- Email: [email protected]
The habit of gypsum and solvent interaction
Journal of Crystal Growth 112 (1991) 445—450 North-Holland
445
The habit of gypsum and solvent interaction Elly van der Voort
*
and P. Hartman
Department of Geochemistry, institute of Earth Sciences, University of Utrecht, P.O. Box 80.021, 3508 TA Utrecht, Netherlands
Received 2 February 1991
A qualitative explanation for the discrepancy between the theoretical and observed habits of gypsum, CaSO
4 2H20, is presented in terms of solvent interaction. The theoretical models predict (011) faces to be large and (111) face to be small, whereas the observed habit shows large (111 } faces and (011) absent or small. It is argued that the hydrated surfaces will behave differently in the growth process. Two kinds of water molecules can be present on the faces. One kind is part of the crystal structure and does not have to be removed before growth can continue. This occurs for [011]. The other kind does not belong to the crystal structure and has to be removed, thereby lowering the growth rate, which is the case for [111]. When organic impurities are present the growth rate of the { 111) and (011 } faces are equally reduced, which is in agreement with the observed habit from solutions with organic impurities. A growth rate reduction is also expected for (120) faces, leading to an elongation into the c axis direction.
I. Introduction There exists a discrepancy between the theoretical habits of gypsum, CaSO4~2H20, as derived by several authors [1—3]and the observed habit, as regards the terminating forms (011) and (111 }. Whereas the theoretical habit is platy (010) with (120) as side form and terminated by large (011) and small (ill) faces (fig. la), the observed habit has large {11.1} faces, while (011) is usually absent (fig. ib). According to Weijnen et al. [2], this discrepancy cannot be attributed to solvent adsorption because the (111) and (011) faces are probably bounded2~ions by water that are atin themolecules crystal. Furthermore, tachedshow to Ca they that the charge density along the reciprocal vectors [111] and [011] gives no obvious reason for a stronger solvent interaction with the (111) face. However, the charge density is not a measure for the strength of the solvent interaction, because it gives no information about the electrostatic potential and field strength near the surface, *
Present address: Philips Research Laboratories, P.O. Box
So this is not an argument to rule out the influence of solvent interaction. Heijnen and Hartman [3] considered the possibility of growth in slices of thickness ~gd011and ~d-115.They found that the difference in attachment energy of both slice configurations of (011) is smaller than for (111), so that the (011) face starts to grow with half slices at a lower super-
—
(111)
(120) (020)
(020)
(120)
N~N.
a
~
Fig. 1. Theoretical habit as derived by Heijnen and Hartman [3] (a) and observed habit of gypsum from an aqueous solution
80.000, 5600 JA Eindhoven, Netherlands. 0022-0248/91/$03.50 © 1991
(ill)
-
(011)
Elsevier Science Publishers B.V. (North-Holland)
(b) (after Simon and Bienfait [1]).
446
/ Habit of gypsum
E. van der Voort, P. Hartman
saturation than (ill). The (011) face thus grows faster with the result that (111) becomes the larger terminating face. However, this explanation is only valid if 2D nucleation determines the growth rate. Christoffersen et a!. [4] showed that growth at moderate supersaturations is controlled by a screw dislocation mechanism. The purpose of this paper is to provide an explanation for the discrepancy by studying the hydrated crystal faces and the influence of the hydration on the growth rate. ,
2. Hydration of the surface The faces of a crystal in an aqueous solution are hydrated. In general these water molecules have to be removed before growth can continue, When the water molecules are adsorbed very strongly, the dehydration of the surface determines the growth kinetics, When the crystal is a hydrate, the situation may be different. Water molecules at the interface between crystal and solution may belong to the crystal structure because they are adsorbed on crystallographic positions. The bivalent cations in hydrates such as MgSO 3. 6H20 [5], MgSO4. 7H20 [6], K2Mg(S04)2 6H~O[7] and NiSO4’ 6H20 [8] have a complete first coordination sphere. The A(H2O)~cation complexes in the crystal are also present as such in the solution. So dehydration of the cation in the solution and at the surface is not necessary and a strong bond between a water molecules and a cation will not lower the growth rate, 2 + ions are surIn the by casetwo of gypsum, the Ca while in the rounded water molecules, solution a much higher coordination is possible. The coordination is far from complete and coordination water of a Ca2~ ion at the surface can belong to the crystal structure, i.e., is present on crystallographic positions, or it belongs to the solution. In the latter case it has to be removed. If the dehydration of the surface determines the growth kinetics, then the “solution water” lowers the growth rate. An application of this hypothesis to gypsum is discussed in this paper. ‘
and solvent interaction
3. Results and discussion —
Figs. 2a and 2b show a [001] and [211] projection of the crystal structure of gypsum with the (020), (120), (011) and (111) slices and the bonds p, q, r, s and t. Bond p is a Ca—SO4 bond in which 2~is bonded to two oxygens of the SO~ ion. Ca In bond q, Ca2~is bonded to one oxygen. Bond r is a short contact between Ca2~and water. Bonds s and t are two non-equal hydrogen bonds between water and the SO~ ion. Table 1 shows the electrostatic bond energies of the p. q, r, s and bonds. These bond energies were calculated based on the crystal structure determination by Cole and Lancucki [9] by Heijnen and Hartman [3]. The charges on the ions were taken as Ca2~,S2±,O~ and H1~2~. Bonds p and q are the strongest because they are contacts between a positive and negative ion charged + 2 and 2 respectively. Bonds r, s and t are much weaker. In this study it is assumed that the slices are bounded by water molecules that belong to the crystal structure. These water molecules are bonded to Ca2~ and SO~ ions in the slice. In figs. 3a and 3b these hydrated slices are shown, which are constructed by moving water molecules along lattice translations. This has consequences —
for the slice energies because bonds are made and broken when the water molecules are moved. If we consider nearest neighbour interactions, only, the bonds made and broken can be expressed in terms of bonds r, s and t, so an energy balance can be made. Table 2 shows the bonds to be made and broken and the energy involved to transform an F-slice into a hydrated slice. In the nearest neighbour approximation (NNA), corresponds the energy involved in this slice transformation to the increase in attachment energy. For the (111) face no hydrated slices were constructed because there are no strong bonds r, s and t between the slices, The attachment energies of the hydrated slices are shown in table 2, as well as the attachment energies that were calculated by Heijnen and Hartman [3]. The attachment energies were also calculated with SURFPOT [10], a program that calculates Coulomb energies of infinite lattice slices, based on the Ewald method [11], and are also listed in
E. van der Voort, P. Hartman
/
Habit of gypsum and solvent interaction
table 2. The agreement between the attachment energies calculated in these two different ways is good, which shows that the nearest neighbour
x ~
approximation can be used in this case. This is because the slice transformation involves translations of water molecules which are electrically
~
~‘
~
~‘~“
~
-
/r~’/q
i~
r~
~
-~
-
)K
~sq~0~)<
x
p~~~cC
a
447
-
x
C
d
020
0
/
/Q~dz21
/
/ ~
7~
/
,,,,/
~ ~
,~,,,.. /
~
t
~
~
~
/
/...
0
/
/
~
/
/
/
e~~d
n-’~ /..Q N
~
~1lc)
~
~
~ N
b
0~
~ —
~
/~
/
~11(b)
d
0111~ Fig.2~ions. 2. F-slices Smaller as established circles areS by Heijnen and 0 atoms and Hartman and smallest [3]. (a) circles [001]are projection H atoms. of(b) gypsum [211] projection with slices with (020)(011)a, and (120). (011)b Large (011)c, circles (022), are Ca (111)a, (111)b and (222) slices. Large circles represent the chain Ca—S0 4—H20—Ca along [211].
448
E. van der Voort, P. Hartman
/ Habit of gypsum
and solvent interaction
neutral. The theoretical habit, that is constructed by taking the central distances of the faces proportional with the attachment energies of the (111) face and of the hydrated (020}, (120) and (011)
does not differ much from the one determined by Heijnen and Hartman [3] (fig. Ia). Although the hypothesis that the (011) face grows in layers of ~d011 does improve the habit, a
faces, is shown in fig. 4. This theoretical habit
better agreement can be obtained by taking the
x ~ C X 0
~
/ /
rx xC
-
-
020
0 0
o
0 -
~~~::c
r-~....,, S
~
~
.,,,(
C
N.) ~
-~
~
~
~r.
0-... N.)
—
/
-.-
/
b
o:~2Q0’
~.
~
0
~ N
~
//d011.
Fig. 3. Hydrated slices (a) (020)’ and (120)’ in a [001] projection and (b) and (011)’ in a [211] projection. See fig. 2 for the explanation of the symbols.
E. van der Voort, P. Hartman
Table I Electrostatic p. q, r, s, and
t
/ Habit
of gypsum and solvent interaction
Table 3 Sites available for strong solvent interaction
bond energies
bond
bond energy (kJ mol~)
Face
Sites
p
1736 1577 249.5
(020)’ (120}’ (011)’
None
88
(111)’
q s
449
2~and SO~ oxygens Ca None Ca2~and SO~ oxygens
90
Attachment energies of the F-faces of gypsum as determined by Heijnen and Hartman [3] and attachment energies of the hydrated slices
mately the same as bond r, is stronger than that with the SO~ ions, the strength of the latter being similar to that of bond s or t. The largest
Face
E
Bonds broken
Bonds made
E’~, (NNA)
(020) (120) {Oll}a
75.8 292.4 286.7
2r+2s 2r+t 2r
2t 2s+t 2s+2t
1997 373.2
(111)
472.4
growth rate reduction is expected for (120) and (111) faces, which have Ca2~ and SO~ ions available for solvent adsorption. The growth rate reduction of the (011) face is substantially smaller than that of the (111) face, so the latter can become the dominant terminal face in agreement with the observed habit (figs ib). When organic impurities are present in the solution, that interact more strongly with the faces than water, they may replace adsorbed water molecules on both crystallographic and non-crystallographic positions. The growth rates of the side faces (01l} and (111) will be affected equally and the theoretical habit like the one shown in fig. la may be obtained. Indeed, the (0111 } face is found to prevail over the (111) face on crystals grown from aqueous solutions with organic impurities [12].
—
-
322.5 -
E’,,
199.6
357.2 311.2
—
hydration of the surfaces into account even when growth in half layers is not considered. When growth takes place from a solution, water molecules are also present on other than crystallographic positions at the surface. As was already presumed in section 2, these water molecules reduce the growth rate. Table 3 shows which sites are available for water adsorption on non-crystallographic positions. The solvent interaction with the Ca2~ions, of which the strength is approxi~~jN~11l)
~
—
3. Conclusions
~
(020) (120)
—
‘~‘~
Fig. 4. Theoretical habit of gypsum based on the attachment energies of the (ill) and the hydrated (020), (011} and (120) faces,
The habit of gypsum crystals grown from a pure aqueous solution can be explained in a qualitative way when solvent interaction is taken into account. The habit of gypsum from solutions with organic impurities can be explained by the adsorption of the impurities. The solvent (water) should also be considered as an impurity, but in the case of a solvate (hydrate like gypsum) one should be very careful. On some of the faces water seems to act as an impurity, lowering the growth rate, and on other faces it adsorbs on crystallographic natural water positions thus leaving the growth rate more or less unaltered.
E. van der Voort, P. Hartman
450
/
References [1] B. Simon and M. Bienfait, Acta Cryst. 19 (1965) 750. [2] M.P.C. Weijnen, G.M. van Rosmalen, P. Bennema and J.J.M. Rijpkema, J. Crystal Growth 82 (1987) 509. [3] W.M.M. Heijnen and P. Hartman, J. Crystal Growth 108 (1991) 290. [4] M.R. Christoffersen, J. Christoffersen, M.P.C. Weijnen and G.M. van Rosmalen, J. Crystal Growth 58 (1982) 585. [5]E. van der Voort and P. Hartman, J. Crystal Growth 106 (1990) 622.
Habit of gypsum and solvent interaction [6] M. Rubbo, D. Aquilano, M. Franchini-Angela and G. Sgualdino, J. Crystal Growth 71(1985) 470.
[7]P. Hartman, unpublished, 1990. [8] M.H.J. Hottenhuis, PhD Thesis, Nijmegen (1988). [9] W.F. Cole and C.J. Lancucki, Acta Cryst. B30 (1974) 921. [10] C.S. Strom and P. Hartmars, Acta Cryst. A45 (1989) 371. [Ii] PP. Ewald, Ann. Physik (Leipzig) 64 (1921) 253. [12] M.P.C. Weijnen and G.M. van Rosmalen, Desalination 54 (1985) 239.
445
The habit of gypsum and solvent interaction Elly van der Voort
*
and P. Hartman
Department of Geochemistry, institute of Earth Sciences, University of Utrecht, P.O. Box 80.021, 3508 TA Utrecht, Netherlands
Received 2 February 1991
A qualitative explanation for the discrepancy between the theoretical and observed habits of gypsum, CaSO
4 2H20, is presented in terms of solvent interaction. The theoretical models predict (011) faces to be large and (111) face to be small, whereas the observed habit shows large (111 } faces and (011) absent or small. It is argued that the hydrated surfaces will behave differently in the growth process. Two kinds of water molecules can be present on the faces. One kind is part of the crystal structure and does not have to be removed before growth can continue. This occurs for [011]. The other kind does not belong to the crystal structure and has to be removed, thereby lowering the growth rate, which is the case for [111]. When organic impurities are present the growth rate of the { 111) and (011 } faces are equally reduced, which is in agreement with the observed habit from solutions with organic impurities. A growth rate reduction is also expected for (120) faces, leading to an elongation into the c axis direction.
I. Introduction There exists a discrepancy between the theoretical habits of gypsum, CaSO4~2H20, as derived by several authors [1—3]and the observed habit, as regards the terminating forms (011) and (111 }. Whereas the theoretical habit is platy (010) with (120) as side form and terminated by large (011) and small (ill) faces (fig. la), the observed habit has large {11.1} faces, while (011) is usually absent (fig. ib). According to Weijnen et al. [2], this discrepancy cannot be attributed to solvent adsorption because the (111) and (011) faces are probably bounded2~ions by water that are atin themolecules crystal. Furthermore, tachedshow to Ca they that the charge density along the reciprocal vectors [111] and [011] gives no obvious reason for a stronger solvent interaction with the (111) face. However, the charge density is not a measure for the strength of the solvent interaction, because it gives no information about the electrostatic potential and field strength near the surface, *
Present address: Philips Research Laboratories, P.O. Box
So this is not an argument to rule out the influence of solvent interaction. Heijnen and Hartman [3] considered the possibility of growth in slices of thickness ~gd011and ~d-115.They found that the difference in attachment energy of both slice configurations of (011) is smaller than for (111), so that the (011) face starts to grow with half slices at a lower super-
—
(111)
(120) (020)
(020)
(120)
N~N.
a
~
Fig. 1. Theoretical habit as derived by Heijnen and Hartman [3] (a) and observed habit of gypsum from an aqueous solution
80.000, 5600 JA Eindhoven, Netherlands. 0022-0248/91/$03.50 © 1991
(ill)
-
(011)
Elsevier Science Publishers B.V. (North-Holland)
(b) (after Simon and Bienfait [1]).
446
/ Habit of gypsum
E. van der Voort, P. Hartman
saturation than (ill). The (011) face thus grows faster with the result that (111) becomes the larger terminating face. However, this explanation is only valid if 2D nucleation determines the growth rate. Christoffersen et a!. [4] showed that growth at moderate supersaturations is controlled by a screw dislocation mechanism. The purpose of this paper is to provide an explanation for the discrepancy by studying the hydrated crystal faces and the influence of the hydration on the growth rate. ,
2. Hydration of the surface The faces of a crystal in an aqueous solution are hydrated. In general these water molecules have to be removed before growth can continue, When the water molecules are adsorbed very strongly, the dehydration of the surface determines the growth kinetics, When the crystal is a hydrate, the situation may be different. Water molecules at the interface between crystal and solution may belong to the crystal structure because they are adsorbed on crystallographic positions. The bivalent cations in hydrates such as MgSO 3. 6H20 [5], MgSO4. 7H20 [6], K2Mg(S04)2 6H~O[7] and NiSO4’ 6H20 [8] have a complete first coordination sphere. The A(H2O)~cation complexes in the crystal are also present as such in the solution. So dehydration of the cation in the solution and at the surface is not necessary and a strong bond between a water molecules and a cation will not lower the growth rate, 2 + ions are surIn the by casetwo of gypsum, the Ca while in the rounded water molecules, solution a much higher coordination is possible. The coordination is far from complete and coordination water of a Ca2~ ion at the surface can belong to the crystal structure, i.e., is present on crystallographic positions, or it belongs to the solution. In the latter case it has to be removed. If the dehydration of the surface determines the growth kinetics, then the “solution water” lowers the growth rate. An application of this hypothesis to gypsum is discussed in this paper. ‘
and solvent interaction
3. Results and discussion —
Figs. 2a and 2b show a [001] and [211] projection of the crystal structure of gypsum with the (020), (120), (011) and (111) slices and the bonds p, q, r, s and t. Bond p is a Ca—SO4 bond in which 2~is bonded to two oxygens of the SO~ ion. Ca In bond q, Ca2~is bonded to one oxygen. Bond r is a short contact between Ca2~and water. Bonds s and t are two non-equal hydrogen bonds between water and the SO~ ion. Table 1 shows the electrostatic bond energies of the p. q, r, s and bonds. These bond energies were calculated based on the crystal structure determination by Cole and Lancucki [9] by Heijnen and Hartman [3]. The charges on the ions were taken as Ca2~,S2±,O~ and H1~2~. Bonds p and q are the strongest because they are contacts between a positive and negative ion charged + 2 and 2 respectively. Bonds r, s and t are much weaker. In this study it is assumed that the slices are bounded by water molecules that belong to the crystal structure. These water molecules are bonded to Ca2~ and SO~ ions in the slice. In figs. 3a and 3b these hydrated slices are shown, which are constructed by moving water molecules along lattice translations. This has consequences —
for the slice energies because bonds are made and broken when the water molecules are moved. If we consider nearest neighbour interactions, only, the bonds made and broken can be expressed in terms of bonds r, s and t, so an energy balance can be made. Table 2 shows the bonds to be made and broken and the energy involved to transform an F-slice into a hydrated slice. In the nearest neighbour approximation (NNA), corresponds the energy involved in this slice transformation to the increase in attachment energy. For the (111) face no hydrated slices were constructed because there are no strong bonds r, s and t between the slices, The attachment energies of the hydrated slices are shown in table 2, as well as the attachment energies that were calculated by Heijnen and Hartman [3]. The attachment energies were also calculated with SURFPOT [10], a program that calculates Coulomb energies of infinite lattice slices, based on the Ewald method [11], and are also listed in
E. van der Voort, P. Hartman
/
Habit of gypsum and solvent interaction
table 2. The agreement between the attachment energies calculated in these two different ways is good, which shows that the nearest neighbour
x ~
approximation can be used in this case. This is because the slice transformation involves translations of water molecules which are electrically
~
~‘
~
~‘~“
~
-
/r~’/q
i~
r~
~
-~
-
)K
~sq~0~)<
x
p~~~cC
a
447
-
x
C
d
020
0
/
/Q~dz21
/
/ ~
7~
/
,,,,/
~ ~
,~,,,.. /
~
t
~
~
~
/
/...
0
/
/
~
/
/
/
e~~d
n-’~ /..Q N
~
~1lc)
~
~
~ N
b
0~
~ —
~
/~
/
~11(b)
d
0111~ Fig.2~ions. 2. F-slices Smaller as established circles areS by Heijnen and 0 atoms and Hartman and smallest [3]. (a) circles [001]are projection H atoms. of(b) gypsum [211] projection with slices with (020)(011)a, and (120). (011)b Large (011)c, circles (022), are Ca (111)a, (111)b and (222) slices. Large circles represent the chain Ca—S0 4—H20—Ca along [211].
448
E. van der Voort, P. Hartman
/ Habit of gypsum
and solvent interaction
neutral. The theoretical habit, that is constructed by taking the central distances of the faces proportional with the attachment energies of the (111) face and of the hydrated (020}, (120) and (011)
does not differ much from the one determined by Heijnen and Hartman [3] (fig. Ia). Although the hypothesis that the (011) face grows in layers of ~d011 does improve the habit, a
faces, is shown in fig. 4. This theoretical habit
better agreement can be obtained by taking the
x ~ C X 0
~
/ /
rx xC
-
-
020
0 0
o
0 -
~~~::c
r-~....,, S
~
~
.,,,(
C
N.) ~
-~
~
~
~r.
0-... N.)
—
/
-.-
/
b
o:~2Q0’
~.
~
0
~ N
~
//d011.
Fig. 3. Hydrated slices (a) (020)’ and (120)’ in a [001] projection and (b) and (011)’ in a [211] projection. See fig. 2 for the explanation of the symbols.
E. van der Voort, P. Hartman
Table I Electrostatic p. q, r, s, and
t
/ Habit
of gypsum and solvent interaction
Table 3 Sites available for strong solvent interaction
bond energies
bond
bond energy (kJ mol~)
Face
Sites
p
1736 1577 249.5
(020)’ (120}’ (011)’
None
88
(111)’
q s
449
2~and SO~ oxygens Ca None Ca2~and SO~ oxygens
90
Attachment energies of the F-faces of gypsum as determined by Heijnen and Hartman [3] and attachment energies of the hydrated slices
mately the same as bond r, is stronger than that with the SO~ ions, the strength of the latter being similar to that of bond s or t. The largest
Face
E
Bonds broken
Bonds made
E’~, (NNA)
(020) (120) {Oll}a
75.8 292.4 286.7
2r+2s 2r+t 2r
2t 2s+t 2s+2t
1997 373.2
(111)
472.4
growth rate reduction is expected for (120) and (111) faces, which have Ca2~ and SO~ ions available for solvent adsorption. The growth rate reduction of the (011) face is substantially smaller than that of the (111) face, so the latter can become the dominant terminal face in agreement with the observed habit (figs ib). When organic impurities are present in the solution, that interact more strongly with the faces than water, they may replace adsorbed water molecules on both crystallographic and non-crystallographic positions. The growth rates of the side faces (01l} and (111) will be affected equally and the theoretical habit like the one shown in fig. la may be obtained. Indeed, the (0111 } face is found to prevail over the (111) face on crystals grown from aqueous solutions with organic impurities [12].
—
-
322.5 -
E’,,
199.6
357.2 311.2
—
hydration of the surfaces into account even when growth in half layers is not considered. When growth takes place from a solution, water molecules are also present on other than crystallographic positions at the surface. As was already presumed in section 2, these water molecules reduce the growth rate. Table 3 shows which sites are available for water adsorption on non-crystallographic positions. The solvent interaction with the Ca2~ions, of which the strength is approxi~~jN~11l)
~
—
3. Conclusions
~
(020) (120)
—
‘~‘~
Fig. 4. Theoretical habit of gypsum based on the attachment energies of the (ill) and the hydrated (020), (011} and (120) faces,
The habit of gypsum crystals grown from a pure aqueous solution can be explained in a qualitative way when solvent interaction is taken into account. The habit of gypsum from solutions with organic impurities can be explained by the adsorption of the impurities. The solvent (water) should also be considered as an impurity, but in the case of a solvate (hydrate like gypsum) one should be very careful. On some of the faces water seems to act as an impurity, lowering the growth rate, and on other faces it adsorbs on crystallographic natural water positions thus leaving the growth rate more or less unaltered.
E. van der Voort, P. Hartman
450
/
References [1] B. Simon and M. Bienfait, Acta Cryst. 19 (1965) 750. [2] M.P.C. Weijnen, G.M. van Rosmalen, P. Bennema and J.J.M. Rijpkema, J. Crystal Growth 82 (1987) 509. [3] W.M.M. Heijnen and P. Hartman, J. Crystal Growth 108 (1991) 290. [4] M.R. Christoffersen, J. Christoffersen, M.P.C. Weijnen and G.M. van Rosmalen, J. Crystal Growth 58 (1982) 585. [5]E. van der Voort and P. Hartman, J. Crystal Growth 106 (1990) 622.
Habit of gypsum and solvent interaction [6] M. Rubbo, D. Aquilano, M. Franchini-Angela and G. Sgualdino, J. Crystal Growth 71(1985) 470.
[7]P. Hartman, unpublished, 1990. [8] M.H.J. Hottenhuis, PhD Thesis, Nijmegen (1988). [9] W.F. Cole and C.J. Lancucki, Acta Cryst. B30 (1974) 921. [10] C.S. Strom and P. Hartmars, Acta Cryst. A45 (1989) 371. [Ii] PP. Ewald, Ann. Physik (Leipzig) 64 (1921) 253. [12] M.P.C. Weijnen and G.M. van Rosmalen, Desalination 54 (1985) 239.