlorate method pp. Copeland and Hayes [36]. A solid the volume of the specimen. men. specific surface area. 249-336) and by Powers [98]. ([17] log = logarithm to the base 10. C. if crystalline. helium. the author's definitions of a few terms must be kept in mind. The bleeding of portland cement paste. (Compare mostly of colloids. particularly calcium with cement gel. and residual cement. Dry ice method: the method of isothermal drying of samples Cement gel: the cohesive mass of hydrated cement in its of cement paste described by Copeland and Hayes [36]. specific surface area. as discontinuities in a mass of cement gel. corrected for water displaced by bleeding. densest state. Appendix 1 Glossary To understand some parts of the text. cc/g. m — maturity factor = fraction of cement that has ditto. Vhc = specific volume of hydrated cement. weight.) Capillaries. drous state. Chem. pp. Portland cement dispersion bj' adsorption of lignosulfonate.) hydroxide. Portland cement paste. porosity as calculated from the volume of the specific volume of saturated paste. 288-11-D4. w= weight of water in fresh paste. Appendix 2 List of Repeatedly Used Symbols c. Bull. overall volume of a material. A colloid is characterized by a high specific cement gel. When there is neither be. Bull. either positive or negative. specific volume (apparent) of nonevaporable capillary porosity. Ps density of solid. the Colloid: a substance in such physical state that its chemical space occupied by water. H. 37. 2. Magnesium per chlorate method: the method of isothermal As applied to fresh paste. specific volume is about 0. practically anhy. (All terms defined in the glossary are italicized. if any. ([17] pp. Its definition. Cement Assoc. Ernsberger and Wesley G. weight of original cement that has become hy. weight of cement in its original state. (1939). Portland [5] H. composed of cement gel. References [1] T. 598-602. These terms are given below. p. Research Lab. and water. 249-336). (1945). ratio of volume of interstices to gross. the sample. Steinour. V= volume of specimen or batch of paste. surface area. the Brunauer-Emmett-Teller surface-area factor drated. cc/g. in mature paste. [3] Fred M. cc/g dry weight. Powers. are included in this definition. if Pore (in cement paste) : space in cement paste that is. Cement paste: the term is applied at any stage of hydration. or Capillary spaces: in fresh cement paste. specimen of paste prepared by the magnesium Ve = specific volume of evaporable water. hardened paste. cc/g dry weight. it is the mixture of cement and drying of samples of cement paste described by Powers water. It includes gel pores. H. and Cement: portland cement in the initial. water. (1945). The solid material is composed Gel: a cohesive mass of colloidal material. but. the crystals are apt to be imperfectly Capillary cavities: capillary spaces that are isolated by organized. Research Lab. cc/g. (See water-vapor pressure and ps is the water-vapor glossary). cement paste and quantitatively. cc/g dry €= porosity. involves a standard method of drying cement gel are identical. Eng. except transient products of initial reactions. pressure at saturation and same temperature. [2] H. France. Vg = specific volume of cement gel. perhaps some other components are not colloidal. 4. 43.

Decem- 36 840-847 (1944). and George Verbeck.. Ind. -8 [23] H.. 47 [57] L. Bull. Amsterdam. Portland Cement Assoc. The rheology of concrete. Am. Bull. Lab. J. tion des Zementes. 132-155 and reference [2]. Carman. Research Council. H. Concrete Inst. Publ. 558. Powers and H. [52] Geo. 249. [50] L. Portland Cement Assoc. Ill. Some unpublished work. in Building materials. and Develop. L. Rate of sedimenta. 51. Grim. in hardened portland cement pastes. Hans F.. p. Research NBS 59 edited by F. Portland Cement the workability of concrete. (1958). 933-992 diffraction analysis. Assoc. Ind. 106. (1941). Proc. [37] L. Forslind. Chemistry of hydration of portland cement. E. Develop. Chem. Research and (1946). 257 (1949). [14] T. 9. C. Chem. 34 of Research Lab. [53] Duncan McConnell. The stoichiometry of the hydration of beta-dical- [13] H. Specific volume of non-evaporable 37. 3. Hunt. Lab. Steinour. II. cement paste. Concrete Inst. 469-504. 901-907 (1944). Weir. E. Cement Assoc. Testing Assoc. The determina- [9] H. 3. 19-20. p. C. Steinour. Blaine. Bemerkungen tiber de Hydrata- [27] Ibid.. Research and pp. Concrete Inst. pp. 1956. Stephen Brunauer. (1941). (1958). Valis. Portland Cement Assoc. No. Portland Cement (1955). Wiler. Steinour. No. 56 No. 39-50. C. Carman. A device for studying Also Research and Develop. 36 618-624. Am. 3. Also Research Lab. pressures up to 10. p. Kantro. Steinour. The stoichiometry of the hydration of tricalcium their elasticity and plasticity. Soc. p. Research and Develop. Capillary rise and capillary movement [55] H. 608 .000 atmospheres. Phys. Can. Weise- [58] [33] Ibid. Am. Blaine and H. Bull. [8] Ibid. H. an international symposium. 482-488. [36] L. durable concrete. Bull. Winterkorn. Concrete Inst. [48] Reference [17]. E. pp. Powers and E. clay-water systems. Copeland. Steinour. Ludwig and S. Highway Research Board Special Report 40. In: Water and its conduction [41] Stephen Brunauer. Copeland. pp. Zement u. Am Research Lab. 47. Eirich (Academic Press Inc.... Swayze. p. 73. E. Flow of Gases Through Porous Media reaction" by T. Research of cements and related materials under hydrostatic NBS 42. L. Surface available to [35] C. [45] Stephen Brunauer.. Proc. 4. p. [38] Reference [17]. p. (1954) or Interscience a ball-mill. L. Eng. [26] Reference [1]. H.. fig. and R. factors affecting the surface area of hydrated [15] Marcus Reiner. [49] Ibid. No. 43 117-140. Soil Sci. Brownyard. 86 (1958). [17] T. and time of finishing flow of water in hardened portland cement paste. Portland Cement Assoc. 72 Portland Cement Assoc. Bragg. Bull. 52 863-874 (1956). chapter 9 Portland cement as determined by wator vapor of: Rheology: Theory and Application. 135-136. Specific volume of evaporable water Proc. February 11. 17. 81-87. p. 60 112-120 (1956). [29] P. Research 308-323. Am. 1946. Properties of portland tion: I nonflocculated suspensions of uniform cement pastes cured at elevated temperatures and spheres. H. Moore (Miss). 52 309-326 (1955-1956).. M. 52. Lab. Symposium. 16. 196-201 (1946-1947). [32] Reference [17] p. Copeland. pp. 194. (July 1959). Zement. C. [43] Stephen Brunauer. J. p. Bull." Ibid. Kantro. See (1956). E.) Ibid. 693. Am. The original Cement Assoc. Kantro. J. L. Bull.. Reprinted as Research Lab. September 7. and Chas H. p. 3. 845-880. Chapter 7. E. C. J. (1944). Copeland. p. Hayes. Proc. E. Proc. Lab. Quantitative X-ray 336. 253-289. Am. Copeland. [20] Reference [1]. Portland for determination of apparent surface area of Cement Assoc. pp. C. pagination is used in Bulletin 22.. C. H. Research Lab. [19] Reference [2]. Hydration in paste form. N. Mann. on the scale resistance of concrete. The effect of atmospheric conditions [56] T. 3-7D. [12] Ralph E. 669-712. Assoc. [7] Ibid. 30. 88. No 75 [11] Reference [9]. 673-687 particles. tjber die Wasserbindung inerharteten of moisture in fine sands. E. physical properties of hardened portland cement [46] Reference [17]. Nat. Electro-osmosis tests on cement cium silicate and tricalcium silicate at room pastes. M. p. C. Finishing and curing: A key to publication. Research and Develop. 22 (1948). 812-4 (1955). Behavior nitrogen in hydrated portland cements. Am. ASTMBull. R. 35-43. Anal. concrete products. Sci. ofsome published researches on the alkali-aggregate [25] P. A. 80 761-767 (1958). Portland Cement Assoc. 135. Materials Proc. Tomes. North Holland silicate at room temperature: I. Kalousek. 840. (1946). D. (New York Academy Press. M. and R. L. 1943). pp. (1956). paper IV-3. 488-489. Ibid. this [21] Reference [17]. 1959). Develop. 696. Assoc. and R. vol. 571-574. 485. Discussion of "Simplified method 847. surfaces and its implications for the properties of [40] Ibid.C. Acad. 70-74. 1956. 497. Portland Cement. [22] Reference [2] and H. (1958) Washington. D. 101-132. Fresh concrete. October 317-331 (1950). Lab. Should portland cement be dispersed?. L. The during the bleeding period. pp. L. and L. Portland (1958).. Copeland. Chem. Hydrated in Publishing Co. York. (See reference [22]). See Part C.. No. temperature. 28-32. 36-37 (Julv 1959). C. Bull. NBS. 1945. and nitrogen adsorption. Steinour. 448-7. Chem. II suspensions of uniform-size angular pressures. Rate of sedimentation: Suspensions tion of non-evaporable water in hardened portland of uniform-size angular particles. Unpublished progress report. p. Bragg. Nat. Highway Research Board [42] Reference [17]. zur Strassen. 491. III concentrated flocculated suspensions (1956). [59] R. 498. Powers and T.. report. H. D. Research and Develop. Portland Cement (1956). edited by Vapors (Princeton University Press. Concrete Inst. Discussion of "An interpretation [24] Reference [1]. Chem. Lab. 41 1003-1015. L. 41. Copeland. Powers. 714-724 (1959). The surface energy of tobermorite. Lab- water in cement paste. D. 52 1-14. of powders. Concrete Inst. paste. [18] Reference [1]. Series 254. L.Y. [51] N. 31. later [44] L. p. unpublished laboratory Portland Cement Assoc. of Publishers Inc. [54] A. Eng. [47] L. p. J. Chem- [34] H. 65. Hunt. 130. Proc. Powers. Also. A. Bull. Studies of the 65 (1956). and R. H. E. Blaine. C. (1944). Research and Develop. M. Bull. [16] E. (1959). New 357-364 (1957). H. Bull. Bull. [6] Reference [1]. Bull. 51. Organization of water on clay mineral [39] Ibid. reference [12] for further description of the HRB [31] M. Part 2. Copeland and John C. Research. and L. Private report. E. Part 2. 47 (1953). 37. 549-602. J. Bull. H. p. 1. Portland ber (1953). [28] Reference [2]. Research Lab. 3. Soc. Lab. A. [10] T. The Absorption of Gases and in soils. p. J. New York.. [30] Paul KHeger. 105 (1959). Beton No. Special Report 40. fig. Pence.

(1958). Permeability of portland cement pastes the Portland Cement Association Research and Proc. L. C. 4-11. Skokie. A. London. — hardened portland cement paste its significance „ Portland Cement Assoc. 95. ConcreteInst. Am. Powers. pp. 29 (1949). Pollitt H. ment paste. Research Lab. A method for following the hydration (1958). A. Powers. Anal. of PCA Research and Develop. [64] L. "The struc. [106] — George Verbeck LTnpublished report (1955). 492. Research erharteten Zement. Czernin. [91] C. 84] J. W. on the Chem. [103] P. Frost action in concrete. Self-desiccation in dioxide on mortars and concrete. Phys. Hunt. [73] T. Bull. crete Inst. [101] Reference [17]. Report [72] T. George Verbeck. Proc. C. [82] J. W. hardened cement pastes. [65] Wolfgang Czernin. 28 dratation des Hants hydrauliques. pp. and H. 585. Chem. Am. Blakey. [97] M. Emmett. H. 485-490 (1936). No. 38-48 (May 1959). Ed. 495. H. [79] Ibid. Chem. p. 16. pages (1959). 534. Kyushu Uni- properties of concrete. ASTM Bull. 932-969. Australian J. 94. Porosity of 176. 47-56 (Jan. No. liquids. Hayes. Con. Some effects of carbon [67] L. H. Ind. Bull. drated cement. Lab. Bull. 1960.. 68-76. [83] W. apparatus for study of freezing and thawing of [63] Ibid. Lab. Taplin. Portland Cement Assoc. Bull. [75] By cooperative arrangement with the Swedish Re. Mr. L. (22 pages). 26. 549-602. at the Institution of Civil Engineers. Testing Materials Spec. Zement 25. Beton. Powers' most inter- von Zementmorteln. 204. Powers. p. 34-39. Zement u. pp. Lab. 476-477. Die Einwirkung der Kohlensaure auf Research Board 27. during 1957. E. reaction in portland cement paste. Grudemo. [80] Max Volmer. Dur. Giertz-Hedstrom. A discussion of cement hydration in Chem. hardened portland cement paste. Sci. (1959). p. p. 285. 53.. Develop. Bogue. 795-797 (1935). 4. and [66] T. C. posium on the Hydration of Cement. Carbonation of hvdrated portland cement. esting paper. 1310 (1952). Hunt. Theory of volume von den physikalisch-chemischen Eigenschaften des changes in hardened portland cement paste during Zements. 2. W. plasticity and diffusion as [81] C. 285-298 (1954-1955). ASTM Bull. (in Japanese). London [96] Reference [17] pp. 10 No. Stockholm Proc. concrete. j. No. Soc. Some physical aspects of the hydration [102] P. C. M. 633-640 (1955-1956). 1951). No. Hayes. E. of the Symposium on the [71] R. Powers. Die von erhartendem Zement [85] gebunden Wassermengen und die Frostbestandigkeit My remarks concern Air. Chemistry of Cements. J. Hawksley. fig. Research. Proc. 158. 25 (1948). February (1955). the settling of suspensions and flow through [78] Ibid. 103 [99] W. of PCA Research and and clays. 1000-1006 freezing. Unpublished laboratory report. Bull. p.. posium. pp. Mann. [104] H. 297 (1953). 68. Eisner von Gronow. Bull. No. Some keit des Gelwassers. Development Laboratories. 205. Bernath. A. Versuche iiber die Reaktionsfahig. Copeland and R. Helmuth. Labs. 28-33 (1941). Structure and physical properties of University of Liege. 29. Research and Develop. of [95] Kinji Shinohara. 13. 1958. p. Proc. 2. Copeland and John C. 16. 10. Portland Cement Assoc. Powers. Rheological behavior of hardened paste Discussion under low stress. 52 (1955). I should like to say first how much 609 . Symp. No. Stockholm. Proc. 1938. 51. 46 (1953). I. Bull. Eyring. Permeability of saturated sands. Bull. 1. J. phology and crystallization properties of calcium 1949). Portland Cement Assoc. The effect of concentration on [77] Reference [17]. The physical structure of hy- Cement Assoc. Die Abhangigkeit [69] Reference [17]. Highway Research Board Bulletin No. Lab. Glucklich. Also Research and Develop. 865-880.. Am.[60] P. Portland Cement Assoc. Pub. Reprint of a lecture presented versity. Lab. 327- 337. Highway Research Board 32. of Portland cement. pp. (1958). The non-evaporable water content of 41. Proc. and R. Soc. 9. p. pp. I. applied to bubble formation.. Helmuth. 178-188 (1947). Fundamentals of the strength of Cement. W. Also Research and Develop. (March 1956). NBS 64A (Phys. Proc. Fisher. Sci. Mann. D. J. Research and Develop. The theory of bubble formation in examples of absolute reaction rates. crete Inst. 90. Appl. 496-498. 6 206-226 (1953). No. pp. C. Copeland. 114-135 (Edward Arnold & Co. soils. 3-10. H. DeWitt. (May.. 283. 1. Carman. Relation entre la r&istance et I'hy- velop. [100] T. Monthly 48. Am. Cement and Concrete Inst. Determination of [86] Reference [17]. 845-864. C. Discussion of paper No. Zement 20.'44. Highway [92] A. Dzulinsky. 262 (1939). Lab. p. tion". Tech. Royal Inst. paper VI-2. 53 (1955). [93] D. [88] George Verbcck and Paul Klieger. Sci. [87] R. Agri. Concrete Inst. 1949). Eng. 302 and pp. Am. An electronographic study of the mor. silicate hydrates. In: Some Aspects of Fluid Flow. Also issued as Research and Develop. Lab. Copeland. Powers. A. Bernal. 87 (1958). C. Blaine. London. Portland hardened portland cement paste. 35-37 effects of aging on the surface area of portland ce- (1959). Leber and F. 704. E. Eng. Beton No. Third Int. for concrete research and its method of determina- [74] A. J. 984-987. Zement 24. Illinois. C. Swed. Giertz-Hedstrom. search and Concrete Institute. Powers and R. [76] T. L. Research and Develop. G. Bragg. M. J. and H. Lab. Research and [89] Paul Nerenst. Also issued as Research and De. (1936). private communication. C. fig. Kinetic der Phasenbildung (Steinkopff. Grudemo was. [90] Reference [17]. 415. 295-308 (1956). Portland Cement Assoc. Bull." by J. Portland [94] S. and T. 505- tures of cement hydration compounds. H. J. The Swelling of Wood Under Stress. Lab. Capillary continuity or discontinuity in cement 419 (1947) pastes. Labs. Ind. (1931).. 37-39 (His Majesty's Stationery Office. Cement Assoc. Tech. Viscosity." [68] J. surface areas. L. Ceram. Portland Cement Assoc.) 163-169 (1960) relation to the curing of concrete. 254. Am. October. Bull. Con- Portland cement pastes. photographs of cement gel. Centre. Sym- pages (1955). Werner and S. Tomes. C. E.. [61] Chas. 1-6. 3. Steopoe. porous media. l\r. Bull. 1939). [98] T. Discussion of principal papers. 54-163 (1947). C. Zement u. L. a guest scientist in the laboratories of M. Proc.. Barkas. 16-19 (1959). 1961). der technisch wichtigen Eigenschaften des Betons [70] T. (1956). Vienna. pp. Calorimeter-strain [62] Reference [17]. Powers. 52. London 1952. Fracture of liquids: Nucleation theory [105] T. 496. Chem. 56. this Sym- Develop. (1958). 329-345 (1959). The physical structure and engineering of the Department of Engineering. d'Etudes de Recherches et d'Essais Scientifiques. Portland Ce- ing this time he obtained several hundred electron ment Assoc..

it seems. however. the information now available.01. number of commercial bricks show ranges in ment of this test. through capillaries in cement paste and that the We. the ratio is remarkably air or to honeycombing. Particularly welcome less by lime water. The clear inference chiefly of importance insofar as it can be related is that it is the dissolving power of the fluid with to the physical properties of cement. so that the curves of both 610 . tractors frequently make concrete with insufficient strength which is usually due to too high a water. at the Associated Portland Cement Manu. position with cement products that the drying men was intermediate between those for pure shrinkage. is similarly water and carbon tetrachloride respectively.78 to 1. Determinations of the often ±0. The method works at the higher pressures and still longer times may well. beyond the range used in practice.I agree with him that there is a need for a closer of interconnected capillaries. except when large open constant. other things being equal. rather of cement and concrete. like to comment on the section dealing with It seems possible that chemical reactions may permeability where it is said that the presence of similarly have afl"ected the results reported by dissolved alkalis in water materially affects its Mr. as described in The space left by evaporated water is measured various papers presented at the Symposium and by the amount of carbon tetrachloride absorbed. Consequently strength is was removed from the carbon tetrachloride. whether these were due to entrained analytical determinations. con. In self-defense we have developed a method of determining in hardened concrete Compared with the complexity of hydraulic what was the water-cement ratio at the time of cement concrete and cement pastes. As everyone present will know. were studied. respect to the membrane that governs the degree Returning now to Mr.02 and convert to xonotlite. the close to that of water. It is. from the writer's experience with the industrial Normal chemical analysis determines the quanti. dence that factors other than viscosity may be I should be interested to have Mr. independent of water-binder ratio and bears a Pure water on the other hand explored a larger simple linear relation to the binder (i. flow. Bearing in mind that the calculated ratio involves chloride would fill the water voids but not enter the combined error in each instance of four separate the air voids. products. combined lime and "soluble" silica in a large During the research which led to the develop.28 with over two- and the effects of entrained air and honeycombing thirds of the analyses between 0. From all cooling in carbon tetrachloride under vacuum. Powers' important. This is consistent with the concept of a network lower drying shrinkage. When the ratio of lime to silica is increased Now the viscosity of carbon tetrachloride is very further. voids were visible in the surface of the specimen. E. This is a penetrable by fluids having the viscosity of water. precipitate constricts the capillaries and opposes facturers Research Department. I should of penetration into air voids. Bessey cement is bad. The air voids on connection between chemical and physical studies the other hand are penetrated by water. the more setting. volume absorbed. Discussion cement ratio. When lime water It is perhaps more interesting in relation to the was used the volume of it absorbed by the speci. Powers. common practice on the part of the contractor to assert that the G. formed by direct reaction of The procedure is to dry a piece of concrete at well-compacted mixes of lime and silica in high 105 °C to constant weight and to place it after pressure steam.2. comments on these facts and opinions. Powers' paper. have some evi. the related almost linearly to lime content over the greater the concentration of lime the less the practical range. who discussed the effect surrounded by a membrane which is impermeable of impurities on cement strength.e. less effective bonding agent and has also. It therefore appears that available surface of silica becomes saturated and a in dry hardened concrete the water voids are dicalcium silicate hydrate is formed.9 and 1. by penetrating into air voids. and setting can be calculated. several liquids were investigated lime:silica ratio of only 0. presumably content over the practical working range. simple hydrous calcium silicate bonded products. It was found that carbon tetra. lime) volume than did carbon tetrachloride. it seems that the initial reaction in their ties of cement and of combined water so that the hardening is normally to form "amorphous" CSH.. giving usually an accuracy of ±0. It is conceivable that precipitation penetrating power and it is remarked that this of lime occurs when water containing alkalis passes may be due to viscosity effects. total water and water-cement ratio at the time of which slowly converts to a tobermorite phase. In these materials the water content used in In that case any carbon tetrachloride taken up by pressing is low. sand-lime bricks. This suggests that an air void is II by Welch and Gutt. and not at all by carbon from this point of view was the paper in Session tetrachloride. It is by virtue to an inert fluid having the viscosity of water of physical properties such as strength that cement but permeable to water to an extent depending on is used in practice so that cement chemistry is its content of dissolved lime. and insufficient to fill the voids in the open voids drained out when the specimen the pressed brick. offer a clearer picture.

C. This bears fronted by every writer on this subject. the values are not shown on much until the period of set begins. The effect of lime content upon drying shrinkage of hydrated calcium silicate bonded (sand-lime) bricks Czernin. Several of its important mental pictm'e of the structure of cement paste properties may have any one of a wide range of with simplicities not requu-ed by the basic concept characteristic values. It is not necessar}' 611 630446 O —62 4 . and they remain separated depart from rising linearity and then drop. This thj'oughout the dormant period. There is a natural tendency to endow a teristics— far from it. either axis. Figure 2. a statement that is true about the characteristics of a particular specimen. producing the stiffening reported by Mr. LIME (CoO ) COWTEKT Fl GUBE 1. The effect of lime content upon strength of hydrated calcium silicate bonded {sand-lime) bricks may appear in one or another of innumerable possible combinations. a paste having a high particle concentration at the Since the actual values for given lime contents. The diffi. and these various values on which the picture is based. different parts of a single specimen may show significantly different characteristics. just as does curves passing through all the experimental points. of cement paste or concrete is liable to be untrue if applied generally. Powers phj^sical factors as well as chemical ones. Czernin per- One of the questions exemplifies a difficulty con. and its consistency does not change very and other factors. The form of curve remains the same Thus. except in the isshown in figures 1 and 2. and cementation probably would begin immedi- ately. right or wrong depending on whether or not one speaks of a paste in which the cement particles are Closure highly concentrated . if a paste is initially soft enough cement grains are well to bleed. Such behavior is typical of pastes in which the cement grains are initially very close together. "normal". For such reasons. Furthermore. which are reproduced sediment that accumulates on the bottom of the from actual data using a particular sand with the container. some questions that deserve careful consideration. paper. Czernin 's argument on this point is for all sands. or class of specimens. on the concepts of gel structure and the evolution culty is that concrete is not a substance that can of that structure dealt ^^^th in section 6 of my be described in terms of a few single-valued charac. and start. Mr. tained to the mechanics of h^'dration. tion. Another question raised by Mr. Czernin raises about another class. is true about one class of pastes may not be true lems of Cement Hydration". The bottom part stiffens. The point to be emphasized is that setting and cementation depend upon T. are dependent upon the fineness of the sand. Czernin cited data showing that cement paste becomes progressively firmer throughout the period we have called dormant. the strength and drying shrinkage versus lime first separated by water. Mr. does not undergo cementa- place. It would not bleed. which the contents at which inflection of the curve takes bleeds at constant rate. but the upper part of the sample. LIWE CaO ) CONTENT (. On the other hand. consistency and a water- cement ratio of about 0.23. Such a paste might have standard. Mr. and what In his remarks entitled "A Few Unsolved Prob.

The constant gel porosity. tables 5-11. at least theless has limitations and uncertainties common in the early stages. the higher the excess from air bubbles and the like. 50. Bessey's statement Development Laboratories. and thus the total volume of voids duced. the effect is detail in a paper prepared for the January 1961 to change the ratio between the volumes of the issue of the Journal of the PCA Research and two classes of voids. tances from the source material is the result of sand-lime brick. strength.to suppose that cement gel grows in such way as pared. . However. tetrachloride and in water." . within the error of the given specimen can hold. The granules were free present. air-filled spaces and the residues of originally When any region becomes filled to this degree. Czernin argues. then the strength should dry gel. Bessey's remarks and data bring out a point deposited on any solid surface that is itself not about empirical strength relationships that is producing gel. gel porosity is not the only factor deter- merely to translate outward the original contours mining the strength of gel itself. Also it is prob. air voids have about the same weakening hydroxide crystals formed early in the process. Being incomplete. At the start. such surfaces would sometimes overlooked. affect viscosity also pull the water molecules into rial distributed in the gel. . Czernin said. effect as an equal volume of the much smaller Since gel substance is insoluble. This in solution very long. and lime. p. the spaces be- surface diffusion. remain the same. I know of no corresponding effect on rate of penetration of a dry theoretical basis for predicting that the porosity specimen was not tested.This observa- Mr." Actually. cement content. and it seems contrary to observation. whatever the amount of unhydrated mate. in 1947. if the properties of the hydration product term derived from other contexts. interstitial space is varied while holding the I have dealt with this question in considerable amount of reaction product constant. The arguments in indicates that such a change does not affect the that paper lead to the conclusion that no single strength.) These experi- not a theoretical basis for such an assump- tion. Bull. show that. water. strength is higher. Dr. that the means that void-size does not have much effect on development of gel one sees at considerable dis. we know that even though the specific volume of dry hydrated cement in the porosity of the hydrated part of specimens is the form of small granules obtained by crushing and same when different amounts of excess cement are sieving hardened paste. The same seems to be true of pressed. but it is nevertheless ably true that when different cements are com. it cannot remain capillary spaces in the hardened paste. water-filled space has become filled with reaction Regardless of where the gel substance is pro. Research Lab. some of the where gel particles can grow. ing the viscosity of water . Portland the hydrated part of the paste. After curing. thus maintaining a clear relationship between strength and gel-space ratio demarcation between cement gel and capillary is more general than the relationship between space. gel substance seems to be Dr. I should like to suggest of the heterogeneous mass of hydration products that viscosity is not likely to be a factor in the should be always the same when excess cement is phenomena discussed by Mr. 43. perhaps of my Symposium paper.72 cc of solid per cc of space. the gel substance being essen. it cannot achieve a density that averages becomes the sum of the volumes of the original more than about 0. but it never- as many other observations. it seems. PoUitt's observations pertaining to the we lower the w/c from 0. product. Pollitt referred to the remark that alkalis to designate adequately the complex process hy "reduce the rate of flow through paste by increas- which cement becomes hydrated. Czernin's final comment pertains to the tion pertained to the rate of flow through a matter of gel porosity. In concrete. or topochemical reaction is able Mr. cement gel does not grow that to most empirical relationships. With. Mr. tween sand grains are filled with three components: tially in the adsorbed state until it reaches a place air. Viscosity present regardless of the amount of excess. we can only say that this has but not necessarily the amount of fluid that a been observed to be true. such as solution. Czernin's excellent pictures as well strength and water-cement ratio. therefore. Am. clear that the apparent specific volume of dry 612 .000 times the normal value (table 13) and yet "If the porosity of the hydrated part remains the it penetrates dry gel readily.4 to the lowest possible penetration of dry concrete by different fluids ratio. Pollitt.. affects the rate of flow under a given driving force. he refers to a "theory of saturated specimen in a permeability test. As ments had to do with the determining of the Mr. 22. 693. It seems therefore that we must the data do not permit direct quantitative com- conclude that gel strength alone does not define parison between the results obtained in carbon strength of a specimen of paste. and-precipitation. water-filled spaces. Before curing. The point is that in be presented by aggregate particles and calcium concrete. Concrete Inst. Mr." This is a logical argument based on the brings to mind certain experiments we reported tacit assumption that the strength of a specimen (Proc. The same forces that same. of cement paste depends only on the porosity of 1947. Water in gel pores experimental determinations. way. If the water content of the we say that the region is full of cement gel. as shown in figure 1 seems to have a very high viscosity. out such a theory-. The empirical of each cement grain. I do know of Cement Assoc. 693. remain at the same level or at least near it when Mr.

3 quantities involved. considering the solubilities and the carbon tetrachloride molecule is about 5. by Mr. The amount of difference also carbon tetrachloride cannot cause the gel to between absorption of pure water and lime water swell appreciably. it is not surprising that carbon observed by Mr.9 .. illustrates the point. and probably not to a difference of penetration into to the degree to which the specific volume of the air voids. and it is theoretically possible for at least the very small air bubbles to Molar Apparent become filled with water.429 action out of doors can be considered practical evidence that air voids do not become filled with water under ordinary circumstances. However. The subject is It does seem likely that the solubility of lime discussed briefly in the third paragraph of section and alkali compounds in water should make more 3 ofmy Symposium paper. the effectiveness of entrained air in pro- Acetone 73 4. that the differences in penetration found on the degree to which the fluid used in the experi. cc cc/g Water 18 1. fluid is influenced by adsorption. and gel pores in each of the granules of the sample. Pollitt was not stated. When a dried specimen of concrete is submerged in water the gel pores and capillary spaces seem Apparent Specific Volume of Hydrated Cement {Cement 20Q) to become filled readily. that the time required for an appreciable fluid (temp. The apparent specific volume depends mostly therefore. Similarly. molar volume of 23°) volume hydrated amount of water to enter air bubbles or any voids cement of that kind is very long as compared with the duration of an ordinary laboratory experiment.395 Indeed. Relative specific however.hydrated cement is considerably higher in organic tetrachloride is unable to penetrate cement paste Hquids than it is in water.0 . Since water. space available to water than to a fluid in which As pointed out in the Symposium paper the these salts are insoluble. It seems. Displacement volume. Pollitt are probably due to differences in ment is able to penetrate the capillary spaces and size of molecule and degree of swelling.408 tecting the cement paste in concrete from frost Toluene. and since factor could be large. 613 . It is our experience. an aqueous mean width of gel pores is estimated to be about solution of lime should find less space than pure 5 times the diameter of a water molecule.0 0. it seems unlikely that this times as large as the water molecule. The following table to the same degree as does water. 106 5.

.

C. II est brievement rendu compte des methodes experimentales d' etude de structures coUoidales. Different types of compounds appearing in the hydration of portland cement are reviewed separately with respect to their microstructural properties. various possibilities of arrangement of their lattice structures are suggested. Au sujet des phases d'aluminate et d'aluminoferrite hydrate ainsi que des sels complexes du type hydrate de sulfo-aluminate. en commen- gant par differentes modifications de I'hydrate de chaux. The problem of ettringite formation is treated in relation to some phenomena observed in cement slurries. type of cement used. Les donn^es experimentales des etudes de structure de silicates hydrates en pate sont comparees aux observations correspondantes des phases de gel dans les pates de ciment portland. il est propose differentes possibility de disposition des elements de structure dans le r^seau cristalline. starting with various modifications of lime hydrate. Resume Le present compte-rendu n'embrasse pas seulement les microstructures de produits en pate durcie mais ^galement la formation de j^'seaux cristallines et les modes de cristalli- sation de differents composfe se rapportant a rK^. Les modes de cristallisation et les principes de la formation de structure dans les phases de tobermorite fait I'objet d'4tudes d^taillees parti- culieres. The general principles governing the formation of lattice structures in calcium siheate hydrates are described and illustrated by reviewing a number of crystal structures deter- mined. Diff^rents types de composes apparaissant lors de I'hydratation du ciment portland sont traites separ^ment en ce qui concerne leurs propri6t4s micro-structurales. and represents a compilation of available data obtained especially in electron-optical studies. ce compte- rendu est egalement une compilation des donn^es disponibles obtenues essentiellement par des etudes optiques ^lectroniques.. Sweden. L'apparition des phases de plaques hexagonales des hydrates d'aluminate de calcium dans les pates est decrite. Les principes gen^raux r^gissant la formation de reseaux moleculaire dans les hydrates de silicate de calcium sont d^crits et illustr^s par la presentation d'un certain nombre de structures cristallines determin^es. Les produits d'hydratation de silicates de calcium purs sont etudids du point de vue special des processus d'hydratation de la pate. 615 . D.\jratation du ciment en general. Paper V-2. Les donnees obtenues lors d'etudes de la structure des pdtes sont en desaccord avec un certain nombre de theories courantes sur la microstructure des pates. Contribution from the Swedish Cement and Concrete Research Institute. with particu- lar reference to the use of electron microscopy and electron diffraction in the examination of cement hydration products. le rapport eau/ciment. Washington. une attention particuliere ayant et6 accordee a I'emploi de la microscopie elec- troniqueet dela diffraction 41ectronique pour I'examen des produits d' hydra tation du ciment. 1960. de meme que le probleme de la formation d'afwillite. The hydration products of pure calcium silicates are reviewed and considered with special reference to paste-hydration processes. Le probleme de la formation d'ettringite est traits par rapport a certains phdnomenes observes dans les laits de ciment. The data obtained in structural studies of pastes are found to disagree with a number of current theories on paste microstructure. and curing conditions. le type de ciment utilise. La necessite d'une revision de ces theories est soulignee et il est fait mention de quelques problemes revetant une importance particuliere pour les recherches futures. fineness of grinding. •Fourth International Symposium on the Chemistry of Cement. The appearance of hexagonal-plate phases of calcium aluminate hydrates in pastes is described. Ces donnees sont illustrees par un certain nombre de microphotographies eiectronique d'echantillons de pate de descriptions diverses choisies dans le but de mettre en lumiere I'influence de facteurs tels que le degre d'hydratation. and a few problems of special importance for further research are indicated. The Microstructure of Hardened Cement Paste Ake Grudemo Synopsis The review covers not only the microstructures of hardened paste products but also the formation of lattice structures and the crystal habits of various compounds related to cement hydration in general. as is also the problem of afwillite formation. selected in order to elucidate the influence of such factors as time of hydration. The need for a reconsideration of these theories is emphasized. Experimental methods of studying colloidal structures are reviewed briefly. and the complex salts of the sulfoaluminate hydrate type. The experimental data from structural studies of paste-hydrated silicates are compared with corresponding observations of the gel phases in portland cement pastes.. The crystal habits and the principles of structure formation in tobermorite phases are treated in particular detail. la finesse du grain et les con- ditions de durcissement. These are illustrated by a number of electron micro- graphs from paste samples of various description. w/c ratio. In dealing with the hydrated aluminate and aluminoferrite phases.Stockholm.

are func. werden mit entsprechenden Beobachtungen der Gelphasen in Portlandzementpasten verglichen. Heller and Taylor [5]. Betreffs der hydratisierten Aluminat. Die Kristallformen und die Prinzipien der Strukturbildung in den Tobermoritphasen werden besonders eingehend behandelt. (in order of time of publication) of which depend on how the particles of solid Kuhl [1]. 616 . tiber die Eigenscliaften von verschiedenen in der Hydratation des Portlandzementes erscheinenden Verbindungen wird einzeln berichtet mit besonderer Riicksicht auf deren mikrostrukturellen Eigenheiten. erhaltlichen Angaben dar. Diese werden durch eine Anzahl Elektronenmikrophotographien von Pasteproben ver- schiedener Typen illustriert. phy of components within the cement hydration thing to do with cement research. die in der Absicht. das Wasser/Zementverhaltnis. the litera- positions. shapes. den Einfluss von solchen Faktoren wie die Hydratationsdauer. written on the subject of the chemistry of cement ables such as. die Zermahlungsfeinheit und die Erhartungsverhaltnisse zu erlautern. the sizes and distribution textbooks. tiber Versuchsmethoden fiir die Untersuchung von KoUoidalstrukturen wird kurz berichtet mit besonderem Hinweis auf die Anwendung der Elektronenmikroskopie und der Elektronenbeugung zur Untersuchung von Zementhydratationsprodukten. Zusammenfassung Dieser Bericht umfasst nicht nur die Mikrostrukturen von geharteten Fasten sondern auch die Gitterstrukturbildung und die Kristallformen verschiedener. and its constituents up to 1956. etc. and com. the initial chemical and mineralogical viewed in a series of principal papers presented composition. there- the amount of mechanical work applied during fore. der Typ des verwendeten Zementes. Die Hydratationsprodukte von reinen Kalziumsilikate werden mit besonderer Riicksicht auf die Hydratationsvorgange in Pasten behandelt. as well as 1956 [6]. matters pertaining to the microstructural prop- edge of the microstructure it would therefore. the initial particle size distribution. on cement chemistry in London. sium [2]. paste is a complex and highly variable system of However. Bogue [3]. Lea [4]. Introduction As is generally recognized by all who have any. most exhaustive account of practically everything tionally dependent on a number of primary vari. place in the reaction of cement with water. sowie auch das Problem der Afwillitbildung. Die Notwendigkeit einer erneuten Priifung von diesen Theorien wird hervorge- hoben. In the present review. only papers dealing more specifically with mixing. Die Bildung und Eigenschaften der hexagonalen Plattenphasen der Kalziumaluminathydraten in Zement- pasten wird beschrieben. the dynamic variables (temperature and water vapor latest developments in cement research are re- pressure). ausgewahlt worden sind. ture on cement has been enriched by a number of and of pores and voids. paper. dass die durch Strukturenuntersuchungen von Pasten erhaltenen Angaben nicht mit einigen. 1952. It can of course be stated quite generally that the and the Proceedings of the Moscow Symposium microstructural properties of this system. at this Symposium. viz. besonders durch elektronoptische Untersucliungen gewonnenen. Es ergibt sich. In addition. be necessary to consider first referred to and commented on in connection with the fundamental theories of the processes taking the discussion of various microstructural problems.^ the Proceedings of the London Sympo- matter are aggregated and linked to each other. vorlaufig giiltigen Theorien tiber die Mikrostruktur von Pasten iiberein- stimmen. erties of cement-hydration constituents will be strictly speaking. of water in different states of fixation. sich auf die Zementhy- dratation bezielaende Phasen im allgemeinen. wobei mit den verschiedenen Modifikationen des Kalkyh- drates begonnen wird. die durch Struktur- untersuchungen von reinen. Taken together. In review of the present-day knowl.und Aluminoferritphasen sowie der zum Sulfo- aluminattype gehorenden Komplexsalze werden verschiedene Moglickheiten ftir die Anord- nung ihrer strukturellen Elemente in Gitterstrukturen vorgeschlagen. Die allgemeinen Gesetze. hardened cement system. und stellt eine Zusammenfassung der. and to refer to a great number of papers dealing with 1 Figures in brackets indicate tlie literature references at the end of this different aspects of the chemistry and crystallogra. Die Ergebnisse. Das Problem der Ettringitbildung wird erortert mit Bezugnahme auf einige in Zement-Wasser-Mischungen beobachteten Erscheinungen. these books provide a of any multicomponent hydrate system. und einige Probleme von besonderer Bedeutung fiir weitere Untersuchungen werden angegeben. in addition to time and the thermo. in Pastenform hydratisierten Silikaten erhalten sind. die die Bildung von Gitterstrukturen in Kalziumsilikathydrate regeln. since the time of the 3d Symposium particles of a manifold of sizes. werden beschrieben und durch eine Anzahl von experimentell bestimmten Kristall- strukturen illustriert.

Some writers have preferred to use chemical symbols (C-S-A-F-H). as a construction material in cement and concrete In general. and the aggregation consisting of small and mostly distorted crystals of such particles to condensed systems. systems of colloidal particles. in order to clarify in some degree the London Symposium. at present. may of Moscow and Berlin. at present well on their way to being overcome. it was emphasized that al- present situation regarding the various possibilities though at that time a comparatively large number of exploring the microstructures of hydrogels. crystallographers as nearly insurmountable. and of cement with water. versity of Aberdeen and the Academies of Science The morphologic aspect. and the spatial arrangement of that most of the compounds formed in the reaction ions and ionic complexes within the unit cells. and XD = X-ray diffraction. and related to the reaction chemical theories of structure formation. being closely related to problems more general problem of structure formation in concerning the physical properties of cement paste.g. mineral names even for ill-crystallized and ill- bols S=S03 and C = C02 are introduced for the defined products exhibiting certain structural or sake of uniformity. there is the morphologic aspect. on the other hand. Some problems of and mineral names will therefore be used alter- nomenclature concerning compounds within the natively or in combination. Definitions and Nomenclature Only materials and paste constituents within C-S-H system still remain unsettled. Bogue's nomenclature of subindices. are only available volving the shape. less well-crystallized particles. In the present review. The reason for this. Second. if not otherwise stated. to Compounds of unknown or unspecified compo. and the term which aimed at reconciling the various names "cement" is used in the meaning of portland and notations used. will be considered in this review. as the case may be Experimental Methods for Studying Microstructure General Remarks Crystallographic Methods The subject indicated by the heading is in itself In the review of the structures of cement hy- a very large one. among other things. of uniform symbols. or in of crystal structures related to the cement hydra- general. micro. e. especially within the C-S-H of crystallography pertaining to the repeat units of system. cement hydration processes will be considered in especially its mechanism of deformation when used later sections of this review. the oxide sym. Further abbreviations used stoichiometric analogies to some natural mineral are EM = electron microscopy. size. It should be recognized that the concept of it would be necessary to develop essentially new microstructure can be divided into two distinct XD techniques in order to attack the structures of (although closely related) parts. are It is evident that the development of ideas con. ED = electron previously described in the literature on mineral- diffraction.. find a completely unambiguous denomination sys- sition are generally denoted with dashes instead tem. by the nature of the It seems that the situation has changed radically interparticle links and the distribution of pores in this respect during the last few years. terized. ogy. haps been given by crystallographers at the Uni- water system. ED). charac. First. possessing large unit cells of low symmetry. and only a few comments will be dration compounds given by Bernal [8] at the given here. It does not seem possible. and studied in magnified images (light culties which were at one time regarded by many microscopy and EM). using refined methods of XD obtained using analytical methods of study of a analysis and applying recently developed crj^stallo- more conventional type. tion compounds had been at least partly resolved. was crystal structure. single-crystal or powder XD techniques have 617 . and distortion of more or' as materials of a very low degree of crystallinity. due to the combined efforts of crystallographers dration constituents is largely dependent on data all over the world. The chemistry and the phase diagrams of hydration most notable contributions in this field have per- systems which are less complex than the cement. in. other compounds of great interest in the field of structure can be considered from the point of view cement hydration. Diffi- and voids. The nomen- the general field of portland cement hydration clature proposed by Bogue [7] some years ago. it was stated. C-S-H. or occurring as natural minerals. and at introducing a system cement. has not been generally In addition to the accepted abbreviations of accepted. cerning the crystallographic aspect of cement hy. rational applications of standard technology. including hydrothermal studied by means of diffraction methods (XD and products of autoclaving at elevated temperatures. Some implications of the serve to bridge the gap between pure and applied results of these studies of crystal structures for the cement research.

specifically. paring and mounting EM samples. which have become the standard X-ray crystallographic is thus themaximum particle thickness permitting equipment in most cement research laboratories. The monochromator Owing to these circumstances. crystal-orienta- especially theclinker minerals themselves. compressed. and major contributions to the advance. screen. However. the temperatures reached in this way. or high accuracy of intensity measurements particles is essential. are often observed the presence of weak reflections which are. and strongly employed in electron-density syntheses. It is mostly difficult to estimate analysis of mineral composition. Even these comparatively low temperatures. this is not nearly higher level than is strictly necessary for observing true with regard to the application of ED tech. in to be reversed or strongly changed in EDdiagrams diffractometer records. either masked by strong of the same sample.crystal powder pattern of a mixture of crystals gives the diagrams have sometimes been useful in the deter. Crystal-structure transformations and 618 . However. and ED methods are therefore particu- for which either high resolution. it a quantitative determination of relative intensities would seem very difficult to avoid selective crystal. high recording larly suitable for studying structures of colloidal speed. since too It would seem that whereas different ways of many uncertain factors are involved. or streaky. between stronger and the special problem of detecting with certainty weaker intensities of reflections. with focusing monochromators. This is also the chiefly because of the high adaptability of this upper limit rather arbitrarily set for colloidal par- type of instrumentation for different purposes. as camera is superior to ordinary diffractometers for measured in XD diagrams. strongly pronounced. or not clearly resolved from selected-area ED patterns can hardly be used for the background ripple. using ED methods. the temperature of at least the thinner ment of knowledge of cement hydration processes particles or particle aggregates in a sample is prob- are probably to be expected from investigators ably well below 100 °C. and when care is taken ining crystal structures related to cement hydration not to increase the electron beam intensity to a have been well demonstrated. have been commonly applied in cement research Since the EDpattern is a practically undistorted in recent years. patterns and to decrease the exposure times. tion characteristics. pecially in combination with the vacuum main- mally incorporated in different types of three. It can thus be stated that reflections nearby. powder layer samples used in mono. steam-cured C-S-H mixes. Some typical uses of the recording It is an intrinsic property of colloidal particles.g. Electron-diffraction single. almost completely absorbed by solid matter of an Miiller tube counter and recorder units seems to effective thickness of the order of 1. e. or tion effects are also. are powder cameras equipped image of a section through the reciprocal lattice. Afurther consequence of the strong interaction chromator cameras. On the other hand. extremely struction of delicate or ill-crystallized hydrate small sample volumes can be singled out and structures. crystal symmetry. which make them less suitable of high-speed electrons with matter is that the for measurements of relative line intensities in sample particles will be heated when hit by the general. This possibility constitutes the main differ- vated temperatures have been studied by in situ ence between XD and ED methods. stage electron microscopes. In this way even minor well-defined crystal specimens for investigation. using XD methods to their full capacity in exam. ED diagrams to be obtained.proved to be sufficiently effective. tained in an electron microscope or ED instrument. restricted in one or more directions This restric- talline compounds. superimposed patterns of different constituents mination of the true unit cells. Other instruments which density distributions of ellipsoidal symmetry. the relationships. the X-ray diffractometer with Geiger. whereas the ED patterns vary from one microcameras have in some cases been success.000 A. es- Facilities for recording ED patterns are nor. of the so-called reciprocal lattice into reciprocal width measurements. to have their crystal lattices of the mineral composition of a mixture of crys. diffractometer are.. constituents of a composite material can be and preparation and mounting of crystals must detected and studied with respect to their diffrac- often be executed under the microscope. and unit tural transformations during dehydration at ele. crystal elements are often observed to be diffuse tion lines frequently appearing in powder photo. in the same way as diffracted X-ray intensities are orientation effects in the very thin. which are capable the ED reflections from very small and distorted of resolving even the very closely spaced diffrac. ticle size. the quantitative analysis if at all crystalline. cells. Struc. Because of the special method of pre- grams from a mixture of cement hydration minerals. with intensities proportional to their relative Microfocus X-ray tubes in combination with amounts. and recording the images visible on the fluorescent niques. under normal conditions. The electron beam in an electron microscope is However. Because of the strong are stUl sufficient to cause dehydration and de- interaction of electrons with matter. and the determination of tion is equivalent to a degeneration of the points particle size from small-angle scattering or line. An XD XD techniques. for a quantitative electron beam. and. as a rule. point to another in a suitably prepared sample fully used to improve the resolution of powder scanned with the electron beam. examined by means of the so-called selected-area great care must be taken to select suitable and diffraction technique.

Although in most cases the effects of heating can be considered negligible. tion rates of the various cement compounds differ. but the many cases sufficient to remove all volatile oxide remainder cannot be differentiated". changes in morphology are frequently observed in ment of statistical methods of synthesizing data EM studies. It has also been necessary to single crystal specimens. According to substance studied. liquid-air cooling. the surface. to leave the original microstructure intact recorded for one particular. To provide components from the specimen and transform it to a representative example of the information some anhydride compound. and imdisturbed hy various treatments. it is clearly incapable of revealing the finer ture or habit takes place and can be observed as a details of paste structures. Disregarding the use of develop special methods of sample preparation devices for tilting the specimen through a limited and EManalysis. In I j lattice restrictions and distortions. stillan indispensable tool in the examination of niques applied to the study of metastable and microcrystaUine mixtures and hydrated paste easil}^ decomposing cement hydration constituents. tively. and a device for tem. and of increasing the intensity of the electron beam suc. Lea can fmther be quoted on some analysis of the hydrate originally present. hydroxide crj^stals can be detected and. and studies of residual unhydrated constitu- be drawn by using the technique of deliberately ents of the course and rate of hvdration. ceeded by a gradual penetration into the particle advantage inherent in this technique. the constitution and ED data would therefore necessitate the employ. the hydration of the coarser particles electron-microscope constructions. pastes. the distribution of larger pores and voids in cement cessively. structures. Asystematic approach to the analysis tacitlyassumed that as long as saturation is main- of crystal structure by means of single-crystal tained in the solution phase. and because of electron beam is completely absorbed in them. graphic examinations of coarsely crystalline struc- pounds of unknown identification can sometimes tures. paste. "who noted that manj^ of the cement par- particles heated by the electron beam cannot be ticles were composite and that hydration pro- easily controlled or measured is evidently a dis. and mostly arbitrary. which is in sionally. and. interesting observations made by BrownmiEer The fact that the temperatures of the sample [11]. and can sometimes be expected to obtained from a sufficiently large number of have occurred even before the electron beam is single-crystal patterns. the EDpattern of obtainable in microscope examinations of cement which can be used for identification and further hydration. whereas it can be used to since it may lead to serious misinterpretations of advantage for such purposes as the surveying of various observations made. although in from the surface. morphologic properties of the sofid hydration 619 . occa- grees C can be reached in this way. and The particle aggregates in hydrated cement very nearly monochromatic. untH a transformation of crystal struc. electron beam. but it seems out the whole spectrum of crystal reflections by that different constructions of electron microscopes rotating a specially selected crystal about one or have only during the last decade reached such a more of its principal crystal axes cannot be em. cement paste specimens. in order to avoid contamina- angular range [9]. Except at. one can Morphology Examination Methods never be quite certain that some unobservable change has not occurred. many cases of minor importance. switched on. It is possible grains seemed to hydrate more slowly than the that such devices will be of great use in future sUicate crystals Acontent of 10-15 percent of research on the microstructural properties of Ca(0H)2 crystals was observed after 28 days". mineralogic and petro- On the other hand. high degree of reliability and versatility that they ployed in ED. or near. materials are mostly so large and dense that the Because of these circumstances. thermally unstable colloidal materials. the above-mentioned effect of various t3"pes of only the edge structures being observable. Lea [10]. The use of electron microscopy in cement hydra- The commonly used XD technique of bringing tion studies dates back about 20 jv. in the case of hardened represents a section of the reciprocal lattice. many investigators have j homogeneous sample frequently exhibit a wide selected to study the looser and more well- range of variation. This circumstance is The ordinary light microscope is undoubtedly obviousl}" a disadvantage of EM and ED tech. and have therebj^ observed. setting of a stationary crystal in a stationarj-. In special therefore. on account of obvious difficulties can be used for analytical work of a more sys- of design caused by the extreme minuteness of the tematic character. owing to the limited radical change in the ED diagram given by the resolving power of visible light. calcium sulphoaluminate. as compared vrith. and each pattern must be dispersed structures obtained in mixes containing interpreted separatel}'^ in order to determine the a large excess of water. a single-crystal ED pattern tions and artifacts. the w/c sj^mmetry elements common to all the patterns ratio normally used in pastes. valuable conclusions on com. single-crj'stal order to facilitate the EMexamination of cement- I patterns by dift'erent crystals within a given water reaction products. the specimen may not be very selective even though the hydra- stage is equipped with electric heating. "in a powdered set cement calcimn Temperatures amounting to several hundred de. or alterna. Some of the interstitial material in the cement perature measurement and control. However.

ordinary methods of grinding replica material. preferably by placing this film in the EM methods. As stated previously. various observations must largely be overcome by structural details relatively unaffected. 21. forming early stage of cement hydration. distor. was identified as having the appearance of rounded In the highly supersaturated lime solution thus platelets or spherulites which were found to grow formed. it can be spread out in a thin and the examination of more-or-less dilute suspensions evenly distributed layer by means of electrostatic can give a great deal of information on the micro. However. protection against carbon dioxide attack. CHis also habit was not given by the various observers. such as prolonged ball-milling or high- introducing the specimen into the electron micro.8 to mixing. CH crystals was momentarily attained. whereas patterns originating from other 3. is revealed by ED diagrams which show only a developed already in the first few hours after strong but very diffuse reflection at about 2. 620 . A technique involving replication of fresh faces. the pended in a suitable organic liquid. whereas alternatively. and it is therefore rather sur. the suspension spread over the specimen support judging from various observations.0 A. preferentially the contour structures in pore sur. IMicrostructures of Lime Hydrate IVIodifications Immediately upon mixing cement with water. 19. will almost inevitably lead scope has been devised by Bernard [14]. hydrated constituents are mostly weak and can be The reason why CH precipitates in amorphous identified with certainty only by means of special structures rather than in its natural crystalline methods [17]. the powder can also be sus- uents in their most well-developed states.product formed cannot differ very much from those dust cloud formed in an elutriation column. In order to improve the dispersion structural properties of cement hydration constit. the surface. Dehydration phenomena such as disturbances of the original paste structure in the crumbling and cracking of the fractured paste form of fractures of crystalline materials. has been used by some investigators For obvious reasons. 23. and also affords an excellent hterature [1. especially detected in fully hydrated paste. including necessary for a successful evaporation of the the present author. exposure of the paste specimen to the high vacuum in the experience of some investigators. and that Macrocrystalline CH takes the form of hexagonal this unstable colloidal solution coagulated into plates or prisms. and also to serious disturbances. a content of 10 to 15 percent of saturated lime solutions formed by rapid evapora- CH crystals being observed after 28 days of tion of water or in the first stages of cement hydra- paste hydration. and other Ca ions come decomposition of cement constituents. and eventually to complete adopted by Stork and Bystricky [15]. owing to the unavoidable tion of delicately built particles. CH crystals are precipitated out at an into each other at the points of contact. on to largely hidden from inspection. however. 22. However. justified only screen. The application of certain that the suspension medium does not results obtained by using methods of preparation produce any undesirable effects of interaction involving high water-to-solids ratios for drawing with the suspended solid particles. Care must be taken. that can be cleavage surfaces of cement paste by evaporation examined by this method. spherulitic aggregates of an amorphous. such as carbon. _ diagram from portland cement paste also contains The amorphous character of these formations a strong and easily recognizable CH pattern. the only crystalline hydrate readily detectable in It was probably believed that in highly super- the microscope. whereas the internal of a thin film of highly resistant. etc. it is necessary to break in later years. surface may occur. Difficulties in interpreting and milling apparently leave most of the micro. 24] (fig. and is likely to cause very little damage to Clearly. of EM specimens. Ordinary conclusions on the microstructure of setting and laboratory ethyl alcohol has been found suitable hardening pastes has also been criticized in the in this respect. frozen-in prising to find that in all the earlier studies of structure. A few percent more may be tion involving rapid dissolution of lime. power ultrasonoration. and chemically texture of aggregates of the colloidal particles is inert. and a drop of assumption of equality of phases is obviously. paste. to make to a rather limited extent. whether formed Ca ions are dissolved from the surfaces of the by precipitation from pure solutions or by the calcium silicate crystals. methods [16]. or of products formed in a paste. It is disintegration of hydrated structures. depending on the from C3S. any procedure used is likely to cause the structure. precipitates from lime solutions. solid CH from the dissolution of gypsum admixtures. 1). and at the edges of cracks. 20. 12. 13]. This method is supposed to give a down hardened paste into a finely subdivided true reproduction of the internal structures of the powder before it can be examined by EM methods. The dry comparing the replica structures with those powder can simply be dusted on to the specimen observed in dispersed specimens using ordinary support film. The XD long chains [18. a high concentration of small nuclei of clinker mineral composition. A method of letting extremely thin layers of The application of more powerful techniques cement paste of normal w/c ratios hydrate directly for achieving a high degree of dispersion of paste on the specimen screens for various times before specunens. amorphous material.

with the The normal appearance of CH crystallized in mixing water is observed to prevent the growth cement pastes and in suspensions protected from of larger crystals. These authors also explain the effect of microscopical observations on paste-hydrated ma- sulfate in the mixing water as an increase in the terials from these studies confirmed the presence rate of hydration and formation of the hexagonal- of considerable quantities of more or less amor- plate phase. they suddenly changed appearance. carbon dioxide for a longer time. for some time in a state of very fine subdivision. CC. owing to lattice disturbances in the superposition of growing hexagonal-plate caused by inclusions of silica groups. two parallel edges. 23]. appeared in the ED diagrams. the be imagined to be composed of disordered introduction of a small amount of a siu"face- aggretates of extremely small CH crj^stal nuclei. is shown in figure 2 a. platelets. CH crystallization of be disturbed by the will Ordinarily. the microstructure of the C-con- presence in even of certain the lime solution taminated precipitates is that of chains of rounded ionic groups other than C anion. or at least very poorly lime mortar as being caused by bond formation crystallized. the precipitates forming in a supersaturated lime solution are observed to consist of well-crystallized particles. exhibiting internal striations for stoichiometric reasons. Similar observations were reported earlier by Eitel [19]. One such plate in aggregates with sawtooth-shaped edges [20. found that certain quantities of CH which. b. of cubic symmetry. Tikhonov Shepinova. 31. Electron- crystals. such as Kantro. or thin slabs with stoicbiometry of C3S and /3-C2S hydration prod- ucts. 22. The identification of 21. and the hardening mass remains the attack of carbon dioxide is that of plates. C. active agent. These are sometimes of comparativeh* large In specimens containing lime solutions which dimensions. but other formations of a similar nature and behavior have also been observed. 32. mostly appearing as rhombic slabs. was given rather recently by Schimmel [25]. but were obviously over- looked by later observers. show that atmospheric Figure 1. A simple. predom- inantly plates or slabs of a more or less regular. but mostly have been exposed to the attack of atmospheric so thin that they are transparent to electrons. These latter formations are obviously very easily decomposed in precipitated in the samples. superimposed reports [29. and apparently correct. CC has been Examples of electron micrographs and ED observed to take the form of regular crystals of patterns of CH plates in cement pastes and calcite. description of the true nature of the spherulites. Brunauer. with similar preparations have been given in various the crystalline angles of calcite. hexagonal shape. and simultaneously the easily recognizable pattern of calcite. may act as an inhibitor of the formation of the CH crystal lattice of regular hexagonal symmetry. part of which could well carbon dioxide attack. being replaced by a ring pattern of calcium oxide. state. Calcium hydroxide chains precipitated from solu- carbon dioxide readily taken up by lime solutions tion in the presence of carbon dioxide contamination. Further heating in a beam of strong intensity caused the CC pattern to disappear gradually. more than 1 ix across. If rigorous precautions are taken to avoid carbon dioxide contamination. He observed that when spherulitic lime aggregates were moderately heated in the electron beam. and Nabitovich the hydrolysis of the anhydrous silicates was [27]. the ED diagrams showing counted for by the data obtained in XD quanti- modified single-crystal spot patterns from oriented tative mineral analyses. The most probable aggregates of calcium oxide. could not be ac- the electron microscope. were expected to be or cracks parallel to these edges [26]. at least those formed from pure lime solutions. On the other hand. in studying the networks of fibrous particles. who also described the hardening process of present in an amorphous. CH crystals is readil}^ made by means of their It can be expected that the process of regular characteristic and mostly well-developed ED 621 . combined with an inhibition of phous-looking matter. 33. or considered to be of an occasional nature. These observations which have been confirmed by the present author. 30. sulfite alcohol waste. and Copeland [28]. explanation for the apparent absence of this quantity of lime was found to be that a minor Similar phenomena have been observed by part of the amount of solid CH resulting from Andrievskii. 34].

An example is given in figure 3 a.l) lattice planes. (a) Calcium hydi'oxide -plate as formed in cem. of calciwn hydroxide plate shown in figure Sa. are observed to precipitate out at an In other diagrams even the basal (0001) reflection early stage of reaction. where some In cement slurries of higher w/c ratios than is plates are observed to stand more or less on edge. normal in paste hydration.91 A lattice planes is occasionally observed (a) • • - • • • 1 • • • • • • w • • • •> • (b) (b) Figure 2. (a) Calcium hydroxide plates formed in a cement pastes and similar preparations. due to the effect of orientation. 69 A.patterns which. b. crystallization which are less cramped than in contain in general only the hexagonal (hk. {b) Electron diffraction slurry of high w/c ratio.O) ordinary paste hydration. cell aH=3. (b) Electron diffraction pattern pattern of calcium hydroxide plate shown in figure 2a. CH resulting in the appearance in the ED diagram of comparatively small dimensions and mostly of reflections from oblique (hk. 622 . Owing to conditions of at 4. numerous crystals.ent Figure 3. very thin. these CH plates often (prism) reflections corresponding to the unit assume the habit of almost regular hexagons.

^^=5. and Taylor [37].0 A. has been observed also by the present author.7 protons. and with crystal water groups accommodated outside this layer.5 A 6 H2O+OH. As suggested by Jones. C for 2 H in the natural mineral hy^ drocalumite.2 A 7. lead to deviations from exactly hexa. within the matrix of the original An alternative principle of arrangement of these structure. supposedly stable in this tem- ture region of the system C-A-H. CsAHe [35]. except AH3 perature range. upon heating in unsatu- The subject indicated by the heading has been rated conditions to temperatures of 100° C or treated in detail in another review article at the somewhat higher. and it is even re. The first transformation occur. and CH. The 5. and Taylor teristic isometric crystal shapes of this compound proposed a model for C4A-aci. F). together with forms intermediate between rated state C4AH19. P". Tlie charac. respectively. a unit cell of about 5. tions replaced by Al ions and with hydroxyl and pensions of pure C3A in water. crystalline compounds which are of greatest -in. garnet structure. and Glasser. Microstructures of Phases Related to Calcium Aluminate Hydrate General Character of Compounds in the transformation.2 and cement. according to Jones and Roberts these. lattice hydrate phase than to the isometric hydro- The only compound stable in the low-tempera. The Al ions are arranged in an ordered pattern with Metastable hexagonal-plate phases are very fre. 2 Al ring upon heating consists in the formation of oriented layers. Dent C~A-F-H System. and Taylor.. Buttler. rapidly mixed sus. Thus. according to the following model.7 A. with the contents of each layer given per The stability of the hexagonal phases seems to in. (ref. structure ported that a stable CaFHe hydrogarnet compound can be derived from that of C4AH19 by substituting does not exist [36]. consisting essentially well known from a number of microscope observa. Al ions for protons in the intermediate double The hexagonal-plate phases have been observed layer of water.ge. as described by Buttler. Analogous compounds are formed in the C-F-H The sequence of la\'ers within the molecular units system. and will be reviewed here only in convert more easily to another hexagonal layer very broad outline.7 A 8. Dent or C4AH19. e. A typical electron 2 Al • micrograph and ED pattern combination is shown 3 H2O+4 OH by Buttler. a distance inter- quently observed to be the products first formed mediate between the orthohexagonal cell distances in the hydration of the aluminate components of of CHand AH3 (gibbsite) which are about 6. giving a unit cell of about 0. orthohexagonal unit cell (ai/Xa^fV3). seems to be the hydrate end member of In reconsidering the crystal structure originally the hydrogarnet series. The ED patterns of the pure compounds are strictly 6 OH hexagonal. The compound C4A-aq. The connnon occurrence of this type of structures presents itself on noting the great sim- C4AH19 C4AH. and including the most satu- C2AH8.7 A. sizes of the order of 1 m [26]. Dent Glasser. crease with the content of F.3 -> C4AHU C4AH7 c„=10. the hexagonal phases seem to Symposium. this Symposium). judging generally assumed to have the composition C4AH13 from the dehydration curves of Buttler. are paratively well defined states of hydration. 20H 623 . of the cliaracter of mixed-layer compounds. depending on the state of hydration. Dent Glasser. has a number of com- terest in connection with cement hydration. -6 H2O+2OH. 4 H2O + 2OH.8 to 6. and to persist over long periods. and Taylor. and in the C-A-F-H system various is illustrated diagrammatically at the foot of the types of solid-solution compounds are possible. in several optical and EM studies on products of hydration of C3A and CaCA. [7] of review paper by Jones.9 A 5. of an octahedral layer of the type occurring in CH tions. and of Complex Salts Glasser.g. suggested [38]. 20H- 6 OH 6 OH 6 OH 6 OH 4 Ca+2 Al 4 Ca + 2 Al 4 Ca+2 Al 4 Oa+2 Al _ 6 0H 6 OH 6 OH 6 OH 6 H2O+OH -6 H2O+2OH 4 H2O+2OH. Ca + 2 Al 2 C2AH8 4 Isomorphous substitutions of different cations for A 6 OH Co=10. of the CH type and unit cell (about 3.g. the CjA-aq.6 A). H2O+4 OH 3 gonal symmetry. have also been observed in EM studies to however with Ca ions in % of the octahedral posi- form in large amounts in e.

On the other hand. but is not stabilize each other. CH in the literature. In similarity to clays. C4AH. but it is compounds are indicated in the following diagrams. in some types of clay minerals.6 A 6 0H( + 2 H2O) 6 OH+2 HoO 2 Al 2 Al 4 OH-f-2 H2O 3 OH +3 H2O 4 Ca 3 Ca 4 OH + 2 H2O 3 OH + 3 H2O 2 Al 2 Al 6 0H(+2 H2O) 6 OH + 2 H2O 3 OH + 3 H2O 2 Al 3 OH+3 H2O 624 . Partial or complete replace- structural element in which the two components ment of Al ions by Fe ions can occur. etc. unilaterally or bilaterally natural to modify these proposed structures in polymerized with (M (4)205)0 layers.-C4A-aq. The formation of CH-type (or (Ca02)n) layers and tetrahedral S2H (or (Si205)n) layers upon dehydration can be interpreted as a layers cannot polymerize.- (Al205)n layer would have a calculated cell size C2A-aq.. C4AH1. approximately equal to the the observations of formation of CsA-aq. and a hypothetical may form. This is being accommodated in the empty octahedral considered to be the reason why no C-S-H min. with layer sequences of. would be isomorplis of the 1:1 kaolinite- usually Si ions). with M(4) such a way that the structure models related to denoting cations in tetrahedral positions (in clays C4A-aq. and C2A-aq. but poly. Separate (M(6)02)n layers occur type clay structure..4 A C3A3H28 ("CAHio") (Roberts [100]) Co=14. the molecular unit layers of clay The C4A-aq.ilarity of the C-A-F-H phases to clay minerals. layers. In the isolated state. without any essential increase of It is generally assumed that octahedral CH the basal spacing.g. the It can be further noted that extra water mole- layer units possess hexagonal symmetry.3 to 14. or segregating in the formof amorphous erals of the clay-type structure have been found.9 ^ C4AH13 .. conceivable that partly filled (Al2q05)n layers in wiiich the compatibility with known data for can combine with partly filled (Cap02)n layers the composition and cell sizes have been taken (q and p denoting degrees of filling) to form a into consideration. mixed-layer compounds hedral AIO4 coordination. It seems usually Al or Mg ions). Such a unit. Possible layer sequences for various C-A-H would probably be liighly unstable. structures originally minerals consist of (M(6)02)n layers. Excessive negative isomorphs of the 2:1 montmorillonite-type clay charge is balanced by introducing a suitable structure. those related to C2A-aq. As is well known. cules can be accommodated in the "holes" of merization and stacking lead to lower crystal (M^Os)!. e. monoclinic or triclinic.= 10. positions. number of protons. cell and sheet. with M(6) suggested by TUley. and in which the positive indicated. with all the tetrahedral sites filled. C4AH7 H2O 6 6 OH +2 H2O 6 OH+2 H. charge deficiency is compensated for by protons. the Al ions misfit of the hexagonal unit cell sizes. because of a too-large collapse of the tetrahedral layers. 4Ca 4 Ca 4 Ca 6 OH 6 OH 6 OH + 6 H2O OH Ca OH C2AH9_7 c. C4A-aq.0 6 OH 6 OH 2 Al 2 Al 2 Al 3 Ca+2 Al 2 OH +4 H2O 2 OH-t-4 H2O 2 OH +4 H2O 6 OH . possibl}^ accounting for aii of about 6. reported orthohexagonal cell size in the layer. alumina hydrate.7 to 10.2 A. Al ions can occur in tetra. Megaw. and Hey [38] are denoting cations in octahedral positions (in clays essentially of the type now described. one molecule per hexagonal unit symmetry.

this hydrated C —A— F— H compounds is likely to be Symposium) seems to be based on more reliable influenced by the presence of varying amounts of XD data. Fratini and Turriziani [102] give a hexagonal due to the slightly larger equivalent ionic radius unit cell of 8.85 A.8 A.2 ^ C4ASHio(^8)(?) "low sulfate sulfoaluminate" [100. further work is needed for the establisli- have been given above have been observed to be ment of the details of this structure. '^(Ih) in the case of described. in an earlier which are incorporated in the C A F com. complex salts exist.86 A Si 6 H2O 6 OH+2 H.5S 2 . 39. 8.2 A^8. all the structures in the group of compounds now lated to this distance (a^fVS.4 to 12. or simply re. sub- different observers. specifically sulfate ions from gypsum. For the_cement hydration system. Some ex. S. — — —H work by Wells. and is thus a hypothetical greatest interest. and C2A-aq. the details of which have not been It can be added that the hexagonal unit cells of confirmed by a complete structure analysis. the generalized oxide formulas two alternatives are given.7 A Co=10. and Taylor hold much less C. as well as of either in the vicinity of aH = 5.Al + C 2 0+4 OH 0 + 5 OH 6 OH 6 OH 4 Ca 4 Ca 3 Ca 4 Ca 2 0 + 4 OH 0 + 5 OH 6 OH 6 OH 2 Al + Si 2 A\ 6 H2O 6 OH + 2 H2O 6 OH + 2 H2O Si OH • Tlio hydrocalumiic stndipd by Buttler. be applied in conceiving hypothetical crystal The structure indicated for ettringite is an structures even for these compounds. A possible Isomorphous substitution of Fe ions for Al ions exception is the "gehlenite hydrate". of which the latter one of which are usually written CsA-CY-Hm and represents an attempt to explain the compara- C3A-(CY)3-Hn. and McMurdie [103]. 625 . 40]. the formation of Schmitt (see discussion of paper by Jones. alternative to the zeolitic structure suggested by amples are given below. Bannister [41].7 A. A similar hexagonal cell unit. but pound structures.0 C2ASH7 C2ASHg • C6AS3H30-34 "gehlenite hj^drate" ettringite _ * c. In the case of phases in the hydro- C4ASHi6(?) ^ C^ASHu C4ASH. forming complex salts of widely more recent studies liave given values related to varying types and compositions [36. Under the conditions prevailing in the reaction (5.0 2 H.9 to 9.7 A. 5. 6 H2O 4 H2O H2O 2 6 0H-h2 H2O 6 OH + 2 H2O 6 OH + 2 H^O 6 OH + 2 ( H2O) 2 AI+S2 2 Al + S 2Al+S 2 Al+S 2 0 + 4 OH 2 0 + 4 OH 2 0 + 4 OH 2 0 + 4 OH 4 Ca 4 Ca 4 Ca 4 Ca 6 OH 6 OH 6 OH 6 OH 6 HoO 4 H.= 12.7 A [37]. the cell size 9. as found by equivalent unit amount of a large number of oxides. Clarke. admixtures. 101] c„=lQA A ^ 9. was found for C4A-aq. with Y denoting a structurally tively large spacing of 12.4 to 12.8 A (41) C4ACHU liydrocalumite or OH ' 00 = 7.5 A 8.7r)V3)given recently for this compound by of Portland cement with water. Dent Glasser. compounds with larger true unit cells. analog of the much-debated Edelman-Favejee The clay-type structure model containing voids model for the structure of certain clay minerals in tetrahedral and octahedral cation positions can [104]. Ob- all the compounds for which structure diagrams viously. It is generally postuhited that two series of such In the diagram for "gehlenite hydrate" below. n.0 A 8. approximate formula unit CiACiHu. for which leads in general to a somewhat larger unit cell. and even S for are of Y molecular layer.96 A of the Fe ion.0 A.0 6 OH+2 H2O 6 0H( + 2 H2O) 6 OH + 2 HjO 2A1 + Si 2 Al A1 + 1. However. This model contains Si — OH apices of tetrahedra directed outwards from the stitutions of H.

The substitutions are gested by Jones that the expanding and labilizing believed to be of the ordered type. Calcium aluminate hydrate plate from a (a) substitutions of F or S for A. but the evidence effect of the introduction of increasing amounts is inconclusive on this point. as for example in a cement-water mixture. 34]. (&) Electron. small quantities in nearly all paste samples the Si04 group requiring less space than the (0H)4 examined.6 and 5. Weaker spots halfway between the stronger ones indicate that the true unit-cell length is twice that given above. The sulfur and carbon ions. C3(A. 32. According to available data.24 to 11. superstructure spots and measurable deviations from perfect hexagonal symmetry cannot be detected.garnet series.70 to 5. in water or in lime solutions. together with amorphous-looking or fibrous C-S-H material. It is probable that similar easily interpreted as in the example now given. respectively. aff= 5. as well as in the majority of crystal plates of this type found in pastes. of Fe ions into the lattice is counteracted by the Plate crystals of the C-A-H type occur in simultaneous uptake of Si ions in tetrahedral sites.40 A are shown. stabilizing effect on the formation of the C-A-F-H solid-solution phases. They can therefore be ex- pected to have a similar. In crushed or ground paste samples the hexagonal shape is of course less evident than in the C-A-H plates formed in suspensions. 626 . 112)3. diffraction pattern with S replacing void tetrahedral groups.62 A. These facts are (b) indicative of a lattice with isomorphous ionic Figure 4. the octahedral sites are only half-filled in these compounds. but the ED patterns are not always as tetrahedral group. the phases within the hexagonal structure group pattern of strong spots may have an irregular under various conditions when silicon ions are intensity distribution. unit cells relatable to a distance between 5. b shows the characteristic appearance of such a plate from a nearly 10-yr-old paste sample. and prove the existence of a superstructure in this crystal. it has been sug. The ED single-crystal pattern corresponds to a nearly hexagonal unit cell. or pseudohexagonal. with six and four posi- tive charges. and the reflections may be present.46 A) and ettringite (5.F. C —A— H and Related Phases in Hardened Portland Cement Paste Of C-A-F-S-H compounds possibly present in hardened pastes. but two-thirds filled in all the others.8 A.55 A). or alternatively. The exceptions are CAHio (5. substitutions influence also the formation of The superstructure spots may be very weak. 10-yr-old hardened paste. composition can probably be approximated by the Weak spots Indicating a true unit-cell length of 11.F) (S. formula C4(A. The of calcium aluminate hydrate plate shown in figure ^a.7-A unit cell seems in most cases to be connected with deviations from exact hexagonal symmetry. as measured in different directions of the pattern with strong spots. most of the compounds now discussed possess hexagonal.S)SHio. This doubling of the 5. The size of the hexagonal unit cells may obvi- ously be influenced also by the degree of occupa- tion of octahedral cation positions. have very small equiva- lent ionic radii and are easily accommodated in oxygen tetrahedra. whereas in ED patterns from hexagonal-plate crystals obtained by the hydration of C3A. only the hexagonal-plate phase has been observed by the author [26. Figure 4 a. It is perhaps significant that in the hypothetical structures indicated above. 33.

The implication of this observation would be that the hexagonal C-A-H phases take an active part in the for- mation of paste structures. taken from a highly porous paste. obtained by dehydration. Ettringite and Related Phases The characteristic crystal habit of ettringite. Particles of the described habit have not been w/c=1. it may ettringite in a sample of old paste. observed in cement pastes examined by the present The spot pattern corresponding to aH=2x5. hy forming bridges between gel structures adhering to different parts of the plate surfaces. the expected transformation to the iso- metric C3(A. the structures causing the initial set are probably of a different type. Unpublished be expected to form pseudomorphs upon rapid XD data for hardened cement paste. without gypsum or surface-active-agent admixtures. The C-A-H crystal plates in old pastes may be secondary formations. i. (a) Tobermorite fibers. and that fibers of tobermorite are adsorbed on their surfaces. An extreme case is shown in figure 5 a.70 A. It is very unlikely that ettringite can be ring pattern from the tobermorite fibers. All these effects indicate the presence of various types of disorder in the corresponding lattices. first. and second. if present in a paste. the isomorphous re- placements and possibly the introduction of additional cation complexes seem to stabilize the hexagonal structure. = The spot pattern. is that of long rods or needles or. corresponding to a„ 2X5. as reported from microscopic examinations. formed in vicin- Watanabe [44] show this compound in the form of ity of calcium aluminate hydrate plates in a highly porous heavy rods or splines. in normal portland cement pastes.F)H6 compound does not take place.. CSH{B). also showed a number of reflections At the London Symposium. On the other hand. Judging from the common occurrence of par- ticles of the hexagonal C-A-H phase even in old pastes.70 A is superimposed by a spot author. Nurse and Taylor considered to be characteristic of ettringite. due to the formation of sulfoaluminate hydrate and so-called adsorptive modifications of hydroaluminate caused by sur- face-active agents [43]. by acting as a kind of template for other structures crystallizing in the vicinity of its surfaces.e. A similar mechanism is regarded by Rehbinder [42] and his associates to be responsible for the process of setting of cement pastes of the quick-setting type. the author. that of hexagonal prisms with sharply cut off end faces. A commonly observed phenomenon is that the C-S-H gel particles tend to develop fibrous structures in the vicinity of the C-A-H plates. However. the plates are rarely observed to form coherent structures linked together by the coalescence of partially overlapping plates. b. paste.5. diffuse or streaky. EMstudies by Iwai and Figure 5. (6) Electron diffraction pattern of material shown in -figure 5a. in well-crystallized (b) specimens. The [17] presented XD evidence for the presence of intensities of the ettringite reflections were about 627 630446 0—62 5 . kept from decomposing in the electron microscope without taking special precautions of cooling the specimen. However. re- sulting from a process of recrystallization following the breakdown of structures developed during the period of setting. As mentioned previously. is superimposed by a spot-ring diagram from the CSH(B) tobermorite fibers.

A hpxagonal plate of CiAHn is shown together with a cement covered with able data seem to indicate that the first stages of splines pseudomorphic possibly of a C-A-S-H compound. that some structural elements are being replaced by others.F) (S. radiating splines or rods. yielding gel-like products. The character of the slurry constituents is shown in figure 6 a. A and S by S. b. in another series of paste samples. 7) This change may indicate . and simultaneously the rod elements seemed to loosen up and unfold (fig. the rod-like elements became covered with a felt-like mass of crinkly foils of CSH(B) tobermorits devel- oping from the surfaces of the cement grains.7 to 3. that account of XD data obtained recently on the the ettringite reflections disappeared after a couple appearance of ettringite and other sulfoaluminate of weeks. Disregard- ing the hexagonal plate of C4AH13 structure which gives the ED spot pattern of figure 6b (correspond- ing to a hexagonal cell aH=5. It may (b) well turn out that the effect of admixtures of sulfate and similar ions on structure formation is Fig UBE 6. in which. in com- mercial Portland cements. phases in cement pastes is given byCopeland and ing to a compound isomorphous with C4AH13. without apparent change in outer form. due to the formation of polymerized silica chain elements. at least some of the avail- w/c=4. 34] of the hydration products of a Type I Portland cement after 24 hr of shaking in a slurry with w/c = 4.. On the other hand. of a habit associated with ettringite. (a) Hydration products formed by 2Jf-hr shaking more complicated than is generally assumed. However. structure development in setting cement paste are 628 . has yet to be attained. Kantro (this Symposium).68 A). in order to indicate an important problem for future research. it seems unlikely that silica groups do not take part in building up these structures. as shown by EM data given by Astreeva and Lopatnikova [46]. Judging from the abundance of these structures. and were replaced by reflections belong. it is seen that the original cement grains are covered with thick. leading to structural rearrangements. It is very tentatively suggested that they are composed of a C-A-S-S-H_ compound of the ettringite variety. Other workers have reported the decomposition of It may be of interest in this connection to describe some observations made by the present author [26. except sometimes a very diffuse ring at about 2. and taking into account the fact that the splines appear to form from the material in the surface of the C-S anhydride particles. Although a coherent picture of these processes tion pattern of material shown in figure 6a.equally strong from 2 days to 1 yr of age of the ettringite upon storing in air or upon heating at paste samples.g.2 A. After a few days of standing in paste form.S)3Hn. C6(A. it seems worthwhile to in- clude these observations liere. since the ED patterns contain no trace of crystalline diffractions. and there are probably alternative ways of interpreting the phenomena observed. (6) Electron diffrac. It is evident that the suggestions now made are of a very uncertain nature. However. upon the structural collapse following dehydration. A detailed observed. It is likely that these formations are only pseudomorphs of an easily dehydrated structure originally present. Taylor [45] 40 to 80 °C. e. a fibrous internal structure is still retained. of a slurry of Type I portland cement. even various forms of pure calcium sulfoaluminate hydrates can be partly deprived of their water of crystallization. in comparison with the normal content of S. amounting to about 2 to 3 percent.

some investigators have a water molecule. C2SH(B) itself in oxygen or hydroxyl coordination polyhedra hillebrandite. such as the well-known Lossier cement. due to their densification and interlocking in fibrous networks. of widely varying type and coordination number. it seems worthwhile to review separately compounds which are in reality unstable under the most essential microstructural properties of the prevailing thermodA^namic conditions to per- various C-S-H materials. each condensation of pairs of CSH(B) and C2SH2 tobermorite phases. CSH{B). the works [48. IVIicrostructures of Calcium Silicate Hydrates General Remarks Arrangements of Lattice Elements in C-S-H Structures Schematic phase diagrams of the C-S-H system were presented at the London Symposium [8. and gypsum. also his paper in this Sym- posium) represents an interesting application of the principles of crystal chemistry and micro- structural development in the sulfate-cement field. The formation of lattice structures from the Although a great number of studies of various structural elements of the C-S-H system. who identified them as ettringite. Ca ions. Under certain rigorously specified conditions of curing. such as C3S2H3 afwillite. seem to tetrahedra being accompanied by the expulsion of predominate. such as rings. A factor of primary importance is the tendency In the products of normal cement hydration. and Katz [20]. high-alumina cement. 629 . i.. paste products autoclaved at elevated tempera. and net- region of the C-S-H phase diagram. still leave their imprint on the developed in France. although of a transient nature and cemxents. However. Asecond factor is that the presented evidence for the appearance of other Ca ion has a great capacity of accommodating compounds. The so-called "stressing cement" developed by Mikhailov [105] (cf. in various hydrated cement samples. 49]. These and other that the groups involved may combine into ciystal- compounds are also likely to be present in cement line structures in a great number of different ways. and its reaction with water is charac- terized by the formation of a paste structure consisting of an expanding framework of high- Figure 7. The struc- tures formed in C3A suspensions with small admixtures of lignosulfonate [24. is commonly recognized changes in the gen&ral outlines of these schematic to be governed by a few simple principles. 42] may be of a related type. and O ions with or C-S-H system have since been executed. of free silica groups to polymerize into larger the phases occurring in the low-temperature complexes. Structures similar to those described by the author as heavy splines. sulfate sulfoaluminate crystals. gave a major contribution to the devel- opment of strength in cement paste. strength. structures developing in th(> subsequent pi'ocesses of hardening. by Sinolnikov [22]. Tobermorite. later decomposing. Further. Similar principles of structure forma- and by the formation of complex C-A-F-S-S-H tion are employed in other types of expanding phases which. Gildart. this composite system of governed by a delicate balance in the composition particles integrates and hardens to a mass of high of the minor anhydrous phases and admixtures. causing certain reason. chains. who observed tliem in such abundance that he consid- ered them to be C-S-H compounds which. and C-S-H gel material accommodated in the meshes of this framework. and C2SH(A) in various prepara. no major without attached protons. foils formed in a cement slurry on standing. have been observed in cement hydi-ation studies by Sliepccvich. diagrams have been introduced. compounds and phase relationships within the Si(0H)4 tetrahedra. the difference in lattice energies of two tures. layers. This cement is a specially proportioned mixture of portland cement. and by Boutet [21]. and have in fact been identified and studied alternative compound compositions is often small. For this and the rate of transformation low. viz. 47].e. tions obtained in the hydration of portland cement A natural consequence of these circumstances is or its pure silicate constituents. sist over long periods without changing observably.

5. sheets parallel-oriented.65 A chain structures (wollastonite). by assuming the was found to be essentially incorrect and was existence of tetrahedral and octahedral voids and replaced by a structure containing metasilicate proton substitutions in the structure of minerals chains of ideal composition SH and a repeat unit of the truscottite-gyrolite group. and OH ions that tend to screen The double chain in xonotlite contains 8-mem- the Si04 groups from each other. possible to obtain a fit with observed data on viously suggested structure containing SisOg rings composition. in which promotes the growth of crystals of macroscopic process 5-membered rings of tetrahedra are size. consisting of two sheets of parallel chains superim- ment of 3-metasilicate chains in layers. both containing discrete of ideal composition CeSgH.3 A along the fibers has been observed to be by them. whereas in the second The C-S-H compounds possessing a more or group more or less orderly trigonal rotations of less clearly pronounced fibrous particle habit have the sheets take place in the process of superposition. if no chain repeat distance is 7. Liebau that such compounds are found among those [48] gives a of minerals. It is natural chains alternate with layers of Ca ions. whereas on tlie opposite side the upon heating. amounts of Ca. drodite [52]. with Ca ions accommodated in the oxygen The structure of the mineral tobermorite. . posed in such a way that the free OH-OH edges 630 .6 A. as worked out chains. Mamedov and Belov [58] that such layers. However. the presence of chains of silica would seem to sheets are stacked with the fiber axes of different be a more likely alternative. Gyrolite. 3-metasilicate chains arranged in a structural unit sponding to different alternatives for the arrange. has been described recently with respect to its occur as the structural elements common to a crystallographic properties. By further edge- The three-dimensional character of such lattices to-edge condensation of SigOi: chains. O. the minimum need of space across the chain is None of these compounds is likely to occur in equivalent to the space occupied by an OH — OH cement paste cured under ordinary conditions. chains. The repeat unit distance of about lite by Mackay and Taylor [61] but was rejected 7. The two the double chain. In a few C-S-H compounds. with discrete Si04 groups. However. The various forms of wollastonite can by Megaw and Kelsey [62]. Further. Although nothing is known with cer- as structural elements [53]. contains discrete be described as polymorphic modifications corre. Mame- most well-known of these compounds are C3S2H3 dov [57] showed that in the structure of xonotlite. tionships probably existing between one or more analogous to tlie (Si205)n sheet occurring in clay of tliese phases and the fibrous constituents of minerals. or similar phases not described here. or about 5. 61]. is C5S2H. which has a structure closely related the okenite-nekoite group with fibrous crystals to that of 7-028. among them several list having a comparatively high C/S ratio. lar crystals and hexagonal or trigonal symmetry C3SH1 is believed to contain isolated S'^Oy groups . isolated Si04 tetrahedra. 56]. wollastonite [54. and the truscottite-gyrolite group with tabu- high-lime end member of the C-S-H system. since the distortion of the sheet necessary common to all fibrous C-S-H compounds. attracted special attention because of the rela. would seem unnatural. seems possible that in the first group the SijOis cles. of polyhedra formed by superimposed layers of approximate composition C4S5H5. a conversion into chain axis are spaced at intervals of about 3. judging tainty about the details of these structures. An alternative structure of the silica sheet. It may be worth pointing out that just as repeat unit to appear was first established in in the case of the hexagonal C-A-H phases studies of the structure of the anhydrous mineral described in the previous section. of composition SigOn. it would be CS. where the C-S-H compounds. etc. in which the silica groups are tendency of polymerization of silica groups is likely to occur as such double 3-metasilicate counteracted by the presence of relatively large chains. (afwillite) and C2SH(A). 55. unit cell. lavers of such double SiOaCOH) ions as lattice constituents. group. observed to occur along the axis. for which the pre. number of C-S-H minerals of low C/S ratio.. The simplest possible silica group formed by stituent of the hittice is present in the form of condensation of 3-metasilicate chains is of course unpolymerized. In the wollastonite structure. the chains are could possibly form in autoclaved cement-silica arranged in layers of paralbl-oriented separate paste products. [60. bered rings of Si04 tetrahedra. calcio-chon. the silica con.3 A. 51]. The [59]. stacked Another hydrate compound of similar type which in various ways together with Ca and OH ions. It can be observed that elements were present in the unheated minerals. of SisOg (Si303(OH)e if written formally as a silicic some difficulties would be raised because of the acid) relatively low content of water of crystallization A characteristic feature of this 3-metasilicate of these minerals. a polymerized sheet of composition SigOij C-S-H structures those of afwillite and C2SH(A) may be built up. It has been suggested by were the first ones to be determined [50. and this is perhaps the reason why of all formed. The to fit it in the observed unit cell would be unduly true nature of the structural elements causing this large. was considered in the analysis of gyro- cement paste. it from the observed fibrous habit of C3SH1 5 parti. leading to a low degree of chain is that on one side of the chain the free proton substitution computed for hypothetical OH — OH tetrahedral edges perpendicular to the lattices of this type.

6 A is the most obvious property common additional Ca and OH ions.3 hj 7. The been emphasized [68. respectively [71]. plained. a distinction is made between tobermorite 0 and 3. on the to 2.7 OH ions. the C/S ratio thus being 5/6. and completely different structure constant.65-A spacing. It seems probable that other liand. However.33 C/S. of rare occurrence in nature and of un. and particle habits of tobermorite phases have Ten Ca ions per cell are accommodated within been reported in numerous papers. The basal spacings may be Another. brandite. the value 11. centered. hitherto observed data in a few simple statements Even in the case of the fibrous C-S-H com. been proposed.5 to 2. The structures of the hydrated the CSH(B) type in the range 0. e. tobermorites are composed of layer-lattice struc- tions is not very likeh^ to destroy the established tures in which fibrous elements can also be dis- agreement between observed and calculated X-ray cerned. and re- gions of regularly cr3"staUine lattice formation may Tobermorite Phases extend over only a few unit cells in some du-ec- tions. compounds are derived from the wollastonite and the C2SH2 type in the range 1. viz. The similar- each double chain in one sheet sharing tetrahedral ities to certain types of clay minerals have also corners with two chains in the other sheet.0. which form a very open la3"er 14 A. low degree of order in the arrangement of parallel- at least in the case of foshagite. 65] postu. compounds formed at temperatures below about late a general formula Ca+nSsHn. 67]. or may increase or decrease with the model for tobermorite has been suggested by ambient water vapor pressure. two different structure models have and especially regarding the colloidal In general. etc. hille. any attempt to summarize the positions. however. to different tobermorite phases of widely varying were left undetermined. other in fibrous particles. by means of a process of diffusion of synthesis mixtures. been found to vary with the content of water dated in lattice positions between the protruding within a certain range of spacmgs.8 to 1. Alternativeh".0 C/S structure by splitting up the layers of sirgle chains [70. both natural and synthetic that the inclusion of these groups in the calcula. with n between 100 °C. and water molecules. On the whole.8 to 1. pseudo-unit cell of probably ortho- or somewhat smaller than the ratio observed in rliombis symmetry and of approximate dimensions natural tobermorite minerals.5 and 1. Further. large variations in the conditions of formation of This structure is believed to expand in a direction the materials studied. foshagite.4 A has been surface network of approximately hexagonal used to characterize well-crystallized tobermorite symmetry. that the parallel to the plane of the layer thus formed. 0. and the basal spacings of which have intensities. among the C-S-H compounds.5 both of these compounds can be described in their phases are present.2 A. but the authors state C/S ratios. The latter properties place of this structure is a silica layer formed by poh^. Gard and Taylor [64. gregated in coherent sheets in la'yerlike particles. results ob. Specifically. such as differences in perpendicular to the layers in the presence of temperature. tobermorite was hardly recognized as more than the name of a differentC/S ratios may imply that the (SH)3 mineral among many others within the C-S-H chains are longer and more isolated from each system. and the composition SiijOa. tion. The scarcity and diffuse character of XD structure..3-A spacing are pointing outwards and are It was soon recognized. The existence of a C-face- arrangement corresponds to a C/S ratio of 2/3. will necessarily leave many observations unex- pounds C4S3H. strikingly similar to the synthetic materials tions in the planes of the free OH —OH edges of observed in the low-temperature region of the 3. water-to- sheets of water molecules entering interlayer solid ratio. for the wollastonite-foshagite- series products obtained in two ranges of C/S ratio.of 7. resulting corrugated silica layer has a unit cell of Observations on the microstructural properties about 11. Owing to and between layers. other limits for these C/S into bands and replacing missing valencies by ranges have been assumed. or that transitional phases essential features as association complexes of exist. These groups can well be accommo. Mamedov and Belov [66. There is even a remote possibility that the estab- 631 ..6 by 3. The essential element type of tobermorite. about 9 to OH —OH edges. Observed variations in particle habits for At the London Symposium.g. 69]. depending on the Mamedov and Belov [63]. composition of raw materials in excess water. ill-crystaUized and inhomogeneous nonequilibrium tained recently in an analysis of the foshagite solids. hillebrandite. the tobermorite phases in a class by themselves merization of two sheets of SigOn xonotlite chams. 30]. replacing all the protons. This C-S-H system. of low C/S ratio. shorter and more ag- known significance in relation to cement hydra. time of processing. quoted by Taylor review paper of (see and ED reflections observed for some CSH(B) and this Symposium) are definitely in favor of the C2SH2 phases are indicative of a comparatively structure containing single 3-metasilicate chains. naturally occurring mineral was in certain features Calcium ions are accommodated in internal posi.. are of the opinion that the structures in batches of average C/S ratio about 1. oriented (SH)3 chains and CH complexes.. and C2SH(B). the common phase rules of xonotlite chains and CH columns (foshagite) or reaction chemistry are hardly applicable to these CH layers (hillebrandite). The positions of 5.

reached in a CH solution of about 3. and by when it comes to recognizing these compounds as Mamedov and Belov. is that of thin sheets or foils.lished agreement between the unit-cell dimensions Finally. the cryptocrystalline substance formed as a metastable phase prior to the crystallization of distinct compounds in hydrothermal reactions of lime and silica at 120 to 160 °C. Gaze and Robert- son [76] report the extensive formation of crinkly foils oftobermorite in mixed portland cement- asbestos pastes produced by low-temperature foaming. is of dubitable significance tures suggested by Megaw and Kelsey. These formations. paragraph. the variation in temperature fortuitous. However. It consists of small. 74. except in the immediate vicinity of saturation where a finely fibrous or needlelike growth is promoted [23]. 23. (a) CSHiB) foils and plates. platy particles. The appearance of foils persists almost through- out the whole range of concentration of CH solution phase. ferent C/S ratios [70]. b. and appear mostly as large conglomerates of crinkly fods. and comparatively flat. tobermorite. Similar ideas on the CSH(B) and C2SH2 struc- tures have been expressed by Taylor [72]. It is questionable if (b) a transformation from CSH(B) to the C2SH2 phase can be achieved in temporarily super- Figure 8. The most commonly observed particle habit of CSH(B). and thickened in the coiu-se of hydration. These particles are also easily deformed. cured at temperatures of 100 °C and above. The crinkly-foil type of CSH(B) appears to be predominant in the pore surfaces of the cement-paste specimens prepared by Bernard [14]. described in the preceding crystallographically distinct species. Different varieties have later been observed in several EM studies [70. together with crinkly foils.1. 75]. C/S ratio about saturated solutions. The first EM observation of such crinkly foils was made by Eitel [19] in a product formed in the reaction of a. One form of CSH(B) is shown in figure 8 a. forming dense. Kalousek and 1. and by Assarsson [73] in a discussion on what he calls "phase B". (b) Electron diffraction pattern of material shown in figure 8a. opaque aggregates.8 mM/liter.4) and cured at 165 °C the paste aggregates consisted of decidedly fibrous or lathlike elements. CH solution with an SH hydrosol.1. The C/S ratio is about 1. and gives a well-developed ED ring pattern of CSH(B)(hk) reflections. precipitated from a Ca(OH) solution containing about Prebus [30] report the formation of a mixture of 3. termed mono- calcium silicate hydrate by Bernard. but in some products of very low C/S ratio (about 0. It seems probable that silica dissolved from the asbestos component and reacting with the lime solution can have contributed to the formation of a product having a C/S ratio in the CSH(B) range. using a special technique of preparation. 26. Similar compounds were also observed [77] in lime-quartz paste mixtures of low overall C/S ratios.33 C/S CSH(B) foils and a high-lime solid in 632 . 1. 30. the high-temperature exothermic peak of CSH(B) and well-crystallized tobermorite is in DTA diagrams.8 mM/liter. estimated to be only one or two molecular units thick in extreme cases. An example of two radically different and height of which has been used to distinguish ways of arranging a given number of ionic groups in a reproducible way between compounds of dif- within the same frame is afforded by the sti-uc. were fre- quently observed to occur together with acicular particles.

b). Similar formations are also possible to replace a limited amount of Si ions by common in some varieties of hardened cement Al ions in tetrahedral positions. lathlike or tubular crystals [30]. ED single-crystal diagrams indicate the existence of a simple (not face-center^'d) base unit cell of dimensions 5. A. estimated from specific- surface-area measurements and from their trans- parency to electrons to be only two or three molecular units thick [28. 633 . straight rods together with thin plates splitting up into lathlike particles. where they can be expected to contribute per C4_5S5H5 tobermorite. considered to be the characteristic liabit Of considerable interest in this connection is the of C2SH2.96 A.5 C/S phase consists entirely of long. 71. in the reaction mixture. which leads to a displacement of. e. 82.25 C/S tobermorite prepared at 150 °C shows long. 4 to 5 percent A pastes.6 A. Fascicular aggregates later sections of this review. of fibers. normally observed in tobermorites. Of special interest is tlie decrease in unit distance across the chains. as in the tobermorite structure model proposed by Megaw and Kelsey.. Increasing the amount to the formation of a rigid gel structure (cf. straight. in a 1. Ribbonlike crystals of appearance are shown by similar Akaiwa and Sudoh and have also been ob- [82]. He found that it was this Symposium).1-A milling C3S in water. A consequence of the more or less regular superposition of layers is that (hkl) reflections and strong basal reflections occiu' in the XD powder diagrams. (a)High-temperahve tobermorite phase formed pounds may account for the general blurring of by boiling of an af willite-CaiOH.4 by 3. The formation of mixed-layer com- Figure 9.2 to 5.g. (b) Electron diffraction pattern of range. bundled fibers. The transformation from CSH(B) foils. 79] (cf.n electron micrograph from a 1.5 C/S ened cement pastes.8. 70. The low-lime phase. and the corresponding 1. 9 a. pared with varying amounts of alumina present also review paper by Brunauer and Greenberg. flat plates. material shown in figure 9a. observed in different constituents of hard. the strong hk=ll reflection from about 3. to tobermorite plates is accompanied by a large decrease in specific surface [70]. which agree well with those obtained from certain natural tober- morites. instead of SH — C — SH.2. 77]. estimated to be of the order of 10 to 20 unimolecular layers in thickness [80. of C/S 0. With increasing C/S ratio the high-temperature tobermorite phases tend to become increasingly fibrous. 23. 77].) slurry prepared by ball- the strong ED reflections in the 2. 81. of alumina in the mixture above about 5 percent below) The tobermorite compounds formed at elevated temperatures (approximate range 110 to 170 °C) generally exhibit more characteristic crystalline habits. forms large. 26] (fig.to 3. served to form slowly during prolonged boiling of afwillite-CH slurry mixtures resulting from ball-milling of C3S in 'water [28. have been observed in the products of study made by Kalousek [83] of tobermorite pre- hydration of C3S and /3-C2S [78.07 A. as is further exemplified in bat'^li processed at 60 °C. to (b) about 2. although reflec- tions with h + k odd ar? generally weaker. but are sometimes indicative of poly- morphic stacking disorder of the same type as is observed in clay minerals.the form of coarse.8 to about 1. A pos- sible way of interpreting tliese data in terms of structure would be to suggest the presence of a layer lattice element of the unsymmetric SH CH — type. ED single-crystal diagrams fully confirm the symmetry and dimensions of the postulated unit cell.

5 C/S tobermoritic material stituents. assumed for itself would correspond almost exactly to a 5 per.resulted in the formation of increasing amounts of CSH(B) and The Megaw-Kelsey C2SH2. and it seems frequently for subcrystalline tobermorite unit difficult to imag'ne an incorporation of extra lime layers crystallizing in the very narrow gel spaces to take place without increasing the unit layer distance to impossible values. already been described. but silica layers of a correspondingly size and to favor the development of small particles high degree of complexity and polymerization are tending to assume the habit of flakes rather than unlikely to be present in the ill-formed structures fibers. thus no alumina would be recent paper by Gard. This lattice may of a cement paste.0. with C/S ratios increasing in the 1. it would be almost impossible 634 . is the most likely alter- ever this may be. of the hydrogarnet compound C3ASH4 (or pos. at least in heating. One of these is the suggestion of the conceivable.33 C/S. containing with mutually rotated directions of chains.2 A unit cell) between the composition approximating the well-known Jeffery ridges of the silica sheet would bring the C/S ratio fornuila. —Hcomplexes in the two structural calculation that in a cement containing alite of a "holes" (per 11. Brunauer. occurring at 1. However. However. and principles of arrangement of structural elements the XD powder patterns exhibited slight changes inherent in this model. An aggregation of the type well apply to certain types of well-crystallized suggested is likely to cause restrictions in particle tobermorite. this and C2SH2 structures and the structure models suggests that the surfaces of tobermorite crystal proposed for tobermorite.5 range Some considerations and evidence on the proba. the postulate of relative independency of reaction and Kantro [84]. C2SH2. The other observation is that of the com- Phases mon occurrence in well-crystallized tobermorite plates of aggregates stacked with the b axes of It is a matter of some interest to speculate on separate layers rotated through approximately the possible relationships between the CSH(B) hexagonal angles. and in addi- belite. [23] are. the unsymmetric SH — CH layer. The accommodation of C-S-H gel phases. some alumina in solution may various morphologic habits and other properties enter into the C-S-H gel phase formed from of the synthetic tobermorite phases. it can be assumed that such native. This postulate. How- polymerized silica layers. which is assumed to contain no alumina tion furnishes data on the modes of transition initially. Two observations seem to be of special the early stages of setting and hardaning. Howison. modes of rotational superpositions would occur tively rigid. or that Pressler. According to the authors. this structure has a compara. self-contained character. it seems that also the open. and Taylor [85] exchanged with the other clinker constituents. more easily explained by a ble incorporation of silica groups in the hydration mechanism of structural change involving the products of the C-A-F and C-S constitutents removal of SH elements from the layers. a method would be afforded of testing of CSH(B) at 1. sibly C3AS2H2) as a second phase. is. seems to have acquired a degree of general validity Certain observations of basal spacings decreasing which may not be altogether justified. existence in C2SH2 of unsymmetrical SH —CH minate ions in the liquid phase may accelerate the layers coupled together by C —H layers. In view of the low rate of hydration of between different particle habits and on transfor- iS-CsS. time curve. therefore.54S16AM. structure model on the whole. it would seem that only a very small amount mations to dehydrated C-S compounds on of alumina is involved in this process. CSH(B) and sek and Prebus set a limit in the range of existence C2SH2. however. One such hypothetical from those of tobermorite without substitutions. once advanced by obtained in a hydrated 0. found an abrupt the cement chemists as a kind of working break in the rate-of-extraction vs. hypothesis for lack of evidence to the contrary. sheetlike habits predominate elements consist of CH layers rather than SH chains. more flexible. rather of cement were presented in the previous section. has If observations of a corresponding limit of substi. It is perhaps fortuitous that Kalou- of the low-temperature compounds. to halloysite among the clay minerals and agree- ing with the considerations put forth in this re- Structure Elements in Colloidal Tobermorite view. cent substitution limit in the tobermorite phase The electron-optical evidence presented in a formed from this alite.0 to 1. that equally well permit a fit between adjacent units the Mamedov-Belov structure model.8 paste. The crystals and can be modified theoretically in various ways of the Al-substituted phase showed a marked in order to fit observed data without violating the decrease in size but were still of a platy habit. This layer has a C/S tution of Al ions could be made also in the case ratio of 4/3. that the presence of alu. the alumina held by the alite up to the maximum value of 2. it can be shown by a simple additional C. C. modification.2 by 7. as described earlier. in CSH(B) particles. It would appear. ture of lime and 1. in extracting lime from a mix- with water of the C-S and C-A-F clinker con.34 C/S in the tobermorite phase. however. It is interest. fully confirms the description given above of the On the other hand. in analogy hydration of belite considerably. than by an addition of CH elements to a layer core Concerning the question of alumina in the of constant composition. As has been shown. Further. in spite of the probable nearly hexagonal structure of the layer of SH existence of one-dimensionally coherent lattice chains of the Megaw-Kelsey lattice model would elements in them.

cement paste differing considerably from those plete replacement would correspond to a C/S of other commentators [8]. We have studies.12. probably of afwillite. calcium oxide. no absolute comparison can be made. together not a fundamentally different conception of the with crystalline CH. discrete is advanced by Taylor and Howison [68]. ball-mill-hydrated in a similar way but omitting seems to afford interesting possibilities for a the subsequent dispersion by ultrasonic treatment revision of current ideas on the C-S-H gel [26]. 4. dealing in particular with products of structure type between pure CH and pure obtained by ball-milling of C3S slurries. a process of coalescence and intergrowth of the silicate chains decreases with increasing C/S ratio. the sug- tobermorite. then C3S2H3. According to this the mechanism of hydraulic hardening as being theory.82. lically active constituent of hardening portland ture of an ill-crystallized composite gel material. specifically the crystal patterns are observed. reflections. in many diagrams dif- arrangements of ionic groups in the structures of fraction spots not belonging to the CH pattern these compounds. the distribution of intensities and liydraulic properties. probably constituting. some- missing. with the purpose of obtaining a confirmation aggregates of rather large and thick afwillite of structural theories or a better resolution of the crystals. Owing to difficulties. afwillite. if developed in further detail. This theory. ment pastes of normal data the expected forma. ED due to the absence of Si (OH) 2 groups described terns from the surfaces of the large agglomerates earlier in this paper as protruding OH—OH edges are said to resemble the diffraction patterns given of the Megaw-Kelsey model structure for tober. and possibly CSH2. gel globules conglomerated in spherical clusters. A few other strong XD reflections Although afwillite is not very likely to develop ' In the opinion of the present author. 5. especially at about 6. However. It seems that this is merely an ges<^ion is made that afwillite could be a hydrau- alternative way of describing the molecular tex. In reports by Swerdlow. However. observations.04. conclusive evidence is still largely about equal amounts of CH crystal plates. Other largely fails to occur. of the appearance of rhombic slabs with tobermorite gel structure. 10). whereas the pat. one In order to verify these observations. are formed as tion of fibrous structures of C2SH2 tobermorite surface layers on the anhydride crystals. representing only one of a few alternative needlelike crystals. structure. Instead of a missing Si (OH) 9 group.14 A. in the ways of reviewing the object matter. 2. lead to a conception of structure formation in sumed to enter nearby sites in the lattice. The fully hydrated product was composed of However. the degree of polymerization of the meta. even at low levels of electron irradiation constituent of cement hydration products. Because of the apparent importance the intimate mixture of CH and afwillite.with a short-range undulation man [87. 3. Buckle and early stages of hydration of calcium silicates. agree with that of afwillite. it seems urgent to make some generally too thick and well-crystallized to give attempts at forming a conception of the various ED effects. those ofEO patterns of C. owing to the essentially non. made in EM studies by Zhuravlev It must be emphasized that the ideas put forth (unidentified reference given by Schwiete and in the last paragraph are of a highly speculative Miiller-Hesse [31]). Judging from evidence obtained in microscope tions by means of diffraction methods. sup. Taylor [29] regard the problem of structure for. and 2. equilibrium in fully hydrated C3S paste. indicated the formation of nature. examined the hydration products in C3S slurries. They assume a structure difficult to reconcile with later findings. 88]. cement pastes. McMurdie. intensity. first C2SH(B). composed of a mosaic of crystal elements of ex. ported and further developed by Brunauer and and a theory was advanced explaining tentatively Greenberg (this Symposium). 635 . or medium strong. However.20. Com. and Heck- tremely small size. the present author ratio of 1. coinciding approximately with some strong.48. discrete particles chain.^ morite. Nacken [86] suggested that in the suc- here a possible explanation of the fact. the relative spacings of theEP patterns shown are in good agreement with colloidal character of its crystals. inEM studies of freshly precipitated hydration products which have not yet had time to stabilize. spherical clusters.75. flaky structures. will meet with great somewhat rounded corners (fig. exemplified cessively proceeding hydrolysis of CjS crystals. which Ca ion and possibly one water molecule are as. hillebrandite. to detect any crystalline properties in such forma. and work. by tobermorite and possibly hillebrandite. that in some ce. plombierite. in later sections of this review. at certain intervals of the extracted fractiors of the small. whereas the afwillite crystals are ing cement paste. it has neverthe. and it is obvious that further experimental times aggregated in large. the product of ultimate structure. reflections Afwillite of afwillite. The XD patterns given by It is assumed that. it can be pointed out here less been mentioned in the literature as a possible that C patterns frequently appear. too much importance should probably mation from a some- in ill-crystallized tobermorite not be attached to these observations which are what different angle. the of some form of the tobermorite phases for the ED diagrams consist of mainly CH spot-ring formation of an integrated gel structure in harden. For lack of numerical data. The Another theory of the relationship between active ingredients of the observed gel structures structure and C/S ratio of tobermorite gel phases were considered to be extremely small. the ionic bond connection is interrupted. in the outer surfaces. as reviewed earlier in this paragraph.

but cally stable product below at least 170 °C. slightly larger than the unit distance in the This result indicates that the presence of a suffi. van Bemst observes that viz. slender.5 could be involving prolonged and intensive mechanical detected in electron micrographs of C3S pastes treatment. can be further supposed that the application It ofhigh-power ultrasonoration may lead to complete peptization and destruction of crystalline struc- tures. An electron micrograph of the natural mineral shows aggregates of par- ticles of fusiform shape. symmetry. The itshould be pointed out that one can never expect restrictions on atomic transport and rearrange- a complete agreement in weak spot patterns of ment in pastes cured at these relatively low this erratic character. Single-crystal ED patterns can be structures incidentally formed. yield tober. Similar habits are exhibited by C2SH(B). of microstructures in the hydrated phase. These authors state that "although per- sistent. Since both C3S and afwillite contain ciired for long periods at room temperature. or breaks up any chain the ends [53]. This mineral consists of long. colloidal or microcystalline toberinorite the silicate constituents are dissolved superficially. has been observed by Kalousek [80] and by Akaiwa and Sudoh [82]. It can be noted indexed on the basis of a unit cell of hexagonal that j3-C2S slurries. It ciently large excess of Ca ions in the mixture is remains to find a lattice structure compatible also required. reviewed here.5. often with foil-like or tubular extensions at polymerizing into chains. with n variably reported to be 1/3. C3SH1. Calcium hydroxide plates and aggregates of afwillite crystals foryned by ball-milling C3S slurry without low as 50 °C. phases. of com- position CSHn. it Crystals of C3SII1 5 produced at high temperatures seems natural to explain the effect of ball-milling showed characteristic habits of long needles or by assuming that it prevents the Si04 groups from rods. radically by applying a procedure of mixing No sign of the presence of C3SII1 .. resembling C2SH2 [82]. After the Ca(0H)2 has had time to crystallization processes taking place in the segregate into relatively large crystals. isolated Si04 tetrahedra in their lattices. giving ED spot patterns indicative of three-dimensional crystallization [23]. Tricalcium silicate hydrate. fibrous or lathlike crystals. C3S producing one extra mole of per CH and tliat CSH (B) is precipitated in the outer 636 . or 1/6. pounds with the same Ca/Si ratio as the starting These results indicate that the nature of the material. with the c axis of length about 7.5 A morite materials in the hydration products [28]. which also forms aggre- gates of broom-shaped or fascicular character.e. the two ideas put forth by van Bemst [71. [29]. 3-metasilicate chain) parallel to the fiber axis. will not be a basis for further discussion. tricalcium course of hydration of cement can be changed silicate hydrate does not form below about 200 °C". GgS and (8-C2S Suspensions and Pastes General Considerations on formula at complete hydration [28]. as is indicated by the observations just described. The XD diagram contains temperatures possibly favour formation of com- the spectra of both CH and afwillite. It may suffice to state that these In general agreement with the theories on compounds mainly the same hydration give cement hydration once developed hi the classical products at temperatures up to at least 100 °C. as observed by Buckle and Taylor subsequent ultrasonic dispersion. Hillebrandite. It is observ- Hydration Reactions ed that C3S hydrates much more rapidly than In matters pertaining to the development The extejisive literature on the reaction chem. similarly treated. C3SH. works of Le Chatelier.. and Tricalcium Silicate Hydrate The morphologic habit of xonotlite. in order to keep silica groups from with this rather peculiar relationship between combining. is of con- siderable interest because of its occurrence in C3S pastes cured at different temperatures after long periods of curing even at temperatures as Figure 10. (i.5 is probably not a thermodynami- of afwillite are not observed in ED patterns. the istry of the hydration of C3S and j3-C2S. 89] can serve as main ingredients in portland cement. synthetic hillebrandite. 1/5. morphologic properties and crystal symmetry. Xonotlite. .

larger amounts with excess water.5 could be reached states that the hydration process is considerably in about 10 days. radiate outwards from the surfaces. reality far exceeding this value. conditions are large floes showing at the edges a feltlike texture favorable for the formation of C2SH2.8 to 3. according to van Bemst. but their optical by long-time shaking of CsS-water mixtures of such and X-ray evidence seems to indicate that the proportions that a C/S ratio of about 1. as.1 A. He hydration of ^-€28 at w/c = 0. more rapidly than hydration at higher water-to. The relative amount of water in a mix is patterns as CH crystals. and the decomposition of the anhydride taminated type. if formed in morite particles. Also present. Another phase. composed of small plates. Kl\eiker. but phase forms directly on the surfaces of residual considerable quantities of the fusiform particles anhydride cyrstals. but occasional larger crystalline hydration tends to increase with increasing w/c fragments were identified by means of their ED ratio. with C3S suspended in a highly super- point (about 20 mM C per liter). characterized as fascicular fiber aggregates with A possible explanation of these phenomena splintered ends. were C2SH2 phase proceeds at a comparatively high large amounts of CH precipitates. reporting solid ratios. with raetasilicate anions par- Observations made by the. active agents in solution can change the rate of is that paste hydration of the silicates proceeds hydration and modify the hydration products.0. Some support were present) consisted of large floes of foil-like for this statement is given by observations of struc- material of a rather open texture. which also acts as an inhibitor of hydrolysis. if once formed. further reaction seemed to especially of C3S. The precipitation of the were also observed. On the other C3S Pastes hand. adhere to the anhydride silicate surfaces. The structure of this material is tenta- hydration. This of finely fibrous particles were predominant. present author on tially and disorderly replacing OH ions in the CH hydration products formed in C3S suspensions in structure. The irregular masses are only one among several factors that determine the almost certainly very ill-crystallized tobermoritic rate of reaction at various stages of the process of matei'ial. This has been shown by Adamovich [24]. strongly resembling C2SH2 tober- would be that the CSH (B) foils. and most probably also the reached in a nearly saturated solution (small but morphologic character. by carbonation of the original structure. All pastes showed the It should be noted. phases of particle habits quite different from those observed under similac conditions at room tem- pended in a saturated CH solution were still perature. in the appearance of a mixture of hydrated solid Corresponding hydration products of C3S sus.- largely foil-like. but probably m saturated CHsolution (about 31 mM C per liter). that other inves. water and in lime solutions are mainly in agree. is changing rapidly with undetermined amounts of solid CH modifications increasing temperatures of curing. appeared as centlv been reported by Berkovich. sumed by van Bemst to be normal saturation Finally. by Buckle and Taylor [29]. and effective Ca ion concentration increases.7 was phase composition. who noticed that complete amounts of lignosulfonate (sulfite liquor). The introduction of certain types of surface- A consequence hereof. covering them with a more or less impermeable coating. as is elucidated in further detail by tively described as intermediate between those of Brunauer and Greenberg (this Symposium). In the case of paste and slurry data from EM studies of C3S and i3-C2S hydration hydration of j8-C2S at 100 °C. 637 . were examined in the electron microscope and hydrolysis. a similar observation in suspensions. same structure. of course. The hydration products are proceed at an exceedingly slow rate. various additions of asbestos and quartz. for a few minutes [26]. ob. The ED patterns were those of calcite. CH.and tobermorite. the fibrous particles of C2SH2 type. with or without admixture of small is made by Funk [79]. The solid phase produced periods at higher temperatures. These aggregates gave indicating that in general the rate (or degree) of no ED pattern. At a these particles are therefore pscudomorphs caused certain concentration of Ca ions in solution. that of irregular masses possibly tigators have arrived at quite different results.solution phase thus formed. giving ED ring ture formations in a CsS-water suspension boiled patterns indicating ill-crystallized CSH(B) (3. have re- served to be present in small quantity. of the C-con- rate. In addition. The products of hydration in C3S pastes cured furnishing paths of easy diffusion of the ionic for long periodsup to more than 5 yr at room tem- groups participating in the process of hydration perature. time.8-A reflections). however. This reaction is fusiform particles consisting of bundles of parallel rapid at first but slows down considerably as the fibers. compounds is again accelerated. whereas at w/c — 3 the product accelerated in the presence of the surface-active was only about 50 percent hydrated after the same agent. This product products of C3S and 13-C2S In^drated under different gave an ED pattern consisting of only one diffuse moisture and temperature conditions and w'tli ring at about 2. but with some foils rolled up to Data on phase compositions of the hydration form long and rather thick fibers. No electron-optical data are given by Buckle ment with the conceptions of structure develop. Tliis treatment resulted and 1. and Taylor on the C3S pastes cured for prolonged ment now described.

were 638 .82 A. a second phase was also observed in about equal amounts. characterizing these products as fibrous tobermorite. 33]. often radiathig from a common center (fig. presumably CH crystals. A large part of the material consisted of irregularly shaped agglomer- ates of small and thin particles. and Kupreeva [90]. A third phase. was also found in appreciable amounts.80 to 1.7.0 A. 11). revealing in many aggregates a skeletal framework of elongated particles (fig. sheets. the formation of CSH(B) in mixed CsS-quartz powder pastes is connected with an increase in strength. that of large and fairly thin plates of CH crystals.8 to 3. This observation seems to indicate that the amorphous-looking masses may be aggregated in and aroimd the meshes of a rigid structure wb. If the dried and powdered C3S paste was shaken with water for a very short time prior to the preparation of EM samples.ich holds them together. age 17 months) produced in the stoichiometric studies made by Brunauer. 12). the material seems to consist of about equal amounts of finely textured irregular masses. Kantro. The mixtures autoclaved at higlier temperatures show complex compositions. and a mostly weak but sharp reflection at 1. Graclieva. With special refer- ence to low-teniperature hydrates. An addition of small amounts of gypsum to the paste mixture is said to influence the development of hydrated structures in an unfavorable way. and Copeland [28] were examined by the author [26. Fibrous tobermorite formed in a fully hardened of long and rather thick needles or rods. Some fully hydrated C3S paste materials (water- to-solid ratio 0. It consisted Figure 11. obviously paste of C3S.1 A. of j3-C2S ing after brief shaking of dried and powdered fully hydrated containing additions of small amounts of various C3S paste with water. some amount of the irregular masses seemed to dissolve or become rinsed away. age=7 montlis.5. it is stated that fibi 'ous C2SH2 and CSH(B) can be identified in materials cured at 20 and 60 °C. The ED patterns from such aggregates contained a diffuse ring at about 2. and of thin plates. Microstructui'al elements very similar to those now reported liave been observed by Kurczyk [91] inC3S pastes. composed of bimdles of fibers or tubularly rolled Water-to-solids ratio=0. hydrated one month at 30 °C. /3-C2S Pastes The products of paste hydration. at 50 and 100 Figure 12.7. However. stabilizers (to prevent conversion to 7-C2S). 32. Skeletal framework of elongated -particles remain- °C and a water-to-solid ratio of 0. It can be specially iioted that the appearance of CaSHCA) seems to correspond to a lowering of the mechanical strength of mixed C3S- asbestos pastes. Apart from larger unhydrated particles. but the decrease hi strength is ascribed to corrosion of the asbestos particles. and showing in ED diagrams at most a very diffuse ring at about 2. similar to those described by Buckle and Taylor. of rather long branched needles. but the character of the changes taking place is not clear.9 to 3. easily recogniz- able by means of their characteristic spotED patterns. On the other hand.

whereas at 100 °C strengtli of such products was veiy low. together with some irregularly sliaped material. integrated structure. It seems natural to assume that the rapid development of a crystallized CH phase at early ages in C3S pastes is the main factor responsi- ble for this discrepancy in the rate of hardening.32 A along the fiber axis). Considerations on Differences in Structure Development and Hardening Processes The evidence available on low-temperature paste-hydrated C3S and 18-C2S seems to show that the pliase composition and particle habits are essentially the same in both materials. The microstructures of this material resembled those observed in cor- respondingly hydrated C3S pastes described m the preceding section. wlio concludes that CH participates in the hardening process. On the contrary. Occasional large. b. wliereas longer and thicker needles or rods were observed. 639 . by an intergrowtli of partially overlapping plates. and. material. (b) Electron diffraction pattern of Ubers shown with no signs of adsorption or bonding of gel in figure 13a. The ED fiber pattern consists of six diffuse. in water vapor at 100 °C it gave a product com- with minor <iinounts of CH crystals. posed of loosely aggregated hun.7.lles of fine rials prep ired at 50 °C were composed of dense needles. the coarsely fibrous or tubular particles also occurred in appreciable quantities. or of formation of continuous structures Water-to-solids ratio=0.95 A.studied by P^unk [79]. with the exception that the amount of crystalline CH is considerably larger in C3S pastes. the crystal plates of CH seem to be easily broken away from (b) the rest of the gel structure. Thoso pastes consisted being liigher in lime. /3-C S paste. It is well known that C3S pastes harden and develop a higli mechanical strength much more rapidly than do j8-C2S pastes. When /S-CaS was hydrated almost entirely of fibrous C2SH2-type toberniorite. thin plates of CH crystals were also ob- served. and practically always Fig URE 13.0. after which period it was found to be nearly 70 percent hydrated [28]. The nuite. hexagonally arranged spot reflections at about 2. distorted plates as the predominating phase. However.7. indicating that some hydrolysis of 13-C2S had taken place.83 A (correspond- ing to 6 =7. This opinion is held by Ershov [92]. the materials produced at 100 °C of a solution pliase during hydration is essential for the formation of a hardening. with irregular aggregates of small. The present author examined dried and ground material from /S-CaS paste (water-to-solid ratio 0. A bundle of such fibers is shown in figure 13 a. that "the basic factor controlling early strength of portland cements should be viewed as the quantitative content of Ca(0H)2 in the products of hydration". (a) Coarsely Hbrous particles from hydrated appear as detached particles with clean surfaces. The C/S ratios in the solid phases varied from 1. and a sliarp streak at 1. Seventy percent hydrated at 17 luontlis. whereas "the basic factor controlling strength gain at later periot's of hydration is the delayed hydration of C2S and the continuous process of densification of gel ' structure These views cannot be directly substantiated by the EM data. in particular. paste-liydrated j3-C2S materials were quite hard.7 From this fact Funk concludes that the pi-esence to nearly 2. hydrated at room temperature for 17 months. It was observed that the mechanical aggregates of thin fibers.

but rather selected from masses predominate. the particulate structure in the interior of the aggregates is a mixture of fibrous or acicular elements and ele- ments of the shape of small and thin.) crystals. 15. exactly as in XD 33. the author finds it convenient to use Figure 15 is taken from a similar paste (w/c 0. higher in /S-CaS and lower in C3S and C3A found to be the case. In at Normal Temperatures rare cases. similar particles have given ED patterns consisting of six hexagonally arranged. 640 . have not been prepared sible in both pastes. mostly hydrated Type I cement paste. not very common. Whenever the fibrous phase is present in noticeable amounts. For the present 1.0-A spots plus a fiber reflection streak at of normal composition consists of a mixture of 1. irregular aggregates of diameters generally ranging between 0. relative rates of development of the amorphous Thealternative explanation of the difference in and the coarsely-fibrous pliases during the early the rates of hardening. w/c 0.65.800 cmVg in cement.55. it can be expected that the were in general found to be very similar to the predominant phases in hardened portland cement Type-I cement pastes. F.1 A. 7Kyrold). in which it is sometimes difficult to discern any definite morphology. As in pure C3S and /3-C2S pastes. showing only a broadened reflection at about 2. S)-aq. and there seems to be reason to hope such particles occur in the image field. Figure 18 shows a the largest part of the cement pastes examined type of aggregate consisting of very fine (less than consists of exceedingly ill -formed.7 w/c=0. for major contributions to the systematic solution Figure 14 shows a representative aggregate in a of important problems relating to this subject to Type I (ASTM) cement paste (specific surface be forthcoming in the near future. leading to a before any opinion can be formed on the true more rapid development of colloidal C-S-H nature of early hardening and subsequent strength phases. heavy bundles of long fibers are seen together with finely textured material. 34. is of course of a very vague nature. the properties of which have already been described. alite and belite. Figure 14.82 A can also be detected. Fibrous or bundled structures are vi- materials which. Tlie ED diagrams from representative aggre. and the fibrous phase has in general a finer and more distorted texture than in pure silicate pastes. The remainder of the pastes consist of relatively large-sized plates of CH and C-A-H (possibly C4 (A. of the order of up to a few hundred A in diameter. reactivity of C3S in contact with water. Gel Phases in Portland Cement Pastes Introductory Remarks to 3. Figure 17 represents gel- pastes would be very shnilar to the gel phases structure elenients from a Type-I V (low-heat) observed in the pure silicate pastes. 30. and which are always present in some quantity. age=5 months.81 to 1. 88] have established in broad outline the patterns. The lightly ground paste materials are mostly found to be composed of rounded. that of the generally higher stages of paste hardening juust first be studied. refiection at 1. the only reflections recorded.81 A. Representative aggregate from lightly ground gates are exti*emely weak and diffuse. This is cement. mostly small.5 and 2 is. from the same available samples of pastes of different description paste as figure 15.55. In figure 16. the impure clinker modifications Pastes made up with other types of cement of C3S and i3-C2S. Such Observations on Pastes Hardened formations are. colloidal prodixcts. The gain in calcium silicate pastes.as is probably the case in hardening li'ne mortars. It is supposed that these aggregates correspond to the initial cement particles. in case cement gel. mainly his own observations on various paste Bji yr old). structure of cement paste materials [14. review. except spot character of various constituents of the hardened reflections from CH and C-A-H plates. a weak but sharp fiber Hitherto published EM data on the micro. but the more amorphous in a systematic way. Since about 70 to 80 percent of portland cement diffuse 3.6. however. The latter phase is predominant. rounded flakes. however. 5 months old). 32. It can be stated that by far (2300 cm^/g. similar to the pattern in figure 13b. These diffraction effects are. w/c 0. and that grinding or milling operations have the chief effect of splitting tlie material along zones of contact between hydrated particles.

The ED patterns ai^e paste are characteristic features. prepared from tiie 8000-cmVg fraction of an ordi- stituent. w/c 0. The large An efl'ect of higher degree of fineness of grinding and thin plates also visible in this micrograph are is exemplified in figure 19.6. together with flaky aggregates. Tlie of Type III (rapid-hardening) cement. Figure 16. Representative aggregates from lightly ground Figure 17. age = 5H yr. w/c=0. age=7H yr.500 of particle elements of tlie gel aggregates in this cm^/g. ago = 7Ji yr. in spite of the fibrous appearance. . w/c=0. originating from a paste crystals of the C-A-H hexagonal-plate phase. 100-A wide) fibers appearing as a common con.6. high in increase in porosity and the exceedingly small size C3S and ground slightly finer t'nan normal (2.65. cement paste.6. w/c=0. to formation of fibrous habits is small. Representative gel aggregate from Type III drated Type I cement paste. yr old).7. Figure 15. 2}^.55. Gel-structure elements from Type IV cement hydrated Type I cement paste. Relatively rare fibers from lightly ground hy. in a paste nary Typ? I cement (w/c 0. Figure 18. age=5 months. w/c=0. 7^2 yr old). paste. The tendency indistinct.

considered to be the characteristic ED 642 . cured at 80 to 50° C for parent crvstal fragments together with CH crystal 15 hours. Some lathlike or rodlike particles short duration (less than a few days) at temper. pressure of about 700 kg per cm^. Microstructurally. The sisted almost entirely of bundled and intertwined. Representative Uber bundles from a paste pre- pared from the 8.7. some irregular. 21 difficult to grind.82 A. Figure 4a is a micrograph previously in this review could be detected. age=3 yr. The somewhat higher degree of crystalliza- chiefly of dense gel aggregates of small flakes.to 3. C2SH2. Representative gel aggregates from a paste pre. tentatively identified by the authors. w/c 0. the diffuse 2. were identifled.000-cm^/g fraction of Type I cement. in addition to In another paste. but under the C-A-H plate. Figure 20. it con. among other compounds. The latter paste was very hard and tion of fibrous structures was noticeable (fig. age = 2J. cement-silica mixture consisted exclusively of very fine fibers. Both pastes were composed a. hydrated for more than 3 yr habits. b). and hi figure 5a. as CsSHi. The ED pattern in figure product indicated. as a rule. w/c=0.000-cni^/g fraction of a Type I cement.5. of Type I portland different from those observed in pastes cured at cement.5. The spot reflections originate at Elevated Temperatures mainly from CH crystals. as shown in figure 20.j. and then kept refrigerated for about 1 yr. had been autoclaved at 215° C normal temperatures.40. while C-A-H plates of the hexagonal C4AH13 type could not be found Hydrothermal curing of pastes for periods of in the specimen.7. the tendency to forma- clumps.70 and 0.35 which had been kept in sealed the gel fragments in this paste and in those shown vials for nearly 10 yr. The complex XD pattern of the former flakes (fig. cured at 177° C. by means of their single-crystal atures below about 150° C does not. The samehigii-surface-area cement was used some C2SH(A). appearing as rather large. plates. a at the time of examination. 0. which agrees with that of well. the 22 b shows a great number of irregularly distributed formation of C2SH(B) (hillebrandite) and possibly spots. No difference in appearance of of w/c 0. nontrans- cement-silica mixture. show electron micrographs of the gel material in The major part of this paste was found to be paste samples of cement and of a 60-40 percent recrystallized.46. Kalousek and Prebus [30] for 24 hr.Figure 19. The resulting paste few rod-like or tubular particles as hillebrandite. The electron micrograph for this illpreparing paste samples witli the exceedingly sample shows particles of various morphologic high w/c ratio of 1. pared from the 8. Figure 5b also shows the ED pattern given by The author examined a cement paste of w/c such fibers. some of them only about 50 A irregular masses composed of small globules or across. w/c=1. autoclaved for 24 hours at 97° C and a crystallized CSH(B) tobcrmorite. ED patterns.1-A ring and the fiber reflec- Observations on Pastes Cured tion at 1. then stored for The oldest pastes examined were two samples about 7 years. b). 22 a. yield products of microstructural properties very A third paste examined. tion is reflected in the increased intensities of with only minor amounts of fibrous phase visible. thin plates as CHor CsSHCA).5 yr. was of course very soft and had a consistency gel aggregates of partly fibrous appearance as resembling talc stone. from the former sample containing. flaky gel otherwise equal conditions.

were bundles of long lods or silicate hydrate). flalces from Type I cement paste about 700 kg'cm^. (a) (b) (b) Figure 21. stituents under the prevailing hydrotherinal gates of thin.4.pattern from large and well-crystallized particles an ED pattern of spot rings. {b) Electron autoclaved at 215 °C for 24 hr and kepi refrigerated for 1 yr. probably be identified as a synthetic clay (Al-Mg- in minor quantities. This compound can this paste is C2SH (A) Other constituents. corresponding to the of a compound of low crystallo graphic syirnnetry. diffraction pattern of material shown in figure 21a. cement paste autoclaved 24 hr at 177 "C and a pressure of together with Ca(OH). w/c=0. and aggre. rounded flakes (fig. b) giving conditions. (a) Crystal fragments believed to be C SH(A).5. (6) Electron diffraction pattern of material shown in figure 22a.33 A. 23 a. identified by ED as C3SH1. (a) Representative fibrous structure from a Figure 22. present . formed from the cement con- laths.0446 O — 62 6 . 643 63. stored for about 7 yr. (hk) crossgrating reflections from a hexagonal It is believed that the compound predominant in unit cell of aH = 5.

thickened in the course of hardening to a dense. (a) (b) Figure 23. present in minor quanti- ties in Type I cement paste autoclaved at 215 °C for 24 hr and kept refrigerated for 1 yr. that at least the properties of C-S-H compounds (given. structures are formed in quantities large not phologic character of ordinary cement pastes do enough to make it assume that a frame- safe to not confirm the theories on paste structure ad. to defer the discussion on various paste constituents from the appearance this subject until it can be more clearly surveyed Experimental Data and Theories The data presented here on the general mor. be used to confirm or refute theories on the 94. and by Strelkov [13]. the employment of replica methods will be justified only if the results can A number of investigators in later years [93. comparative EM stripped off or isolated by dissolution of the base studies of this kind. have not yet been attempted It is not surprising. thus imparting a steadily increasing pressed by Rehbinder [42]. probably synthetic clay (Al-Mg-silicate hydrate). which were based on the the hardened paste. cement. 96. that the replica studies EM replica techniques is in a state of rapid have met with considerable difficulties of resolving evolution. direct observation. liave a nearly amorphous structure even after ture was visualized as a network of fibers which many years of curing at normal temperatures. 15] have reported electron-optical micro structural properties of hardened paste. The opinion held by these authors is. and then as is known to the author. evaporated onto the observations of dispersed paste particles. As far surfaces of cement paste samples. mechanical strength to the paste. Observations on Cement Paste of their impressions in the replica film. In the by Replica Methods opinion of the author. but it seems that in ordinary pastes the fibrous briefly stated. 95. the paste struc. it seems advisable.5 and aggregates. considering the general in any great extent.4. For this reason. 97. information then available on the microstructural It has been shown previously. sible for the strength and dimensional stability of According to these theories. \v/c=0. using both replication and material. This observation is contradictory to the views ex- feltlike mass. also by this author). The particular branch of lack of structure in cement gel particles as revealed research on paste microstructure represented by in direct observations. work of fibrous structures could be mainly respon- vanced by Bernal [8] at the London Symposium. observations on thin replica films. According to in developing the theories originally formulated the data given in this review. among various pastes examined by the author seem to many others. such a model can be by Baikov on the course of paste hydration of applicable to special cases of cement hardening. of amorphous based on data obtained in direct EM and ED carbon or other materials. the finer details of paste surfaces and of identifying at the present time. (a) Laths of C5SH1. that the first products of hydration 644 .

1955) Russian. Stationery Proceedings of the Third International Symposium Office. to establish any functional relationship be. These views were based on observations of siderably. and that these products are. S.. far-reaching theories on the general physics between structure and mechanical strength. and M. Budnikov.. fibrous microstructure is associated with high [99]. However. Therefore. of these links between surfaces.. in mathematical Another question pertinent to the relationship terms. However. of surface lattice structures. low w/c ratio. and the relative tween microstructure and. e. unlikely contribute much to the structural to A reconsideration of cmrent ideas on paste rigidity of certain types of pastes is also partly microstructure is obviously needed. In view of the great variabil. possibly act. In other cases highly colloidal struc- evolution of large surface areas in materials tures are observed in paste products of high possessing hydraulic properties has been par. [3] R. Moscow 1956. 645 . permeability. but the experi- opposed to some views on cement hardening put mental material on the systematic influence of a forth tentatively in a recent report by the author number of different factors must be enlarged con- [34]. The evidence on this point given ity of microstructures in different cement pastes. as initial w/c ratio and porosity.Y. F. preponderance of different types of forces at or other important physical properties such as various stages of hardening. N. New York. 1956). gel particles. etc. edited by P. (in ed. in order to develop. before any definite statements can be hydration processes in cement slurries. However. English translation by M. Zement-Chemie (Verlag Technik. Considerations on Microstructure in Relation to Physical Properties It is evident that there are still a great number chemical bonds (valence forces) formed in the of problems concerning the microstructure of process of a direct superposition and intergrowth cement pastes remaining to be solved. Lea. made. as is demonstrated by crystal elements (e. Crystallographic 1952). M. as far as is known uct a monolithic structure of intergrown crystals. mechanical strength. M. Moscow. M. etc. converted into excess of liquid phase may be essentially different crystalline products of continuously increasing from those formed during setting of fresh paste of degree of crystallization. cement gel.. W. Port- 14] F. 1956). emphasized by these as is also strength. whether the presence of any particular habit of the specifically cement paste. 2d Butt. The application of creep under load. Corbin. Datafor the Calcium Silicates. land Cement Association). fibers) in a paste structure Powers and by L'Hermite. Royak. (Edward Arnold Ltd. it seems colloidal masses. it would seem that in ing via intermediate layers of water molecules. It has been reported that paste prod- characterization of cement gel is the enormous ucts consisting of. References [1] H. The Chemistry of Portland Cement. However. (Reinhold Publ. Yu. London 1952. in their review papers is more favorable for the development of strength of this Symposium.. leads to disintegration.g. mechanical strength. Yushkevich. Berlin. to the author. in that the structures formed in the presence of an later stages of prolonged hardening. H. These the ultimate result as much as does the habits of links or bonds between surfaces may be largely of physical nature (absorption forces). well-crystallized tobermo- specific surface area developed in the collapsing rite plates show high strength. has not yet been studied with re- The statement that the fibrous constituents are spect to microstructure. in the literature related to the subject is rather it seems that the only property remaining for a confusing.. 1956). e. Bogue.g. Corp. The Chemistry of Cement and Concrete. whereas pastes anhydride structures due to the penetration of composed of fibers or crinldy foils may be poorly water through them. Obviously such quantities but also the nature of links formed between sur. forming in the end prod. Heller and H.g. London. M. [6] Reports of Symposium on the Chemistry of Cements. . O. with the present set of for future cement research is to establish the nature data. Taylor.formed during setting are precipitated as Jiighly taking all available evidence into account. it is not only the development of surface. are also factors which influence and strength development in a paste. One of the objects it would be very difficult. [5] L. (H. London.. P. The importance of the integrated. than others. which material. In stUl other cases a ticularly stressed by Powers and by Brunauer [98]. Literature^ on Structural Materials. is and the mechanical behavior of hydrogels. Kiihl. while a recrystallization of the structure authors. [2] on the Chemistry of Cement. it results obtained in structural studies on more is not necessary to know in detail the molecular well-crystallized materials may prove helpful in microstructures of various components of the solving this problem. curing and mLxing faces that determines the properties of cohesion conditions. (State Publication of 2d ed. further research on the hardening of pastes the but in view of the dimensional stability of hardened formulation of some fundamental theories on the paste it seems that some parts of the interacting influence of the shape and size of gel particle surfaces must be connected by means of strong elements must also be included.

9. Congress Ind. and W. W. H. S. 7. and M. Akad. Acta Cryst. L. [6]. and D. J. No. and tetra- 1178-87 (1943). Portland Cement Assoc. Rev. Astreeva and L. [2]. Roy. G. The system of lime. T. D. J. Buckle. SSSR. Beton. pp. Lwow Polytech. 41. Am. 361. W. 463-6 (1956). of woUastonite. E. Maddrell's salt. 1959. J.Os 13H2O and the related 183-200. see Ref. D. Dent Glasser. 107. (The University of Chicago Press. Electron-microscopic Crystals from portland cement hydration. Kalousek and A. 80. [29] E. of Concrete Re. [8] J. (in Ref. 8. Composition „ Khim. F. Zeit. and normal and steam curing conditions. 0. Steinour). 526-527.. Crystal chemistry [56] J. (Frankfurt). V. Andrievskii. Tokyo 1956. Antrim. and H. Considerations on concretes in general. 206. [4]. Taylor. 124-32 (1958). pp. J. Priklad. Acta Cryst. Nabitovich. Gard. Gard. 5. W.A. Electron-optical investigation [52] E. 5. E. S. On the structure of the hydration the electron microscope in unit-cell determinations. III. Chem. 198 ff. W. Sliepcevich. Zhur." microscopy of cement pastes in the course of [38] C. Tilley. Travaux. A. Mag. [50] H. 23. W. No. 3-9. calcium aluminoferrite. [20] C. Discussion on [42] P. [26] A. Am. 25-9. calumite. see Ref. Prebus. Gard. 5. H. The microscopic structure of Swedish). Kalousek). 3. and I. Eitel. Mag. Shepinova. M. sodium polyarsenate (in German). „ (in German). Nauk SSSR (1955). 22.I. Taylor. 351-8. 9. [4]. M. p. 646 . 16-28 (1959). D. Electron. S. Mamedov. Grudemo. J. [40] L. [16] J. compounds. Asia and Oceania. [28] S. A. Bystricky. and on cement concretes in particular (in French). (unpublished). Chem. 693-7 (1939). [30] G. 16 (July 1959). Changes in structure and properties NBS. see Ref. Beton. Proc. Mag. Antrim. des Materiaux. [25] G. 23. Miiller-Hesse. Soc. Morphology Nature. F. Study of hardened cement [39] F. [45] H. paste by means of the electron microscope. French). nicum. Cement [47] G. Proc. Proc. pp. Taylor. [10] F. Liebau. Liebau. Grudemo. 121-6 (1959). Radczewski. and water from 50° to 250° C. The use of the stereoscopic tilt device of [33] U. 10. see Ref. (in Russian). 193-210. W. Iwai and K. S. Chem. Gildart. Lea. Appl. Portland Cement Assoc.. E. Solovyeva. Zement u. [41] F. L. Study of the Electron-microscopy study of lime hydration mechanism of structure formation in cement (in German) Zement. [44] T. Zement u. [15] J. stoichiometry of the hydration of /3-dicalcium [54] Kh. Lea. Chem. mechanism of their hardening. Soc. hydration of portland cement (in Russian j. 154-72. Stolnikov. Swed. and E. County 507. Dissertation. Phys. [9] J. L. Crystal structure of )3-wollastonite. S. F. 761-7 (1958). 75-92 (1954). 1012-3 (1958).. 489-97 (1942). Recent data on calcium silicate hydrates. [34] A. H. Mamedov and N.. Soc. D. 1954). 752-4 (1955). Chicago. J. (1953). W. 394-400. Taylor. R. 26 (1955). Eitel. Schimmel. H. see Ref. E. Tsement. Stroitel. Specimen preparation for electron micros. Thilo. Physico-chemical concepts on the "Cement hydration at ordinary temperatures" mechanism of setting and hardening of mineral (H. D. Illinois. Nauk ture. [13] M. A new procedure for studies by electron 42. 35 tricalcium aluminate.Microscopy Proc. alumina. Ceram. a new mineral from Scawt Hill. Reports on electron-optical studies. 719-25 (1950).Remarks on the systematics of crystal crystal formation during hydration of cement structures of silicates with highly condensed anions clinker minerals. A. 1-11 (1950). L. Bogue. see Ref. Buckle and H. 324-329 (1936). Electron microscopy in cement research plasticizers (sulfite liquor) on these processes. 16 (July 1959). J. and other properties of tobermorite and related [57] Kh. V. 37-9. 16 (July 1959). Electron microscope studies of [48] F. Studies of the hydration of portland [22] V. of the action of surface-active reagents on the Bruxelles 1954 (Ind. 39. 361-7 (1956). F. The hydration of [55] K. 1955). No. V. Nurse and H. [24] A. products of cement (in German). Doklady Akad. 477-91 (1952). Research [12] A. Satalkin. hydrate. Chem. F. pp. Tech. E. (1943). zur Strassen. Am. Danielsson. 14-8 (1956). dicalcium ferrite. (in German). Bannister. Bernal. N. Bessey. 43. Brownmiller. Ind. pp. The Physical Chemistry of the Silicates. Clay Minerals Bull. Stockholm 1938. and L. R. M. [18] O. 173-9 231-45. pp. 52. and H. Symp. Peppier and L. 216-60. No. phase during hardening of cements. p. the hydration of calcium silicates (in German). [6]. 818-23 (1958). 27th Internat. F. 11-5 (1957). 1st and calcium carbonate with an electron microscope ed. Materialy.. The 1351-5. (in Proc. Ya. Tolliday. Stork and V. (1957).uid No. V.. Mtiller. [46] O. search. Jones. Dornberger-Schiff. [43] E. Boutet. Strelkov. Russian). Rehbinder. and the [37] F. — [7] R. [53] E. Russian). of cement paste and concrete when hardening [36] H. SSSR. Crystal structure silicate and tricalcium silicate at room tempera. E. Inst. Megaw. binders. Katz. Zement.. and adsorptive effects of surface. Changes in the true state of the liq. [51] L. Inst. A. Co. Eitel. Tichonov. „ No... An electron-microscope investigation cement. hydrated portland cement. The structure of afwillite. phology of calcium silicate hydrates. Bernard. Heller. Studies on 4CaO Al. The calcium aluminate complex salts. Russian). 24. An study of Portland cement hydration products electron microscope study. Crystal structure of xonotlite (in phases. Hey. A calcium analogue of the hydration and hardening of burnt lime (in of chondrodite. Grudemo. 31. see Ref. Discussion on "Cement hydration at and Concrete Inst. B. [6]. Ceram. Concrete [35] R. On tricalcium and /3-dicalcium silicates in pastes under the structures of (3-wollastonite. [6]. Kantro. copy. Regional Conf. 607-13 (1934). Lopatnikova. Adamovich. 1958. The structure of dicalcium silicate a- [27] A. The structures of cement hydration [32] Yearly Report of Research and Development Division. Taylor. (American). 87-91. Brunauer. Soc. Copeland. Watanabe. natural mineral hydrocalumite. Wells. L. Hydro- hardening (in French). 1st. Segalova and E. L. Buttler. F. Brit. Chem. Zement-Kalk-Gips. [21] D. Cements. Beige. 182. Belov. 23. 134-8 (1957). Am. R. F. Acta Cryst. A. des Materiaux. Gullkornet. Eng. Buckle and H. 724-8 (i952). M. H. 163-72 (1959). 1957-58 Ca3(Si030H)2 2H2O. [17] R. Megaw. No. G. J. Ettringite from Scawt Hill. pp. 208. [2]. Am. 73-92 (1956). 33-5.. Taylor. Beton. see German). Investigation of calcium hydroxide [49] W. [2]. Laboratories. A note on the nomenclature of the [31] H. 3. Schwiete and H. Tricalcium silicate hydrate. suspension and effect of admixtures of hydrophilic [19] W. 125-37. Physik. active admixtures. 288-91 (1957). Appl. 311-18. Chim. Mineral. elevated temperatures" (G. 28. Rev. see Ref. [14] P. June 1958. J. [6]. 183. of hydrous calcium sihcates. L. 138-53. Microstructure of cement paste (in [11] L. Chem. Miner. Electronographic studies on the mor. On the chemical nature of cement under load. hydration products (in German). Mineralogist. E. Stockholm. (1959). No. Zement u. pp. see Ref. and properties of calcium hydrosulfoaluminates (in [23] A.

Chem. 896-902 (1956). 362-3. [90] T. J. 31. [85] J. Robertson. F. Doklady Akad. Favejee. A further investiga. 25 57-61 (1955). Chem. soc. Discussion on the reactions of (July 1959). pp. Taylor. Nauk SSSR. (1959). Mag. [98] T. Swed. Research and Develop. 7 543-6 (1957). A. No. and R. J. S. 22. the system CaO-SiOz-H^O. H. L. J.. 163-5 (1958)... I. the reaction Osterreichischen Zementfabrikanten. Yoshii and G. 64. W. 24 1654-7 (1954). Fellowship. Chem. Kurczyk. H. Mag. L. Belov. W. 233-48. [3]. Electron-microscope studies on hydrated [71] cements: Morphologic studies of carbon replica French). Soc.. No. 1-5 (1957). 9. Kupreeva. 102 417-31 (1940). Ch. M. Nature. S. Contribution to the study of the [64] J. Kheiker. of mica-like calcium silicate hydrates: rkenite. Hydrothermal reactions between cement (in German). and F. Unbroken tobermo. Saji. Gard and H. McMurdie. F. Am. [62] H. Belov. V. gyrolite. Nature. X-ray Chem. L. Foshagite. Nauk SSSR. hydration of pure calcium silicates (in French). 151-8 (1959). I. Kalousek. Gyrohte. Fundamental factors in the drying (Stratling's compound) (in Italian). Ca8[Si60i. Central Research Institute of Industrial tobermorite. I. images (in German). Ershov. Roberts. (1958). 120. Appl. [65] J. W. Tech. 40. W. Ricerca sci. Semento Gijutsu Nenpo. 0. [75] R. Hydrothermal reaction [104] C. S. Goldman. Crystal structure [84] E. NBS. H. nekoite Investigation of the Franke method of determining truscottite. Lab. 121. Mikhailov. Truscottite. Assarsson.. Compound formation at ordinary temperatures. Bull. 381-3 tween the montmorillonoids and the calcium silicate (1958). [59] J. [99] S. 989-1011. Mamedov and N. Nov. Edelman and J.](OH)6 (in Russian).. 71. see Ref. Zement-Kalk-Gips. u. Mag. [89] A. Portland Cement Assoc. Mag. 1954. Bull. A. Akad. Structure and physical properties of [74] T. X-ray. Schippa. 11. Am. (in [96] W. Heckman. Belov. Forschungsinst. p. Brunauer. On the physical properties of hardened [73] G. Doklady Akad. Moscow. I. L. F. p. Semento Gijutsu lonite. Gijutsu Nenpo. Beton. knowledge of a hydrated calcium silicoaluminate [80] G. 21-2. Czernin. Z. [105] V. Gracheva. Electron microscope studies on hydrated [95] Roy. [100] M. Am. see Ref. Clarke. 8. 361-70 (1957). [94] K. Minerals Bull. Beiges. (1956). [92] L. V. free calcium hydroxide and free calcium oxide. C. Scotland. Crystal structure 2CaOSi02 (in Russian). D. 11. Turriziani. cement mortar (in Japanese). Phase composition of the foshagite. Silicates inds. (in Russian). 41. analysis of the calcium sulfoaluminate hydrates [79] H. W. 213-8 (1957). Electron-optical studies of tricalcium [67] Kh. J. Research NBS Mineral. Gard and H. Portland Cement Assoc. Taylor. A. 6. L. 74-80 (1957). calcium silicates hydrates (in Japanese). Funk. calcium hydroxide and amorphous silica. McMurdie. 123. A. M. E.C. New York. 390-1 (1956). Pressler. S. 333-51 (1955).Y. cement hydration at elevated temperatures (G. Science Bulle- silicates. 1-22. [103] L. 1-43 Nenpo. G. 94. Soc. F. Mackay and H. Inst. 98-111 (1956). Phys. Zement u. (in Russian). Crystal chemistry of hydrous calcium (in Russian. Portland Cement tobermorite. [78] H. [88] R. Crystal structure No. 1959. Taylor. Crystal structure silicate hydration. Kalousek. D. 5. On the crys- products of the system calcium oxide-silica-water tal structure of montihorillonite and halloysite (in Japanese). Wells. electron-microscope. D. 25-6 (1957). article in The Science of Engineering rite crystals from hydrated cement. Taylor. J. 22. and dehydration study. F. Am. [101] N. Zement-Kalk-Gips. 1-6 (1958). 450-7 (1954). Electron microscopy of hardened cement [69] A. [102] N.. Some observations and Develop. S. 1950. Research [77] R. Turriziani. Concrete Research. 264-78. [86] R. 183. [60] A. W.[58] Kh. L. hydrates of the tobermorite group. S. 14-23 Z. 1-15 (1958). & Sons. of tobermorite (tobermorite phases) (in Russian). Mamedov and N. J. F. cements: Observations on crystal phases and pores [70] G. Research Inst. 1957). 48-54 (1957). anorg. 119-23 (1958). W. F. Gaze and R. 636 ff. No. Relationships graphic structure of clinker on the properties of between calcium silicates and clay minerals. Nacken. John Wiley Research. N. Concrete Inst. Gard. 372-5 (1958). H. E. Mag. Grudemo. van Bemst. W.. man). Taylor. Constructions. Brunauer. Czernin. 30. tion. Ceram. Ceram. Doklady Akad. [81] J. Electron-microscope study on hardened mental data for the sorption of water in montmoril. Ver. Gard and H. Nauk Hesse [3i] SSSR. Berkovich. A. Semento Soc. Self-stressing reinforced concrete. (1958). [6]. V. 16 (July 1959). Some experi. Czernin. 3682-90. Fratini. Effect of phase composition and petro- [68] H. Am. Robertson... 177. Nauk of foshagite. Some aspects of the physics and chemis- [76] R. F. No. shrinkage of concrete block. between 120 and 160°C. see Ref. J. Taylor. 123. Mamedov and N. Contribution to allgem. Stockholm. Kalousek). 223-8. Powers. Howison. F. 5-20 (1956). of Concrete Materials (edited by J. Mineral. Properties and forms of hardened portland cement paste.. (1956). F. Taylor. Mineral. W. chim. English summary). S. peratures of 21 °C and 90 °C. Chem. 0. 1-92 (1955). 3.. Bull. Taylor and J. F.. Ricerca sci. see Ref. H. G. Swerdlow. Washington D. W. [93] W. 639. Mineral. Mitt. 28. Assoc. V. and H. 80. 31. [2]. Howison.. H. Mackay and H. W. Kelsev. van Bemst. Bogue. [87] M. Gille. on calcium silicate hydrate (I) tobermorite. Crystal structure of and N. Concrete Inst. and dehydration. [3]. A. The new silica radical (Si60i5)n.. Clay cement paste. 291. Note on Research in Progress. Substitution of aluminum in lattice of tins. L. reported by Schwiete and Mtiller- of hillebrandite. 276-93 (1957). 69-78 (1953). 647 . 7-12 (1956).. pp. W. Krist. Study of the system CaO-Al^Os-HoO at tem- tion of tobermorite from Loch Eynort. 43. June 1955. V. 720-3 Anal. L. Setting — — hardening expanding (in Ger- [61] A. 62. 12. Zement u. unit cell. try of cement. Nov. Mamedov and N. /3-Ca SiO below 120° (in German). Beton. Inc. composi. Gard and H. 901-3 (1958). H.. of No. J. Doklady SSSR. Kalousek. Tobermorite and related phases in (in German).water system: analogies be. F. 10. Fratini and R. New calcium aluminate hydrates. Kantro. 30. and H. S. A. 171-3. 11. [97] W. Mag. Akaiwa and G. Belov. F. Crystal structure of Hydration of tricalcium silicate. 80-91 (1953). Paper [63] Kh. J. The hydration of calcium silicates. The products of the action of water on (in Italian). 741-3 (1958). Bogue. 16 [72] H. Nauk SSSR. Gaze and R. and D. Sudoh. [82] S. Mag. Mineralogist. Hydrated calcium silicate. Sudoh. of Concrete Research. Taylor. W. The silica. 121. Megaw and C.. Doklady Akad. 30 367-406 (1943). pastes (in German). [83] G. Okenite and nekoite Synthetic compounds related to tobermorite: An (a new mineral). [91] H. products of hydration of 3CaO SiOj and j3- [66] Kh. Cement and Concr. F. Lab. Taylor. H.

L. The primary purpose of our work this series.1-trichloroethane and sub. samples taken at shorter tunes. The frozen sample was then more slowly. the particles cannot be distinguished. electron diffraction in the future.4 by weight using a procedure previously of the cement starts immediately upon contact described [4]. clusters of acicular particles have analysis) are: developed. During compositions of which (by X-ray diflfraction the second hour. We shall try to identify these plates by ranged from 1 niin to 2 months. microstructure of hardened paste by following the Figure la is a micrograph of unhydrated cement development of structure in the paste from the 15754.8 1.6 review of the inicrostructure of hardened pastes. Pastes were prepared from two commercial Figure 4a is an electronmicrograph of a paste cements. The particles some data on the morphology of fresh pastes. lot Nos. A/F 9. These same pastes will be reexamined has been to see if we can learn more about the later for this latter purpose. to illustrate the development of structure in the times difficult to discern any definite morphology". Their identity was established by their prepared by grinding the dried samples in a syn. 648 . in which it is some. the surfaces of the cement grains transferred to a vacuum desiccator and the ice being partially covered with a product having the was sublimed.0 and in his review has pointed out certain areas Ferrite. time of the first addition of water to the cement The magnification is the same as that of the until it is completely cured.1. Tie tre tment of of the particles. 15754 and 15669. He has concluded that "by far the large -t part of the A series of electron micrographs of pastes made cement pastes examined consists of exceedingly from each of the two cements has been selected ill-formed. i^. 1. Hydration of 0. % 25. figure 3b. the hydrated layer is thick enough were sealed in air-tight containers and at the to cover the surfaces of the original grains of cement prescribed time the hydration process was stopped 15754 with the small. colloidal products.9 1. % 6. and show Pastes were mixed with a water-cement ratio little evidence of surface hydration. the phase of cement 15754 after 2 hr of hydration. Schulz 15754 15669 Mr. Within 1 min after adding water to atmospheric pressure and poured into covered the cement.cicular crystals of gypsum are present container was closed. — Discussion Cement lot No. for each time of curing was selected to represent amined do not support the theory of paste struc. except that the pastes were mixed at with water. Grudemo.22 where little or no work has yet been reported. have a characteristic form and are easily reccg- iSpecimens for the electron microscope were nized. Samples to be taken before setting occurred After 14 min.44 7 5. The hexagonal thetic-sappliire mortar. so for the most ture proposed bv Bernal [1]' nor the views of part the series does not give a complete representa- Rehbinder [2] and Strelkov [3]. The micrograph He further concludes that the data he lias ex. the major products of hydration. Curing times after mixing square. Copeland and Edith G. small plates develop on the surfaces plastic containers for curing. with pestle. E. almost square plates. figure by placing the container in a freezing mixture of 2a. % 53. figure 3a. be associated with grains of cement are present. lb is a micrograph of cement 15669. jecting the suspension to ultrasonic vibration for After 1 hr the layer of hydration products on two min. suspending the plates are larger than those observed in paste ground material in 1. onlv 14 min.9 56. We can now present hydrated pastes which follow. and within 6 min small rodlike par- the paste after it was mixed depended upDn wheth. A drop of the suspension was then the grains of cement 15754 no longer appesirs to be placed on a carbon-covered parlcdion film sup- composed of small plates. samples were to be taken after If same appearance as that described above. The appear- rated in a current of air free of carbon dioxide and ance of the grains of cement 15669 hydrated for the specimen was transferred immediately to the one hour is about the same as in pastes hydrated microscope. electron diffraction pattern. The liouid was evapo. Ferrite %. Grudeino lias presented a comprehensive Alite. colloidal masses of the paste.. the nature crystallized phases was of little interest. er samples were to be taken before setting or after Some of the plates are hexagonal.1 29. so no of the structures formed during the setting of effort was made to include their development in cement pastes. Belite. along with the platelike particles. The hydration of cement 15669 proceeds dry ice and alcohol. others appear setting occurred. At the prescribed time the probably formed during the quick freezing of the sample was crushed and then dried in the vacuum paste in the sample preparation. figure setting occurred excess water was added before the 2b. ticles appear as the small plates increase in number.0 CsA. for the We have done some work in one of the areas of purpose of this work the formation of well- ignorance pointed out by Mr. have smooth edges and sharp corners. tion of hydration processes. Clusters of acicular particles that do not seem to ' Figures in brackets indicate the literature references at the end of this paper. Thepe crystals desiccator. For example. The form of ported by a copper grid.

15669. 15754. Figure 1. Figure 2. (a) Cement No. 15754. (b) Cement No. (a) Cement No. 649 . Unhydrated cements. 15669. Pastes hydrated 14 min. (b) Cement No.

650 . 15754. 15669. 15669. ' Pastes hydrated 2 hr. I Figure 4. Figure 3. (a) Cement No. Pastes hydrated 1 hr. (a) Cement No. (b) Cement No. (b) Cement No. 15754.

particles seem also to be getting broader. identical in appearance with now completely covered or completely hydrated. These rolled foils do not appear plates are striated as if they were fibrous. the clusters in pastes of cement 15754 which can be seen at thin edges of the aggregates. figure 7b. Figure 6b shows a ance of this slieet is very much like that of crepe small aggregate or a particle from a paste of cement paper. pastes of cement 15754 ever. from a specimen of paste made from fibrous appearance. shows clusters. 651 . lathlike particles. shows that after 6 hr hydration the edge of the aggregate appears to be a continuous clusters have become larger. thin plates After 4 hr. After 16 hr hydration the clusters of acicular No acicular products are present in the paste of particles have disappeared completely from the 15669 after 2 hr of hydration. 15754. The the particles do not have as sharp points as others. in that some of that of the particles seen at earlier ages. The outlines of larger acicular-shaped particles have appeared. Figure in pastes of cement 15754. The texture of surface is the same in all these — the edge of the sheet is smooth that the striations particles. The irregular aggre- the hexagonal plates are larger than before. but the shape of some of the The surface texture of these plates is the same as particles seems to have changed. The surfaces of these 18-C2S and C3S. In pastes of cement 15669 clusters of flakes are now evident. The thin edges of the irregularly shaped acicular clusters that looks very much like the aggregates that one can see seem to be composed rolled "foils or tubes that are found in pastes of of thin plates or flakes. Pastes hydrated 4 hr. and field is shown in figure 7a. and makes them appear to be aggregates are not separated individual fibers. the product formed in the hydration of cement Alternatively these clusters may have been broken 15754. The mass seems to be a single may have started on very small grains that are star-shaped particle. still the blunt particles are shaped more like laths. Cement No. and broader difi^raction line around the particle indicates that platelike particles can be seen in the same cluster.These may have precipitated from solution. How. a typical continue to form on the surfaces of the grains. 8a shows one such aggregate with a pronounced Figure 6a. After 24 hr hydration. in addition to the acicular particles one cannot be dispersed sufficiently to show individual occasionally sees a particle embedded in the particles. Figure 5. but many lathlike particles and figure 5a. This micrograph is slightly underfocus. figure 5b. of 15754 after 2 hr hydration. The small platelets hydration products of cement 15754. have grown larger. paste of 15669 hydrated for 16 hr. and the Acicular particles. or 15669 cured 6 hr. The thin foil on the upper cement 15754. and look lathlike or rodlike particles imbedded in the much like the acicular particles present in pastes clusters can also be seen. from larger clusters formed on grains of cement. (b) 15669. The appear- of exceedingly fine fibers. but the individual sheet even though striations are seen on it too. (a) Cement No. gates appear to be formed by flat.

Pastes hydrated 16 hr.Figure 7. (b) Cement No. 15754. (a) Cement No. 15669. 652 .

The striated sheets are more 24 hr hydration. Thin regions of the aggregates paste of cement 15754 hydrated for 2 days. There is one important of particles shown in figure 9b has a thin sheet difference between the two sets: in pastes of attached to it as if it were part of an outer layer of cement 15669 many particles appear which are the roll that had been torn away. In other fields relatively large that appearing in this field. shaped like rolled sheets seen in younger characteristic appearance of the hydration pro. are shown in figure 10. A significant number of tubes cluster of acicular particles can be seen. pastes made from cement 15754. and sometimes thin sheets of than can be seen in pastes hydrated several years. in other or rolled sheets. are embedded in the aggregates. regions of aggregates again appear to be made of Occasionally one can find small thin. (b) Cement No. figure 10a is a micrograph The progress of hydration as seen in these of paste from cement 15754. can also be seen. 15669. Figure 8. 15754. 10b that of a paste micrographs suggests that the silicates in portland from 15669. It will be With longer curing times it becomes increasingly interesting to discover whether this difference is to disperse the pastes sufficiently to see diffi ^ult the result of the high belite content or low C3A much detail in the structure. (a) Cement No. fields particles similar to rolled sheets. Figure lib is th'n edges. hydration is a mixture of rolled sheets or tubes The main characteristics of pastes of cement 15669 and the irregular striated sheets seen in pastes of are about the same as those of pastes of cement 15754. In the field shown. so predomi. The clusters of acicular and lathlike particles to be composed of sheets much like those seen in are stillpresent in pastes of cement 15669 after other micrographs. Thin foils are visible with sirface from a paste of 15669 cured 28 days The thin striations similar to those found in younger pastes. pastes. Pastes hydrated 1 day. ducts seen in all these pastes. Tne surface similar to those found in pastes of pure C3S and texture of this sheet is similar to that of the i8-C2S. The appear very much like the product that can be aggregates are too thick to show detail except at seen in less well-hydrated pastes. Pastes cured 7 days content of cement 15669. a pronounced in 10b. at first. flakes. The effect of the difference in the rates of hydra- The paste of cement 15669 after 2 days tion of these two cements is evident in this wcrk. fragments like striated sheets. Figure 11a is a micrograph of a typical field from Figure 9a is a characteristic micrograph of a a paste of 15754. figure 8b. whereas these particles were not found in striated sheets. were found in other fields. Not much structure can seen be in cement hydrate to foi'm. The particle at the right of the aggregate 15754 at an earlier age. These clustei's of acicular particles 653 . These have the same particles. No more details can be seen at 28 days hydration n\nt in C2S pastes. clusters of acicu- 10a except that thin parts of the aggregate seem lar particles.

654 . (a) Cement No. (b) Cement No.Figure 10. 15754. 15669. Pastes hydrated 7 days.

of the structure involves some sort of optical nism of hardening. The absence of acicular particles and the presence of laths and In discussion of the microstructure of hardened sheets. 1956. approach. 183-200. Mann.. there necessarily enters the partic- indicates that growth of the particle transverse ular concept of something liappening in between to the axis of the needle formed initially must the cement particles. The clusters become larger and denser. cles are individually in the size range of the wave 655 . It 1952. Figure 11. Yu. in the more hydrated pastes cement paste. the binders. direct perception phase during hardening of cements. The surface texture can be explained is a concentrated suspensoid of independent solid by strong orientation of the silica chains within particles in water. grains becomes the central feature in investiga- and M. C. [3] M. (in Russian) (State Publi. pp. because (a) some 20 percent of the cement parti- Proc. and H. M. Within a time inter- val of minutes to hours the mass has become a single solid unit with very substantial shear References resistance. The transformation results primarily from chemical reactions within the mass. or flakes. I. and it seems logical to suppose that interlocking and growing together of the clusters of acicular Levi S. pp. ibid. and the mecha. 285-298 (1954). D. Strelkov. Concrete 'inst. are formed at about the time of the initial set of Discussion the paste. (a) Cement No. Moscow. M. shear resistance of the bulk the sheets. Moscow. M. Powers. S.. 15669. 51. Proceedings of the Third International Symposium on the Chemistry of Cement. paste. Permeability of portland cement paste. In the light of these observations. The structures of cement hydration known that the chemical reactions continue long compounds. P. Pastes hydrated 28 days. 15754. L. C. 216-26. pp. Physico-chemical concepts on the occurrence in good concretes after many years in mechanism of setting and hardening of mineral service. Changes in the true state of the hquid Being mechanical in nature. Am. A. O. Rehbinder. The cement mix initially take place. 1956). edited concept of something happening in between the by P. Budnikov. Copeland. Royak. (b) Cement No. is known that cement still unhydrated is of common [2] P. after the arbitrary time defined as final set. 125-137 of Reports of Symposium on the Chemistry of Cements. London. J. The problem is especially elusive [4] T. tions of the microstructure of hardened cement cations of Literature on Structural Materials. Yushkevich. Brown particles causes final hardening. Bernal. E. Butt. Hayes. It is [1] J. mass is substantially zero.

nevertheless. with Fe203. the various minerals in cement. better. the continuous phase. The initial pected to be different from that of the earlier structure. The question is easily subject to answer by direct by something happening between the grains. mineralogy recognized. one may inquire as to the manner to the more common concept of cementitious bond of long time hydration of these larger units. however. to say nothmg of more specific identification. Observation of cement still unhydrated after the mass of fresh concrete in the form comes into many years in concrete of course means observa. existence as a competent structural unit through a tion of the larger particles or. Though never in equilibrium with the unhydrated cement That is because tiie initial clinker structure is minerals. cement components and water. in conclusion. one colorless (or dark. essary to identify the secondary substances. it is relatively easy to recognize Since 75 percent of the cement is above this size the remains of progressively smaller particles. it follows that most of the cement particles. approaches an equilibrium composition. iso sible. that is. It is no trick at dration processes usually centers about reactions all to reveal the centers or cores of unhydrated clinker whicli. observations certainly indicate a high degree of and without evidence of mecluinical distortion or interaction between the products of hydration of volume change through the process of alteration. for the writer. in part on the fact that this profound and pervasive chem- because of their relatively large size. in brief. It is the size of these larger particles that clearly It isonly in the size range of 5 to 7 m and below defines this pseudomorphous hydration. to offer a In considering the niicrostructure of hardened thought relating this pseudomorphous hydration cement paste. the manner of interaction may be ex- observed to be perfectly preserved. These initial minerals may be com. The interpretation of necessarily specialized subject pseudomorphs of former C3S or C2S are variable. preparation. Significantly. visual microscopic observation of the initial and the altered material. and for its significance. it is unnec- follows that separate distinction of the fine parti. such residuals thereby that the alteration must be pseudomor- are formed only in good concretes. or polished sections of the hard- of disequilibrium is extreme. For also is the limit of resolution by microscopy it this purpose. Far more ical alteration occurs without significant changes in importantly. Ake Grudemo tions) recognized as C3A and one a dark brown (or bright. are always per. and likewise to be affected bv the pro- C'S and (b) nests of orbicular C2S. though with progressively lesser clarity and se- particularly the larger individuals. reveal also the clinker mineralogv. curity. together mak - gressively lesser availability of the'liquid. of course. for reasons or under conditions not known. initiallv is water. Initially the range powder mounts. recognized as such. material the initial boundary of the particle of clinker. which from location. in some variation. showed (a) angular crystals of period. and mu. Once that the particles tend to become monomineralic. the external boundaries of designed capacity. It is hardly more difficult to perceive in the altered in what happens in between the grains. it is because of very slow imposed external dimensions. in the chemical reaction between the different pletely altered. and perhaps not too long a time. if the cement particles cretes that must be called good because of many in the mix ever lost significantly their mechanical years of satisfactory and unimpaired service in resistance or rigidity. can be sepa. The reason that these particles fulness of concrete as constructional material rests can persist for many years is. and refractive index can be identified as extends the range. rately perceivable. That is to say that. on polished sections) recognized as It is clear that the niicrostructure of hardened C4AF. as such. for that gen- erally is the period of set. yet these shapes. The fuid in the cement mix. range. As noted earlier. The cement minerals are un stable in contact with water. incidentally and by techniques well through this disequilibnum period. Discussion of hy- ened cement paste or concrete. and further the clinker This manner of hydration holds. yet more structural significance. which thereby is called "interstitial material". The use- larger particles. Clement particles actually are clinker particles. The electron microscope of course notably tropism. con phous. with interest primarily known. The liquid in time. It is appropriate. ing up some 80 to 85 percent of the whole. the mass could not be expected to remain fixed. It is appro- suspensoid concentrations the distances between priate further to designate or to label the manner the grains are of similar magnitude. by means of thin sections. enclosed within the (c) remaining 20 to 15 percent. and which Closure shows (d) itself to be made up usually of two com - ponents. the The structure of tlie altered material is a perfect development of structure in cement paste is 656 . but at the disadvantage of C3A-6H20 or.length of light or below and (b) at the usable pseudomorph of the initial clinker. A number of tual relations are seen to be perfectly preserved. These. can be difficult. on polished sec. as a hydrogarnet. The best defined component is and the intervening spaces ranges into the impos. features. It is indicated access of moisture. Since this of alteration as pseudomorphous hydration. the interstitial material. paste is conditioned by the processes taking place ceivable. That cles. residuals of complex sequence of chemical reactions.

and Taylor.closely related to similar phenomena taking place liydration. of a modified single-crystal up and depoljnuerization of the metasilicate chains. fibrous or rate-of-solution experiments. structural elements are oceedingly small. probably especially in what regards the microstructural substituting for silicon in tetrahedral positions. and structur. of crystal size was observed. if such structural problems in the independently written a process occurs. clearly identifying these particles as being They arrive at a silicate chain model resembling structurally intermediate between a disorganized the one given by Bernal at the London Symposium. diffusion- review papers seems to show that. that the incorpora- needlelike particles are commonly found. Alternativel_y. in the case of well- given more ably and much more in detail earlier crystallized. Buckle be promoted by a high water. most describe the change taking place as probably of them in tlie truly colloidal size range. In this connection Also Mr. it case. and probably leaves most of broad sense. It was a of aluminum ions from the C-S-H gel phase to pleasant surprise to find that a comparison of the the C-A-H plates. and in C3S pastes. Powers. There are in hardened pastes in the case of the crystallization discrepancies and disagreement on some points. but the tendency to fibrous growth mately in unsymmetrical 1 1 layer units instead of : is quite noticeable in most pastes. The habit of thin plates chains from the tobermorite layer. conclude from certain cement pastes. in which an important feature is the presence of Similar particles are seen occasionally also in hydrogen bonds connecting silica groups in the ordinary cement Dastes. and seems to the symmetric 2:1 layers. has had occasion to deal with a tobermorite (G) observed in the vicinity of C-A-H number of problems connected with micro. of course. platy tobermorite. the hardened paste is a remarkably stable must be stated that the hardened paste does not system. the Gard. in which there is a short-range variation The phase now described is generally recognized of structure between the extremes given by low- to be mainly a calcium silicate hydrate. ingredients. of a C/S lime tobermorite. resulting ulti- predominates. Howison. and its modification with increas- lattices. some of them well crystallized. where a diminution in the Symposium. C-S-H gel. than the C-S-H component of and Greenberg. and calcium hydroxide. finely ground cement. after the paste has reached a certain age. I found it necessary to formation of well-crystallized fibers of tobermorite. of an inhomogeneoiis C-S-H by a high C3S content in the cement. amorphous material of composition CH. I would like to draw in cement paste of ordinary water-cement ratios your attention to a number of conflicting ideas. ances in C-S-H to disappear. Brunauer least partly. especially made by Brunauer and his coworkers. which are somewhat better crystallized. such as in liigli-porosity pastes with in the hydration of the pm-e components. and of caused by the gradual stripping off of metasilicate ill-defined outer shape.5 or 2. steamcuring at not too high temperatures. it is still only a local. smiimarize what is known about the microstruc- Thegenerally lower degree of crystallinity of tures of a number of materials of interest in cement cement gel particles can probably be ascribed to hydration. Clearly. and well-crystallized tobermorite. all of them tend to converge to give the gel unaffected. It can be imagined that the but in such cases the need of further research is material required is taken from the residue of clearly indicated. Such reviews have been was shown by Kalousek. in general. plates of hexagonal crystal symmetry. fall in the latter category. the question of 657 . others Concerning the question of the lattice structure with severely restricted or deformed crystal of tobermorite gel. In other words. slowly structure. ratio somewhere between 1. The segregation of In view of this 1 will confine myself to touching such particles may be associated with a transport upon a few specific points of interest. of pastes. by and Taylor advanced the theory of the formation. Kurczyk and Schwiete give a model of high-lime ally related to tobermorite. the picture of the existence of which has been evidenced in studies paste microstructure that has emerged. This is clearly brought tobermorite consisting of low-lime tobermorite out in electron microscopy studies of C3S and C2S layers alternating with complete layers of CH. In any out of electron microscope studies? First. in his unique we must also consider the phenomenon of increased representation of the pliysical properties of fibrosity and more regular crystallization of cement paste. at distorted or disorganized structures. material. tion of aluminum ions in the lattice. the What is then. as well as myself. recrystallization certainly proceed at a very slow tion. at least in a controlled reaction. and to give the essential features of structural disturbances caused by the incorpora- the cement liydration chemistry in general. causing the lattice disturb- data given and the opinions expressed on micro. type. finallj^. as aspects of this subject. In the pure pastes. crystallizing out in pastes. one and the same reasonably well-resolved picture Similar slow processes of segregation may occur of paste structure and properties. therefore. In special cases of paste direction of the chain. there may be an extensive In my paper. However.5 and 2.cement ratio. of large CH plates. which in tion of excessive amounts of lime in the tobermorite electron diffraction diagrams show a series of gel structure is accompanied by a gradual breaking diffuse spot patterns. but is a composite material containing several speed. Most particles or aggregates of particles ing C/S ratio up to 1. and the possibly occurring processes of consist of an integrated mass of uniform composi.

future work on the development of colloidal the course of development of the micros tructure hydrate structures. that is. are obviously decisive without much further work. There are many indications. lecular layers of well-crystallized tobermorite but remain as very important problems for future sometimes overlap. and the mechanical cement clinker. parallel. not with their fiber axes in research. systematic way with respect to microstructure. rather than fibrous or that I have made recently on similar materials needlelike crystals. only to disappear at later stages. it seems to be an develop.the tobermorite gel structure cannot be solved and other admixtures. Copeland and Schulz. taneously at comparatively high rates of speed. well-crystallized tobemorite gels. Transient phases may experimenting chemist. as de- ment of the structural elements proceed simul. attempts should be made to during the early period of cement hydration. together with the effects of sulfate behavior of construction elements. must concern ourselves with the formation and as perhaps the continuously and uniformly spatial distribution of different types of bonds increasing stiffness and strength of the cement between the surfaces of gel particles. Ish-Shalom and Dr. are in very good agreement with those made by The ultimate paste structure is obviously condi. They noticed that unimo. among others some It can be stated. rheological properties of different pastes. in less stages of paste hydration was presented by Dr. factors in the development of microstructures in In this connection. at least. and Taylor. Schulz. as a conclusion. To the paste might indicate. Such a process of over. we hydrated phases is not a straightforward process. in correlate data on microstructure with the observed which a number of processes involving rearrange. scribed in the paper by Dr. that the development of the of the mechanical properties of cement paste. that in apply- derived from electron microscope studies of ing our knowledge of paste structure in the study cement slurries. of particles of Copeland and Mrs. volved here have not yet been studied in a very Howison. The extremely difficult problem to treat theoretically balance in the solution phase between the amounts the relationship between bond formation in a of the more easily hydratable compounds of colloidal system of particles. Greenberg. 658 . However. but rotated at approximately hexagonal A valuable contribution to our knowledge of angles to each other. I would like to point out setting pastes. the internal microstructure at different early lapping might account for the appearance. the relationships in- another interesting observation made ty Gard. It also seems to me that in tioned by the prehistory of the material. Some observations the habit of small plates.

Eine gewisse Anzahl Fragen stehen noch offen und folgende Vermutungen wurden angestellt: 1st der innereVorgang des Schwindens eine wasser-zusammenziehende Erscheinung abhanerig von deni Gleichgewicht der Wasser. celle des agregats et la composition du beton. Contribution from Laboratories du Batiment et des Travaux Publics. wieder in Gang tretender Wasseraufnahme des Zements? Dem Referat folgt eine speziell zusammengestellte Bibliographie mit 230 Referenzen. Paris. Resume Ce rapport constitue un document de mise a jour quant aux connaissances actueUes sur la deformation du b^ton. The following phenomena have been examined: hygrometric. Fra. Paper V-3. Zusammenfassung Dieses Referat gibt den heutigen Standpunkt der Kenntnisse was die Verformung des Betor s -anbetrifft.ice.und der Dampfphase oder befindet sich der Sitz dieses Vorganges in der Kristallphase in der Form von physikalisch gebundenen Wasserschichten? 1st das Kriechen ein zeitlich unbegranzter Vorgang obwohl mit bestandig verminderter Geschwindigkeit oder ist das Kriechen eine begranzte Grosse? Sind Kriechen und Schwinden die beiden Erscheinungen eines selben Vorganges oder sind diese beiden Vorgange verschieden und in Wechselbeziehung stehend? Welches ist die Erkliirung des Kriechen was die Struktur anbetrifft? Hangt das Wiedereinsetzen des Kriechen unter Einfluss von Wasser nach einer pseudo- Stabilisierung unter trockenen Verhaltnisse mit der Wasserbewegung zusamnien oder mit der. Die folgemden Vorgange wurden behandelt Feuchtigkeitsgehalt. chimiques et : dynamiques (action des forces) en relation avec la nature du ciment. D. or does it arise from the crystalline phase in the form of physically bound layers of water? Does creep continue indefinitely though at a continuously reduced rate or is creep a phenomenon of limited amplitude? Are creep and shrinkage two aspects of a single phenomenon or are they two distinct phenomena linked by correlation? What is the explanation of creep on the scale of the structure? Is the phenomenon of the resumption of creep in water after a pseudostabilization under dry conditions linked to the movement of the water or to the resumption of hydration of the cement? The report has appended to it a specially prepared bibliography containing 230 references. in connection with which certain hypotheses have been put forward: Is the internal mechanism of shrinkage a hydroconstrictive phenomenon due to an equilibrium between the liquid phase and the vapor phase. Volume Changes of Concrete* Robert G. Woods). : chemisches und dynamisches Verhalten (Wirkung der verschiedenen Kraften) in Verbindung mit der Beschaffenheit des Betons. L'Hermite Synopsis The purpose of the present report is to give an up-to-date account of the present state of knowledge on the deformation of concrete. and the composition of the concrete. Washington..G. and dynamic phenomena (action of forces) in relation to the nature of cement. Ont 6t6 evoqu6s les ph^nomenes suivants hygrometriques. 1960. *Fourth I vternational Symposium on the Chemistry of Cement. thermal. thermisches. (This paper was presented at the Symposium by H. thermiques. of aggregates. Un certain nombre de questions restent pos4es a propos desquelles ont ete emises certaines hypotheses : Le m^canisme interne du retrait est-il un ph^nomene hydro-constrictif d<i a un 4quilibre entre la phase liquide et la phase vapeur ou prend-il son siege dans la phase cristalline sous forme de couches d'eau liees physiquement? Le fluage se poursuit-il ind^finiment quoiqu'a vitesse continuellement reduite ou le fiuage est-il un ph6nomene d'amplitude limit^e? Le fluage et le retrait sont-ils deux aspects d'un meme phenomene ou sont-ils deux ph^nomenes distincts lies par correlation? Quelle est I'explication du fluage a Techelle de la structure? Le phenomene de reprise du fluage a I'eau apres une pseudo-stabilisation k sec est-il li^ au mouvement de I'eau ou a la reprise de I'hydratation du ciment? Le rapport est suivi d'une bibliographie specialement r^unie et qui cpntient 230 references. A certain number of questions remain unanswered. 659 630446 O — 62 7 . chemical.

which appears to be due to a settling by gave the following shrinkage results for a portland gravity and the action of capillary forces [6. 6. since form of a de-aired slurry. absolute volume of test piece. cement. the time of its mixing in an excess of water in the The nature of the cement is of importance. and we shall do no variations in structure. The results of tests w/c 0.300 X 10"" with a content of 500 kg/m^ The by a thermal dilatation due to the release of heat quantity of mix water has an influence on con- produced by the chemical reaction of setting. produced by to enter deeply into details. vents internal evaporation. These figures. and indicate ti e In general. thermal. applied loads. ventilation at 0. shrinkage at an early age increases with In air. rate of flow of air. cretes. the effects to which these factors areas still to be explored.000X10"" [5]. 1.000 XIO"*^. in any case. following results were obtained: Several optical methods have been used to make such measurements [3. kg/m\ 1. While the halfway between mortar and pure paste. temperature 20 °C. Thermohygrometric Variations in the Volume of Young Concrete (First Shrinkage) The setting and hardening of a hydraulic binder gave at the same time 850X10"". the addi- often preceded by a slight swelling. Figure 1 shows the ap- anhydrous components and the water.50 0. after 8 hr: without forced ventilatijn. and then diminishes. attempt to make a synthesis. 660 .000 XIO"'' for ventilation speeds of 7 to 8 a structure is established. their com.600 1. however. This pheno. The influence of plasticizers is marked. For air at 100 °C he ob. by the 1. course of hardened cement. produced In a necessarily limited account it is not possible by temperature changes. has shown that the volume of nydrates is always The shrinkage of concrete at an early age is less less than the sum of the absolute volumes of the than that of pure paste.600X10"" with a content of 350 kg/m^ During the first hours this swelling is augmented and 2.55 0.000 1. the apparent For a series of concretes. the volume of anhydrous portland cement [1. Dutron's is of the order of magnitude of the loss of water up tests confirm these results and indicate a shrinkage to the time when the grains enter into contact and of 14.6 m/s.550 1. the loss in volume ventilation at 1 m/s. the first cause is dominant and. produced by more than examine the state of present knowledge. then the law of shrinkage m/sec of air at 20 °C. assumes a different form. produced little known. yet will enable readers to discover the elements waUe they may be studied separately. passes through a maxi- shrinkage before setting begins.300X10-^ R. cement and concrete begin to show a the water/cement ratio. poration.350 water: humidity. temperature. absolute volume diminishes. Rilem-Cembureau mortar having a cement content It is higher for aluminous cement.45 0. it slows down to assume the habitual initial shrinkage. the first shrinkage volume of the paste cured in the presence of an varied between 800X10"" with a content of 200 excess of water increases in a continuous manner. Introduction Variations in the volume of concrete may be binati")ns have a complex aspect that is as yet due to a variety of factors: hygrometric. 4]. Since the binder mum. portland cement 7. chemical. by a variation in water content. A shrinkage is in fact influence of the speed of ventilation of air at 50 equally to be observed in the absence of any eva- percent relative humidity. A broad bibliograp] y give rise are not cumulative or commutative. dynamic. this shrinkage can give rise to cracks.60 show that during the hours shrinkage depends first on the conditions affecting evaporation of the mix shrinka?cXlO* . A bleeding of the water in part opposes the first 7-cm layer of pure paste without forced ventilation shrinkage by producing a wet surface which pre- ' Figures in brackets indicate the literature references at the end of this paper.100X10"". 2]'. requiring further study.300X menon has been brought out by Le Chatelier who 10"" supersulfated cement 10. cement gives 4. show the volume. 7]. blast-furnace contraction in absolute volume. It is observed that the tained figures of the order of 19. is accompanied by a variation in its apparent which differ by an order of magnitude. At 7 days pearance of the shrinkage curves and indicates this contraction is of the order of 12 percent of that shrinkage is increased with the proportioning. When a pure cement paste is cured from importance of ventUation.700X10-". does not yet have any other than a capillary For a concrete proportioned at 350 kg/m^ the cohesion. time the surface becomes dry. Rapid shrinkage generally manifests itself between 1 and 2 hr. the whole undergoes a with a speed of aeration of 1 m/sec. After 6 to tion of these having the effect of increasing the 10 hr. The initial shrinkage can be explained only in On slabs of pure cement paste 2 cm thick the part by the loss of water. 7.000X10"".

6 to 1. 1:2 mortar cracks having lost less water.000 XIO-^ and ous while it is at the same time in evolution. According to Dutron [11] the quantity is thus ment.8mm /m 1/2 3/4 1 2 3 4 5 6 78 Time in hour* Figure 1. panies the swelling which occui-s with an absorp- tion of water greater than that which would Hygrometric Swelling of Cements and correspond to the increase in volume. at a speed of 1 m/sec with air at 50 percent relative sions contrasts with a reaction of the adjoining humidity. The mortar having 1:1 portland parts composed either of the parts of thevolume cement cracks after 20 min. tion of the water either by protective membranes This is why cracking tests have been devised. after setting this risk diminishes. We shall therefore speak first of this quite variable. for 661 . This first shrinkage can have awkward conse. Shrinkage from 0 to 21^ hr after placing. cements of this kind have at includes swelling due to physical and chemical 3 yr a linear swelling of from 0. the cracking of • in the first hours.200 X10-^ We may note that the term shrinkage generally Generally speaking. for example. the test consists in grinding a flat is as we have seen rather different. The adding of a plasticizer accelerates the A familiar illustration is the cracking of the cracking. in weight after curing in distilled water.000 days. even following linear elongations [7]: at 100 days. having a structure that is continu. at 2. There after 1 hr 10 min. Thermohygrometric Variations in the Volume of Concrete in the Course of Hardening This dimensional variation is usually called urement being made at 24 hr after mixing. For a from that dealt with in the preceding paragraph portland cement paste we found. The first cracking can be counter- ing on their proportions. 2. This result Concretes therefore means that there is in fact contraction in the liquid-solid combination at the same time as an A cement cured uninterruptedly in water from increase in the apparent volume. It constitutes a different phenomenon swelling continues at a reduced rate. that of concretes or mortars depend. The initial meas. 7000 ent P°*"i. or by artificially keeping the One of these consists in applying a 2-cm coating to surfaces moist [9.5 mortar does not may result cracks on the surface or in depth [8]. and 1:2. 1.4 mm m / Concrete a\ 350 kg |cm3 " 1.5 percent. crack. The increase the time of its molding swells. but the corresponding risk appears only pavement of concrete roads. An increase in weight accom- phenomenon. ^6. for the decrease in diaien. 10].300X10-". the surface of a concrete beam. (curing compounds). For concrete. or by the support. 2. their water/cement ratio. fixed heads. this shrinkage. however. depending on test piece held in place at its broadened ends in their nature. causes independent of the hydroconstrictive move. at 1. acted by preventing or slowing down the evapora- and the addition of products like plasticizers. plaster. etc. The behavior of cements. the in that it applies to a material that is solid. though porous. and ventilating it quences for building.6mm/m 6000 4.000 days.

one in which the surface swells much cement (at 275 days): aluminous cement. Blondiau [15] who used mainly the autoclave pastes. It is to be noted that of sulfates [13]. At the outset it often appears to have nate in the cement and a formation of sulfoalumi- = the form A Q log t where A is the linear expansion. Is such swelling bound to lead to disaggrega- The swelling of concrete in water is much less tion? Not if it becomes stabilized below a limit than that of a neat paste.000 days. 160X10"«.cements.T. and Q a constant.M.). is between 3 and 5 percent. A portland slag cement swells method. X f)ortland. and. and it increases with the magnesia content. It should not be without previous dissolution of the latter. The volume of the sample. A mix of portland 662 . The former case is. for a cement content of 500 kg/m^. For concrete.000 days is 100 to 120X10"^ with an increase the measurement of vibratory modulus of elasticity in weight of 1 percent at 1. Bogue. composition itself of the curing water or in that rom 1to 1. who draws a number of conclusions. and disappears test advocated by the A. There swelling in terms of time is difficult to define is a combination of this salt with the lime alumi- precisely. 300 more than the interior. H. cement alone. Dutron [11] It seems that disaggregation is due to a dis- has studied the influence of the nature of the continuous swelling in the volume on a micro- cement and demonstrated a greater expansion of scopic or macroscopic scale. Bogue [12] out over 6 yr have not reached a stable value. it is possible with elsewhere in this symposium (N. flaking. this figure 2 yr is three times greater in sea water than in corresponds to that given by Le Chatelier's distilled water. the silicates and aluminates. Attempts made on very fine sheets The boiling-water test long advocated by Le of cement appear to tend to make this phenomenon Chatelier has the advantage of bringing out very disappear. Plum et to separate the factors due to the normal hydration al. The law of where there is an attack by lime sulfate. attenuated on mortars. the latter cannot be derived from tests even A thoroughgoing study of swelling in terms of extended over several years. 230X10-*'. It thus seems that swelling is indeed a three times less than a portland in a 5 percent phenomenon of chemical origin that accompanies magnesium sulfate solution. 12 to 16 percent of the swelling of ordinary portland cement after about volume of the anhydrous cement. slag cement. the effect of which is especially moreover. the following: expansion increases with the free tude of the experimental errors. lime content. within which the mechanical properties are not portioned at 300 kg/m^ the linear expansion at afl^ected. nesia. that cement results. nesia makes cements more sensitive to the action sist over a longer period. however. provided the treatment is prolonged for at least 5 hours. These differences are greatly Such is the case where there is free lime and also reduced in the case of concretes. but this varies with in the crystalline form. The shrinkage appears to be due to deduces therefrom a justification for this method. ordi. Tests of paste carried composition has been made by R. mortar [2] due to the presence of expansion nuclei. Chemical Expansion of Cements and Finally an expansion may be produced by the Concretes reaction between the cement and the aggregates. Another highly important contribution is that and the grain size. aluminous cement and a lesser expansion of slag for example. it is subjected to consider- 10"®. including the annual variations being of the order of magni. may there are nuclei of expansion within the volume. able compression and disaggregation through nary Portland. hardening.000 days. however. and it might be considered to be an rapidly the presence of free lime and free mag- "effect of volume". like R. The latter case is brought about when Mixes of slag and portland cements. The presence of mag- although the increase in weight appears to per. The which shows a diminution at the beginning of mortars have an expansion of 150 to 300X10"^ disaggregation.S. and regular clefts are formed in the swells indefinitely. It may likewise be due to mixes of cements from those due to another influence either in the reacting chemically [16]. 240X10"". but if there is a limit. of the mix. the internal desiccation brought about by such He indicates in particular that the addition of rapid hydration that the supply of outside water pozzolanic materials to the slag markedly dimin- cannot compensate because of the impermeability ishes the swelling. the proportions. Although the hygrometric swelling may be We merely call attention to it here for it is dealt explained as having a chemical origin. nate. A^. H. For a concrete pro. rapid-hardening portland. If the sulfate combines with the aluminate t the time. give an expansion greater than that of portland discontinuous and distributed microscopic nuclei. the swelling phenomenon stabilizes appreciable between one month and three months. It is very visible on neat of L. much more rapidly than for neat cement paste. Goggi [14] indicates that ex- at the beginning of hardening a shrinkage under pansion increases with the content of aluminate water may be observed [11]. the dimensions of the test piece. he cements. which would correspond to of the product itself aside from its fundamental an increase in voids of 8 to 10 percent of the components. swelling deduced from this equation. for accelerating for concretes except in the case of aluminous the phenomenon. the effect is much slower. but It practically stops between 6 months and 1 yr. It is possible to know this fact through 1.

temperature. At 20 days for example the coefficient in microns per meter note that the desiccation of a concrete previously and per degree. water 35 50 75 100 content. Hygrometry in 7» Tests on neat paste [17] show a very marked influence of the curing hygrometry preceding the Figure 2. We can write an approximate but approximately the same age under the same simple function: conditions. to diminish with time.cement and aluminous cement gives. For sure given by Kelvin's expression log 6. age. The coefficient of expansion after long maturing in water is in both cases between dd 10 and 13X10-^ Q{i-d){d-ei). Coefficientof expansion of a portland cement test according to the age of the concrete at the paste in relation to water content. increases from 14 for cured in water increases its coefficient of expansion and that the modulus of dilation of a saturated a humidity of 35 percent. 20 a swelling in water at the end of a year three times the average of those of the two components. 15. and 10. In tion of the surface tension of the water with the any case. dt It interesting to note that the average coeffi- is cient of expansion on the return between 60 and Moreover. These phenomena of temperature. the variation in coefficient of 26X10"^ at the age of 20 days pressure corresponding to a variation in temper- (between 45 and 60 °C) for a previous curing at ature dt is: 50 percent humidity and a coefficient of 10 X Add 10"^ at the age of 200 days and beyond for a dt ~' e dt previous curing in water. which depends on the previous state of curing. the maximum de Now -77 is nU for 6= 1 and nil for 6. this coefficient is a function of numerous factors such as composition. If we assume a swelling pres- stimulated by the increase in temperature. -di. the extremes are a where 6 is the relative humidity. This phenomenon is probably due to the balancing of the water with its vapor at the the factor of physico-chemical transformations same temperature. The matured cements thus structure but solely on the presence of water. the water in the voids with the temperature The coefficient of expansion is not independent produces a swelling pressure. In the case of concrete. a humidity coefficient of 39X10"^ occurs between 45 and dt 60 °C at about 40 days for curing at 50 percent of equilibrium below which there is no longer any humidity. have a lower coefficient of expansion than cements There is probably an interference between dilation cured in humidities between 50 and 75 percent and the phenomena giving rise to the hygrometric which at every age present a maximum that movement. the corresponding figures is reversible and does not depend on the solid are 14. There must therefore be a that which is measured between 45° and 60° maximum of swelling for an intermediate value (outward) the dilation-temperature curves for . to 22 for a humidity of concrete is independent of its previous states of 50. etc. for a slag cement it is 25X10"'^ at free water. At 200 days. FinalW the has been observed by various authors. For a portland cement having a high initial strength. It is therefore clear that the phenomenon percent. the average coefficient measured cannot occur in a saturated medium nor in the between 60° and 20° (return) is always less than absence of free water. and drops back to 12 for a humidity of 100 curing. Figure 2 relative to a portland cement shows an increase with the water tity of water under pressure. C. (70 percent according to S. for 30 percent of the former and 70 percent of the latter. it is always lower for curing in water [19] or that the increase of the vapor pressure of than in air. including expression of the instantaneous thermohygro- D Bonnel and F. Harper [18] who in addition metric swelling coefficient has the form: 663 . Meyers [20]). This phenomenon Eq is the water contained for d=\. moreover. the swelling A must be' proportional to 20 °C of a paste aged 200 days varies with the the product of the swelling pressure by the quan- quantity of water included. S. But outgoing and for return cannot be superposed this hygrometric swelling does not constitute a except for a cement cured in water for a very state of equilibrium and must disappear through long time. ^= sdE^ ^ where content followed by a decrease. moment when it is carried out. ^XIO®. It can be imagined that the diminu- appears. f=A a normal portland cement. L. if the coefficient of expansion diminishes temperature diminishes the shi'inkage pressure with age. 15- Thermal Dilation of the Concrete \ Thermal dilation is generally expressed in the 10 form of a coefficient of expansion.

It also appears that the coefficient of expansion diminishes with the cement content in a manner that also corresponds to the law of Loss of woter mixtures. For a young paste. Relation between loss of water and shrinkage of siliceous and Seine limestone aggregates had an a cement paste. The variations in beyond this the deformation is proportional to the terms of age and of the curing medium are reduced loss of water for an old cement and follows a in comparison with those that are found for the certain curve for a concrete that has not completed neat cement paste. that of the aggregate 5. Bonnel and F. Mitchell [22] who have achieved results similar to ours — concretes with siliceous gravel. 8. This role has been recognized and studied by various authors. beyond 200 to 250 °C we get a contrac- which passes through a maximum for 6= — tion due to dehydration followed by a fresh swelling from 400 "°C on [23].E. sandstone. If the limit humidity of the free water. 29. For a mortar ity. 28.2. is 0. the relation takes the form OCD. tion cycles. The consider- able importance of the role played by the aggregate in thermal expansion is apparent. It should be noted that the indications pro- -j=Qa-e)id-e. 24. Harper [18] and L. and that of the cement 14X10"^. then placed in a dry atmosphere. there is only a slight shrinkage. that of the pure paste. on that of the aggregates. it is curvilinear (OB). desiccation-bumidification show a residual shrink- age Os (fig. the gel by a crystalline phase.40. When the There is thus a relation between the movement of equilibrium is established. 30]. or a concrete cured in humid air Powers [21] finds an equally plausible explana. taking account of the shrinkage of a cement paste or a concrete that has moduli of elasticity of the pure paste cement had no previous desiccation. 664 . A cement paste. 26. Residual shrinkage after desiccation-humidifica- deviations that this approximation entails.6. 11. Replaced in water. Va the volume of aggregates and Vc that of cement. granite.70.) vided here concern only relatively low tempera- tures. 3) for an advanced age rium. The delay in investigators have concerned themselves with this this transfer between the gel and the capillary relationship [1. of which the abscissa The coefficient of expansion of concrete depends corresponds to OC. crushed slag. A similar concrete having quartz aggregate with a coeffi- cient of expansion of 11. 4) taking account of a hysteresis of which the effect may increase or decrease according _ aaVaEa-\- VaEa+V. Various of water between the two phases. or in water. water determines an additional dilation that For a compact neat cement paste the relation disappears after reestablishment of the equilib. will tion in saying that there is a capillary water and shrink and at the same time lose weight through a water fixed in the gel that is constituted by the evaporation of water. a change in temperature the water and shrinkage that does indeed appear destroys this equilibrium and requires a transfer to be a cause-and-efTect relationship. The experiments that we its hardening. Repeated cycles of and of the aggregate Ea.5X 10"^.3X10"^ gave for the Loss or water concrete 11X10"^. 25. C. of beginning desiccation increases. expansion coefficient of 8. have made show that the coefficient of expansion But we are now concerned with an initial obeys the law of mixtures. 6X 10~®. autoclaved under steam pressure have a coefficient the curve OB tending toward OA when the age of expansion independent of the degree of humid. appears to be linear (fig. they hydrated phase. taking account of the unpredictable Figure 4. For a concrete with limestone aggregate at 400 days of curing in air. The same author indicates that cements (OA). which would be due to the replacement of and a concrete. A third concrete with mixed Figure 3. 27. 10. we find the maximum for 0=0. the entropy of the capillary water swell and regain weight through imbibition.9X 10"^.7. including D. and that of the gel being different.8 X 10"''. where aa is the coefficient of expansion of the aggregate. the coefficient of expan- sion was 6. which corre- Shrinkage and Desiccation sponds approximately to the tests and shows in any case that the reasoning is plausible. 13. Up to a certain loss of water.

It might seem that we here (a) a free water having little or no influence on have a phenomenon of capillary evaporation in the dimensional variations. what we have is different that govern the latter. Beyond a certain degree of desiccation that a relatively slight drop [32]. an arbitrary figure. This "adiabatic" desiccation. time or of temperature. Let us evaporable water [31]. be the humidity. (Lucas [33] indicates practically can be obtained only by treatment at a limit of 85 percent for pure portland cement a temperature above 200 °C. Atmosphere in which 0 = 0. Figure 6 shows curves of forms of water in regard to the shrinkage variation in weight of concrete cured in different evaporation relation: humidities. Loss of water as a function of relative humidity. water linked to shrinkage. phenomenon shows that hydration has the effect There is in concrete a certain quantity of water of diminishing the internal humidity and leads to that can be evaporated under these conditions. in figure 6. only a negligible affinity for water. for complete saturation leads to condensations In the diagram of figure 5. 665 . which the latter would be zero at 100 percent (b) a physically held water that is primarily humidity and total at 0 percent. however. shows that the phenomenon is much more com- (c) a water held at a higher degree that can be plicated. Experience involved in these variations. The effect of " diminution of humidity by hydration can occur ^i/aj. It can probably be explained as free water. the internal humidity can descend to 95 percent. All the It is interesting to know that a sample of cement water will eventually evaporate. both having a mechanical origin [28]. relative humidities. 6 Figure 6. it is interesting to know the laws possible transitions.99 (we shall evaporated only under exceptional conditions of assume 9=0. is higher than that which corresponds to evapora- tion above a certain hygrometric degree. and is the value AB of figure 5. is comprised between tion. increases more rapidly in relation to loss of water. 9=99% \ \ \c u e= 75% 9 °. and not 9=1. 100% A ^ Time Relative humidity. It is difficult to define the law of this evapora- tity of mixing water is such that [32]. It begins with the wet sur- thus seems that the potential energy of hydration face where the speed of evaporation is constant. non.er only when the quantity of water chemically 0 D v. the loss of water in impossible to control experimentally. Since shrinkage is for the major part conditioned All this seems to mean that. after a year. the same consequences as a progressive desicca.ou % 1^= ct5% 0 e. The water cannot subsist in the liquid shrinkage after a slight initial swelling [1]. This state in the presence of an indefinite volume of air. Let us assume a spongy medium depends on the volume of the large voids in the with large pores and such that the solid part has paste. This effect is appreciable for pure an effect of adjustment compensated by an effect pastes that contain little free water. Weight change of concrete cured at different Figure 5. free water whose quantity weight constant. It tion in terms of time. This quantity.to the nature of the concrete and the degree of absorbed becomes greater than the quantity of humidification. The limit appears to be around 40 percent relative humidity for portland cement. here we terms of the degree of humidity of the curing have a discontinuity) may determine an evapora- medium can be expressed schematically by three tion or an increase in weight but never leaves the straight lines: AB. curve A of curing in water and the curve 6=99%~ should not modify the hardening when the quan.) For ordinary concrete such relation can be attained in which shrinkage an effect is scarcely measurable. on condition that it is less than able coating during curing nevertheless shows saturation.99. of fatigue. saturate it with water and place it in air. CD. whatever may protected against any evaporation by an imperme. 100%. leaving aside by evaporation. 6 [31]. BC. ^S. it seems that a new paste at 150 days.

Between certain limits. latter increases when the vapor stress diminishes. monotonic function when 6 increases. Equation (1) should then be written: dryness: 0<O.then it diminishes until the concentration at the dE surface reaches a value in equilibrium with air. having a positive value for ^=0 and a value. 0= 0. If we call the bound and the evaporated water: water Ec. from fied and that represents a coefficient of diffusivity. It is difficult to find a solution to this equation. so that in order to take both into and q{E) a function of E that remains to be speci. free water are not totally separate. a movement of diffusion of the internal water toward the surface begins. Ec=eEc when dixir) equilibrium is reached (fig. a surface is (j^^ = where n is the direction In a one-dimensional medium.99~1 the speed of evaporation is not (2) zero. appears to show that the transfer as the vapor phase is much slower than the transfer as the X a normal geometric coordinate on the surface liquid phase.E dt Finally. the movement of of the normal. at least for the volume once the surface has reached a sufficient most part. the preceding expression than the preceding for 9=1. An experimental verification of water at every point is defined by an equation this has been made on very fine sands [34]. not zero. the relative humidity of the surrounding medium. For 0=O.99. one observes that the speed of evaporation. less By making t. becomes: All this is very theoretical and in fact presents dE = only a phenomenological aspect. account we might write eq (1) in the form. it slowly gets into equilibrium with the E=Ei\l-e->^^^'] relative humidity 6. whether it be capillary. In a first approximation. 666 . in a third period. ^={E^-E){£j{A-Be)\{r) (4) If D is the designation given to the average d dimension Z>= ^'^^j"^^ surface > and K a coefficient of in which now diffusivity peculiar to the solid. they partly tion is proportional to the quantity of remaining overlap. may write We an expression still valid in a more general way if X more simply in the form X -^t where U IS we write being a decreasing '^=(j^ ^(^). Experience shows that the mean evaporation of a piece corresponds to an ex.. ~~ -QiE^-E)Al. dry barrier would oppose a transfer of humidity. both in its structure and in the mechanism of ment of the water outward occurs within the liquid evaporation which is itself heterogeneous in the phase and not the vapor phase.99 and d=OA for example. pression of the form: hence the function 7 must include a term in 6. adsorbed. Let us go back to eq (2) which indicates the By operating empirically. fi^) between 1 and 2. and would have as a consequence that a perfectly It is in fact this third phase that becomes all. or colloidal. 5). for concrete is a heterogeneous medium free water and appears to indicate that the move. the saturated state: For a three-dimensional medium: dE= dE d'E d'E d'E\ \Q{E^-E) + Q'E]L'E dt (1) dt dx'^ dy'^ dz' / with Q'«Q. However the with X(t)=- di evaporation of the bound water and that of the This expression shows that the speed of evapora. important for the solid pieces in which the water Indeed the quantity of water that passes through from surface wetting is in small quantity com- pared with the free water of the pores.(£j{A-Be)tKud\{r)='^ 7(0 can be replaced by a function ^(^^'^^ tfiat is or to say that the time necessary to obtain a deter. which dE (PE —=q{E)-^ where E is the quantity of water. £'=^Jl-e-><^']. we observe that -. but one that it is ' {E^-E)(^\{r) (3) interesting to consider for it will guide us in the dr realm of bound water. (5) mined evaporation is proportional to a certain function of the average dimension. it seems that eq (3) where 7 a function of time t elapsed since the is can be written in the more precise form: beginning of the operation and Eis the evaporated dE water. It is a Fourier's equation with variable coefficients.

may be determined by placing where Ei the evaporated water. . we shall write: tudes for each value of t. which indicates the quantity of hydrated cement also increases )667 . bound water diminishes when the humidity in- creases. we find that the cor- relation is approximately linear. E^ the total is the test piece in a saturated atmosphere and free water. --E. <i> The function 7?(r) remains to be defined. power of T. identical loss of water gives a greater shi'inkage at We must now go on to the relation MEc). it is different depending on the composition. and we slight. This result. and p a coefficient of shrinkage. An sponds to the evaporated free water Ei.If 1. (7) Tim« Figure 7.. in a first approximation. dt v{r)=. and figure 8. Even more generally. can be explained if we recall that the water variation. whereas for a concrete the variation of K is the volume or average length variation.000 days than at 7 days. at a given moment. This maintaining it long enough (fig. 34] that n is close to 2. we shall write: ^^. surprising we represent in juxtaposition figure 6. to let keep a linear form. tending toward a constant value with age. P is small compared with ^^_^^y/m with ' m close to 3. The diff'erence corre. m is above 12. varies with age a and increases with time t. We shall thus write: where K(t) is a decreasing function during harden- ing.99°/.9^35% Experiments show [31. we could write: " E=Ec^{d)[l-e-^^'^] «oti£ 6. Volume change of concrete cured at different relative humidities. For a pure paste. Determination of bound and free water The only distinction that can be made in relation evaporated. . look for a correlation between these two magni- Finally.. dE to eq (4) is that is annulled for 6= \ and £'=0. 7).{l-d)-E](^^ Kr). It is easy to represent it in an empirical manner as a W e= 75 V. or '?(^)~{ ^0 ' -^^t.) E. the curve that corresponds to the permanent Experiment shows that the coefficient of shrinkage maintenance in water. but it is possible. Ec. K itself is a function of age a since the concrete hardens and «*>.>E^ The quantity of bound water evaporated. A similar test piece such as the action of carbon dioxide and that of placed in water takes on greater weight and joins the internal stresses produced by slu'inkage. A=p(Ei-Er. but in the present state of knowledge it is J difficult to do better. The gain of relationship leaves out of account other influences weight corresponds to Ec. . — ^ ! 1 / »0>T where Ec represents the bound water and t-- so that the speed of evaporation is: f air dE '-=[E.gj50°/« becomes less permeable. We do not want to say here that a better approximation is impos- E'. which gives at first sight.{l-e)\l-e L J sible. we can write: (6) ^^Z. as Time for the variation of K. By analogy with the evaporation of free water but knowing that the evaporable part of the •Mn\»r. we find that Figure 8.

therefore.3 and 0. all the more so as the mean dimension D is greater. in other words on the two variables t and {t-\-a). Figure 9. We shall say more beyond which d is constant is modified. Drying shrinkage m relation to -previous curing. in water is important the shrinkage can diminish We may note that this gives for the differential after a certain time. the final shrinkage remains the same An important phenomenon to consider is the and proportional to (1 — 0)^. A ~' :[A. Time A consequence of the theory that has just been put forward is that the final shrinkage in atmos. the shrinkage curves. tion. 3). finally: This proposition appears more correct but leads to an equation difficult to integrate in an analytical ^=A„X0)[l-e manner. to [11]). that the physico-chemical swelling referred to above persists even in dry air. the re- versibility is not complete even after a prolonged bases. A permanent deformation 5 subsists. (10) It seems that they can be given the same form dt as that which we had recourse to for the evapora- and the speed of shrinkage is less. which appears to justify eq (10). It is. such that — varies between 0. and if we admit the linearity of the final representative. By contrast with what having to make calculations on hypothetical we observe for the variation in weight. so that. possible that in shrinkage in the form: solid pieces the development of internal stresses may lead to deformations irreversible plastic (8) whose resultant opposes contraction. the form of the shrinkage curve would be the same as that of the curve of evapora- tion of the bound water. This function must depend on the age a of placing in dry air and on the time t. Wehave already given a function of shrinkage depending on the time in which we find in the e If we consider the proportionality to the differ. ) with time and we can admit the hypothesis that phere 6 must be independent of the previous curing p is proportional to the volume of the hydrated up toan age a (this is confirmed by Dutron phase or. We have observed this element: phenomenon after several years of curing [7]. but young concretes diverge slightly (fig. "^'^^'J In any case. dt It immediately suggests a plastic phenomenon. where x is a function to be determined experi- but it is not certain that this similarity is in fact mentally. In order to avoid reversibility of shrinkage. So it is for aged con- cretes. age rate at advanced ages can partially mask the Finally. moreover.„(^)-A](gyx(T) We are entitled to wonder to what this is due. Figure 9 simply by taking water as a variable that p is which results from our tests shows the shape of proportional to E.. it must be possible to express the law of phenomenon. It may be supposed that the final shrinkage somewhat --QEn{t)EAl-e)[l~e-^^^^] (9) diminishes with the age of loading. Finally: A(0=A^(e)[l-e-**^'] (11) io\ ^ \ r" 3k- 10(A 10001 1 1 (in >o9 days ' dHr) \dry where x(t) = \ air di \elry \ \air \dry >v In the case of application of eq (8) x{t) = v{t) : (see eq (7)). Only the speed of shrinkage from the age the volume of gel formed. It will be further remarked that if the measured swelling with T=( t. 668 . We shall see later that it cannot be affirmed that this last relation is correct. it appears possible and especially prudent to write simply: immersion in water. power another function ^{t) which governs the ential element only. the quantity of water ab. if one prefers to reason from the gel. If such were the case the defor- shrinkage in terms of 6: A„(^) Ajv/(1 = ^) where — mations would be due to a shrinkage pressure is a reference value for 0=0. indicating.. We see that the low shrink- sorbed by hydration. especially as E^it) is inexactly known. this leads us to write: whole phenomenon.6.

rich mortar a marked shrinkage at 99 percent of This theoretical part which we conclude here- with and which rests on our personal investigations relative humidity. these do not appear to be affected by the shrinkage in an appreciable manner before the stresses that would enter into play. that swelling in water has been studied in particular b}. alternations of drying and moistening. In favor of this explanation we observe that the residual shrinkage disappears when concretes have undergone a prolonged hardening in water (180 days for example) before placing Relafiv« humidify. In addition to which we shall see later that carbonation. they carry out shrinkage measurements in atmos- This point of view appears to be confirmed b}' pheres free of carbon dioxide which may distort the fact that concretes cured for a very long time the results in a manner that is unexpected and (more than a year) in water show a shrinkage. On the other hand. but it seems that it may be at least partially answered by the introduction during shrinkage of additional links due to the progress of hydration in the condensed medium.95 appears to be approximately moreover. evaporation and shrinkage are approxi- surface. It is found that the shi-inkage in dr}' air cements such as slag give in neat paste and in of prisms of pure paste having section ratios of 669 . (1 — 6). We believe that this slu-inkage is due to the inner desiccation caused b}" hydration [35] must be compared with the studies of other which produces a shrinkage without variation in authors. which seems indeed to show a of mortar placed an atmosphere of carbon in cessation of hydration in these very low humidities. the coming together of the crystallites that oppose swelling. An explanation must therefore probably be sought elsewhere. It seems. We see that Finally it may be said that the assumption of this influence is far from negligible and must be linearity between certain limits of humidity appreciable even for atmospheric air. tends toward 0 for 0<25 percent. and [11]. that is to say to the creation of new links through air at 6= 99V. This increase is zero for saturated atmosphere. and Carlson Finally it is necessary to explain why certain [42]. mately linear functions of 6. It It is observed. AD represents the swelling in It appears moreover that carbonation diminishes water in relation curing at 99 percent rh. A natural factor that plays an important role in shrinkage is atmospheric carbon dioxide. which generates a structure insensitive to moisture. Glanville [41]. increases the residual Figure 10. Now. d=OA and 0 = 0. that is really proportional to our laboratory.resulting from inner stresses but which woukl modify the rupture conditions. The theoretical shrinkage curve properties of the hydrated phase of the cement. 9 in dry air [1]. dioxide at the same relative humidity show an HaUer [36] clearly brings out the linear relation increase of shrinkage and an increase in weight. The curves enon should modify the aspect of the theoretical obtained in terms of 6 are in general of the form curve of shrinkage and give it different shapes given by figure 10 in relation to the saturated according to the experimental method used. difficult to control. below 40 percent. the curves assume a somewhat irregular appearance. it is not certain that this It is therefore not surprising that such a phenom- is always the case for shrinkage. Ross [40]. Those of Lucas [33] appear to indicate weight of the parts furthest removed from the that between relative humidities of 40 and 99 percent. shrinkage considerably. at the moment of the return to the wet medium. Tliis is such swelling being always possible in the presence the proof of the presence of a supplementary solid of free water so long as any subsists in a part of structure that does not have the hydroconstrictive the volume. dimension of the sample and thereupon decreases. than 40 percent. moreover. BC is thus modified by the swelling s which is We therefore recommend to experimenters that greater as the value of 6 is higher. that this effect increases with the acceptable for evaporation. state in water. Prisms [38]. tliis is now the procedure in when placed in air. The straight line i of figure 10 which represents the initial length passes below D.Meyei-s [37]. whereas it is more or less observed in Dutron's tests 0=50 percent where it reaches a maximum. Haller [36]. the swelling then becomes negligible in The volumetric coefficient has been studied by relation to the shrinkage that subsequently occurs. Combination of swelling and shrinkage. The question has not yet been answered. to the reversible part of the slirinkage and subse- But nothing tells us that for 0<C99 percent there is quently gives it a greater stabilit}' during the not a swelling being deducted from the shrinkage. between the evaporated water and the humidity. but this relation does not appear precisely for the approximately double the normal slmnkage for shrinkage. becomes more pronounced after desiccation at less Leber and Blakey and Verbeck [39].

In each category. shrinkages in water of different test pieces of the same con- increase in the following order ordinary portland.35 percent. Practically.355 cm^/g to 2. paste at 1. According to Centre d'Etudes des Liants Hydrauliques.1 to 2 is in the ratio of 1. for aluminous cement. apart free surface in a plate made impermeable at all the from exceptional desiccations. Dutron [11] indicates for portional to the ratio other things being 0^ . medium are concerned. free water). lower than 50 percent rh. 670 . According to Cocagne [35] an ordinary portland gives 0.117 to 0. Tests on samples 4 x 4 x 16 cm at 50 percent rh and temperature 20 °C. we find that the values aluminous cement. On cement on shrinkage by bringing out the increase the other hand. Paris.5 for section ratios of 1 to 8.280 we wrote above that the exponent u was between cm^/g for the same portland cement. the shrinkage at take u=2 as a first approximation. and portland with high align themselves along a single curve in relation early strength. the shrinkage increases with the propor- reaches 8 cm in depth at the end of a month tion of the latter. order of magnitude can and its The part played by the composition of the be roughly fixed as proportional to the square of concrete is very important in creating shrinkage. Roughly speaking. Karapetian [79] indicates The same author shows that in portland and slag that for a block dried at 50 percent rh desiccation mixtures.169 percent in 90 days when the appears to tend toward unity [80]. Figure 11 gives a number of and 60 cm at the end of 10 yr (the term "dried" average curves relating to cements manufactured probably means that the concrete has lost its in France. 0. low humidity. slag. Ross's exponent u close to 1. : crete tested by Ross [35]. Graf [43] seems to have been the first to concern Our tests confirm this point of view insofar as himself with the influence of the grain size of the atmospheres of high humidity are concerned. Shrinkage of some cements in pure paste. the distance to the surface of evaporation.5 while Karapetian's results tests show how the shrinkage propagates from a [79] give a value close to 2. it seems that we may other surfaces.22 percent shrinkage in neat necessary to produce a given shrinkage is pro.25 percent. a distance of 4 in from the surface is still only one-third of the surface shrinkage. This delay Composition is considerable. This means that the time variable. ordinary portland a figure of the same order of equal as far as the composition and the curing magnitude.000 days. Lafuma [45] likewise 1 000 FiGUEE 11. the Haller [36] has shown that shrinkage increases exponent u of formula (3) and the following from 0. This is why specific surface increases from 1. finds a linear relation between the specific surface We may note that Weil [81] gives shrinkage and shrinkage at 90 days. If First of all the nature of the cement has a con- we compare the shrinkages as well as the losses siderable influence. Jaspers [44] 1 and 2.3 to 2.3 at 1 yr and tends values in terms of the dimension which bring out toward 2. a portland having high early strength 0. for shrinkages carried out at a of the latter when the specific surface increases. they are very to the variable t . At 200 days.

moreover. on the other hand. 11.045 percent for -=0. the shrinkage at sideration. Many authors have water.47. 671 . certain authors [11] mean that beyond a certain quantity of mixing indicate a marked increase. who has established this formula theoretically. Numerous investigators among those Portland concrete with a cement content of 350 already quoted have taken this factor into con- kg/m^ and Seine aggregates.17 statistical point of view have shown that the percent while the losses in weight are less for the w — variation of possible for the range of cement highest specific surfaces.5 percent of added CaCl2 and 165 percent for maximum spread of 10 percent. its increase amounting to 110 percent for 2. The cement content.].55^ and aggregates of different mineralogical isto say that the variations are of the order of magnitude of the experimental errors while the natures. 10 percent [50]. (Pickett This increase of approximately 50 percent persists [49]. are approxi- surface. the different authors are interpreted. /=volume of fines. But the most specially studied influence was that of calcium where e=volume of water.52.500 cm^/g obtained with the same ing several thousand concretes studied from the clinker the shrinkage increases from 0. . have indicated. is less at normal consistencies. .70 when the tests of CaCl2 equal to 0. In the case of the aggregates and that of the cement paste.05 to 0. where the — ratio .88—0. sodium nitrate at 6 ^t=yie+c+v+f)" percent doubles the shrinkage. Another semiempiric expression proposed by an addition of 3 percent of CaCl2. indicate a rather important increase in ^^^^ where Hca is a coefficient that represents the rela- shrinkage " with the ratio J =—c /).4 of concrete: percent in the mixing water increases the shrinkage at 1 month by 40 percent. a concrete. chloride habitually used as an antifreeze and = volume of voids. c logical nature of the filler. shrinkage an influence on shrinkage amounting to less than increases with the volume of the hydrated phase. apart from a chemical reaction of expansion. The mineralogical nature of the it is considered after a day of hardening.Another series of tests cover- from 2.brings out a variation of the same kind on mortar : bilities of placing. hardening accelerator [47]. the first shrinkage being on the contrary influenced by aggregate has a marked effect on the amount of variation of this ratio as was said at the beginning of the shrinkage as may be expected from the above this paper and as Pickett [49] confirms.53 .5 percent of the weight of cement. We find 5 months from the introduction of a quantity of that n is between 1. c=volume of cement. Figure 12 shows the results of tests w on a concrete with a cement content of 350 1. Sodium carbonate at 0. For a formula. shrinkage with the cement/powder ratio. cementV But this tion between the coefficient of compressibility of increase occurs with neat pastes.24 with a creases. including Haller [36] and Cocagne --Ac 'c+[l-{c+e+v+J)]nc [48]. The latter fact confirms what we had already established.7.57.20 and 1. and 0. etc. Cer- tain authors. an increase in The same is obviously not true for neat paste. Which c c 0. w — expressed in the preceding formula by the varia- limits of correspond to the extreme possi- tion in the modulus of compressibility. has a The introduction of soluble salts into the mixing very marked influence. the incorporated water becomes free with mixtures of cement and basalt powder [1] water without appreciable influence on shrinkage. without great variation to 2 percent of CaCl2. Dutron indicates for the shrinkage appreciable increase. studied it [48. w — seems to addition of mineral powders to the cement The mately proportional to this result c may have variable effects. on the contrary. This variation is quite adequately 1 . Our tests carried out water.05 percent for -=0.048 kg/m^ with an identical quantity of water = for -=0. us [31] has the form: Let us now examine the influence of the quantity of mixing water in mortars and concretes. and this volume does increase with the specific As the losses in weight. then from this quantity the shrinkage suddenly in- finds ri. that with an contents extending from 250 to 500 kg/m^ had ideBtical quantity of evaporated water. The experiments that volume of the aggregate is written o = [l — (e-\-c-\-v-\-J)] and if a = 0 we obtain the shrink- we have made [7] have shown a marked increase in age of the neat paste A. There w is accordingly an influence exerted by the minera.000 to 5. has the effect of increasing the shrinkage [46]. If the ?. It seems that a usable empirical expression is The salts dissolved in the city water contribute an here possible.000 days was 0. 0. and this expression be- the shrinkage of an ordinary portland measured at comes transformed into A6=Ac(l— a)". 36. our tests have shown a very slight For a siliceous aggregate we find that Mca= influence both on shrinkage and on swelling when w — 0.) Our measurements on siliceous Seine aggregates (silex) gave 7i=1.= 1..

or erally expressed as a decimal logarithm of the height of water in cm corresponding to the suction F'. The force F in this case of a reduced pressure exerted on the latter. This idea appears to have been put on the quantity of water e already contained in forward initially by Freyssinet [51]. If the quantity of water diminishes. for if 6 diminishes.40 in order that the shrinkage may be stabilized approximately within 0=0. suction measurements. the device using suction on the liquid phase is no longer usable. the last layers being too tightly bound to the solid to be extracted. Swelling and shrinkage of concrete with ble water. Calculation shows without dif- ficulty that if we write e=ei-{-ej{l — ff). and for a temperature of 20 °C where T is the absolute temperature. that is to say beyond atmospheric pressure. The mass of water thus conserved may be suffi- V ciently small to be confused with the hydration water. and has a somewhat artificial aspect. How is shrinkage to be explained? The answer The other hypothesis that assumes that the to this question appears not an easy one. water is deposited on the solid crystallites having One of the first theories is that of shrinkage great surface development in adsorbed layers is stress (or of swelling stress) in which the liquid likewise plausible. at equilibrium.5+logio (2-logio e). pended in the void above a solution of sulfuric Kelvin's equation. it must itself be placed under the same stress as evapora- Figure 12. We know that a porous In order to determine F it is assumed that the body placed in the presence of free water absorbs stresses that can be developed are capillary this water by exerting a tension on it that depends stresses. and l>7^=6. develops forces of attrac- initial state. preventing the penetration of the water by means pletely abandoned [19]. ^F=logio H. by reaction. moreover. This hypothesis remains to be proved different aggregates. which is contrary to experience. Beyond pF='i. The evaporation of the outer phase is placed under stress in relation to its layers. The suction is gen- of the menisci that limit the liquid phase. If e( is a kind of nonevaporable water. V the total volume of the solid phase. such a mechanism is not in contradiction Theories with the hypothesis that presents the hydrated phase in the form of a gel.5. 612 . gives acid of known density which. Neverthe- less. and finally the shrinkage would be: ^=ATe log d. The value of the —h-= — log e suction is then given by Kelvin's formula. It has since this body [52]. the least bound. hence of hydration water or water fixed in such a way that it cannot escape for very low humidities. We may then attempt to get an idea of where x is a coefficient of compressibility of the the magnitude of these forces of attraction by solid. which makes it possible to set aside the artifice of capillarity. and A must pass through a maximum for the decreasing d. F{e). Van der Waals forces. Let F be this stress. — must be greater than 0. the radii do We then use a drier in which the sample is sus- likewise and F increases. e the liquid tion on the inner water layers. and under these conditions the calculated shrinkage varies but slightly from linearity up to 0=0.25. e likewise diminishes. F=AT log e. which place the adjacent The shrinkage should be: solid in compression. But such a relation can be considered valid only between certain limits. This force F can be measured by been adopted by various authors and almost com. We should be proportional to the surface tension a then find that F is a function of the water con- of the liquid that fills the capillaries and inversely tained in the body at the moment when the proportional to the radii of curvature and r2 measurement is made. gives a relative humidity 6. by liberation of the volume.

4=40 kg/cm^ The elastic deforma.We then obtain for the cement a curve of the form given by figure 13. between 1. In concretes hydrated under heat. whereas a treatment at 185° for 8 hr probably less sensitive to the hydroconstrictive under a pressure of 10 atm of saturated steam movement. by centrifuging. cml Let us further assume that the water that remains at this humidity represents 50 percent The hydrostatic This shrinkage is then of the order of magnitude of the volume of the cement. Volume Variations of Concrete Under Load Under a progressive loading applied at a given T is shown as the ordinates and A as the abscissas age. concrete undergoes a deformation. In his work on the structure of hydrated ^=80X10-^=0. and we get a curve of deformation. and the suction forces cannot alone varying with the humidity of the medium. One can also attempt to extract the water and measure the necessary force by proceeding according to one of the two above methods and.1 and 0. whether of tension or compression. suction forces alone. 13). a straight line geneous stress. Powers [21] appears to be of a completely changes the phenomenon. But highly sensitive to humidity whereas the treat- whether the freedom is allowed or the mortar is ment by steam under pressure gives a crystallized kept in a closed mold. of layers of water the thick- measured is more than 0. or one with. It is to be regretted that very few attempts have been made to connect suction with the variation in the volume of the cement. call this deformation A measured in a part uni. He bases this expla- measured upon drying after cooling becomes very nation on the fact that cements treated in this slight. this gives i^=100 kg/ Figure 13. further.0 and 2. This is no longer a simple phenomenon of adsorp- tion forces is of the order of one-tenth of the total tion but of layers of zeolitic water of dimensions deformation. and this effect may be favorable to the creation of a residual shrinkage. 14).008%. Bernal indicates [54] the presence of tobermorite in fibrous form and. Warer ContanI. • cent of evaporated water. Now. We first different opinion and finds an explanation by observe in neat paste a very considerable swelling saying that for setting at normal temperature during the treatment (between 0. Let us of a system of rectangular coordinates (fig. It is nevertheless possible to examine whether the order of magnitude of the suction forces is suffi- cient to explain the shrinkage deformation. shrinkage deformation corresponding to the suc. A treatment at 80 °C crystallizes mainly in the form of slabs and no for 3 hr does not markedly modify the resulting longer in the form of fibers.0 percent of the shrinkage manner have a coefficient of thermal expansion measured on an untreated control concrete [53]. of that indicated above under the influence of the pressure on the solid part is then at the maximum: ^=100X0. we find that the shrinkage and not a colloidal phase. Forrn of curves of wetting and drying of cement. The pF{e) relation found then gives a curve that is above the previous one (fig. we should longer sensitive to variations in humidity. is Bernal's opinion that we must look here for An interesting remark to be made derives from the origin of the shrinkage properties of cement. cements. OA is tangential to. OE over a certain length (an inflection at the 673 . tobermorite cretes treated by heat. It is find well then to look for the origin of this sensitivity. however. the observations of shrinkage on cements and con. The part of ness of which varies from 14 to 9 A during drying. is that drying de- velops a higher 'pF than moistening. It is then possible that the origin of the shrinkage is to be found in large part in the crystalline structure and that in the case of treat- ment by autoclave the crystals formed are no With £"=200.2. independent of humidity. in the direction The order of magnitude of the deformations normal to the fibers.1 percent. It explain the shrinkage. What we observe. the hydrated phase appears in a colloidal form cent) if the test piece remains free to dilate. we know that the silicate crystals hydrated at high temperature tion that corresponds to the saturation of the paste is of the order of have a different structure and a greater volume stability.6 per.000 kg/cm^ and a=0. the quantity of water being equal. such slabs being shrinkage. Let us assume that we have reached pF=5 for 50 per. This curve formly stressed and T the corresponding homo.

DoFormaHon fA) 674 . crR=—-^ Ar This property is verified when the maximum load is less than 6/10 of the breaking load. is the same as the one found by means internal microfissuration). OcFormaHon (A) Figure 14. Stress-deformation relations for concrete. Because of this it is called us remove the load. The magnitude Ag A. in 5 a total deformation St. there remains a deformation and. — elastic deformation Ab=At—Ar. line is the same as that of the original tangent OE. curves can thus be established (fig. while A2' A2 is the residual creep. cfe=——- Ae Figure 16. of the measurement of the speed of propagation After having applied the load as far as A. the residual deformation parts to operate through successive and increasing diminishes by a slight amount and tends toward loadings with intervals of unloading. After complete removal an elastic and reversible part and a plastic part. 16): the the instantaneous reversible deformation is called curve of total deformationAy. From the stress is partially decreased and load is again which we obtain three values to be considered: applied. The modulus of elasticity thus defined. 15). the normal to that of the applied total stress. We can recognize — under load. the reversible total deformation Aj A". close to a straight line but very slightly incurved As the total deformation of a concrete includes in the opposite direction. while the permanent residual deformation or Finally. the deformation 5 is measured in the direction tained under load a sufficiently long time. Stress-deformation relation on cyclic loading. it is necessary in order to recognize the relative after a time of rest. —the plastic Poisson's ratio. — or residual deformation A^ and the curve of and the reversible deferred deformation A[ A['. or fundamental —the elastic Poisson's ratio. of the load. one obtains a series of loops that finally stabilize in the form of a very flattened —the total Poisson's ratio. if for a compressive or tensile stress plastic deformation is A[' If the solid is main- . The unloading curve is "modulus of dynamic elasticity". CT=—-r- At loop HH' practically identical with a straight The inclination of this straight line (fig. the deformation increases from A to A' and reaches latter has a sign which is the contrary of the the value As. a residual deformation If instead of completely unloading the test piece 8r and a reversible elastic deformation 8e. is the creep— longitudinal deformation A. let of shock waves [55]. this is the sign of a preexisting modulus. plastic. and elastic deformation of concrete. FiGUBE 15. If the load removal is sufficiently rapid. Three A'l'. Curves of total. and the process is repeated a certain number of times. beginning of the compression curve can sometimes be observed. the curve of plastic Aj— Ai.

that of quartz 800. the ratio of viscosity depending on A. The experimental work done on this The formation of gel or of crystals cannot occur subject is rather limited. for ex- sively examine the elastic deformation of concrete. that the curing during the early E stage is predominant. the modulus of the mortar of volume B. of dynamic methods (speed of sound) it brings out K should be higher. Vivian [58] hardened concrete varies between 200. but remains definitely it is possible to examine its relative variation in lower.000 kg/cm^.5.97 for J=0.33. It varies with the actual qualit}^ to §= hi 0. but this lack of linearity 99 450. Water 450. This E ' A+M+0 behavior appears to show the presence of a viscous where E^ is the modulus of the aggregate of volume elasticity. The modulus E^ conforms approximately a to the law of mixtures for a relatively compact not greatly differ. for the square root of the compressive strength. A curing neglected.000 kg/cm.000 kg/cm^. cylinder).000 kg/cm^. is the load. Le Camus [56] must be.000 and gives for the modulus Ea a formula that leads 600. being higher for concretes with sUiceous aggregates than for concretes with lime- for = 0. If we define the modulus of elastic T is to be noted that a concrete cured in dry It deformation or secant modulus Ea—-7. In addition. It to one-tenth of the rupture load. ^=0. 0 is the volume of the voids. and 0.75 for ^=0.10 (it seems that for young concretes. Relative humidity Modulus of of curing elasticity Elastic Deformations medium Experience shows that Ab(T') is not a straight % kg/cm^ 35 340. These preliminaries having been effected in it is greatly influenced by the humidity of the order to narrow the definitions. and its slow plastic per m^. to modify It is recognized that E remains constant up substantially the subsequent increase in the modu- lus that proceeds as for a curing in water. on the other hand. diminishes in an almost linear manner till it The modulus of dynamic elasticity of a normal reaches one-half the breaking load.for . permitting the creation from has found that -^=0. H 0. (fig. increases with age during hardening in accordance that of a river siliceous aggregate (Seine aggregate) with a law that is approximately proportional to 630. in com- The Federal Laboratory of Zurich gives the pactness. ample.92 £1 of the aggregates.30. The order of magnitude of the modulus of a For a given concrete.25).95 " for ^=0. with sUiceous aggregates. for the authors have in voids from which water is absent.90 for A R the beginning of a more compact structure that subsequently becomes consolidated. This sensitivity. 000 crete ages. is at least as great as that observed in tests E ^^^_T/R for mechanical strength. in most cases. to separate the reversible in water for a week followed by exposure to dry elastic deformation from the total deformation. air does not appear. it . modulus does increase. Our tests have shown. close to 0. It is therefore probable that the mode of terms of the stress T and its relation to the hardening of a concrete cured in dry air does not classical instantaneous breaking load R (on lead to the same structure of the hydrated phase. 000 line and that concrete has an elasticity that is 50 365. 000 not perfectly linear.varies between 55. the following deformation or creep in terms of the variables moduli at 200 days: already considered for the study of shrinkage.000 kg/cm2. curing medium. We "75 see that all these results do stone. 000 is not very great and diminishes when the con. 000 75 385. it confers l-T/R upon the measurement the advantage of appl3nng to a volumetric mean. and limestone. the modulus of elasticity hardened mortar is approximately 200. 50 percent of the breaking load increases to join the modulus of dynamic elasticity [59]. 0. on a concrete containing 350 kg of cement its rapid plastic deformation. and in physical or chemical decomposi- formula: E^^ 1 tion. we shall succes. the modulus Ea measured at AE^±MEr.20. 16) air and subsequently exposed to humidity does not recover the modulus of elasticity of the same as the inclination of the straight line OA that joins concrete cured uninterruptedly in moist air. the curve of elastic deformation approach. ing a straight line for increasing speeds of stress. to which is added the variable stress T.675 630446 O —62 8 . and 0.94 for ^=0. Santarella a localized modification such as a crack capable of [57] indicates that the modulus of elasticity forming a transmission barrier. The modulus of elasticity of concrete is in- fluenced by the variations in composition.2. and in the case of the use with i^=0. When concrete is loaded at concrete increasing speed. then. El. bearing in mind the scat- ter. the the origin to point A considered on the curve.10.

The deformations we are here considering are those that parallel to the axis of the compressive stress. close to 0. In addition. the microphone records multiple limestones and 700. 50 deformation A/j which corresponds to the rupture Tr in the sense usually given to this term. By using the dynamic method. For a hardened concrete. found that Poisson's ratio varies between 0. that is to say at the maximum load that the solid can bear.30 0.very soft limestones to 250. then Rapid Plastic Deformations increases to reach 0.30.10 0. irreversible deformations begin to occur continuity of the structure. Poisson's ratio remains constant up to T^. it is further possible to set up an apparatus for measuring the speed of the shocks along a diameter (normal to the direction of the load applied). This being said. At first we test piece reaches the breaking point (fig. 18). This transverse swelling appears to be We shall use the term rapid plastic deformations due to the establishment and the dilation of to distinguish such deformations a priori from microfissures. Changes in mechanical and sound Figure 17. indications of internal ruptures that disturb the yond this. The transverse swelling is likewise characteristic. the transverse deformations are wholly reversible and drops beyond this. What we then laboratory. In any case. which are crackings. let sion below of the breaking load.18 0. the creep that we In general. 10 0. varies little between 1 min and 1 hr in the light of measure is no longer a simple plastic deformation since there is added to it the effect of a disorgani- the deformation that subsequently appears in the zation. noises are heard. increase in frequency and in intensity when the tention in only a few studies [56. The test piece Load OaFormahon ('^p) Figure 18. a speaking is preceded by a "pre-rupture" which time which corresponds to the practical duration appears rather early. These ruptures have which are due in part to microfissuring. 60]. this cracking and enlarges the cracks which.000 kg/cm^ for average being loaded. find that for a stress below half the breaking load For a load below Ti the speed of sound is constant. They The Poisson's ratio of concrete has received at. 0. it seems that no measurable instan- shall later consider takes these "instantaneous" taneous plastic deformation is found in compres- deformations into account.000 kg/cm^ for very hard lime. it is higher as the concrete is younger or less millimeters and their thickness a few microns.5 and beyond shortly before rupture. passes through a maximum Tr corresponding to a r 0. 05 0. course of time [62]. The crackings are thus to within the precision of the measurements. feeble sounds. and beyond a load very different stones such as marbles. lead to rupture. characteristics on loading. few figures obtained by Le Camus: This curve is concave toward the abscissa. 2R Ajf In the course of a test of this kind. The plastic deformation. we see that the rupture properly are due to a loading of duration less than 1 hr. 61]. strong. the privileged direction of which is creep produced over a prolonged period. Their length is a few 0. moreover.10 and Rusch. that is to say the sounds that are produced in the concrete under the action of loading. Here are a us examine the curve that represents deforimitions of this kind in the coordinates T and Ap (fig. between 50 and 75 percent of of a loading and unloading experiment in a the standard breaking load.02 is The increase in the load augments the density of currently found. 676 .30 they join.20 0. 17). Ah 0 0. Be. and ourselves.20 ±0. [60. Curve for rapid plastic deformation. it is possible to make a number of supplementary measurements on a fairly slender prism. as we nevertheless obtain a higher ratio. Freudenthal [82]. 22 0. In addition to the longitudinal and transverse deformations it is possible to place at mid-height a microphone connected to an amplifier and a device capable of recording the sound characteristics. It is been observed by Berg [63]. To sunnnarize.

it must be confessed that we do not the upper straight line as far as A. If we record both types of reactions beyond 75 percent of the tensile strength exceeds Fji=CC' in figure 20 we see that the maximum a quarter of the elastic deformation. by means of strain gages. amount. 10 percent at 30 sec. As sudden the reaction in M' is greater than in M. then less. equaling zero Todd [68] on the contrary indicates a few experi. zero. 30 0. On neat paste. If there were a sudden break at a given moment. but in In tension. For concrete supplied with a steel guide in the proximity of a reinforcement. This time the concrete was not and Fisher [67] indicate a total linear deformation bound to the tube except at its extremities. concrete in tension is improved by the presence In order to check this. it is broken.^i Blakey [65] proposes an expression T=E /^e'^'^m where is the deformation at the F and ~=Ea^a maximum of the curve. 55 As outside of the steel tube. it is replaced by a micro. Under these conditions the tube tests showed that up to 90 percent of the breaking stretched. These reaction of the free concrete. we reaction of the concrete does not become equal to know practically nothing about it. composed of a tube of steel within which concrete Tests of the plasticity of concrete have been was poured [31. All this shows that the apparent plasticity of fissuring that may look like plastic deformation. but the rupture. the plastic deformation 677 . Glanville [41] has given deformation curves of concrete under loads maintained for variable periods from 5 to 60 seconds on a concrete con- taining 350 kg of Portland cement per m^. an inside thread ensuring made by bending. formation and negligible up to 10 percent' of Yet not only is there no sudden break. 855 0. Our tests and those of Grassam was indicated. Smith and Young [66] system) (the concrete no longer A _T reacts. occurs near A and equals zero shortly bar. fig. the stress must be F=A{Ea'^a-{- hr. ^ =5 percent at 15 sec. however. Deformation under heavy loads has been specially studied by Shank [64]. 62]. that is to say its involve prisms containing a longitudinal steel strength. Elongation of concrete in a threaded steel tube. It occurs at 7 hr for -™^=0. At about 30 percent of the breaking load. etc. and it is recognized that the rupture thereafter as soon as cracking has crossed the or cracking elongation of a concrete is increased section. loaded at 28 days. If fla is the section of the T tube having a modulus of elasticity Eg and fit the For 7p^=92 percent the rupture occurs before 1 -t R section of the concrete having a modulus of T elasticity Et. we should find a drop CC. The CC reactions of the concrete were not exceeding 8 percent of the elastic deformation. This result may suggest plasticity. tion in terms of the age. We find that the propose the formula: 7f-=T- R Am-t experimental form is OAB. The tensile stress the strained surface of joists poured without was applied to the tube by means of threaded reinforcement. strain gauges being placed on adherence between the two. up to 85 percent of the breaking load in tension. along with the concrete. then bends. 0.905. shortly thereafter. Theauthorpro. The plastic deformation increases with time. but this time also without ments in which the plastic elongation obtained suddenness. in the central area. The elongation A was measured on the At 95 percent of this load. As for the plastic Poisson's ratio. 45 0. the concrete should mstantly be equal to zero. we have used a device of a reinforcement. We may mention that F (a perfectly elastic of inclinations —=Ea^a-^Et.. the curing medium. and the reaction of We know only that it is less than the elastic de.and 15 percent at 60 sec. then the same up to A. know a great deal about rapid plastic deforma. 88 0. 19). at 3 months. It is identical with But finally. 905 Tr Elonsorien (a) As Figure 19. His paper appears to lend itself to the interpretation provided by the following table for measurements at 1 hr: T 0. our sleeves. cracking cannot occur. The curve that represents the relation ft Eb^h)- poses representing the rapid plastic deformation in FIa) must be included between two straight lines the form Ar=cT^'"-. rapid plastic deformation tests are order to decide the matter another experiment few in number. it was separated from it by a sheath of foam with the plastic deformation preceding rupture rubber. to the same load no measurable residual deformation occurs.

To build a sealed room where the atmosphere is fixed at 50 percent humidity and into which one will enter religiously every 20 yr does not appear to me to constitute an expense incompatible with the financial means Figure 20.9^ it amounts to 10 per. 678 . the one due to creep. it continues to become deformed. country like the United States of America. this slow plastic deformation that we call creep. after 20 yr 1. it becomes after 2 yr 1.20. Behavior of concrete unrestrained in a steel used in other fields for scientific research in a tube compared withthat in a threaded tube. Without concluding as to whether what close to rupture (^jr=0. let us remove a load from a was 2 to 5 percent of the elastic deformation at test piece after a time t. 21). We say that it appears to approach to which the load applied has been greater. Creep — Slow Plastic Deformations which is a consequence of the applied load. It is the shrinkage test piece are both placed in water. The test piece becomes 1 hr and 6 to 11 percent at 4 hr. and there is no chance of my finding out. Such measurements cludes in part the shrinkage that the concrete should be undertaken on tubes. Davis [69] some 30 yr ago have not yet brought out this stabilization.26. These are a(«»l guid* the longest tests known. The tests begun by R. The permanent deformation. deformed in the opposite sense instantaneously. The few experi. but it is possible binder may constitute.14. that errors due to the friction of the supports a permanent deformation remains. would have undergone during the same period.36. no one knows what happens to the concrete at the end of 100 yr. ments made in torsion [56] seem to show that then it continues to be deformed for a certain time plastic deformation appears earlier than in tension. finally. we have is a deferred elasticity or a reaction of the elastic aggregates on the plastic phase that the cent of the elastic deformations. after 10 yr 1. Creep in 1 yr being taken as a unit. (fig. mains under the influence of this load over a long If at the moment f'>^ the creep test piece and period. Tii»i« Figure 21. we shall say that. But this in- may account for this figure. re. a limit without being really able to affirm this. The test piece tliat has been subjected It appears to approach a limit to the extent to to a load swells more than tlie one that was cured which the load is maintained constant for a very without load and more markedly so to the extent long period. thus appears to be the difference between the perma- When concrete.33 and after 30 yr 1. nent deformation and the shrinkage. after having been loaded. they swell. after 5 yr 1. Having said this. Creep and shrinkage as affected by conditions of loading and storage. I therefore suggest that Davis' measurements be carefully continued by his successors in order that our scientific heirs may some day be able to settle this important point.

F. We shall mention further. this time beiig t=n. creep but which does not seem able to maintain pirical form to the curves obtained. however. R. There is found the concept although for very young concretes.t) -\-K2 where a is a simple equation of viscosity. is lessthan half the breaking load. It appears not impossible to introduce such an influence into the function F. and we shall come back to the T/lt simple equation: gives the function A{t) = T ^^^-p^in which m and n are two constants. have not accepted the assump- crease in deformation A toward a limit value which tion of limited creep. and p is 1905 [70].-A)K find that $ is the sum of a constant and of a function of time. for if one takes into account the form of the deformation curve there Lee [72] gives a differential expression: is nothing to prevent looking for its derivatives. both simply 1907.^ ^. a and C are constants characteristic of the solid. a. a purely loga- rithmic form. K. In any indicates the relation A = Kf". r. dash. Freyssinet in two different creeps Ai-|-A2=A. m. however. We shall not endorse this proposal case the magnitude ^ has a decreasing value which if need be can represent the beginning of with t. E. the time at the end of which A reaches believesit is necessary to add a term of consoli- half its final value A^. Many authors have tried to give an em. All load and other things being equal. What we have is the superposition of precursors. are the times of relaxation of three t2 This leads us to write: visco-elastoplastic systems considered. It that: to derive a simple general equation from is difficult it. There is therefore a term inde. the usual area and of use of concrete. they appear. authors. under constant which also leads to an exponential form of A. creep properly speaking being the sum of two and viscous deformations. Such are the rough outlines of the phenomenon A=a{l-e-")-\-^e''"'(l-e-"") that we shall examine in detail. and Straub [77] in particular can be reached after several dozen years. For concretes loaded beyond 1 week. among the constants. Torroja and Paez the age of loading from the beginning of setting. and dry friction in parallel and in series. f/As A NT^ 1 /A A / N Creep in Compression Thomas proposes an equation -77= dA A [76] Most creep curves terms of time have an in dt (B-\-ty asymptotic shape. and T is the applied load. so that: bibliography will be found in'^the paper presented before the Munich symposium in 1958 [62. W. The final value of the creep dA . dF is thus A = mT where m is called ratio of creep and T the applied stress. jS. McMillan in 1915. [74] use a more complete model with springs.Freudenthal and Roll [83] also use a rheologicai model which leads to an expression: dA dt = (Am— A) (^^^+^2^ when theapphed load A=C [l-e"'^]+ra [2-^'~'^-e"^]. to show an in. h. A fairly complete viscous. r _(k. If we set aside the instantaneous plastic deformation i>l hr.] dA dAl . it appears pots. Lorman [71] itself indefinitely. We may note that this expression can be written dation of form -1= which would tend toward a ill the differential form: finite limit after a few days. iog?±l+K2()"| pendent of the load which represents shrinkage. we ^=(A. or ^=Ki{a. Finally. McHenry [75] gives a relation composed of two exponentials: 679 . The first known study of creep is the one published by Woolson in where a the age of loading. it is convenient and probably correct to A=Ajl-e-^<'>] represent creep by an equation stating that the whence: speed of deformation at a given moment is pro- portional to the deformation that remains to be dt ^ ' dt accomplished and to a function ^(0="^ which Ross [73] uses a simple rheologicai model and gives: -takes into account the state of the concrete at the moment considered. Hatt in 1907. tm. Delarue [78] of period.

is proportional to the limit. deducting shrinkage. etc. is possible to separate basic creep from a deforma- creep measured 70 days after loading {shrinkage or swelling tion favored by the movement of water. The same holds for concretes cured in water (fig. Relation of to negligible for higher pressures [62] (fig. 500 1000 We shall see further that it is possible to have DeFormaHon In ij|ni (-^xlO*) a different conception of shrinkage in which it Figure 23.0005 and 0. On the other hand. The curve is tangential and for a long tune close to the straight line log -K log from which it subsequently separates [62]. However. after prolonged creep under con. the creep. 23). sity of stress below 50 percent of tlie breaking stant external conditions. it seems that the latter can be considered as practically ac- Certain authors claim that the final deformation ceptable for engineering calculations for an inten- of a concrete. If we set down log ^"^ ^ as ordinates and log ^'^i^ as abscissas. or one-third of the breaking load. considerable for slight loads. Portland cement concrete loaded at 28 days. when the influence of the l+t latter. Certain tests suggest that the total deformation a+f becomes linear and proportional to the load. it can be allowed that below 100 kg/ cm^. This allows calculation of Kx and K2. CM o 150 • 100 Figure 24. Maniillan [84]. is proportional to the applied load to within less than 10 percent. we get a diagram like that in figure 22. Creep at 50 percent rh. a But in view of the slight deviation that this Size of Load involves in relation to linearity.0015. We have found for a series of concretes that Ki is in general between 0. the linearity appears less clearcut (fig. included) load: Glanville [41].60 and 0. for curing in dry air. Others appear to be certain of the contrary: de la Pena [86]. 24). L'Hermite [31]. Neville [85]. becomes Figure 22. .80 and K2 between 0. much lower than K^. 25). "3 i- i 50 0 500 1000 1500 2000 DoPopmaHon in |j|m |Ai_x1o' 680 . Our experiments have shown an almost perfect proportionality for samples cured in such a way that there can be no exchange of water with the outside up to 50 percent of the breaking load. etc. log shrinkage included.

the results of experiments of which we have had The mode of curing that precedes loading like- knowledge and of experiments which we have wise has a great influence. R. but nothing in the data 175 njm at 350 days. The fourth column of the preceding table gives the result of the calculation which indicates isa decreasing function. The relation ^"^^l curve. Figure 26.400 Air 50 percent 2. we may indicate that after 1. is that A„=A.(l + (^A. 3 <567e9C 20 30 ^0 50 60708090100 ajy in doys (log.600 days the following creeps were obtained on a portland cement concrete loaded at 28 days at 65 kg/cm^: Relative Total Shrinkage Calculated humidity creep without creep load ii/m film nlm Water 180 -120 200 Air 99 percent 250 -100 240 Air 70 percent 1. drawn from a good correlation.i(l + (?A. The interdependence or at least the corre- Age of Concrete at the Moment of Loading lation between creep and shrinkage appears posi- tive. the creep (shrinkage being deducted) from 550 to this must approach a limit. Mode of Curing Davis [69] was the first to bring out the im- portant influence of the humidity of the curing medium on creep.„ = A„„. the shrinkage with the corresponding humidity.) the creep after several years of loading is less as the age at which the loading has been effected is where A„i is the creep in the absence of shrinkage. Curing in water for A„(7) 28 days instead of in dry air (50 percent rh) reduces way with It seems that the logarithm of the age. Variation of creep in terms of age of loading. It the mechanical strength R reached at the moment is believed that the total creep deformation varies of loading and to add a correcting term to the inversely with the mechanical strength of the con. in any case. What is certain. but the precise form of this relation remains to be found. Q a constant depending on the concrete and A^ more advanced.) 681 . Deformation. carried out. in relation to stress.100 +1. We observe the enormous influence of the hu- midity.200 2. empirical formula already established: crete at the time of loading. A. aside from the scatter.)/(/0 Figure 26. that curing in water before loading may greatly A„(a) the relation diminishes in a practically linear diminish the outer creep. Let us call A„(7) the creep after several years This relationship indicates that creep in a con- of a concrete loaded at 7 days and A„(a) the creep stant atmosphere is a linear variable of the abso- after an equally long time of a concrete loaded at lute shrinkage measured in relation to the swelling the same rate at any age a. By way of example.450 +750 1. Davis has shown shows that. It is possible to bring in at present available allows us to establish it. coord. We have proposed [31] relating these two Experiments made on this subject are relatively phenomena by a formula: few in number when great differences in age are involved. including shrinkage. 100 Shrinkog* + crasp Figure 25.

which it would be presumptuous to try likewise obtained hy Hansen [87]. to specify at this point but which must take into however. He obtained two families of concretes: by steam. In The influence of temperature on creep has been order to check this fact. were The effect of the cement content has been replaced in dry air for 70 days. of drying produce an increase in creep". produce a rapid drying. These un. Kordina [91] produced eight concretes Tests by Seaman [88] have borne on the of identical cement content with different aggre- influence of the preliminary treatment of concrete gates. He also indicates that creep increases especially In any case. (4 to 7 days). and beyond question a considerable creep for neat systematic experiments. After hardened in moist air at 30 °C. and so forth. a lessening compensated proximately 180 n/m per 150 kg/cm^. and quartz. ratio.300 n/m. Similar results were content. as against 400 n/m of total creep at 20 °C. it varies between 1. including Powers. and following the concomitant variations of shrinkage is stabilized after 2 weeks. 100. it would be interesting to pursue at the beginning when the size of the aggregates investigations of this kind. namely 50 n/n\ at 90 days. cement content increases. the creep values (shrinkage included) at 25 yr for 50 swelling in water is compensated.in the case considered. However. The influence of the nature Without load. basalt. the concrete swells in water. generally speaking.100 from this conception to that of swelling pressure m/ui. 50. the swelling their modulus of elasticity is very low or when they is reduced and varies inversely with the compres- can be subjected to actual plastic deformations. The nessed a general shiinkage. independent of the grain size plays a role as well as the water/cement load but without creep. shrinkage under a given load We call attention also to Ross's original article diminishes when the strengtli of the cement [89] in which he proposes treating the concrete increases (the creep under this constant load 682 . afterwards the deformation curve 200 kg/cm^. quartz. Next. After Composition of the Concrete this period. remains parallel. We find upon examination of these test the samples were again placed in water for 90 results that there is a lessening of creep when the days: theie was again creep amounting to ap. to consider another aspect of the account the grain size of the aggregates and the phenomenon as it is indicated by figure 27. Moreover. The difference for 45 °C that Some swelled. For 28 days. properly speaking. 1. granite and elasticity 30 percent lower than that of a concrete basalt. The treatment in autoclave at 180 the one includes aggregates of the following types: °C under steam pressure gives a modulus of siliceous gravel. for before loading by heat and vacuum in order to instance as a first approximation /( 7?) =^//?.where /(/?) is a variable to be determined. who indicates that this of samples having stood for 600 days in dry air factor is appreciable in the first days of loading under a series of permanent loads: 0. shrinkage independent of the load. put forward by certain authors.700 m/hi. 150. A fresh by the increase in shrinkage. it is exposure to dry air for 28 days gave rise only to a possible to observe a new increase for rich mortars. 27 and 28). appear to show that concrete composition with minimum creep which — creep as it is usually considered the difference is likely to be found with an aggi-egate. but persists to 60 days under a load of 25 kg/cm^ the absolute difference between the swelling of the applied at 8 days of age without exchange of water nonloaded control and the deformation under with the outside (coated test pieces) is 160 n/m load was positive and spectacular. 1. others continued to contract. Placing in a damper But an unexpected phenomenon was brought atmosphere at the moment of loading determines out at the last RILEM symposium in Munich a swelling that at least partly compensates the (Oct.7 and 2. It is not far percent rh. studied by Davis. This seems to mean that there is a merely as a first attempt. which must be considered pastes. We then wit. When one tries the 7X10~®. slirinks of the aggregates is in fact really appreciable when in dry air.3 X lO"*'. the deformation corresponds to that of shrinkage. increases. It may be said that under a pressure The tests made by Davis [69] give the following p equal to 100 kg/cm. the creep of the sample exposed to a compression of 150 kg/cm2 was 300 ii/m (figs. marble. the second seems that this creep is very rapidly stabilized family gives a creep which varies between 6 and (in approximately 30 days). It is possible. It appears that. having a between shrinkage and deformation increases — special mineralogical composition (quartz or very rapidly through humidification and remains stable hard limestone) and for a rather high cement during subsequent drying. the other includes the sandstones. sandstone. and others. 1 yr of curing at 65 percent rh. 2. sive load. The influence of tlie nature of the binder has There is probably a relation with the structure of been studied by Neville [85. We then placed them in water. approximately after treatment. 1958): "alternations of humidification and subsequent creep. the creep becomes very slight ing to the water/cement ratio. still loaded. we used a certain number studied by Serafim [90]. This diminution of without load. the first family gives the creep of the treated concrete is approximately a specific creep in relation to 1 kg/cm^ which one-quarter that of the other. granite. 2. the hydrates. 92]. the test pieces.600 n/m. But at 70 days. We may note further that the creep silico-calcareous reaction with introduction of increases only slightly but in a manner correspond- powdered silica. nature of the cement. Under load. Glanville.

water after 600 days of loading. Initial recording at first placing in Creep at this time: Okg/cm2: 280 M/m 50 kg/cm2: 1. Series loaded at 7 days.800 Wm 150 kg/cm2: 2. Deformations due to alternations in humidity in relation to the deformations of test pieces loaded under the same stresses and cured continuously at 50 percent rh. E 300 150ke/cm2 \l00kg|em2 0 kg/cm2 100 150 200 Tim« in day* Figure 27.000 M/m 100kg/cm2: 1.900 ii/m .

the coordinates log (Aj. 10 However this may be. Effect of Load Removal When the load is removed from a test piece that has been placed for some time under a permanent load. inversely proportional to the equation A = A„(1 — e~^') where A„ is the total strength). If we The dimensions of the test piece have a bearing unload test piece A. moreover. 21). during a visco-elastic period. in the absence of evaporation or water ab- sorption. The magnitude involved is thus appre- ciable. When the recovery is effected in the same medium as the loading. 684 . time as the shrinkage. it is whereas works such as gravity dams where the negligible). Representation with aluminous cement has a lower creep than portland. followed by a further deferred recovery. 29). ibility of creep (fig. negative (if there is recovery of weight. appears the most probable). the phenomenon appears Figure 30. the unloading curve after on the speed of creep and perhaps alsoon its final time ^2 is obtained by subtracting from the value. Let us now The Influence of Volume consider a creep curve after age curve B. — A) and t gives straight that creep increases with the fineness at the same lines (fig. comparisons should be made with the adiabatic shrinkage or basic shrink- age. The influence of the The question arises as to what this recovery is chemical composition of the clinker has not yet due to:perhaps to a reabsorption of water from been studied and is a subject well worth looking the surrounding air.being. 30). Let us consider a creep curve obtained after age a^: curve A. it is indispensable that in 1h Id 2d 5d lOd Time in days experiments relative to the influence of the dimen- Figure 29. creep is independent — 100 of the dimensions. perhaps again to a reaction foundations should use concretes with a relatively of the elastic aggregate phase on the viscous higher creep. 1000 Results of the same kind were obtained by Karapetian [79]. so that the parameter r which defines the time of equa. Elastic aftereffect. for it corresponds to an reversibility of creep. perhaps to a movement of the water structure must adapt itself to the movement of the in the hydrated phase. o Pure paste loaded at 320 kg|cm2 Gwosdew [93] has studied prisms whose dimension A Concrete loaded at 157kg|cm2 ranged from 1 to 8. sional scale on shrinkage. For an intensity of stress D Concrete loaded at 157 kgjcm^ and put back equal to one-half the breaking load. However. that augments the instantaneous elastic recovery by 15 to 30 percent (fig. creep must be of the form r = with u included between 1 and 2. The supplementary creep that accompanies drying occurs after a period proportional to (^j^ where u varies between 1 ^ 1 and 2. Davis' tests [69] show at 10 yr creeps that X Pure pasta loaded a\ 300 kg/cm2 decrease with the dimension for curing in air. Representation of the hypothesis on the to be markedly viscous. it undergoes a partial instantaneous recov- ery. Indeed certain works like prestressed con. The author indicates further that recovery and Q a constant. the creep after in «ater aPter 1 200 days of load and 700 days a few years was approximately three times greater of non-load in air. Only under these conditions can useful comparisons be obtained. weight that we have made have been absolutely crete bridges require a concrete having low creep. for the small test piece than for the large one. tion seems to be a problem of construction that has McHenry [75] gives a hypothesis on the revers- not yet been sufficiently considered. it appears certain that for concretes ^^=: cured within a membrane preventing any exchange \a of water with the outside. The adaptation of the binder and hydrated-cement phase (this last assumption the composition to the liability to plastic deforma. but the measurements of into. The Russian school advocates the use of the parameter '''—^\T)) analogy with what has been indicated above for shrinkage.

I should like to see 500 "iSOkg/cmZ this lead to a few experiments such as. for river sUiceous aggregates. making allowance for shrinkage. and crushed quartz. particular that the residual deformation after esis is a convenient approximation. changes in length of a concrete and indicates in It may be said. above 45 uncertain for creep in dry air (^=50%). as the load applied has been greater (fig. the placing in placed under the same conditions. This — *50 k9|cm2 procedure might give indications as to the influence 0JkflJcm2 of water on viscosity and an estimate of the forces £ 300 of elastic recovery. to be published in the Annales de I'Institut Technique du Batiment et des Travaux Publics).30. 200 Transverse Deformation of Concrete (Poisson's ratio) S*ri»» loaded al 7 day* al 8=50V> We have noted that under instantaneous load- unloaded at 144 day* at 8= 50 V. and the weight measurements do not on this account it appears to have been verified in several cases appear to indicate an increased reabsorption of [94]. 33). Certain authors consider it to be 2000 equal to 0 or in any case much lower than Poisson's ratio relative to elastic deformations. This is a very puzzling phenomenon postelastic effect is rapid and viscous. unloading and placing in water is much less than We may say further that if the concrete is the maximum deformation. 31). We have observed that it gives an acceptable water. It therefore seems that creep does occur with a lessening in volume and is governed hy the elastic skeleton. which incidentally is not alwaj^s the same per unit of length along the axis and the side of a prism. son's ratio". 685 .5.prdinates of curve A those of curve B. The kg/cm^). If this hypothesis is confirmed the conclusion is of great interest. on the contrary. If we call 100 200 300 400 500 600 Duration ot imm«r*ion In day* <rp = ^ the ratio of the longitudinal and transverse Figure 31. showing thereby that plasticity is a phenom- enon which occurs without change in volume. we call this "the plastic Pois- unloading. Our tests made recently and very carefully to E 1500 clarify this question have shown the existence of a plastic Poisson's ratio very close to or equal to the purely plastic Poisson's ratio for instantaneous deformations (fig. It is also to be observed that this swelling approximation for some 50 days at a maximum appears only if the load applied exceeds a certain when the creep occurs in water but that it remains magnitude (in the case of figure 31. whereas and invites further scrutiny. But. ings the reversible transverse deformation was 10 100 Imm«r**d in water at 315 days to 30 percent of the longitudinal deformation and that this ratio diminishes with age. Defor mation of a concrete loaded at 7 mation to the longitudinal deformation in a bi- unloaded al 144 days.2 (Mamillan. finally. that McHenry's hypoth. 1000 In our experiments we have found it close to 0. to form C. granite. univocal wav. For rolled quartz and basalt he obtains a value close 500 to 0. Figure 32 shows the creep does not conform to a simple law of viscosity. Kordina [91] has found Poisson's ratio varying with the various types of aggregates. and all the more water after unloading gives rise. Up to 200 days. to a higher elastic recovery. it swells observed that if placing in water under load more than a test piece that has never been loaded gives rise to an increase in creep. the values are grouped around 0. This hypothesis is attractive in its siinpUcity. however. for example. This con- clusion is at all events in contradiction with the classical theory of plasticity which indicates a-p= 0. •100kB|em2 the reloading at the moment of placing in water 400 to compare it with the effect of unloading. for sandstone the Poisson's ratio becomes less. and. for it would make it possible to relate the transverse defor- Figure 32. It will likewise be placed in water after removal of the load. Swelling due to water irnmersion after plastic deformations.20. this ratio appears to hold. and submerged at 316 days.

However this may be. Tests between elastic and plastic deformations. We may recall in this connection the tests deformation was the same as that resulting from by Bingham and Reiner [96] on cement joists 1 600 days of permanent load. is a continuity an acceleration in plastic deformation. however. creep occurs with a We wanted to find out if a concrete already lessening in volume. Instantaneous deformations of test pieces.000 percent of that of the nonloaded control concretes. A concrete adapted under permanent as a result of creep. however. mation under repeated load do not have a cumula- not anticipate an appreciable increase in strengths tive action. on the side and 34 in. Age of loading: 30 days. Thus a test piece loaded for 1. This by a maximum of 6 percent. Washa and Fluck [95] have microfissuring for the transverse modulus of operated for more than 10 yr and have shown that elasticity had not been lowered in the course of the at the present date the increase does not exceed 5 treatment). days was unloaded. increase exists and must be confirmed by experi. There. and because of this in load. the creep that it to a permanent compressive stress of 126 kg/cm^ is possible to measure is very slight and is limited and the other to the same load repeated at a rate of by rupture to a short time. repetitions the plastic deformation had increased loaded test pieces and the loaded ones. wliile one can. After 10 million repeated loadings the again. placed at rest. After 3. in length subjected to a 686 . repeated loads in order to ascertain whether the ment of a viscous phase it cannot occur without deformation by repeated load was of the same an increase in compactness. (We already knew that it was not produced by ments on strength. but it appears at least that observed was the same as for 28 days of permanent the creep is more rapid. The laws.000 repetitions). and reloaded On the other hand.600. After 24 hr of treatment It is possible to see that the creep is proportional (720. Davis' tests Creep under permanent load and plastic defor- (verbal communication). which appears undeniable. result appears to be confirmed by R. the permanent deformation to the applied load. in. heating through mechanical hysteresis). Curing: 50 percent rh. As the have been made by Le Camus [56] on two groups of tensUe strength is very low as compared to the identical test pieces. Creep Under Repeated Loads Creep in Tension and in Torsion —Combined Stresses If instead of leaving a permanent load resting on Tension creep devices are more difficult to make the test piece one repeats the same load at a rather than those for compression. Apparently this nature as deformation under permanent load. One of these was subjected compressive strength. the increases that are observed load is no longer apt to be deformed under repeated below 50 percent of the breaking limit are always load (if we neglect the 6 percent indicated above positive and clearly show the effect of an increase which perhaps is due to shrinkage resulting from in compactness. 500 times per minute.000 repetitions (5 days) the a bent beam the neutral fiber rises toward the effect was identical with that of 180 days of static compressed surface and subsequently descends load. rapid rate (several times per minute). one observes appear to be similar. deformed by creep was sensitive to the effect of and if this phenomenon is produced by the displace. the modulus of elasticity under repeated stresses. Finally. At the end of 3 million increases from 20 to 30 percent between the non. -• longitudinal contractions o___o transverse dilatations in center X X transverse dilatations at extremities 100 — \ X \ e^* \ <N 75 <^x ^» E \ S \ V ^ 50 T ^ \ \ \ \ \ Dilatation \ Contraction +100 0 100 200 300 DeFormations (m/"^) Figure 33.

and we shall begin by asking a few questions. This acteristics evolving with the degree of hydration. but that the shrinkage along the axis is the same for having cohesion and rigidity. The author believes that from this viscous creep. the proportion of the exactly for equal loads of contrary sign. The elastic skeleton follows the and of compression at 90°. the binding paste is out over 400 days showed that the permanent composed of a padding of crystals forming no deformation is proportional to the applied couple. which conforms to the conservation of the Le Camus [56]. Under these conditions. Irreversible deformation after unloading at tion. Figure 34. a result which unloading by virtue of its association with an appears to extend the application of the linear elastic skeleton. plus a part of the of sides. which is characteristic of a young creep that has been called secondary shrinkage science. Square Apparently. movement in the direction of the forces applied ments of this kind that I have come across are with the transverse deformations tliat are peculiar those that were carried out in my laboratory by to it. Some of these theories are simply ex. The aggregates are in most Tim« cases much more rigid than the binding phase and have little susceptibility to plastic deforma. The paste is a gel. 687 . that the two forms and compressive creeps compensate each other of binding paste coexist. hence a viscous body. which their plane were stressed on the other two pairs last can include the plastic creep. one can only separate the reversible test pieces left free in the direction normal. the effect of external forces is to put this gel into The simplest combination of stresses is pro. to the authors. but in con- tests were ended too soon to bring out the possible tinual evolution by virtue of hydration. apparently not produce a torsion. longer a viscous body but a plastic solid with a Another observation not devoid of interest is deformation which is viscous. to be sure. tests on compressed cjdinders subjected to a Can several forms of creep be distinguished? lateral stress. second in relation to the first increasing with Dukes and Davis [97] have carried out creep time according to conditions of hydration. The only experi. not well defined. Leaving Theories and Conclusions on Creep out of consideration the creep and the possible evolution of the concrete during unloading. others are quan- elastic creep. by contrast with the primary creep or visco- planatory or phenomenological. This separation is what was done test could be deduced the linearity of deformation by Freudenthal and Roll [83] and by Glucklich [99]. damp medium is approximately proportional to Viscous creep must be in part reversible upon the difference in the chief stresses. Plastic creep cannot be reversible upon un- but the interpretation of the tests is difficult in loading so long as the reaction of the elastic this case. determining for each age of unloading the irreversible deformation. And unloading does have a law of deformation. in the case of creep under double stresses. appears to be of the order of 1/6. the speed of creep becomes constant. relation to time has at the outset the form of a Why and how does the cement paste creep? parabola and then becomes linear in relation to According to certain authors the binding time. movement in such a way that it fills the voids in vided by torsion which is a combination of tension the concrete. according character of viscosity (fig. The method has the advantage elastic Poisson's ratio observed by different of eliminating shrinkage. As Total »train it is difficult to summarize them all and in any case impossible to bring them all together in this report. dicated in figure 34. titative and are aimed at permitting calculation and extrapolation for the needs of builders. The longitudinal creep deforma. since the latter can authors insofar as creep is concerned. Creep thus appears to have its seat in this different ages. The deformation curve in binding phase and in its bond with the aggregates. hence stabilization and the hygrometric conditions were of variable viscosity. in relation to load and a very low value of the proceeding by unloading at different ages as in- plastic Poisson's ratio. for that matter. these two char- the loaded and the nonloaded test pieces. In this case there would have to be an indefinite tion measured after 3 months of loading in a viscous creep and a limited plastic creep. the Theories on creep are numerous and varied if not curve obtained represents the irreversible part of contradictor}". Where does creep liave its seat? Concrete is composed of two essential parts: a skeleton of aggregates and a phase composed of a more or less hydrated cement with generally an excess of water.to part of creep from the permanent part. 29). skeleton is less than the stress applied upon load- Ross [98] has carried out very interesting tests ing. which is the case.bending moment. Poisson's ratio. we shall examine their main features in what is to follow. identical shrinkage can occur only if the tensile It is possible. Tests carried According to other authors.

we consider that the first term is negligible compared to the second. 32). —a wholly reversible visco-elastic creep: Concretes cured without a possibilitj^ of ap- preciable shrinkage nevertheless have a marked ^^=5(0(A„-A. further in these calculations. however.e-^<^ shrinkage under load whUe the deformation re- mains the same. 99. finally. Their practical use. It is probable that placing in must be deducible from the unloading curve. and return atmosphere leads to a higher evaporation of the enclosed water as compared to a loaded test piece.„. the two phenomena appear more precisely linked to a form of the hydrated phase that leads both to a similar evolution. This law may be the combination of three Without any doubt the concretes that have the phenomena: greatest shrinkage have the most marked creep. measurements of strength and of modulus of 688 . But noth- ing. Let us not forget. and temperature? This question is both simple and complex. This relationship hydration that gives rise to a new creep.„. that there can be creep The interaction of these phenomena which we without shrinkage. on the other hand. in any case. From the hygrometric versible: point of view there is certainly a dependency between creep and shrinkage. when placed in water. so that: the hydrated phase for a concrete cured dry or to a continuation of hydration. with or without load. Is creep indefinite? There is nothing. undergo deformation. This is not very different from what McHenry hypothesis does appear to be confirmed by the proposes [75]. If. over a Now B{t) must be a function similar to A{t) to period of drying that follows the placing in water within one factor. which endows the latter. creep (adiabatic curing without exchange of water) This means. This procedure leads to calculations that are extremely complicated and at the same time The basic creep may be a factor of comparison that eliminates the dimensional parameter but does of great interest. we find: a swelling without load and a supplementary A=^(0+A. remains uncertain [83. proves that creep in a dry Kelvin justifies the use of models in which one combines springs.„. This phase would then have as its law where the second term includes the viscous elastic- of deformation: ity. For order to know whethera{t) approaches zero for t approaching infinity. water leads either to an increase in the volume of Its form is viscous but depends on age. there must subsist a permanent viscous phase not included in the F{t) with A(0 = ^^+Ajl-e-^<"] visco-elastic behavior associated with the elastic skeleton. thus: in the elastic recovery for a test piece that is unloaded and placed in water (fig. This hypothesis would likewise explain the increase where /(o) is an increasing variable of time. friction. humidity.„— A)F(f). through the lessening of its viscosity and its rigidity. 27 and 28). one would have to know the the moment.e-'^<''«']- cured in water is less than for dry curing. not appear to have any other effective value. with a new possibility of deformation. A number of authors have qualify in rheology as laws of Newton and of considered this creep as a basic creep. dash-pots. it does not seem possible to go much evolution of creep over a period of tens of years. in addition. that of a viscosity depending on time both for the one and for the other. moreover. 100]. We must then suppose The first term between brackets represents that curing in dry air does not permit a complete the continuous part of the creep and the second hydration and that it is the resumption of this term the part limited to A„p.e-^'"+A.) A. is In come back to the equation ^=(A. while it is not certain that this relation is biunivocal. In — an irreversible viscous creep: other words. however.„.„. it is probably more correlative than functional. . — an elasto-plastic creep that is partially re. valves. and this appears contradictory. that the creep of a concrete A(0=[A. Is creep linked to shrinkage? What do we know about the law of creep with constant load. For this to be so. we where a{t) a decreasing function with t. that a concrete cured dry over three types of a long period may. the latter effect not being separated during loading. to prevent seeking This question highly important and not yet is an empiric variable resolved.e-^^". The form of this function (figs. But to say that creep is a shrinkage under load is in a sense a play on words.r^]-h[^^+A. We If we simply superpose the have seen. = A.

finally. We might represents the additional deformation for curing describe this in a simplified and phenomenological in air in an atmosphere of relative humidity 6. This means that the viscous creep must be approximately proportional where H represents the volume of hydrated binder. C. the same time as the speed of hydration. But we have no knowledge represented in volume. Hansen's surmise is very ingenious. of any such experiment. of view with what has just been set forth We see. on evapora. and taking ^(t) = 0 as a constant. which accordingly would We said earlier that creep may increase in- have the effect of separating certain groups of definitely so long as a viscous phase subsists. The A„„=^logf^ explanation for this behavior is not easy to find. but it does not explain the cessation of creep in a period of where yt^ represents the relative hydration at the drying after wetting. the speed increases at the second effect of viscosity for ciiring in water. T. second the viscous creep. and a viscous and visco- The term between brackets represents the first elastic creep of which. and particles and thus diminishing the rigidity. as tion. or Ho=H„iod where Hmo is movement. If we assume that the relative hydration does this author. We can compare this point resumption of creep upon placing in water.6. atmosphere.=/3[log^^-hlog ^] limited. its of hydration. if this fact remains confirmed moment of loading compared to the total hydra- by other experiments.99 and 0. It disappears jnore rapidly in curing in tion. we find: ment that it produces is a factor in softening of the binding phase. 689 . no matter what may be the state of curing. But instead of saying. But this mechanism does not. herein may be found matter for broad interpreta- Pickett [49] suggests that creep constitutes an tions. that the detailed explanation of the mechanism of creep on the scale of structure and write: Viscous creep of the has not yet been given without possible challenge. in mechanical strength. drying should give a higher and more rapid creep. the medium in general. It is dependent upon time. as the hydrated phase is more considerable. In the case of curing in constant atmosphere hence of previous hydration. —the increase of visco-elastic recovery by and creep in a closed medium is less than in a dry placing in water. Such a hypothesis. lias less creep than that cured in a closed medium. completed. and hydration? water than in dry air where the hydration is con- Creep increases when evaporation occurs in siderably slowed down. is homologous to the variable H indicated above. T. The influence of the absolute volume on creep tory phenomena such as: is undeniable. in a present a resumption of creep after drying and sense. tion. explain the we wrote ~-"=a(<). This calculation has been made without any I hope I may be allowed an approximation other pretension than to show the comparability suggested solely by the drafting of this report and of its principle with the results already known which so far has benefited by no direct verifica. forth the hypothesis that creep is encouraged by temperature. There may thus be a visco- plastic creep of which the speed is all the more A. whether it be degree fresh humidification. C. that the diffusion of water favors is proportional to 6. cured in water. to (l—d). Newton type disappears only when hydration is How does creep depend on humidity.20 for 0 included dt H H dt between 0. apparently. In fact it any movement of dampness alternate wetting and . Hansen [87] puts in rigidity. it is generally explained by the —the more rapid stabilization of creep in a rapidity of evaporation. way such as: dA^Ajt) B{t)dH Now 1 —9 varies between 1 and 1. which is not in contradiction with the the first term indicating the plastic creep and the experiments.dynamic elasticity. given equal hydration. We should then have to — the resumption of creep by humidification know whether a concrete hydrated permanently and its stoppage by drying. accleration of shrinkage. greater. We have seen. that the creep deformation under load the final viscous deformation is: that is added to that of shrinkage without load does not lead to loss of weight. and to propose to investigators a line of reasoning. that is to say when the evaporation is resumes upon placing in water. we find that however. Concrete cured in water damp medium for a lesser final value. in the latter case. An excellent interpretation of it is to be found in This would mean that concrete is a thixotropic a recent report by Ross [101] and we believe that material. we might suppose that the chemical the degree of hydration at time t 0 of the piece= phenomenon of hydration by the molecular move. in water for a long period presents or does not The maturing of a concrete represents. makes it How does creep depend on volume? possible to interpret several apparently contradic. or therefore to a measurable supplementary loss of water.. etc. it dry air. however.

La Louviere.. L'Hermite. J. 4. Bonnell and F.. is large H the surface and lead to a concentration of stresses dH toward the core. recentes sur le retrait des ciments et batons. Materials. Haller. . 1956). Grieu.. Soc. p. Publ. Nou. Brownyard. We know that a preliminary creep in a moist medium less pronounced than the curing in moist air diminishes the final creep in surface creep. 55. 690 . nomenes de dilatation qui se prfeentent dans les [30] J. Shuman.O. [35] R. 1 (Jan. Proc. Crowley. 465 (Dec. 1952). Initial thermal expansion characteris- [6] H. Theorie de I'expansion des Hants hy. Etudes exp^rimentales shrinkage of concrete block. Develop. 742 (July. Research Dept. 7. nonpublished document. Concrete Inst. Leber.. 1959). [34] CoUoque sur le mouvement de I'eau dans les corps 521 (June 1954). Thermal expansion tests on aggregates. Concrete Inst. Durand-Claye and P. Research and Develop. beton.Jan..June 1934). publ. Zurich. L'Hermite. Effect of cyclic wetting [9] P. tech. The pieces bulky pieces. G. Bogue. 3 (March 1955). de Belgique (April. Labs. Blakey and I. 1103 (June Soc. C. London (1951) and J. materiaux construct. Bur. [26] F. et Rech. Davis. [24] E. Fellowship Paper No. Proc. When desiccation reaches the stresses [102. T. Building Research. The latter then undergoes a and is slight. p. Contribution a I'etude du retrait des [17] L. Portland Cement Assoc. Lewis. Am. Ann. No. (1949). 193 (Nov. Melanges de ciments pour Portland cement pastes. publ.R. Paris. Le controle de I'efficacitd des curing com. p. part II. p.P. Ess. 503. Proc.. p. E. the bulky pieces creep tion. Concrete cement mortars. Bat. Inst. C. Alexandrovsky. Here this process is progressive but leads sated there would result a delay in the deforma. properties of concrete. On the hysteresis of the shrink- Trav. Rev. L'Hermite. [20] S. [10] G. and References [1] R. Comptes-rendus des recherches du No. Grieii. Contribution h I'etude de la stabilite de (1955). Lerch. Blondiau. tech. 4-1 (1958).. 35 Sclent. Schwinden und Kriechen von Mortels und ous references). J. [5[ R.S. J. Am.. 1930). Reinforced Concrete. stability of concrete masonry units. Research [14] G. paste. bfitiment trav. age-dilation curve of concrete put through cycles [12] R. et T. L. Blakey and R. Meyers. Nat. federal essai materiaux. J. Ann. lab. bat. Studies No.. [22] L. Finally. The early shrinkage characteristics [21] T. Anns..S. J. p. Mag. Portland Cement Assoc.. La Technique Routiere. Mitchell. b^timent et travaux publics (1958) and Ann. 1957). Shideler. C. J. K. Ceram. Effect of time of [27] F. Self desiccation in [16] L. Goggi. inst. Fundamental factors in the drying [7] R. L'Hermite and J. p. Powers and T. Autogenous volume change of concrete. 1929. C.. Belgique (1957). Station. L.P. p. and J. Causes and control of volume change. Harper. If the effects were simply compen. 320 (1933) (numer. Am. volume des ciments. No. and drying of restrained mortar specimens. 13 (1947). Am. No. Beitrag zum Studium des Volumenbestan. Testing Aug. Creep and shrinkage begin on core. [28] S.. 40. Concrete Inst. [19] T. A. 1957). Thermal expansion characteristics of [2] H. 1947). Dutron. Civil Engrs. Studies of the velle contribution a I'etude du retrait des ciments physical properties of hardened portland cement et betons. Soc. tech. R. Melbourne. Lab.. Inst. Shrinkage in sealed atmosphere) of concrete block. 1956). Res. Bragg. p. E. Org. Idees actuelles sur la technologie du (Sept. 80 (1957). Bd. batiment trav. hardened cement paste and of concrete. 963 (1953). pounds. Washington. inst. Dilatometre a mortiers et ciments. batiment trav. Bull. Le retrait des ciments et betons. tech. Zement-Kalk-Gips. Lafuma. les Assoc. p. [32] L.C. Two explanations are then possible. 549 (Jan. Ann. Rev. [3] B. 90 (1958). Chefdeville. J. p. Vironnaud and N. Div.V. 1940). Ann. Bull. Portland Cem. Saxer and H. Testing Materials. Research Development Lab. Studies on the volume stability of of drying and wetting (in Russian). 33 (June 1958). [23] M. C. 1937). Soc. Research Board. at the beginning. but as the compensation does not occur for less because they are better hydrated. S. tics of insulating refractory concretes. Dept. [33] M. p. Association Fran?. Le retrait des ciments et betons. Bull. Paris (1959). L.S.. p. Note on early shrinkage of application of sealing compound. Bruxelles. Labs. Comm. Am. materiaux (Dec. poreux. Assoc. p. ponts et chaussees. [18] D. Rech. S. Am. Shacklock. piece is. Dutron. Debray. Ind. E. 1958). Feret and M. search. 1949). of Concrete Re. Theinvolves the internal first a closed space. 797 (Feb. sion of concrete. Dutron. 1959). Inst. mesures et resultats sur le beton. A. neat cements and concretes. 29 (Jan. Cahiers de la Recherche. dry air. Venuat. 810 (1888). Beton. the slower its mean evaporation is. R. p.. 34 (Feb. Lab. Build.. p. H. ponts et chaussees. L'Hermite. 79]. The bulkier a rapidly stabilized final plastic deformation. Am. Nat. Van Than. to similar effects. p. Am. Australia. Lucas. Res. Portland Cement [13] L. Documentation techn. Powers. Note sur ph6. C. Testing Materials Proc. 3 (March 1958).M. Constructional Rev. J. p. Spinder. 1956). W. 344 (1958). Investigation of the moisture volume pStes de ciment portland. 193 (1950). Paris [15] L.. [29] T. 1955). Concrete- Portland cement pastes. p. Bull. Paris (April [25] E. publ. Soc. Standards. 33. 9. p. [11] R. 223 (Feb. Paris. Plastic shrinkage. 383 [31] R. The thermal expan. Paris trepied optique. Highway drauliques. Report C 2. if not the closer. p.. H. there results likewise a lessening in protected from evaporation by surface humidifica- the final creep. Copeland and R.. The physical structure and engineering of hand placed concrete. inst. [4] R. Research and Paris. J. Powers. Testing Materials accelerer la prise. batiment trav. Ann. et T. The second explanation derives tion or by a protective device maintained for a from the interpretation of what we have just said sufficiently long time must have a lower and more in connection with hydration. [36] P. publ. Paris (Nov. Am. Toennies. J. Burnett and M. (Oct. Materiaux.. digkeit der Zemente. it comes to curing in complementary. D. Measuring shrinkage 1952) (see p. H. p. the hydration is far advanced. [8] W. 248 (Dec.

Bull. Studies (G. Am.Simmons. Thuilleaux. [68] J. [74] E. size and shrinkage. p. 1956). (London). Inst. Proc. p. C. mor. Concrete Inst. L'Hermite. Am. Tech. C. [58] A. Freyssinet. 11. [86] C. H. 58. Leber and F. Milano (1931). Inst. Creep and creep [56] B. Paris. Amsterdam (1954). No. No.. E. module of elasticity. p. J. Rev. Comparison of three methods for deter. Eng. [83] A. p. Concrete Inst. Serafim. 8. Reiner. du ciment. 24.. 20 (1955). Am. M. Stockholm (1958). ciments. Inst. publ. [82] A. p. M. p. R. Croney. Effect of carbon dioxide on hydrated deformation and creep of concrete?. C. Testing Materials Spec. Civil Engrs. J. The structure of cement hydration [81] G. Ultrasonic pulse velocity. Work Nev. Importance relative des erreurs de [76] F. 803 (Feb. les phenomenes de retrait et de plastique et du fluage du b^ton? Discussion. tures. Berlin time creep and shrinkage tests of plain and rein- (1933). 1956). Paez. Soc. p. Role of cement in the creep of mortar. (1931). Plastic flow of concrete at high overload. RILEM Bull. R. 54. E.) Research Paper (Sept. Creep and shrinkage in restrained concrete. R. et Constr. D. (Aug.). Todd. Civil Engrs. Publ. Quest. 1955). RILEM Bull. Graf. 205 (1958). S. 1959) (Study of the Poisson's ratio in the dynamic [89] A. Techn. Paris (Nov. H. Woolson.S. Theorie generale de la prise des Hants [78] J. 1958) and Am. Shrinkless and creepless concrete. [85] A. Raphael. Concrete Inst.. J. Vivian. Structural Eng. Constructional J. batiment trav. 1959). Roll.. No. Road Res. p. (London). D.. di Cemento. Ernst. Concrete Research Inst. Testing Mate- crete. 1957). 1934). publ. E. Soc. Am. de calcium. Paris (Feb. [50] L. R 46 (1951). Creep and shrinkage of plain reinforced [47] M. 691 630446 O — 62 ——9 . 1957). A. Holland.O. Civil Engrs. (June 1940). Eng. M. p. deformation lente des betons et mortiers. materials. coarse aggregates. Bd. Ultimate flexural Tech. 60 (July 1959). p. Portland [65] F. 43. The determination of tensile stress strain C. Eng. Trans. J. (Feb. [67] N.. of of concrete. No. Bridge. p. [55] R.-Sept.. 58 (March Concrete Inst.. Civ. Pub. Paris. p. Proc. Am.. Work Rev. Am. Labs.R. S. Recherches experimentales sur la recovery of concrete under high compressive stress. des J. Stock- [60] M. G. [54] J. D. Marcotte. Lafuma. nerie et beton arme. bati.. forced concrete. C. tech.. publ. Coleman. 1960). Fisher. [71] W. 87 Eng. Glanville.. 461 (Jan. Acad. 1951). and P. [45] H. Hansen. Mat. What do we know about the plastic [37] S. Etude du retrait du b^ton [53] R. 613 tech. sur le retrait et le fluage du b6ton. Karapetian. Proc.S. (Nov. 356 load. Inst.S. E. [51] E. 4 (July 1959). Rev. I'dprouvette sur le retrait et le fluage du b^ton [52] D. J.. Les lois regissant les liaisons entre 1101-20 (1958). D. July 1959). ponts et chaussees. Davis. Matras. J. deformation du beton et du beton arm. W. p. Ann. [72] C. Testing Materials. J. Philleo. cement and concrete. 33. Cahiers de la ciment. Pickett. [79] K. "La plasticite du beton sous les charges". Paris (Feb. p. p. Testing Materials Proc. tiers et betons. 1951). Paris). p. Berg. Shrinkage and creep of specimens of [59] D. [66] G. Mag. publ. Natl. Young. No. Am. p. Effect of aggregate on shrinkage of con. Essais (1954). Research Dept. 15. Civ. 9. Delarue and R. Brit. 1951). materiaux construct. technologiques la surface specifique et les proprietes [70] I. 295 (Sept. [90] J. Ann. 12. M. [38] I. Engineering. Cocagne and Y. Congres du Beton Manufacture. RILEM Bull. J. batiment trav. curves for concrete. Wiley. p. centre d'etudes superieures de I'inst. Pub.. 327 (1937). Le retrait des produits en beton de sous les climats d'Afrique du Nord. tech. Soc. Swedish Cement [87] concrete.S.. The stress strain relation of concrete. E. Centre d'Etudes Superieures de I'lnst. Technol. intern. (1957). Troxell. Le Camus. materiaux construct. 493 (Feb. Influence of temperature on the creep son of dynamic and static measurements. 321 (April 1934). Preprint Concrete Research.. Roy.. (June 1958) (in French in recherches des lab. Ross. The theory of concrete creep. batiment suction of moisture held in soil and other porous trav. p. Building Materials. Determination of creep strain of strength of concretes made with thirteen different concrete under sustained compressive stress. Weil. 1934). [61] J. C.. [80] J. Effect of the straining rate on the thin section. J. 169 (1945-46). 963 (March 1959). Concrete Inst. Poisson's ratio of concrete: a compari. 729 (April 1953). A. RILEM Symp. inst. 255 flow of concrete under pressure. [62] R. J. Sciences U. Concrete Inst. Kaplan. p. Grassam and D. Izvest. p. London (1952). inst. L. Carlson. D. [73] A. Berthier. rials. C. 15 (July 1959). Some effects of carbon [63] O. A study of the creep of concrete. [44] M. Symposium on the Chem. Ross. Resistenza ed Elasticita di Calcestruzzi RILEM Bull..R. Santarelli. J. Hoepli.M. publ. (Aug. p. F. Blakey. Smith and L. Ann. Ann.. Concrete Inst. Paris (1928). [75] D. compressive strength and elastic properties of T. inst. p. p. dosage dans la confection du beton. Jaspers.. 1069 (1943). Ross. 4 dioxide on mortars and concrete. Research and Develop. Poisson's ratio. 95. R. Quelques observations concrete in M. [39] G. 1949).. Paris. R. p. Thomas. Nauk Armyan.R. Bull. Blakey. [64] J. 69 (1933). [69] G. Creep of concrete. Munich. dynamic holm. 57 (Jan. 1949). (1958) (will be published in RILEM Bull. Versuche iiber das Schwinden. p. L'Hermite. publ. 7. Am. Diskussionsbericht 124 (1940). p. Paris. Cement Assoc. 70. Am. [77] L. Torroja and A. 42 (1946). 597 (Feb... Am. 1 (1956). M. M. Concrete Inst. A new aspect of creep in concrete and [49] G. Am. cement. and March 1950). Concrete Inst. test). News. Eng. 38 (Nov. J. Influence des dimensions de ment trav. its application to design. Concrete (1950). Freudenthal and F. de la Pena. Carbonation of hydrated portland J. Am.. publ. L.. Shank. Association Frangaise Rech. p. 1950). Long [43] O. 96 (Jan. S. Retrait du beton de ciment au chlorure and prestressed concrete. Shape. (J. Akad. [42] R.S. Bruxelles Recherche. G. 32. Freudenthal. Straub. Nouvelles recherches sur le retrait des 4th Congress of Large Dams. Watstein. analysis based on stress strain curves of cylinders. 261 (1934-1935). 21 (March 1959). McHenry. J. 1956). Delarue in Que savons-nous de la deformation hydrauliques. 581 (Feb. M. D. Neville. p. 1943). Vironnaud. Seaman.. Lorman. and the [88] F.. 1953). magon. RILEM BuU. Verbeck. M. Am. Am. [40] A. batiment trav. Stress strain curves for concrete. The creep or flow of concrete under with electrical strain gages. batiment trav. Tests on concrete [41] W. batiment trav. North sur le retrait des ciments.. R. [84] [57] L.. 1126 (Dec.. Bernal. Mamillan. The (translation from Russian by the Lab. techn. The Inelastic Behavior of Struc- minating Young's modulus of elasticity of concrete. Rock Prod.. 233 (Feb. 459 (1905). Lee. 201 (March 1955). Some remarkable tests indicating H. Third Int. Paper No. Influence des dimensions et des contraintes compounds. Congr. Retrait et fissuration des ciments. No. Soc. [46] E. RILEM Bull. and R. istry of Cement. Struc. Paris. London (1952). Proc. Soc. Set concrete and reinforced [48] J. Meyers..

. Flow of concrete under the curing in steam and hot air (in German). 1950). Applied Math. bStiment trav. Fluck. Paris). Des qualites rheologiques du beton [94] S. under sustained combined stresses. and volume changes of concrete.. Creep of cement. Bull. Am. J. trav. Taylor. 577 (Oct. Brown. Kesler. 98 (1956). Washa and D. 55. E.. H. 60 (Dec. G. 1956) Le fluage . Am. A. p. O. p. Blakey. M. (1958). Effect of continuous loads on concrete. J. Plastic yield. C. Proc. may be found here). 327 (Oct. The influence of cement on creep of [98] A. 6. Ann. F. RILEM properties of cement on the creep of mortar. 34.. EyroUes. Concrete Inst. (1955). Influence of calcium chloride on the et les caracteristiques physiques et m^caniques du shrinkage of cement mortar (in Russian). Eng. p. Creep and creep recovery of cement (translation by the Lab. beton. R. Testing Materials. cement. Flow of concrete under J. Civil Engrs. Civil Eng. Backstrom. Dutron. Jeffery. and Mach. p. Belgique. Bingham and M. and A. Ass. Acta Tech. Etude experimentale du fluage du b^ton. M. retrait de quelques betons. A. Jerusalem (1950).96A(0. Faber. A. Am. Bau- T. 4. 5th Cong. Paris (1957). Reiner. Dukes and H. E. Testing (1933). 53. Am.. Duhrkop. Prove di ritiro su calcestruzzi integri e setacciati. Ass. C. Int. stress. Inst. No 1 pub. Concrete Inst. Bridges Struct. Bureau of Reclamation. Kh. Erzen. Gwosdew. Investigations of Portland ce- C. Hamilton. Inst. plastico viscoso dei calcestruzzi. Applications de la theorie du fluage. 40 (May 1949). Collyns. Antra. M. 1956). behavior of hardened in RILEM Bull. R. Physics. H. Dusseldorf. Ric. Univ. C. 1951). 1936-Feb. Munich. F. Prelim. E. 1959). [92] A. (Feb. Berlin planung und Bautechnik. Civil Eng. Proc Am. E. 837 (1931). Hung. 354 (1934). 863 (April 1956). Eng.. Soc. H. R. Report No. publ. Techn. (Nov. Castiglia. S. Bossi. 1955) (The formula A(e) = 0. S. 1952). The rheological properties R. F. 1956). Arnstein and M. Some properties of mineralogical character of aggregates on the creep concrete under sustained combined stresses. p. Indagine sperimentale sul comportamento to prestressed concrete. 303 (1928). Prestressed Concrete. Das Kriechen und die Verdichtung trav. p. Sci. Deformations lentes du b^ton et du b4ton Romande. of concrete. Babuska. J. 88 of concrete under sustained stress. Proc. Soc. Davis. 16 (1952). Experiments on the influence of the [97] C. cement paste under lowstresses. 91 A.S. Acad. [100] I. Deformations elastiques. Ross. Proc. Glucklich. 12 under two dimensional stressing. Proc. Anderegg and A. 36 (1956).... Reduction of shrinkage and creep of concrete by R. R. Bernal. Additional References (Documents consulted but not mentioned in the report) F.. Concrete J. L. 1945). 44. (Sept. F. Materials Proc. 21 (Jan. Some volume changes Centre d'lnformation de I'lndustrie Cimentiere Beige. E. No. p.R. Rheological Symp. Testing Materials Proc. (Oct. p. Suisse R. Busch. Volume changes in concrete. 233 (May 1953). Ann. Concrete Inst. Tidskr. Neville. Plastic flow R. M. Proc. Davis.S. Davis. Experience de relaxation des contraintes which shrinkage is restrained. Some creep theory Concrete Inst. 33. des beton. Concrete A. 1942). Munich. E. Concrete Inst. An expression of creep and its appUcation C. Am. Kriechverhaltens von Beton in Bauwerken. 339 (1923). problems ibid. of concrete from sonic properties. Anderegg. M. Die Bestimmung des Elasticitats und ments. pamp. Dunod. 52 (March 1959). The effect of change of moisture content [96] M. hardened portland cement pastes. (1937). elasticity of concrete. Arnan. 1945). Tek. ibid. Kordina. Torino. p. Reiner. Proc... Research Council Israel. Bat. 39-40 rials Bull. Ch. E. and M. D. Plastic flow of concrete C. 78. W. 30. F. Berthier. 9. Testing Mate. Clemmer. J. Davis and E.5) log 5 (Feb. Flow of concrete under sustained compressive 17. Beer.. 335. Eng.. VDI action of sustained loads. Am. W. 104 of the creep of concrete under a sustained load. Sci. p. Crystallo- sustained compressive stress. 317 (1937). Boulder Canyon Project. p. J. Davis. Prediction of creep behavior reinforced floor structures. & Inst. V. Exp. [102] G. 13 (1949). Preprint (1958) (will be published [99] J. Pickett. Preprint (1958). Davis and J. [91] K. plastiques et de H. Station. E. [93] ZNIPS. 693 RILEM Bull. Soc. Materials. Bolomey. Testing not yet published document. Ed. and H. considering creep (in Russian). W. p. D. Belgique (Dec. Itaha. (June 1951). Zurich (1952). Stavebnicky Casopsis. Am. p. Electrica. Experiments on the creep of concrete concrete and mortar. Copeland Specific volume of evaporable water in ' ory. 591 (Nov. Moscow (1955). S. Davis. Das plastische Verhalten das Betons. Duke and H. 80 (Aug. Influence of water cement ratio on mortar in G. mortar and concrete. F. Soc. Elastic stress state the..). RILEM Svmp.. Mag. Proc. Munchen 1958. E. 1955). Some properties of concrete relieves highloads stress concentration. 28 (1958). E. 113 (1957) (English summary). of Concrete Research (Oct. T. 1937) Le fluage . 13. 44. SiUcates inds. Betons krybning. "Les deformations du beton-Retrait". Ross. publ.. Blanks and D. A. of Concrete (March 1958) and Tests on the influence of the Research (June 1954). 37. Proc. Testing graphic research on the hydration of Portland Materials. E. Bourenine. No. [95] G. Soc. I fenomeni di adattamento nei cal. 888 (1944). Balazs and J. Reiner. Am. 205 (August 1956). 169 (July 1942). Dawance. 55 (March 1959). Aroutiounian. Soc. Am.. Ind. Concrete Zeitschrift. Atti e Rassegna Tecnica. D. J. McHenry. Rheologie Theorique. Effect of sustained [101] A. Testing Materials. A note on the maturity and creep of loading on compressive strength and modulus of concrete. 27 (1927-28) 692 . Creep of concrete (in Russian). pubL. 48 53. K. in mortars and concretes. 37. (May 1950). concrete and their effect on design. Mem. p. Denver (1949). Ja. Am. Am. Soc. Inst. Am. Essais de fluage du beton. Davis. Mag. Plastic flow of cement and cement mortar-stone. J. N. M. arme sous Taction des charges permanentes.. RILEM Bull. 34. of Illinois. Int. Loading tests of Chang and T. Killian. Teinowitz. B. P. Eidelmann. 707 (1930). Ind. p. G. Davis and H. Paris. shrinkage and other problems of cestruzzi giovani. Am. hochwertiger Betone. cement- (Jan. p. No.. Z. Am. pont de Rabat-Sale. 1-2. Reiner. E.. 0. RILEM Symp. 888 (1944). Slovania. Concrete dans le b^ton precontraint. Proc. 1958). J. mortar. 4 (1940).

Concrete Inst. 41. Zement-Kalk-Gips. batiment trav.. publ. Testing Materials. Plastic flow of concrete. Proc. Am. 128 Soc. Fundamental factors in the drying shrink. Time tests of concrete. Gwosdew. Concrete Inst. Italia (July-Sept. pub. Les deformations plastiques du beton R. p. 31. Creep of concrete under the effect of S. English summary). 85 (1956). Am. 1082 (1940). Div. lab. Nat. (Nov. L. Rock Prod. (1930) Une revolution dans les . pub. Proc. Highway Research Board. 13. Soc Testing Materials. in reinforced concrete. Maney. Magnel. 233 (Nov. Recherches experimentales sur le fluage des materials. Extensometer readings in a reinforced concrete W. Gehler and H. Club by the Lab. Jackson and W. W. Proc. and C.3). Akad. Paris). and E. Smith. Lisboa.S. Danish. Uber das Schwinden und Kriechen des Proc. 17. p. b^ton sous charge constante. 1951). 137 (Jan. Kubo. on Chemistry of Series B. Jones and F. Giangreco. Kalousek. Pisetti. 6. Betonstein Zeit. Japan (May 1952). Friedrich. beton (translation from Russian by the Inst. Creep of steel and concrete in relation to No. Temperature varia- in concrete structures. 1934). 693 . A. No. reinforced concrete beams. I. Henk. "Influence of mat6riaux. 421 (1907). in Recherches (1951). age of concrete block... Goldbeck and E. 7. Etude exp^rimentale sur les deformations du research. p. L. Studies in Engineering. S. 302 (1916). Paris). 12. Sci. Izvest. B. R. 127 (1919). Janni.. No. G. Report of com- 1689 (1916). Proc. Concrete Inst. Moore. G. C. Proc. tests data. K. Ed. Vattenfallsstyrelsen.. E. Hrennikoff. 1956). 1934). W. Liege (1930). curing. Concrete Inst.S. non arme (translation from Swedish by the Lab. B. Am. E. changes of concrete masonry units.. Concrete Inst. No. Hatt. Ind. Einfluss G. Mercadi^. p. 2 (1946). Maney and M. Am. Proc. 195. 5 des deformations elasto-plastiques. 293 (Aug.. Civil Engrs. Nielson. batiment trav. Acad. 541 (1940). 108 (1958). Svenska Betong Ed. Fuller and C. p. Proc. Concrete Inst. Soc.. 1955). The physical structure of B.. sand gravel concrete. O'Heir. E. Station. 3 (1915) A study of column . 61. Lynam. L'Acqua.. la plasticite et le fluage des 31. Plas. Kellermann. G. Hummel. 229 (Jan 1946). A. Proc. Betongens Krympning og flytning. A. Purdue Univ. et du beton arm6 soumis h des efforts r6p6t6s. B. The effect of testing speed P. Highway Res.. Cong. Portland cement mortar and concrete. S. Levi and G. 39 (1925). Mech. B6ton. Eng. Charpentes.-Dec. Concrete Inst. of Minnesota. Incremental compression test for cement F. Con. 148 (1952). Trans. Vom Schwinden zementgelundener Massen. H. W. Giertz-Hedstrom. Researches in Evidence that shrinkage dominates time yield. Am. Armyan. Shrinkage and cracking of cementive E. (1934). Int. 84 (1943). Experiences sur la theorie du fluage du Univ. foreningens (Oct. Peabody.. Testing G. Ing. T. p. S. A. materiaux de construction (translation from Russian F. 47 (Sept. 80 design. Lord. Proc. A. 1949).. W. Am. No. Con- on small movements in concrete.. R. 79 (1953). Eng. 1955). A. J. Eng.S. Menzel. F. Am. Proc. Tests of large reinforced Testing Materials. I. T. Nauk. Ill (March 1956). A. Bull. Strength and volumn change of steam cured of prestressed concrete. No.M. K. 1942). Press. Akad. Am. S. Shrinkage and creep of concrete. 36. 3. Eyrolles. calcestruzzi. 209 (June 1955). Concrete (Oxford Univ. Glanville. G. Am. p. p. Chem. Cements. Am. Czaban.S. Izv. No. Am. St. The theory of concrete creep. Shrinkage and time effects Moscow (1955). Shrinkage of concrete as a factor in compressive B. Swedish. Volume changes in 2. Am. Svind hos beton (In methods of measurement and results). tion and moisture retention of concrete during 15. 36. (July. Notes on the effect of time element in loading 485 (1948). W. Discussion.. II retiro e lo scorrimento plastico dei Proc. 22. C. Mongiardini. sur la resistance. Smith. Proc. Theories of creep of concrete. A. Concrete Inst. M. R. Meyer. Fonts et forced K. rech. Am. 24. D. L. 150 (1921). Ernst. C. Louis (March 1916). Betons. p. Betong. V. Concrete Inst. Lagaard. ibid. 129 (1957). Shrinkage and plastic flow C. 1954) (description of E. 123 (1937). 2. publ. Stress increases in com- des Zuschlages auf Formanderungen und Spaimungen pressive steel under constant load caused by shrink- im Betonwerkstein. ZNIPS. 252 (Dec. Melville and R. S. lightweight concrete units. 1950). Lofquist. tech. 40. L'evolution future des propri^tes des Arm. Am. Proc. Neville. concrete structures. 1939). W. 6. 380 (1936). E. E Freyssinet. L. Freudenthal.. Meyer and K. p. Versuche uber Elasticitat. M. Proc Am. Kalk-Gips. high stresses (in Russian). 1954). Kalousek. F. Etude sur les anisotropy on the creep of concrete" (in Russian).. 5th Cong. Soc. and Contr. Symp. Ziems. Science Betons. Travaux. Congress Appl. defomations lentes des ciments en retrait. 44. Le Camus. Time tests of concrete. Plastic behavior and creep of cement mortar. age. Zement. Proc. 12. techniques du b^ton. steel stress. Prehm. Effect of plastic flow and volume changes on concrete slab. Fluage. Pr^contrainte. concrete. English summary). Richart. RILEM Bull. Stockholm (1946). Betrachtung iiber Gefiigespannung im Beton. Ost Betonvereins p.. A H. R. Plasticity. Nauk. prestressed concrete. 670 (Aug. 9-10. mittee 313. Creep effects in the analysis of rein. R.. J. R. L. (London) (1947). 50 2z5 A. Proc. Exp.. Am. Proc. Bdtiment trav. 1021 (1941). 52. Concrete Inst.. Ness. 10. publ. J. 125 (Nov. hydrated cements. F. G. 42. Proc. K. HolHster. ciments. 59 (1957). No. Relation of shrinkage to moisture content in method of testing). McMillan. Ferguson. J. concrete slabs. No. V. Lea and C. Am. Concrete under sustained working loads. A. No. 1604 (April 1958).. Growth and Movement in Portland Cement (1916). p. batiment trav. Morris. crete Inst. Kalousek. Lee. Beton. Proc. 505 (1938)... Soc. Howard and D. 24 (1945-46). b^timent trav. 45 (1917).S. Travaux. Symp. Nordisk Betong. 1954) (very interesting study on p. Lyse. Work of the Building Research Station building over a period of one year. concrete block. Soc. publ. 179 (1943). p..R. Am. Beton og Jernbeton W. Eng. Dunod. p. Paris (1936). Landbo.. Ann.S. 324 C. G. Some rheological properties of concrete (in Proc. 1379 (Dec 1956). Stockholm. Recherches sur le co'nportement du beton Aug. Concrete Inst. Cong.. Quelques notions relatives a la theorie I. L. Paris (1951). C. Neue Erkennisse uber das Schwinden des seiner Messung und seinen Auswirkungen. Paris). Concrete Inst. Liege. A. Soc. A. H.R. structure of hydrated components: tobermorite and Tests of a new method for evaluating volume xonotlite). Comparative shrinkage data for heavy and on strength and elastic properties of concrete.. (Sept. Lorman.. Method of designing a rectangular reinforced C. M. Tekniska Meddelanden. Proc. Am. Civil Engrs. (Jan. Am. First Jap. inst. (June 1930). Amos. ticitat und Schwinden von beton. Plasticity and temperature deformations M. Materials.. Board.R. C. 53. 24. Concrete Inst. S.. Proc. Berlin F. 33. V. 947 (April 1957) (accelerated Saxer. Goldenblat. Karapetian.

(1956). 15.. G. p. p. Testing Materi- H. Williams. 233 (1920). Concrete Inst. 445 (1922). Beton-u. H. 32. Rodriguez.. Paez. Peters.A conception of the creep of unreinforced (translation from Russian by the Lab. Thomas. 52.. No.. Wilson and R. Young. Proc.. A rational conception of creep in concrete. Einfiuss der Nachbehandlung auf die am A. 137 (1934). genio civile (Sept. Torsional resistance of concrete. I: Lyse. Betonstein argamassas e betoes por efeito da sua contracgao Ztg. p. in Mechanical Properties of Non Metallic Stahlbetonbau. Trav. Research. Kluge. Research. Am. Shibata. 1953).. 45. 69 (1933) Investigation . Wagner. Some determinations of the stress defor- Betons auf den Spannungsverlauf. Congress for Large Dams (1951) Soc. Soc. Fundamental factors in the drying shrink- Edif. M. 1959). 270 (1934). The Structural Eng. W. R. 803 (Feb. of K. G. 43 (1938) Concrete . p. G. Station. Trans. Shrinkage and plastic flow T. Batiment concrete and an estimation of the limiting values. in Engineering Reinforced concrete column investigation. W. 43. Fonts et Charpentes. 145 E. Symposium on shrinkage scale method. Am. Soc. Chem Ind. Am. Bond creep and L. K. Ross. de Sousa Coutinho. Plastic flow in plain and reinforced concrete Lisboa (1954). 91 (1935). The mecha. Structures. Concrete Inst. Peltier. 181 (1934). 8. (Jan. Tests of reinforced on the vibration of concrete. Determination of creep strain of concrete Proc. Am.. Slater. Paris). Oberti. Am. L. p. 324 (1916). Concrete Inst. Inst. and Contr. Rianitzine. (June 1954). Civ. pub.. Deformazioni anelastische del calcestruzzi della S. and C. H0gskole Avhandl. No. under sustained compressive stress. 309 (Sept. Bauzt. F. plastischen Verformung von Beton. Ohro and T. Concrete Inst. 33. 54. M.. Flow of concrete under sustained load. F. H. 44. En'g. Wettstein. 1957). Bull. y Edif.-Oct. G. Pub. Pub. 462 (Oct. The Structural Eng. Inst. 42. E. Proc. publ. creep and shrinkage of H... Reiner. J. Tech. Proc. Engineering. Univ. "Effect of temperature on the stress-defor- A. and D. Prelim. Proc Am Bull. Sergev.. Fonts et Charpentes. 27. 53. Staehle. 20.. continuous beams.. Chem. B. 27.. J. R.. 142 (1938). 33 (1939). R. Proc. span reinforced concrete arch bridges. Pub. Am. E. E. Reologia del hormigon. H. No. Ernst. Torroja Miret. Pub.. D. Seaman. 1939). Smith. S. RILEM Symp. Trans. Am. T. E. C. A. Sci. The Hague. Miinchen. 149 (1936). creep data. Kemp. I. Plastic reinforced concrete flow of thin concrete. E. titles). On the flow and extensibility of concrete. Proc. H. 5th Cong. Exp. 1934 (1935). Proc R. Richart.. Pub. city et le fluage des mat^riaux de construction F. (London). A. p. Shrink. The effect of shrinkage on the deformation of F. Am. Eng. Kreidler. Laboratory tests of three Concrete Inst. R. Butterworths Sci.. Long time volume changes (1958). Extensibility and cracking in concrete. Prelim. in Recherches sur la resistance. 1.. age of concrete block. Torroja and A. 99. G. Washa. Proc. Woolson. Civil Engrs. No. Publ. E. Moscow (1955). Fluck. Eng. Soc. The me- Munchen (1958) (will be published in RILEM chanics of plastic flow of concrete. No. W.. Les bases de la th^orie g^n^rale du fluage mation of concrete".. Lisboa. Am. London A. Brown. C.). Proc.. Inst. 279 (1932). plasticizing agent on the creep and shrinkage of 0. 5-6. Las deformaciones del hormigon per F. Madrid (1949). 81 (1939). W. B. Proc. Paez. U. The effect efecto de las cargas. nism of drying of solids. Am. Plastic flow in concrete arches. Richart and T. London (1949).. 1. 3 549 (Jan. Nurse. Straub. p. Am. Versuche zur Bestimmung des Einflusses der als. Lisboa. ZNIPS. Der Einfluss des Deformationseigenschaften des G.. 1343 (1934). Whitney. 1955). Concrete Inst. Das kriechen unbewehrten Betons. 16. No. J. Proc. Some remarkable tests indicating "flow" of betonbau (March 1950). L. 122 (1953). Proc. Schwarz. 119 (Feb. Schweiz.. Civil Engrs. 11 (Jan. I. 459 (1937). 155 (1956). Am. Prelim. 57. Gior. A rheological model for concrete. Essais de fiuage rapide de batons. Contribution for knowing the influence of a Tek. 35. Deformation and Flow. The plastic flow of concrete. C. Ind. E. of plastic flow in rigid frames of reinforced concrete. concrete under sustained loads. compression. Brittle Materials. RILEM Symp. concrete. 37 (Jan. 475 (1949). A fissurabilitade dos cimentos Schwinden beteiligten Wirkungskrafte. chap. Richart and R. W. (London). Engrs. Fonts et Charpentes. y E. Shuman. Dehnung und Verfestigung. F. Determination of the drying shrinkage of age and creep of concrete. Concrete Inst. 33. Moscow (1955). Concrete Inst. 459 (1905). Concrete 237 (1947).. E. 1958 (will be Berlin (1958) (very numerous references: 208 published in RILEM Bull. dimensional stressing. ZNIPS. Richart. K. No. S. concrete. Plain and reinforced concrete arches. R. Deformation of steel reinforcement during diga di Osiglietta. Vogt. 42 (1942). H. Lyse. Mag. Concrete Inst. in Recherches sur la resistance.. Ohio State Univ under slowly increasing tensile load. C. (translation from Russian by the Lab. and after construction. No. p. Taylor. concrete columns under sustained loading. Eng. 1930). Pub. Am. 229 (1946). T. A. Stability of concrete structures submitted to shrinkage eff'ects on reinforced concrete... 677 (1931). cement mortars: a small and cracking of cementive materials. Rapid and long-time tests 5th Cong. 12. mation relations for concrete under repeated and p. 5th Cong. 77 (1957). Proc. Am. R. G. Seed. J. L. la plas- M. Engrs. No. Newitt. Concrete Inst. Am. la plasti. Chem. Staehle. Lewis. Soc Civil 1946). continuous loadings. Concrete Inst. Inst. 479 (1932). arches. Beton-u.). 195 plastique. M. 571 (1916). Shank. 2. 27 (1949). News. 28. Eng. Creep and shrinkage in reinforced concrete C. ticite et le fluage des mat^riaux de construction XII (1949) On volume or isotropic Flow as exempli- . 1956). Quirke. Soc. 248 (1955). La d^formabilite du b^ton sous Faction de la the 4th Int.694 . H. . tlber franzosiche Versuche zur Frage des Illinois Bull. F.. Washa and D. Am.. Pearse. Am. Tec. BStiment fied in the creep of concrete. concrete under pressure. 30. p. Peabody. 314 (1937) . Concrete Inst. Deformation of concrete under loading. Appl. Proc. of Concrete Res. F. L. Ostlund. Proc Am long time loads. Theuer. of prestressed concrete. Concrete Inst. p. London. A. Tal. Reinforced concrete column investigation. Proc. W. Concrete Inst. Probst. Lisboa.. 28.. Inst. Oliver. Construe. 4.. (1956). 1947). Rao. The Structural Eng. 30. White and H. . Heitman. Nac. A. The elasticity. Butterworths Sci. Plastic flow of reinforced Experiments on the creep of concrete under two. The creep of portland blast-furnace cement G. Stahl. F. 47. J. and G. Richart. Zeit auf Festigkeit und Verformung. slabs. Constr. on reinforced concrete knee frames. (London) (1947). Extensibility of fresh concrete J.. Lab. Staley and D. structures. Can. of the 4th Int. 18. E. and T. Congress for Large Dams (1951). of R. (in Portuguese). Norges P. trav. C. Testing Materials. NBS. of Portland cement bars. W. Riisch. pub. Olson. p. (1937). Paris).. Nordisk Betong.

. Specimens Molds Form and Dimensions Figure 2 shows the mold employed. and so it is possible to reach the final equilibrium in a short time. because it seems a proper method for studying among other things the relationships. 1959). All the molds are made of stainless steel in order to pre- The specimens are pipes with the dimensions vent any corrosion. of the the anomalous failures observed when these pipes The side wall of the molds is a tube divided in are made of constant wall thickness throughout two halves lengthwise. which amounts to 9 times that of the 4 X 4 X 16-cm prisms and 30 times that of cylinders 15 cm in diameter and 30 cm in length. fineness. we use mortar specimens having the form of a tube with a very thin wall. The surface per unit of volume of these pipes is 10 cm^/cm^.) . de la Pena Introduction J The shrinkage and creep of the gel constituted by cement and water are better studied when: (a) No gravel and coarse sand are present. I because the stones restrain very much the free : deformations of the paste and put it in tension. and The specimen has . Figure 1.) creep. (Dimensions In mm. The results of the first tests carried out with these pipes were submitted to the RILEM Colloquium upon Shrinkage and Creep of Mor- tars and Concretes held in Munich. which would be more harmful given in figure 1. Dimensions of pipe specimen. 1 (b) a thin section. In the testing procedure described here. Discussion E. These two halves are their length and tested in compression. Specimen mold. Just a few more results are now available. . because the differences in humidity and temperature be- tween its core and surface give rise to internal stresses. of cements and their shrinkage and (Dimensions in mm. Using pipes of such a high specific surface speeds up the process of exchanging water between the specimen and the surrounding air. The paper was published in Bulletin No. etc. Torroja and C. 3 of RILEM _4b (July. The walls of the tubes are a than in others because of the thinness in this case little thicker at their ends in order to prevent specimen wall. Figure 2. but _5a we think the procedure may be of interest to the attendants at the Symposium.between the composition.

follows the shortening of specimen. block that rests on the upper end of the pipe to gently pressed and tapped with the tool and made descend as the pipe shrinks. . immersed in water at 20 ±1 °C until just the tain a flow of 40 ±5 percent on the RILEM flow moment of their test. In the tests referred to in this paper we have Curing used a common portland cement and a very pure siliceous sand in which the particles ranged from The mortar pipes are left in their molds for 24 mm hr in the same room where they are molded. one on each end