CEMENT and CONCRETE RESEARCH. oI. 4, pp. 13-29, 1974. Perganon Press, Inc.
Frintec in the United States
THE EFFECT OF SAND GRADING Ox THE
CALCIUM SILICATE BRICK REACTION
wok teen ay
creek TRASEE MP PEE eed gory any
Spats, oy Gt tan
(Communicated by HF. Ws Taylor)
‘Received duly 24, i973
Calciua silicate brick specimens withdifferent sand
gradings were autoclaved at 1.20 W8/a? gauge steam
pressure for periods of 2, 4, 6) 8, 12 and 16 hours
Ind tested for compressive strength and drying
inkage. Their composition has been investigated
by essy diffraction and differential thermal
Results were in agreoment with existing knowledge
that the reaction was faster, the higher the content
of fine sand in the speciaen: The development of
Lobermorite as indicated by X-ray diffraction was
not related to any other properties of the specimens
cxamined. Drying shrinkage was correlated with the
‘amount of hydros!licate formed but the strength
(ose saxtmun w with a gap grading) w
ore probably r 1c dogree to which the
Conenting saverial filled the voids:
Kalkeandatein-Probestilcke ait verschiedenen
Keravertel lunges wurden bei 1210 atll Daspfdruck
wiirend Perioden von 2, 4, 6, § 12 und 16 Stunden
[Fehiirtee and dann auf brackfestigkett and
Trockenachwindung geprife. Ihre Zusamnensetsung
wurde mittela RBntgenbeuging und Differentialtherm-
‘nelyse unteraueht-
Die Ergebniase stimmton mit der bekannten Erfahrung
iberein, dage die Unsetzung umso achneller vor gich
geht, Jo grosser der Feinsandgehalt des Probestiickes
ine.” Diewdurch RBntgenbeugung angezeigte Entwicklung
den Tobermorits var mic kesner der anderen Eigen
‘aften der untereuchten Prosestiicke verbunden” Die
Trockenschwindung war nit der Menge, des neugebiideten
Hydrosilixate verknipee, aber es ist wahracheinlicher,
Gass die Featigheit (deren Maximum mit einer
Kornvertefiung ohne Mittelkorn erzielt wurde) vor dem
Auagaes abhing?, womit daa Bindemittel die Hoblrdune
erfiliie. .1" Vor, 4, Mo, 1
SAND LIME BRICK, REACTION, SAND GRADATION
aTRopverroy
In experizents on sand grading there have been two main
approaches to defining the size distribution. The most complete
definition is by means of grading curves. Tt has been found
Gieicult, however, to translate these into simple parancters
against which the properties of speciens can be plotted.
Although natural sands should give straight Line distributions
(and hence capable of being defined by two paraneters) on log-
norsal or Rosin-Raanler nets, they do not alvays do s0, and the
curves are further complicated when blends of
Inds are studied.
‘The second approach is to consider the sand in toras of
three size ranges, normally defined by outs at only two sieve
sizes, and to plot the results on triangular graph paper. Despite
its lack of precision, thia approach has been extremely useful in
Laboratory experiments and has given a method by which independent
investigators have arrived at similar results,
From the practical point of view, improvement of sand
evading is best achieved by blending and to be effective this
needs sands of different gradings than the original. It is
unlikely that i would be necessary to blend more than three
components together, 40 that this method of representation is of
considerable convenience.
Previous papers from the author's laboratories (1,2) have
considered sands in terms of their fractions above 0.6 mm, between
0.6 and 0.15 am and below 0.15 aa. The limita for the middie
Fraction were chosen because aost calciua silicate brick making
sands in England have the bulk of their material within this
range and it was also known that the higher surface area fraction
below 0.15 am was sore reactive and had a large effect on the
properties of the bricks, Thus, blends plotted in terms of these
Eractions tended to emphasise the effect of particles coarser
(e.g. crushed flint) of finer (e.g. fine sand) than normal.
German workers have algo used similar triangular diagrans
but with the Limits at 0.63 and 0.10 mm, These are of the sant
order, the precise values in each ease representing sieve sizes
normally used for concrete aggregates in the two countries.Vol. 4, Wo. 1 15
SAND LINE BRICK, REACTEON, SAND GRADATION
The Limite chosen are satisfactory for normal sands, but are
not necessarily suitable for silicate concrete where fractions
such ax 1.0 - 0.1, 0.1 - 0-01 and 0,01 - 0,001 mm, as used by
Chernyi and Tonilina (3), aay be more appropriate.
In the present work, the autoclaving time as well as the
sand grading, was varied. It was thus necessary to present the
results ina series of triangular diagrams, one for each
autoclaving time. The tine factor was included to see whether
differences in grading had a major effect on the rate of reaction
in the bricks.
SPECIMEN PREPARATION AND TESTING
‘The specimens were prepared from commercially supplied
fractions of washed quartz sand (as used in the preparstion of
standard mortar bars (4))with nominal size ranges of 1.2 - 0-6,
0.6 - 0.3, 0.3 = 0.15 and 0-15 - 0.09 mm and a white hydrated
Lime with an available CaO content of 65.4%.
AlL mixes contained 7% by weight of Ca0 based on the total
available calcium oxide plus sand content. The grading of the
snd was altered by changing the proportions of the three size
components of over 0.6, 0.6 - 0.15 and below 0.15 mm. The
0.6 ~ 0.15 am component was composed of equal proportions of
0.6 - 0.3 and 0.3 ~ 0.15 mm fractions. Individual gradings ss
shown in Figure 1 are represented by letters so that they can
be referred to in the text.
The water content requiresent was deterained empirically
so that it was at a maximin without damaging apecimens in the
autoclave. The values shovn in Figure 1 were deterained by a
preliminary experinent.
The components were mixed dry in a Hobart mixer for one
minute followed by four minutes with the requisite amount of
water added.
15 test cylinders 51 mm high by 38 am dismeter and four
prisas 152 x 38 x approx. 32 mm moulded at 36 MN/a? vere pre-
pared from each mix by a method described previously (1). Six
cylinders were used for compressive strength testing, #ix forVol. 4, Wo. 1
SAND LIME BRICK, REACTION, SAND GRADATION
coarse
oe Sem
sepium ine
‘SAND SAND
015-06 ‘below O35 mm
FIs, 1
Teiangular diagran showing the biends of sands used in the
preparation of the specizens. Gradings have been lettered
Borvreference. ‘Open figures indicate the percentage water
contenes of the mixes.
Frost resistance testing and the remaining three were spares
analysis. The four prisas were used for
drying shrinkage deterainations,
‘The specimens were autoclaved at 1.20 MN/a? gauge steam
pressure for perioda of 2, 4, 6, 8, 12 and 16 hours. Nine sets
Of apecizens were autoclaved at a tine 0 that tuo separate
autoclavings were required for the complete range of gradings
for a given autoctaving tine.
compressive strength tests were carried out on specimens
which had been soaked An water for 24 hours. The load was
Increased at a rate of 21 KX/ain and the results were based on
means from six cylinders.
The frost resistance test was based on the method described
in D.T-X. 108 (5) in which specimens were dried at 105°C, soakedVol. 4 No. 1 7
‘SAO LINE BRICK, REACTION, SAND GRADATION
An water and subjected to 25 freeze-thay cycles. They were
‘then tested for compressive strength and those whose strength was
ed
over S0f of the original strength were considered to have p.
the test. Bulk density and water absorption were deterained
by weighing the specinens at the appropriate stages in the frost
resistance test and determining their volunes by displacenent in
The drying shrinkage test was carried out according to the
procedure in B.5.187: 1955 (6) in which specimens were dried at
50°C over a aaturated calcium chloride solution. All specimens
autoclaved for a given period were tested together.
‘The phase analysis of the samples was carried out by X-ray
powder diffraction and differential thermal analysis, ax
described elsewhere (7,8). The K-ray analysis wes carried out
with 3 hour exposures on an 1148 mm diameter powder camera
using nickel-filtered copper radiation at 30 KY and 20 mA. The
differential thersal analysis vas carried out on
Standata 615 apparatus with a heating rate of 10°C/ain.
Stanton
RESULTS AND DISCUSSION
‘The results have been presented in the form of triangular
Alagrans. Th order to facilitate visual comparison between
Aiagrans, and because of space requirenents, in sost cases only
high, medium and low ranges of values for each property have
been indicated.
Differential Thermal Analy:
Figure 2, showing D.T.A. curves for specinens autoclaved for
specimens autoclaved for § hours, illustrates the form of the
traces encountered. The caleiun hydroxide, quartz and high
tomperature hydrosilicate peaks are easily distinguishable. A
small carbonate peak occurred in all samples and in many there
was & broad exothermic peak at about 200°C which probably
Andicated loss of interlayer water from the hydrosilicate.
Fros earlier work (7), it vas assumed thet the size of the
exothernic hydrosilicste peak increased with the anount of
hydrosilicate present, at least in the carlior stages in the(a
PRRFEL
THT |Vol. 4 No. 1
SAND LIKE RI
REACTION, SAND GRADATION
HYDROSIUCATE PEAK. HEIGHT
¢
s
a A
‘ ‘
wr A wh fA
tec tm
1D 50-100mm
Beet
to. 3
‘Triangular diagrans showing results for the hydrosilicate peak
height in terms of proportions of coarse (C), medium (M) and
Fine (F) sand in the mix
LHYOROSIUCATE PEAK TEMPERATURE
or 50°C
Desa
Bi scree
Fie. 4
Triangular diagrans showing results for the hydrosilicate peak
temperature in terms of proportions of coarse (C), medium (M)
fané fine (F) sand in the mix2 Vol. 4, Wo. 1
‘SAND LINE BRICK, REACTION, SAID GRADATION
reaction. The growth of these peaks is illustrated in Pigure
3, where it can be seen that the size of the peak increased in
size faster, the more fine sand there was in the sample. These
peaka reached 2 maximum height of 120 - 150 mn after which no
significant growth occurred.
tthe change in the form of the hydrosilicate peaks has been
described previously (7,8). Initially, a broad ill-defined
exotheraic peak at about 870°C is foraed and as the reaction
proceeds the temperature of the peak drops, a double peak occurs,
and then @ sharp well-defined peak is formed at about 830°C. The
temperatures of the peaks in each trace are indicated in Figure
4, their behaviour following that of the peak height.
A histogram of the temperatures of the hydrosilicate peale
in Figure 5 shows a preponderance of peak temperatures at £30°C,
‘This would tend to indicate some very distinct change at this
AIG.
Distribution of temperatures
Of the exothernic hyarosili-
Sate peaks.” Unshaded
portions of the columns
Tepresont subsidiary peaks
20 on larger peaks.
8
of
sho #70 240 abo azo m0 ATOVol. #5 No. 1 a
‘5980 LIME BRICK, REACTION, SAND GRADATION
temperature and, by inference, a distinct composition of the
C-S-H phase. By contrast, the temperatures of the peaks in the
range 880 - 830°C were more evenly distributed suggesting that
before the 830°C peak was formed the composition and/or the
structure of the hydrosillcate was very variable although previous
work had indicated that C-S-H(I) could be distinguished from
C-S-H(TE) according to whether or not the peak temperature vas
below about 850% (9).
The occurrence of the low temperature endothermic tinter-
layer-water' peak (not plotted on triangular diagrans here) wa
less consistent than that of the other peaks. In general, st
tended to be sore prominent and had s higher temperature when
there was 4 well defined exotheraic peak.
The presence of free lime, as indicated by the D.T.A.
traces, ie shown in Pigure 6, The abgence of calcium hydroxide
PHASE ANALYSIS
¢ ¢
2h an
?
c
tm ?
Ti ee me by xd, Bp tobermorte ciftetion nes
[Tl] tee ime by dite. } no cashor free ime
EE] coh citrction ines troction Ue
Fic. 6
‘Triangular dlagrana showing the phase analysis remults in terms
Of proportions of coarse (C), megium (W) and fine (F) wand am2 Wor. 4, fo. 1
‘SAID LIME BRICK, REACTION, SAID GRAOATION
peaks for gradings Q and R at 2 hours autoclaving is not indicated
on this diagram, however. The contours for the height of the
Pree Lime peak tended to follow the inverse of the contours for
the height of the hydrosilicate peak
Double peaks occurring #5 shoulders on the Lime peak and
attributed to C,SH(A) (8) vere recorded in isolated cases,
‘suggesting that this phase can sonetines occur.
Rav Diffraction Analy:
Not all the K-ray results were in entire agreement with
those from the differential theraal analysis
ALL films showed strong Lines due to quartz (i.e, the
les) and the only other components
ected sand in the a
recorded were calcium hydroxide and hydrosilicates of the
Lobermorite group. On no Films were both Lime and hydrositicate
Lines recorded together.
Im sone cases, neither calcium hydroxide aor hydrosiitcate
Lines were recorded. This did not mean that there was no binder
present; merely that most of the Lime had reacted but there vas
‘an inguffieient asount and/or crystallinity of the hydrosilicate
to give diffraction lines. To label such hydrosilicate as
tamorphous! is probably incorrect. These phases lack erystalli-
nity and represent 4 small proportion of the sample. Their non~
identification is probably due to a limitation of the technique.
‘the results for free Lime (in Figure 6) indicate how slowly
it reacted with the coarse sand and, in general, bear out the
ples after 2
D.T.A. results, The distribution of Lime-free
hours autoclaving as deterained by X-ray diffraction, however,
id not agree with that detersined by D.T.A.; gradings Ly M and
X showed no calcius hydroxide Lines on the K-ray Fila while by
D.T.A. gradings Q and R were the only ones without a Lise poak.
‘The lack of calcium hydroxide Lines on the X-ray fila coineided
with the appearance of C-S-H Lines at these gradings. In all
other instances the D.T.A. resulta for free lime shoved the sane
trends as the X-ray results but had » higher sensitivity.
The C-S-H lines first appeared at gradings Land M afterWor. 4, Ho. 1 2
SIND LINE BRICK, REACTION, SAND GRADATION
2 hours autoclaving and after § hours autoclaving the gradings
showing C-S-ll lines formed a band across the diagram from
gradings L and 0 to grading G. The absence of C-S-H lines from
2 = 12 hours autoclaving for gradings Q and Rwa unexpected, 23
the D.T.A. results indicated the formation of C-S-H(I) and both
atrength and shrinkage increased over this period. The absence
of these Iines over five auc
considered co-ineidental and attributed to bad saxpling or sone
auch cause. After 2 hours, no free Line was recorded for the
ive autoclave cycles cannot be
gradings and it aust therefore be considered to have reacted.
Tt is concluded that the al
due to a structural effect which did not affect the thermal
properties of the sample.
nce of these diffraction lines wae
After 12 hours tobermorite appeared in samples whose gradings
were represented by a band acroas the triangular diagraa and by
this time most other gradings, with the exception of thos
containing a high proportion of fine sand, gave C-5-H lines on
the filme. The toberaorite lines were particularly well marked
For gradings B, F, G, J and K.
The occurrence of tobermorite for this purpose was defined
when the diffuse C-S-H line at approximately 3% split into a
doublet with an extra Line at 2.98 R (9,10). This change wee
tise accompanied by en increase in intonaity of the HLA Lines
The tobermorite, therefore, representa » nore crystalline
variety of hydrositieate.
Scanning electron aicroscope studies on other caleium
silicate brick specimens (not yet published) suggest that the
appearance of the extra diffraction line is due to well-formed
crystals growing into the larger void spaces in the specimen.
These crystal add nothing to the strength of the speciaens but,
because of their better crystallinity, give stronger X-ray
reflections which do not necessarily represent increased
quantities of binding material,
Brving Shrinkage
The drying shrinkage results in Figure 7 showed « consistent.
pattern with the highest shrinkage values for a given autoclaving2 Vol. 4, to,
SAND LIME BRICK, REAL
ON, SAND GRADATION
DRYING SHRINKAGE
1 beiow o120%
TB oca0-o0asx
Th or ons:
FIG. 7
Triangular diagrans shoving the drying aheikage ceaulte in
teraa of proportions of coarse (C), wedium (N) and fine (F)
sand in the #13.
aux waree waren
oenty ABEORFTION ABSORPTION
by weight by volume:
5 ¢
+ & Se
[C1 blow 1850 kg/m? below Okg/100kg below 20ms/h00m?
EE] ersotcin? rosie 20-25 moon?
Hover 1950 kim? ‘over 15 kg/00kg ‘over 25 3/1003
‘riangular diagrams showing bulk density and water absorption
Tomita in terms of proportions of coarse (C)y medium (W) and
Fine (F) Send in the aix
1Vol. 4 No. 1 8
SAND LIME BRICK, REACTION, SAND GRADATION
tine corresponding to the highest proportions of fine aggregate
as reported previously (2). The shrinkage increased with
autoclaving time up to 12 hours but appeared to decrease after
16 hours (this decrease requires confirmation and could possibly
be attribuced to incorrect drying oven conditions).
‘The drying shrinkage (other than for the 16 hour results)
appeared to parallel hyérosilicate growth as indicated by the
height of the hydrosilieate peak on the D.T.A. traces.
bulk Density and Mater Absorpt:
‘The values for these properties, shown in Figure 8, were
based on means for all apectnens with a given grading. Maxieun
bulk density occurred when there war a high proportion of coarse
aggregate with the finer sand and the lime f4iling the voids
becween the coarse particles sn the green bricks. The water
absorption, expressed both by weight and by volune,, showed
similar relationanips with respect to the grading
After 4 hours most specizens had a satisfactory frost
resistance (see Figure 9) but specimens containing only coarse
aggregate alvays disintegrated during testing.
The triangular diagrams in Figure 10 show that in the
Anitial stages of the reaction, the strength increased fastest
in specimens with high contents of fine sand, 10 Mi/a” being
attained within 2 hours in many instances, The major pattern,
as reported in previous work (1, 2), could be seen after 8 hours
autoclaving where strengths of over 20 N/a? were recorded with
gradings E, T, 3, Land M. The maximum recorded strength wae
26 MN/m®. These strengths are lower than those reported
previously, because of t)
lower moulding pressure used in the
current investigation.
‘The actual maximim strength values after 2 and 4 hours
autoclaving occurred with gradings L and 0, whereas at 6 hours
and longer autoclaving times the maxima tended to occur at
gradings H and L. Thue, for practical purposes, the optima26 Vol. 4 Ne. 1
SSMIO LIME BRICK, REACTION, SAND GRADATION
A
AAA
Rost RESSTANCE
>
Fon
Fis. 9
Telenmular diagrans showing frost resistance results in terms
ek proportions of coarse (c), mediua 0) and fine (F) sand in
the aie " « a 7
compressive stesnord
ove maint
1B w-r0nnint
Hh cwr200in?
FIG. 10
‘Triangular diagrans shoving compressive strength results in
Teeaso# proportions of cosrse {e), sodium (¥) and fine (F)Wor. 4, to. 1 a
‘INO LINE BRICK, REACTION, SAND GRADATION
grading does not vary significantly with the period of autoclaving.
Differences in strength remults between 8, 12 and 16 hours
autoclaving were not great and the variations do not necessarily
represent aaxina and minima of strength with time.
Specimens containing only coarse sand (grading A) vere very
weak, @ strength of 4 MN/a? never being achieved, whores:
arading B gave strengths up to 14 MN/=?
Ie the triangular diagrams for compressive strength are
studied it will be seen that the area of maximim strength
corresponds neither to that of maximum bulk density, nor to
maximin formation of hydrosilicate as indicated by differential
theraal analysis but ies between these two areas.
Tt seons probable, however, that the strength is related
An gone way to the degree to which the cementing material fills
tthe available pore space in the specimen. The data in this
paper is inauéficient to test thie directly. If, however, the
ratio of the height of the hydrosilicate peak on the D.T.A. trace
to the water absorption of the epecimen is taken az a measure to
‘nM eo fe. mm
/,
aa aN a +
en
ao,
‘Triangular diagrane shoving the ratio of the hydrositicate
peak height to water absorption in terms of proportions of
Soarse (C); sedium (M) end fine (F) sand in the mix.2 Vol. 4, No. 1
SAV LINE BRICK, REACTION, SAND GRADATION
which the void space is Filled with conenting material, then
there is sone correspondence between the contours for this ratio
(Jn Pigure 11) and those for compressive strength.
Te is not clear how the X-ray results for C-S-H and
tobermorite are related to the strength. The patterns on the
triangular diagrams show some similarities to those for strength,
Both may possibly be related to void apace in the specimen and
the degree to Which It is £Lled, if the hypothesis of Gundlach
et al (11) is relevant in this situation.
concLustoxs
1. Differential thermal analysis results are in accordance
with general experience that the reaction proceeds
Eastest when there is more fine sand in the specimens.
The size of the hydrosilicate peak increases with the
consumption of free lise and the peak tenperature indicates
a constant phase for the end product
2. Koray diffraction analysis showed maximum growth of C-S-
Linea (and, after 12 hours autoclaving, tobermorite Lines)
for gradings represented by @ band across the triangular
diagram. Specimens with a high content of fine sand
showed a rapid reaction with the lime but did not show
any C-S-i diffraction Lines on the X-ray fflm.
3. The drying shrinkage appeared to be related to the amount
of hydrosilicate formed, as indicated by differential
thermal analysis.
4. Maximus strength was achieved with a gap grading and
corresponded to neither maximun bulk density nor to maximum
hydrosiicate formation. Tt was probably related to the
degree to which the hydrosilicate cementing material filled
the voids, The optiqua grading for strength was found to
be the sane for different autoclaving times under practical
conditionsVol. 4 No, 1 2
SSAIO LINE BRICK, REACTION, SAND GRADATION
ACKNOWLEDGMENTS.
‘The author wishes to acknowledge the contributions aade to
the work by Mes. G.R. Smith, M. Baker, A.G. Cruttenden and G.R.
taylor,
This paper was presented at the Third International
Symposium on Autoclaved Calcium Silicate Building Products in
Utrecht on 26th September 1973.
REFERENCES
1. GLB. Bessey and ¥.J. Purton, Autoclaved Calcius Silicate
butlding Products, pris. Society of Chealeal Industry,
Tondon (1967).
2, J.M, Coldrey and H.X. Lee, Chem. and Ind., (8), 260 (1970).
3. Tel, Chernyt and T., Tonilina, Sbornik auch, Rabot.»
Yauch, Ieeled, inst. Seroit: Materdalov Belorus, 8.5.R.,
(Sy 85 (959).
B.S, 4551: Part 2. British Standards Institution, London
(2330),
5. D.T.N, 108, Deutachen Normenausechuss, Berlin (1962).
6. BLS. 187. Britieh Standards Institution, London (1955).
JM. Coldrey and M.J. Purton, J. Appl. Chom., 28, 353 (1968).
8. M.3. Purton, J. Appl. chem., 20, 293 (1970).
9. Mad. Purton and J.M. Coldrey, Trane. Brit. Ceraa. Soe, 59,
03" (1970)
20. Ya. Mt. Butt and L.N, Rashkovich, Tverdenie Yyazhushchileh pri
Povyshennykh Temperaturakh, ind edn., p.Ol, Stroiizdat,
Moscow (1965)
11, H. Gundiach, £. WBretor and G. Rademacher, Tonindustric
Zeitung, 93, 107 (1969).