See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/258170344

Effects of phosphoric acid and phosphates on magnesium oxysulfate cement

Article  in  Materials and Structures · April 2013

DOI: 10.1617/s11527-013-0202-6

CITATIONS READS

18 779

6 authors, including:

Chengyou Wu Yongshan Tan

Qinghai University The University of Sheffield
41 PUBLICATIONS   284 CITATIONS    23 PUBLICATIONS   168 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Magnesia-based materials View project

All content following this page was uploaded by Yongshan Tan on 17 May 2016.

The user has requested enhancement of the downloaded file.

Materials and Structures
DOI 10.1617/s11527-013-0202-6
ORIGINAL ARTICLE
Effects of phosphoric acid and phosphates on magnesium
oxysulfate cement
Chengyou Wu • Hongfa Yu • Huifang Zhang
Jinmei Dong • Jing Wen • Yongshan Tan
Received: 24 December 2012 / Accepted: 21 October 2013
Ó RILEM 2013
Abstract Magnesium oxysulfate (MOS) cement is a
promising material for fire-proofing and insulation
applications. Here, we have studied the effects of
phosphoric acid and phosphates (H3PO4, KH2PO4,
K3PO4 and K2HPO4) on the setting time, mechanical
strength and water resistance of MOS cement. X-ray
diffraction was used to examine phase composition,
and analytical reagents were used to prepare samples
of the new phase found so that it would be examined
by chemical and thermogravimetric analyses. Adding
phosphoric acid and phosphates can extend the setting
time and improve the compressive strength and water
resistance of MOS cement by changing the hydration
process of MgO and the phase composition. A new
C. Wu (&)  H. Yu  H. Zhang  J. Dong 
J. Wen  Y. Tan
Qinghai Institute of Salt Lakes, Chinese Academy
of Sciences, No. 18, Xinning Road, Xining 810008,
People’s Republic of China
e-mail: wuchengyou86@163.com
C. Wu  H. Zhang  J. Dong  J. Wen  Y. Tan
University of Chinese Academy of Sciences, Beijing
100049, People’s Republic of China
H. Yu
Department of Civil Engineering, Nanjing University
of Aeronautics and Astronautics, Nanjing 210016,
People’s Republic of China
H. Yu
Department of Civil Engineering, Qinghai University,
Xining 810008, People’s Republic of China
•
subsulfate phase 5Mg(OH)2MgSO47H2O (517
phase) is formed, which is needle-like and insoluble
in water. Phosphoric acid or phosphates ionize in
solution to form H2PO4-, HPO42- and/or PO43-, and
these anions adsorb onto [Mg(OH)(H2O)x]? to inhibit
the formation of Mg(OH)2 and further promote the
generation of a new magnesium subsulfate phase,
leading to the compact structure, high mechanical
strength and good water resistance of MOS cement.
The improvement produced by adding phosphoric acid
or phosphates to MOS cement follows the order of
H3PO4 = KH2PO4  K2HPO4 [ K3PO4.
Keywords Magnesium oxysulfate cement 
Compressive strength  Water resistance 
Phosphate  Phase composition
1 Introduction
Magnesium oxysulfate (MOS) cement is a kind of air-
dried cementitious material, similar to magnesium
oxychloride (MOC) cement, which can be prepared
from aqueous solutions of magnesium oxide and
magnesium sulfate [1–3]. The outstanding properties
of MOS cement, including its light weight, lower rate
for steel corrosion than occurs in MOC cement, good
temperature resistance, good fire resistance and low
thermal conductivity, make it a promising material for
applications including producing light insulation
board, electrical insulating materials and fire-proof

materials [4]. Compared with MOC cement, which has
drawn much research interest recently [5], little
commercial or industrial attention has been paid to
MOS cement, mainly because of its lower mechanical
strength. Therefore, if the mechanical strength of
MOS cement is improved, MOS cement will also be an
excellent cementing material, which may replace
MOC cement in some cases, especially for applica-
tions at higher temperatures.
The mechanical strength of MOS cement depends
mainly on the type and relative content of the crystal
phases in the cement. According to Demediuk and
Cole [6], four basic magnesium salts that can contrib-
ute to the mechanical strength of MOS cement exist in
the ternary system MgO–MgSO4–H2O at different
temperatures between of 30 and 120 °C: 5Mg(OH)2
MgSO43H2O (513 phase), 3 Mg(OH)2MgSO48H2O
(318 phase), Mg(OH)22MgSO43H2O (123 phase),
and Mg(OH)2MgSO45H2O (115 phase). Kahle [7]
reported that 513 phase and 318 phase only could be
obtained by curing cement samples under saturated
steam condition when the molar ratio of MgO and
MgSO4 was fixed at 5:1. Urwong and Sorrel [8]
studied the crystal phases present in MOS cement and
suggested that metastable 115 phase and
MgSO44H2O could also be detected at ambient
temperatures, along with 318 phase, MgSO4nH2O
(n = 7, 6, or 1), Mg(OH)2, and MgO. They also
suggested that it was impossible to obtain MOS
cement products containing more than 50 % 318
phase, and that this was the reason for the low
mechanical strength of MOS cement. These studies
suggest that adjusting the crystal phase composition
may be an effective method to improve the mechanical
strength of MOS cement.
It is well known that MOC cement has poor water
resistance because its strengthening phases are soluble
[9, 10]. To improve the water resistance, soluble
phosphates have been used as modifiers in MOC
cement [11]. Although adding 1 % H3PO4 to MOC
cement can decrease the compressive strength of
MOC cement by 20 %, the water resistance improves
greatly, with the softening coefficient after immersion
in water for one month increasing by about 50 %.
Soluble phosphates are commonly used in Portland
cement as retarders, based on their ability to form
complexes with hydration products in the cement-
water system and to affect the hydration reactions
[12–14]. Phosphates have also been studied extensively
Materials and Structures
as barium sulfate inhibitors [15, 16]; in these studies, it
has been proposed that phosphoric acid groups may
replace sulfate sites on growing crystal face of barium.
Additionally, the addition of soluble phosphates to
MOS cement has been reported in some patents [17,
18], but the mechanism of the effects of soluble
phosphates on MOS cement has not been reported yet.
We suggest that soluble phosphates are able to
affect the crystal growth of basic magnesium salt by
complexing in some way, and hence may be able to
change the phase composition, and improve the
compressive strength and water resistance of MOS
cement. Therefore, phosphoric acid and phosphates
have been added into MOS cement as modifiers in this
paper, and the effects of these modifiers on setting
time, crystal phase composition, compressive strength
and softening coefficient of MOS cement have been
investigated in detail. Additionally, the mechanism by
which phosphoric acid and phosphates modify MOS
cement is discussed.
2 Materials and experimental procedures
2.1 Raw materials
The magnesia used in this study is light-burned
magnesia (LBM) obtained by calcining magnesite
from Liaoning, China at 750 °C, and the chemical
composition of the LBM is listed in Table 1. Because
active MgO (a-MgO) can hydrate and convert to
magnesium subsulfate in magnesium sulfate solution
within the setting process of MOS cement [19, 20], it is
necessary to know the content of a-MgO in the LBM so
that the material mixture ratio and its effects on the
properties of MOS cement can be compared between
cements made using different LBM sources. The
content of a-MgO used in this work was determined
to be 58.50 % by the standardized hydration method
mentioned in Dong et al.’s report [21]. Magnesium
sulfate (MgSO47H2O) was analytically pure and
purchased from Tianjin Hongyan Ltd., China. Analyt-
ically pure phosphoric acid (H3PO4), potassium
Table 1 Chemical compositions of light-burned magnesia
Component MgO CaO SiO2 Al2O3 Fe2O3 CO2
Content (wt%) 80.20 1.30 6.07 0.15 0.41 11.87
Materials and Structures
dihydrogen phosphate (KH2PO4), dipotassium hydro-
gen phosphate (K2HPO4) and potassium phosphate
(K3PO4) were selected as additives for MOS cement.
2.2 Specimen preparation
The molar ratio of a-MgO/MgSO4 of MOS cement
prepared from LBM was fixed at 7.0. To prepare neat
MOS cement specimens, MgSO47H2O was firstly
dissolved in water to form a 25.00 mass% MgSO4
solution. Secondly, 0.5 % (by weight of LBM) of
additive (H3PO4, KH2PO4, K2HPO4, or K3PO4) was
mixed with the MgSO4 solution to form a clear,
uniform mixed solution. Then, 100.0 g LBM mechan-
ically stirred with 101 g of this mixed solution to form
MOS cement paste. The MOS cement paste was cast in
20 9 20 9 20 mm3 PVC molds, sealed and cured at
20 ± 3 °C for about 24 h before demolding.
2.3 Specimen analysis
The compressive strengths of the cured MOS cement
samples at different ages were tested on a testing
machine using a maximum force of 100 kN at a loading
rate of 10 mm/min; six samples were tested for each
composition and age. The crushed cement was reduced
to a power (D90 \ 35 lm) for crystal phase composi-
tion analysis by XRD patterns, which were collected on
an X’ Pert PRO (PANalytical) diffractometer with
CuKa radiation (k = 0.15419 nm) over a 2h range
from 5° to 70°. The microstructures of the Sorrel
cement samples cured in air for 28 days were charac-
terized by scanning electron microscopy (SEM, JSM-
5610LV) on fractured surfaces after gold coating. The
thermogravimetric curves were recorded with a Q600
differential scanning calorimeter. The samples were
heated from 40 to 1,000 °C at 5 °C/min. The chemical
compositions of the raw materials (as shown in
Table 1) were obtained by X-ray fluorescence (XRF)
on a X-ray fluorescence analyzer (pw-4400-40,
PANalytical).
To evaluate the water resistance of the MOS
cement specimens, specimens cured for 28 days in
air were dipped in water at 20 ± 3 °C, and the
Table 2 Setting time of
Samples
MOS cement with different
additives Initial setting time (h)
Final setting time (h)
compressive strength of the samples after different
immersion times in water was measured and used to
calculate the softening coefficient (Rf) as follows:
Rðw; nÞ
Rf ¼ : ð1Þ
RðA; 28Þ
Here R(w,n) and R(A,28) denote the compressive
strength of specimens after immersion in water for
n days and the compressive strength of specimens
cured in air for 28 days, respectively.
3 Results and discussion
3.1 Effects of additives on setting time
The setting time of MOS cement is related to the MgO
hydration rate and phase crystallization process.
Table 2 shows both the initial and final set times for
the MOS cement samples with different additives.
Additives, especially H3PO4 or KH2PO4, can retard
the initial setting of MOS cement. For instance, the
initial set time was extended to 6.2 h from 2.5 h by
adding 0.5 % H3PO4. The extent of the retardation is
in the order: H3PO4 = KH2PO4  K2HPO4 [ K3
PO4. A previous study has also found that phosphoric
acid (H3PO4) is slightly more effective at retarding set
than phosphates [22]. Additionally, it is notable that
the time interval between the initial and final set times
of the MOS cement sample is shortened by the
additives. This is probably because the additives
change the MgO hydration process, and promote
crystallize of a new kind crystal phase that crystallizes
readily within the MOS cement.
3.2 Effects of additives on compressive strength
and phase composition
The effects of additives on compressive strength of
MOS cement samples are shown in Fig. 1. The
additive-free control sample has low compressive
strength—about 30 MPa after 28 days. H3PO4,
KH2PO4, K2HPO4 or K3PO4 can improve compressive
strength of MOS cement to different extents; these
Control With H3PO4 With KH2PO4 With K2HPO4 With K3PO4
2.5 6.2 5.8 3.0 2.8
5.0 7.5 7.2 6.0 5.6
 Materials and Structures

Control H3PO4 KH2PO4 (a)

80 K2HPO4 K3PO4 X-X phase H-Mg(OH)2 M-MgO 7-MgSO4 7H2O 6-MgSO4 6H2O
X
Compressive strength (MPa)

M
X X X X X XXH XX X X M11H18
60 M10H18
M9H18
M8H18
40 M7.3H18
M6.7H18
H
M6H18
20 6
M5H18
7 M4H18
M3H18
7 6 M2H18
0 7 14 21 28
M1H18
Curing time (days) 10 20 30 40 50 60
o
2Theta ( )
Fig. 1 Compressive strength of MOS cements without (con-
trol) and with different additives (H3PO4, KH2PO4, K2HPO4, (b) X
and K3PO4, respectively) X-X phase M-MgO 7-MgSO4 7H2O

X-X phase H-Mg(OH)2 M-MgO C-MgCO3 S-SiO2 X

X
X X X XXX X XX XX X
X M M5H28

C X M5H20
X M
X X H S X XH
XX H H with H3PO4
M5H18
H H
X H with KH2PO4 M M5H14
X H M
H 7
X H M5H12
with K2HPO4

XH
H H
10 20 30
2Theta ( )
Fig. 2 XRD patterns of MOS cements without (control) and
with different additives (H3PO4, KH2PO4, K2HPO4, and K3PO4,
respectively)
effects are different from the influences of these
additives on MOC cement [11]. In particular, by
adding H3PO4 or KH2PO4 into MOS cement, samples
with compressive strengths of over 80 MPa after
28 days can be obtained, nearly triple the strength of
the control sample.
XRD patterns of the control MOS sample and the
samples with various additives (H3PO4, KH2PO4,
K2HPO4, or K3PO4) are presented in Fig. 2. A new
crystal substance, denoted temporarily as X phase,
occurs in the MOS cement containing additives, and is
most prominent in the samples with added H3PO4 or
KH2PO4. This new crystal phase is not a basic
H 10 20 30 40 50 60
H with K3PO4 o
2Theta ( )
H Control
Fig. 3 XRD patterns of MOS cement samples with different
40 50 60 70 molar ratios of MgO/MgSO4 (from 1.0 to 11.0) at the same
o water content (a) and with different water contents at the same
molar ratio of MgO/MgSO4 (M = 5) (b)
magnesium salt previously found in any ternary
MOS system [8]. Because of the correlation between
compressive strength and the content of X phase in the
MOS cements containing different additives, the
formation of X phase must be a major factor in the
improvement of mechanical strength of MOS cement
by the additives.
To identify the chemical composition of X phase,
many MOS cement samples with different molar ratios
of MgO/MgSO4 (from 1.0 to 11.0) at the same water
content and with different water contents at the same
molar ratio of MgO/MgSO4 (M = 5) were prepared
from analytically pure MgO and MgSO47H2O, and
H3PO4 was used as the additive (representative XRD
patterns are shown in Fig. 3). When the molar ratio of
MgO/MgSO4 was 5.0 and the water content was at least
Materials and Structures

4.97
100 1
75 , 97.89%)
143
92 91.45% 405
90 106
DSC
80 0

Weight/%
75.35%

mw/mg
70 299

59.33% -1
60

2.40
950
TG 786
9.36

50 1023
2.90

45%
40 -2
200 400 600 800 1000
Temperature
10 20 30 40 50 60
2Theta ( )
o Fig. 6 TG/DSC of X phase (M5H18)

Fig. 4 Characteristic diffraction peaks of X phase 4.97

8.14

100

8.25

80

9.36

75

10 20 30 40 50 60 70

2Theta

Fig. 7 XRD patterns of decomposition products of X phase at

different temperatures

Fig. 5 SEM image of MOS cement sample with the molar ratio
of MgO/MgSO4 of 5.0 (M5H18)
18, the MOS cement contained only X phase. Figure 4
shows that the d values corresponding to the main
diffraction peaks of X phase are 9.36, 4.97, 2.90 and
2.40 Å. Additionally, the SEM image of the MOS
cement sample (M5H18) shows that X phase is
homogeneous, needle-like crystals (Fig. 5).
To identify the chemical formula of X phase, the
M5H18 sample was acid-dissolved and the molar ratio
of Mg2?/SO42- was determined as *6.0 by atomic
absorption analysis; this means that X phase can be
expressed as 5Mg(OH)2MgSO4nH2O. Moreover,
TG/DSC analysis was used to determine the number
of crystal waters in X phase, and the result is shown in
Fig. 6. To see if the mass lost before 106 °C was the
loss of free water or crystal water, XRD patterns of
decomposition products at different temperatures were
analyzed (Fig. 7). The peak at 9.36 Å shifts when the
temperature is higher than 75 °C, indicating that the
mass loss after 75 °C is loss of crystal water. The
number of crystal waters in X phase can be evaluated
from the mass lost between 75 and 786 °C and the
reaction:
5Mg(OH)2  MgSO4  nH2 O
! 5MgO þ MgSO4 þ ð5 þ nÞH2 O, ð2Þ
ð 5 þ nÞ  M H 2 o
 100 %
5  MMgo þ MMgSo4 þ ð5 þ nÞ  MH2 o
¼ 97:89 %  59:33 % ¼ 38:56 %: ð3Þ
The mass loss from 75 to 786 °C was 38.56 % from
TG analysis (i.e. the residual masses at 75 and 786 °C
were 97.89 and 59.33 %, respectively). Therefore,
‘‘n’’ in 5Mg(OH)2MgSO4nH2O can be calculated as
6.97 = 7, using Eq. (3). Thus X phase is
5Mg(OH)2MgSO47H2O (517 phase).
 Materials and Structures

Fig. 8 SEM images of MOS cements (prepared by light burn magnesia) with different additives a control, b with H3PO4, c with
KH2PO4, d with KH2PO4 and e with K3PO4
Materials and Structures
1.0
1d
3d
7d
0.8 14d
28d
56d
0.6
Rf
0.4
0.2
0.0
Control H3PO4 KH2PO4 K2HPO4 K3PO4
Fig. 9 Softening coefficients of MOS cement samples without
(control) of with different additives (H3PO4, KH2PO4, K2HPO4,
and K3PO4, respectively) after immersion in water for different
time
Figure 8 presents SEM images of MOS cement
samples (prepared from light-burned magnesia) with
different additives. The control samples are mostly
flaky Mg(OH)2 crystal with a loose structure, which
results in the low compressive strength of MOS
cement. This is consistent with Beaudion and Rama-
chandran report [3] in which he pointed out that high
porosity and loose structure could lead to low
mechanical strength in MOS cement. However, the
micro-structure of the MOS cement samples clearly
changes with additives. There are many needle-like
crystal whiskers in MOS cement samples containing
H3PO4 or KH2PO4. According to the XRD analysis in
Fig. 2, these needle-like crystal whiskers are probably
517 phase. This suggests that the needle-like 517
phase crystal whiskers in MOS cement induce a
compact structure in a similar way to the effect of 518
phase [5Mg(OH)2MgCl28H2O] in MOC cement; the
final result is that MOS cement containing H3PO4 or
KH2PO4 has high mechanical strength.
3.3 Effects of additives on water resistance
Figure 9 shows the softening coefficients of MOS
cement samples with and without additives. The
softening coefficient of the control MOS sample
decreases sharply to less than 0.2 as the immersion
time increases to 14 days, and these control samples
disintegrate after immersion in water for more than
14 days. Additives clearly improve the water resis-
tance of MOS cement, and the degree of improvement
is in the order: H3PO4 = KH2PO4  K2HPO4 [
X-517 phase H-Mg(OH)2 M-MgO C-MgCO3 S-SiO2
X
M
X
X C X M
X S
with H3PO4
H
with KH2PO4
H
with K2HPO4
H
with K3PO4
H
Control
10 20 30 40 50 60 70
o
2Theta
Fig. 10 XRD patterns of MOS cement without or with
additives after immersion in water for 56 days
K3PO4. The MOS cement samples containing H3PO4
or KH2PO4 have excellent water resistance, with the
softening coefficients of samples containing H3PO4 or
KH2PO4 being 0.89 and 0.85, respectively after
immersion in water for 56 days.
The water resistance of cement materials mainly
depends on the stability of the constituent phases of the
cement in water. For example, it is well known [9, 11]
that MOC cement without any additives has poor
water resistance, because the 518 and 318 phases in
MOC are soluble in water. Figure 10 displays the
phase compositions of MOS cement with and without
additives after immersion in water for 56 days at room
temperature. A small amount of 517 crystal phase is
seen in the control sample after immersion in water,
demonstrating that 517 phase is essentially insoluble
in water.
The micro-structure of the control sample after
immersion in water (Fig. 11a) also shows that some
needle-like crystal whiskers have appeared. Formation
of 517 phase crystal whiskers in the control sample
after immersion in water is likely to be because of
reaction of relatively slow-hydrating low-active MgO
with a small amount of surplus MgSO4 in the cement
samples, because no magnesium subsulfate was
detected in the XRD pattern of the control sample
before immersion in water. However, the low water
resistance of the control samples can be explained by
most of the unreacted MgO in the control samples
hydrating to Mg(OH)2 upon immersion in water. This
reaction causes the cement structure to become looser,
and even crack, as shown in Fig. 11a, because of the

Fig. 11 SEM images of control sample (a) and sample with
H3PO4 (b) after immersion in water for 56 days
different densities of MgO (q = 3.5 g/cm3) and
Mg(OH)2 (q = 2.4 g/cm3) [23].
XRD patterns of MOS cement samples made with
H3PO4 or KH2PO4 indicate that little Mg(OH)2 forms
after immersion in water for 56 days. This implies that
the additives, especially H3PO4 and KH2PO4, inhibit
the hydration of MgO and the formation of Mg(OH)2,
making the MOS cement stable in water and explain-
ing the high values of softening coefficient.
3.4 The effect mechanism of additives on MOS
cement
We have shown above that additives (H3PO4,
KH2PO4, K2HPO4, or K3PO4, respectively) can affect
the setting time, and improve compressive strength
and water resistance of MOS cement. To understand
the mechanism behind these effects, we have carefully
Materials and Structures
12
Water
11
Water+KH2PO4
10
pH
9
6
0 10 20 30 40 50 60
Time (min)
Fig. 12 pH changes of MgO hydration in water with and
without KH2PO4. The molar ratio of MgO/H2O = 4 and the
dosage of KH2PO4 is 0.5 wt% of MgO
examined the MgO hydration process. Figure 12
shows the evolution of pH during hydration of MgO
hydration with and without 0.5 % KH2PO4. MgO
hydration in water without KH2PO4 can be divided
into three stages. In stage 1 (pH-rapid rise), one
magnesia hydrated layer with positive charge forms on
the surface of MgO and OH- is released into the
solution (Eq. 4), which results in the rapid rise of pH.
In stage 2 (pH-slow decrease), Mg(OH)2 forms and the
concentration of OH- decreases (Eq. 5). In stage 3
(pH-slow rise), the solid Mg(OH)2 ionizes slowly and
releases OH- back into the water (Eq. 6).
MgOðsolidÞ þ ðx þ 1ÞH2 O
 þ
! MgðOHÞðH2 OÞx ðsurfaceÞ þOH ðaqÞ ; ð4Þ
 þ
MgðOHÞðH2 OÞx ðsurfaceÞ þOH
ðaqÞ ! MgðOHÞ2ðsolidÞ ;
ð5Þ
MgðOHÞ2ðsolidÞ ! Mg2þ 
ðaqÞ þ 2OHðaqÞ : ð6Þ
In contrast, MgO hydration in water with KH2PO4
can only be divided into two stages. The first stage is
similar to stage 1 of MgO hydration without KH2PO4.
The maximum value of pH during this stage is only
10.67, less than the 11.70 observed for MgO hydration
without KH2PO4. The pH remains stable during the
following stage. Phosphoric acid and phosphoric salts
can react with magnesium and other M2? cations
through complexation reactions [13]. They are also
known to poison the nucleation and growth of barium
sulfate crystals [15] and inhibit the hydration of Fe2O3
and Al2O3 surfaces via direct surface adsorption [22].
Similarly, it appears that H2PO4- ions adsorb onto the
Materials and Structures
517 Phase MgO
144 h
72 h
48 h
24 h
12 h
8h
6h
4h
2h
1h
0.5 h
10 min
10 20 30 40 50 60 70
2Theta ( )
o
Fig. 13 XRD patterns track of MOS cement with 0.5 wt%
KH2PO4 (by weight of MgO) prepared by chemical agents (the
molar ratio of MgO:MgSO4:H2O = 5:1:18)
magnesia hydrated layer to form a metastable com-
pound [H2PO4 ? Mg(OH)(H2O)x-1], restricting the
reaction of the hydrated magnesia layer and OH- and
thus inhibiting the formation of Mg(OH)2. Moreover,
the presence of this relatively inert hydrated layer
magnesia with adsorbed H2PO4- further retards the
hydration of MgO, extending the set time of the MOS
cement.
When MgO hydrates in magnesium sulfate solution,
OH- ions released by MgO hydration may react with
magnesium ions and form [Mg(OH)(H2O)x]? in solu-
tion (Eq. 7). However, the MgO hydrated layer with
more positive charge is so unstable that it can react with
OH- to form Mg(OH)2 (Eq. 5), and this can be
regarded as a competitive reaction to the formation of
[Mg(OH)(H2O)x]?. The MgO hydrated layer is stabi-
lized by adsorbing H2PO4- and forming
[H2PO4 ? Mg(OH)(H2O)x-1] on its surface
in the presence of KH2PO4. This formation of
[H2PO4 ? Mg(OH)(H2O)x-1] increases the concen-
tration of [Mg(OH)(H2O)x]? in solution and decreases
the probability of forming Mg(OH)2. As the
[Mg(OH)(H2O)x]? concentration increases in the
solution, 517 phase forms, according to Eq. (8). As
shown in Fig. 13, XRD scans of M5H18 samples with
curing times of 0.5–144 h indicate that Mg(OH)2 does
not form at all during the entire hydration process for
this cement sample, and that the content of 517 phase
increases as the amount of MgO remaining decreases.
The 517 phase emerges at a curing time of *8 h, which
is close to the final set time of 7.2 h. Additionally,
because only a small amount of H2PO4- (0.5 % of the
mass of MgO or LBM) can affect the hydration process
of MgO in MOS cement, the H2PO4- must be released
again after the formation of 517 phase so that it can will
continue to adsorb onto new hydrated MgO layers and
inhibit the formation of Mg(OH)2.

Mg2þ
ðaqÞ þ xH2 O þ OHðaqÞ
 þ
! MgðOHÞðH2 OÞx ðaqÞ ; ð7Þ
 
5 H2 PO4 ! MgðOHÞðH2 OÞx1 ðsurfaceÞ
 þ
þ MgðOHÞðH2 OÞx ðaqÞ þ 4OH 2
ðaqÞ þ SO4
! 5MgðOHÞ2 MgSO4 7H2 O þ 5H2 PO
4
þ ð6x  10Þ H2 O: ð8Þ
To understand why H3PO4 and KH2PO4 can inhibit
MgO hydration more effectively than K2HPO4 and
K3PO4, nitric acid, sulfuric acid, boric acid, and oxalic
acid were also added at quantities of 0.5 % into MOS
cement to see if acidic materials have a stronger ability
to improve the compressive strength and water resis-
tance of MOS cement. Figure 14 displays XRD
patterns of MOS cement samples made with acidic
materials after a curing time of 28 days, and shows that
adding acidic materials does not change the phase
composition of the MOS cement. Therefore, the
remarkable improvement in properties of MOS cement
produced by adding a small amount of H3PO4 or
KH2PO4 cannot be attributed to their higher acidity
compared with K3PO4 or K2HPO4. The effectiveness
of H3PO4 and KH2PO4 may be related to the ions that
they form in solution: H3PO4, KH2PO4, K3PO4 and
K2HPO4 ionize in solution to form H2PO4-, H2PO4-,
HPO42- and PO43-, respectively, and these anions
adsorb onto [Mg(OH)(H2O)x]? to inhibit the formation
of Mg(OH)2 and promote formation of 517 phase. The
different negative charges on H2PO4-, HPO42- and
PO43-, affects the strength of the repulsive interaction
with the superfluous negative charge on the
surface of [HPO4 ? Mg(OH)(H2O)x]- or
[PO4 ? Mg(OH)(H2O)x]2-, and may mean that the
adsorption capability of HPO42- and PO43- onto
[Mg(OH)(H2O)x]? is less because of their greater
negative charge and repulsive interaction. This would
be consistent with the order of effectiveness of the
phosphate compounds being: H3PO4 = KH2-
PO4  K2HPO4 [ K3PO4. However, experimental
evidence for this mechanism is still needed.

X-517 phase H-Mg(OH)2 M-MgO C-MgCO3 S-SiO2
M
H emerges after immersion in water, and this
H C
S H M
H formation of Mg(OH)2 loosens the MOS cement,
M
with HNO3
with H2SO4
with HBO3
with H2C2O4
Control
10 20 30 40 50 60 70
o difficult for HPO42- and PO43- to adsorb onto
2Theta
Fig. 14 XRD patterns of MOS cement samples with different
acids as additives
4 Conclusions
The effects of phosphoric acid and phosphates
(H3PO4, KH2PO4, K3PO4 or K2HPO4) on setting time,
mechanical strength and water resistance of MOS
cements have been studied in this paper. We have
concluded that:
(1) Adding phosphoric acid and phosphates (H3PO4,
KH2PO4, K3PO4 or K2HPO4), especially H3PO4
or KH2PO4, can extend setting time and improve
compressive strength and water resistance of
MOS cement by changing the hydration process
of MgO and thus the final phase composition.
H3PO4, KH2PO4, K3PO4 and K2HPO4 ionize in
solution to form H2PO4-, H2PO4-, HPO42- and
PO43-, respectively, and these anions adsorb
onto [Mg(OH)(H2O)x]? to inhibit the formation
of Mg(OH)2 and further promote the generation
of a new magnesium subsulfate phase.
(2) This new magnesium subsulfate phase has been
demonstrated by elemental atomic absorption
analysis and TG/DTG analysis to be
5Mg(OH)2MgSO45H2O (517 phase). SEM ana-
lysis shows that the 517 phase has a needle-like
morphology; SEM analysis of samples after
immersion in water shows that 517 phase is
essentially water-insoluble, and thus is responsi-
ble for the more compact structure, higher
mechanical strength and better water resistance
of MOS cement containing H3PO4 or KH2PO4.
Materials and Structures
(3) MOS cement without any additives has poor
water resistance, because much flaky Mg(OH)2
eventually causing disintegration
(4) Compared with HPO42- and PO43-, H2PO4-
produces greater improvement in mechanical
strength and water resistance of MOS cement,
probably because the superfluous negative charge
on the surfaces of [HPO4 ? Mg(OH)(H2O)x]-
or [PO4 ? Mg(OH)(H2O)x]2- repels the anions
in solution, and makes it more
[Mg(OH)(H2O)x]?. The monovalent H2PO4
anions would be less affected by this interaction,
leading to an order of additive effectiveness of:
H3PO4 = KH2PO4  K2HPO4 [ K3PO4.
To achieve large-scale application of these MOS
cements, in areas such as substitutes for wood and
building materials including light insulation board and
ferroconcrete, the durability and stability of these
modified MOS cements, including the long-term water
resistance, carbonation resistance, corrosion rate of
embedded steel, fire resistance and heat preservation
properties, require further systematic investigation.
Acknowledgments Partial financial support from the One-
Hundred Talent Project of the CAS granted to HFY (B0210),
and the Qinghai Province Science and Technology Tackling
Key Project under Grant 2008-G-158 are gratefully
acknowledged; this project was also supported by the National
Natural Science Foundation of China (Grant No. 51178221).
References
1. Demediuk T (1952) A method for overcoming unsoundness
in magnesian limes. Nature 70(4332):799
2. Demediuk T, Cole WF, Hueber HV (1955) Studies on
magnesium and calcium oxychlorides. Aust J Chem
8(2):215–233
3. Beaudion JJ, Ramachandran VS (1978) Strength develop-
ment in magnesium oxysulfate cement. Cement Concrete
Res 8(1):103–112
4. Yu HF (1991) Magnesium oxychloride and its application.
China Building Material Press, Beijing
5. Chau CK, Li ZJ (2008) Microstructures of magnesium
oxychloride. Mater Struct 41:853–862
6. Demediuk T, Cole WF (1957) A study on magnesium
oxysulphates. Aust J Chem 10(3):287–294
7. Kahle K (1972) Mechanism of formation of magnesium-
sulphate cement. Silikattecnic 23(5):148–151
Materials and Structures
8. Urwong L, Sorrell CA (1980) Phase relations in magnesium
oxysulfate cements. J Am Ceram Soc 63(9–10):523–526
9. Yu HF (1992) Study of new type water-resisting magnesium
oxychloride cement. J Chin Ceram Soc 20(4):1203–1205
10. Li HQ, Li GZ, Yu YZ (2008) The influence of compound
additive on magnesium oxychloride cement/urban refuse
floor tile. Constr Build Mater 22(4):521–525
11. Deng DH (2003) The mechanism for soluble phosphates to
improve the water resistance of magnesium oxychloride
cement. Cem Concr Res 33(9):1311–1317
12. Griffin JLW, Coveney PV, Whiting A (1999) Design and
synthesis of macrocyclic ligands for interaction with crys-
talline ettringite and demonstration of available mechanism
for the setting of cement. J Chem Soc Perk Trans 1
10(2):1973–1981
13. Ramachandran VS, Lowery MS, Wise T, Polomark GM
(1993) The role of phosphonates in the hydration of Portland
cement. Mater Struct 26(161):425–432
14. Billingham J, Francis DTI, King AC et al (2005) A multi-
phase model for the early stages of the hydration of retarded
oilwell cement. J Eng Math 53(2):99–112
15. Black SN, Bromley LA, Cottier D et al (1991) Interactions
at the organic inorganic interface-binding motifs for phos-
phonates at the surface of barite crystals. J Chem Soc Far-
aday Trans 87(20):3409–3414
16. Al Rohl, Gay DH, Davey RJ, Catlow CRA (1996) Interac-
tions at the organic/inorganic interface: molecular modeling
of the interaction between diphosphonates and the surfaces
of barite crystals. J Am Chem Soc 118(3):642–648
View publication stats
17. Wallner CE (2010) Cementitious veneer useful for forming
laminated tiles/panels in building applications comprises
magnesium oxysulfate compound comprising e.g. magne-
sium sulfate; and cementitious composition comprising e.g.
hydrate calcium sulfate. USA patent US2010222457-A1;
US8182605-B2
18. Kishimoto T, Yamamoto S (2009) Modified magnesium
oxysulfate fibrous particles, useful e.g. in polyolefin com-
position, comprises magnesium oxysulfate fibrous particles
treated on their surfaces with nucleating agent comprising
e.g. phosphoric acid compounds. USA patent US200929
2047-A1
19. Birchal VSS, Rocha SDF, Ciminelli VST (2000) The effect
of magnesite calcination conditions on magnesia hydration.
Miner Eng 13(14–15):1629–1633
20. Li ZJ, Chau CK (2008) Reactivity and function of magne-
sium oxide in Sorel cement. J Mater Civil Eng 20(3):239–
244
21. Dong JM, Yu HF, Zhang LM (2010) Study on experimental
conditions of hydration methods of determining active
magnesium oxide content. J Salt Lake Res 18(1):38–41
22. Bishop M, Bott SG, Barron AR (2003) A new mechanism
for cement hydration: solid-state chemistry of calcium ni-
trilotris(methylene)triphosphonate. Chem Mater 15(16):
3074–3088
23. Salomão R, Pandolfelli VC (2011) Citric acid as anti-
hydration additive for magnesia containing refractory ca-
stables. Ceram Int 37(6):1839–1842