Rheology and Setting of Alkali-Activated Slag Pastes and Mortars: Effect of Organic Admixture

(ProQuest: ... denotes formulae omitted.)

INTRODUCTION

The worldwide need to reduce the energy used and the greenhouse gases emitted during cement manufacture has led to the pursuit of more eco-efficient materials, such as alkali-activated slag cements. These alkaline cements perform better than the equivalent portland cement products in terms of initial strength; heat of hydration; and resistance to acid, seawater, sulfates, and reinforcement corrosion.1-4 When the activator used is a waterglass solution, however, these materials are characterized by high shrinkage and uncontrolled setting, which translate into poor workability.1,3

Previous papers5,6 have shown that the addition of shrinkage-reducing admixtures to waterglass-activated slag mortars reduces their autogenous and drying shrinkage significantly, partially solving one of the chief problems encountered in the use of these materials. The uncontrolled setting exhibited by waterglass-activated slag pastes and mortars is due to the formation of a primary C-S-H gel in the early stages of the reaction as a result of the bonding of the Ca2+ ions present in the slag to the silicate ions in the waterglass solution.7,8 However, no solution for controlling setting in these cements or for improving their workability has been put forward to date.

While the rheological behavior of portland cement pastes and mortars has been investigated in depth,9 the rheology of alkali-activated slag materials has not yet been explored, despite the importance of such research for an understanding and explanation of the behavior of these cements and concretes during on-site placement. According to the literature,9-12 portland cement pastes and mortars behave like Bingham fluids, which means that their rheology is defined by two physical parameters: yield stress and plastic viscosity. A number of papers8,10,13,14 report that the presence of high-range water-reducing admixtures (HRWRAs) reduces these rheological parameters with a greater effect on yield stress than on plastic viscosity, and this enhances workability of the product.

By contrast, studies of the rheology of alkali-activated slag pastes and mortars are scarce. Based on results of minislump tests, Quing-Hua and Sarkar15 found that waterglass-activated slag pastes, like portland cement pastes, exhibit structural breakdown. They noted that, like the particles in portland cement pastes, slag particles form flocs when in contact with the alkaline solution. Such floc formation is consistent with the early formation of the primary C-S-H, mentioned previously as the cause of uncontrolled setting. When subjected to stress greater than the interfloc interaction force, these flocs break down, rendering the paste more fluid. Empirical tests such as minislump tests are not good indicators of rheology9 and no rheological model has yet been formulated for alkali-activated slag mortars.

Again, from minislump tests, Puertas et al.16 concluded that the inclusion of vinyl copolymer and polycarboxylatebased admixtures had no effect on the fluidity of waterglassactivated slag paste. Nothing has been reported, however, about the effect of organic admixtures, such as those commonly used to prepare conventional concretes, on the rheological parameters and, therefore, the workability of alkali-activated slag pastes and mortars.

Consequently, the primary objectives of this study were to understand the rheological behavior of alkali-activated slag pastes and mortars in relation to the effect of the nature of the activator on their rheology; it also aimed to establish the effect of one shrinkage-reducing and four HRWRAs on such behavior.

RESEARCH SIGNIFICANCE

Alkali-activated slag cements offer ecological advantages over portland cements and can attain high strengths with reduced emissions of greenhouse gases. Their performance, however, is limited by their relatively poor workability, uncontrolled rapid setting, and high drying shrinkage. While a previous paper showed that shrinkage reducing admixtures can reduce both the autogenous and drying shrinkage of waterglass-activated slag mortars, no previous research on the rheology of mixtures containing these binders has been reported. This research aimed to establish the effect of HRWRAs on the rheology and setting of pastes and mortars to develop formulations and processing conditions to enable these materials to achieve greater use.

EXPERIMENTAL PROGRAM

Materials used

A Spanish ground-granulated blast-furnace slag and a portland cement Type CEM I 42.5R17 were used and their chemical composition is given in Table 1. The specific surface of the slag and the portland cement was 325 and 360 m2/kg (1370 and 1515 ft2/lb), respectively; the vitreous phase content of the former was 99%.

Two different alkaline activators were used: 1) a waterglass solution ((Na^sub 2^O.nSiO^sub 2^.H^sub 2^O) + NaOH) with a SiO^sub 2^/Na^sub 2^O = 1.0; and 2) NaOH solution, in both cases with 4% Na^sub 2^O (by slag mass).

Five different chemical admixtures were employed-four HRWRAs: polycarboxylate-based (PC2), melamine formaldehyde derivative (M), naphthalene formaldehyde derivative (NF), and vinyl copolymer (V), and a shrinkagereducing agent (a polypropyleneglycol derivative [SRA1]).

Table 2 shows the physical and chemical characteristics of these admixtures. Dosages of admixture ranging from 0 to 2% (liquid by binder mass) were included in the mixing solution.

Tests conducted

Rheological tests on pastes-The rheological behavior of alkali-activated slag and portland cement pastes was determined along with the effect of different dosages of Admixtures PC2, M, NF, V, and SRA1 on paste rheological parameters. The admixture dosages used were 0, 0.3, 0.5, 1.0, 1.5, and 2.0% (by mass of binder), chosen to cover both low and high dosages.

Alkali-activated slag pastes were prepared by hand using a spatula mixing 50 g (1.75 oz) of slag with the respective activator solution (waterglass and NaOH, with 4% Na^sub 2^O) containing the appropriate percentages of admixture. The liquid/solid ratio (l/s) used to prepare the slag pastes was 0.5, whereas portland cement pastes were also prepared by hand using an l/s of 0.4, reflecting normal practice with these materials and their different water requirements.

Immediately after mixing, the pastes were placed in a rheometer fitted with a helical impeller18 and subjected to a five-cycle measuring procedure. During each cycle, shear rate was first kept constant at 200 s^sup -1^ for 2 minutes preshear, then ramped from 0 to 10 s^sup -1^ in 1 minute, from 10 to 200 s^sup -1^ in 1 minute, and lastly from 200 to 50 s^sup -1^ in 1 minute (making a total of 5 minutes from the start). The next cycle started immediately at 200 s^sup -1^ and test cycles were completed at 5, 10, 15, 20, and 25 minutes from the end of mixing.

According to the literature,3 waterglass-activated slag pastes suffer an uncontrolled setting. To deeply explore this negative property and to study the effect of the mixing time on the setting time and on the structural breakdown of waterglass- activated slag pastes, they were subjected to a constant shear rate of 200 s^sup -1^ for 30 minutes and the shear stress was determined according to a method first used by Tattersall.19 In addition, the setting time of waterglass-activated slag pastes prepared with an l/s of 0.5 and different times of mixing (3, 10, and 30 minutes) were determined according to the Spanish standard.17

Rheological tests on mortars-The rheological behavior of alkali-activated slag and portland cement mortars was studied along with the effect of 1% dosage of organic admixtures on their rheology.

Waterglass- and NaOH-activated slag and portland cement mortars were prepared using a mixer according to the standard EN 196-1:200517 using a siliceous sand with a maximum particle size of 2 mm (0.08 in.). A binder:sand ratio of 1:2 was used throughout. The l/s for the waterglassand NaOH-activated slag and portland cement mortars were 0.58, 0.50, and 0.42, respectively, showing the different water requirement in mortars.

The mortars were placed in a rotational viscometer20,21 immediately after mixing. The rheometer is used to obtain flow curves that have units of torque instead of shear stress in contrast to the rotational rheometer employed for pastes.14 Mortars were subjected to a five-cycle measuring procedure. In each cycle, the speed of rotation was ramped from 0 to 3.3 rev/s in 2 minutes, held at 3.3 rev/s for 30 seconds and then ramped to zero over another 2-minute period. The next cycle started immediately and testing cycles were completed at 4.5, 9, 13.5, 18, and 22.5 minutes from mixing, with no initial preshear. Additionally, waterglass-activated slag pastes were subjected to a constant shear of 3.3 rev/s for 30 minutes.

RESULTS

Paste rheology without admixtures

The hysteresis cycles plotted in Fig. 1, obtained when the test described in the Rheological tests on pastes section was run on waterglass- and NaOH-activated slag and portland cement pastes, show the nature of the activator solution to be a determining factor in the rheological behavior of alkaliactivated slag pastes.

With waterglass, the area of the first hysteresis loop was much larger than the area of the following four (Fig. 1(a)) and, in particular, much larger than the area of the hysteresis loops found for NaOH-activated slag or portland cement pastes, whose five cycles virtually coincide (Fig. 1(b) and (c)). The shear stress values found for waterglass-activated slag pastes were much higher than the values observed for the NaOH-activated and portland cement pastes.

These observations suggest that waterglass-activated slag pastes are much more highly structured than the other two, to the extent that the shear energy imparted during preshear and five cycles testing is insufficient to break down the structure, whereas the NaOH-activated slag and portland cement pastes are fully broken down by preshear alone. This could be explained by the formation of the primary C-S-H previously mentioned.

Lastly, the rheological behavior of waterglass-activated slag pastes differed completely from the behavior of the other two binding systems studied in respect of the down-ramp results. In the former, the down-ramp results describe a curve, indicating that these pastes fit not a Bingham-type fluid (refer to Eq. (1)), but a Herschel-Bulkley model (Eq. (2)),9 where t is the shear stress applied to the paste, is the shear rate, t0 is the yield stress, and a and b are the characteristic parameters describing the rheological behavior of the paste.

... (1)

... (2)

On the contrary, in both the NaOH-activated slag and portland cement pastes, the down-ramp shear stress values fit a straight line, confirming that both conform to Bingham rheology (Eq. (1)). This behavior of the waterglass-activated slag pastes is exactly what would be expected for a Bingham material whose structure is still being broken down by the shear imparted throughout the cycle. A Bingham material whose structure is broken down to equilibrium during the test cycle gives a Bingham straight-line down-ramp,9 whereas one whose structure is incompletely broken down gives a curved down-ramp, which may be fitted to the Herschel-Bulkley model with an index b greater than 1. It has been long known that thixotropic materials give downramps that can be fitted by almost any model.22

Figure 2 also shows that, whereas in waterglass-activated slag pastes, the yield stress declined progressively throughout the test, this behavior was not recorded for the portland cement or NaOH-activated slag pastes. Moreover, the initial yield stress in portland cement pastes was double that for the respective NaOH-activated slag pastes. In waterglass-activated slag pastes, the yield stress was similar to the portland cement paste values, although as the test advanced, this parameter steadily declined and ultimately reached values lower than those observed in NaOH-activated slag pastes. The plastic viscosity of NaOH-activated slag pastes, in turn, was double the portland cement paste viscosity. These changes also suggest that the waterglassactivated slag pastes are continuing to be broken down by the shear energy imparted by the tests.

Waterglass-activated slag pastes were subjected to a constant shear rate of 200 s^sup -1^ to determine their structural breakdown (refer to Fig. 3). Figure 3 shows an initial decline in paste shear stress during the first minute. Shear stress grows very quickly thereafter, peaking at 160 Pa (0.02 lb/ft^sup 3^) after 3 minutes, when the paste floc structure ruptures and the shear stress then declines progressively to the nearly constant values approximately 44 Pa (0.005 lb/ft^sup 3^). Table 3 shows that increasing the mixing time (using machine mixing) from 3 to 10 minutes retards initial setting by over 40 minutes and final setting by over 4 hours from the start of the mixing procedure. When machine mixing is extended to 30 minutes, initial and final setting are retarded by 2 and 8 hours, respectively, compared with the pastes mixed for 3 minutes.

Paste rheology with admixtures

The effects on rheological parameters of adding between 0 and 2% of HRWRA-V, M, NF, and PC2-as well as shrinkage-reducing admixture-SRA1, are discussed in the following.

As in the pastes without admixtures, waterglass-activated slag pastes with organic admixtures showed breakdown with a Herschel-Bulkley type down-ramp and a progressive decrease in yield stress with time as the number of test cycles increased. The inclusion of the vinyl copolymer (V) (Fig. 4(a)) and the shrinkage-reducing admixture (SRA1) (Fig. 4(d)) increased the yield stress by up to 80% without affecting the progressive decrease with time. The presence of a polycarboxylate HRWRA (PC2) (Fig. 4(e)), however, increased yield stress to nearly twice that of the control. This increase, however, was observed in the early stages of the test only, for at later times, the value was barely different, as the yield stress with PC2 fell to that of the control. Lastly, the inclusion of the naphthalene (NF) (Fig. 4(c)) and melamine (M) (Fig. 4(b)) admixtures had little effect on the yield stress of the pastes.

As in the pastes without admixtures, NaOH-activated slag pastes with admixtures exhibited Bingham behavior and a progressive increase in yield stress with time. In this case, the admixture that reduced the yield stress value most significantly was the naphthalene derivative (NF) (Fig. 5(c)) by up to 80% with a dosage of just 0.3%; with dosages of over 0.3%, however, the paste was observed to segregate. The other HRWRAs reduced paste yield stress slightly, although this parameter rose during the test, particularly with the melamine derivative (M) (Fig. 5(b)) at dosages of 0.3, 0.5, and 1%. Lastly, the shrinkage-reducing admixture (SRA1) (Fig. 5(d)) had no effect on the yield stress of the paste. Plastic viscosity was essentially unchanged in the presence of the various admixtures, remained constant with time within experimental error, and is not plotted herein.

In portland cement pastes, the inclusion of admixtures did not change the rheological behavior from Bingham type. The presence of HRWRAs decreased the yield stress by over 90% (refer to Fig. 6). In this case, 1% dosages of the polycarboxylate (Fig. 6(e)) and vinyl copolymer (Fig. 6(a)) led to paste segregation, as did the use of 1.5% dosage of the melamine (Fig. 6(b)) and naphthalene (Fig. 6(c)) derivatives. On the contrary, the shrinkage-reducing admixture (Fig. 6(d)) increased the yield stress by up by 30% above the value recorded for portland cement pastes without admixtures. Finally, plastic viscosity was unaffected by the polycarboxylate and shrinkage-reducing admixtures, but dropped slightly in the presence of the other admixtures. Both yield stress and plastic viscosity remained essentially constant over the time scale of the experiments.

Mortar rheology

The hysteresis cycles for alkali-activated slag and portland cement mortars tested as described in the "Rheological tests on mortars" section are shown in Fig. 7. This figure shows graphs of the torque T exerted by the paddle against speed of rotation N. This presentation of the flow curve follows that described by Tattersall and Banfill,11 where for a Bingham material

T = g + hN (3)

where g is proportional to yield stress and h is proportional to plastic viscosity, the constants of proportionality for the viscometer being capable of estimation by a routine calibration procedure.21 In this case, the absolute values are unimportant, and the data are left in units of torque and speed. As observed in the pastes, the nature of the alkali solution is a decisive factor in the rheological behavior of alkali-activated slag mortars.

With waterglass (Fig. 7(a)), the flow curve exhibits significant hysteresis with a curved down-ramp fitting the Herschel-Bulkley model (Eq. (2)) with an index b greater than 1, whereas with NaOH solution (Fig. 7(b)), the systems, like the portland cement products (Fig. 7(c)), behaved like Bingham fluids, with a straight down-ramp.

The yield torque g decreased during the test in all three binder systems (refer to Fig. 8). This means that, in all cases, the number of linkages broken, that is, flocs separated, grew with the amount of time shear was applied and that the preparation method imparts insufficient shear energy to the mortar to break its structure down to equilibrium (note that there is no period of intensive preshear in the mortar cycle). The yield torque in portland cement mortars was higher-up to double the value recorded for alkali-activated slag mortars.

Additionally, Fig. 9 shows that the torque in a waterglassactivated slag mortar subjected to constant shear of 3.3 rev/s decreases gradually to a constant value of approximately 0.05 N-m (0.037 lb-ft).

The effect of admixtures on the rheology of mortars was studied at a single dosage (1% by binder weight). Waterglass- activated slag mortars containing admixtures exhibited Herschel-Bulkley behavior. All the admixtures except the polycarboxylate HRWRA reduced the yield torque g by up to 50%, although after 9 minutes, the presence of the admixture induced no significant further change in its value (Fig. 10(a)).

Except for those containing the polycarboxylate HRWRA (PC2), the NaOH-activated slag mortars behaved like Bingham fluids (Fig. 10(b)). In these mortars, the greatest decrease in the yield torque was obtained with the naphthalene HRWRA (Fig. 10(b)); the negative values for this rheological property generated by a 1% dosage of this admixture are, however, an indication of mortar supersaturation and the need for a lower dosage. The NaOH-activated slag mortars containing polycarboxylate HRWRA (PC2) conformed to the Herschel-Bulkley model (refer to Fig. 11).

Melamine HRWRA reduced the yield torque in NaOHactivated slag mortars significantly at first but the value rose rapidly as was observed in the equivalent pastes. Lastly, the vinyl copolymer HRWRA reduced the yield torque in the first 10 minutes, whereas the shrinkage-reducing admixture had no effect.

The portland cement mortars containing HRWRA PC2 could not be tested because the 1% dosage caused mortar segregation. The inclusion of melamine and naphthalene derivatives, however, reduced the yield torque of portland cement mortar by up to 80% and 73%, respectively.

Finally, the effect of admixtures on plastic viscosity differed between the mortar types (Fig. 12). In NaOHactivated slag mortars, the inclusion of the various admixtures, and in particular, the naphthalene derivative, raised plastic viscosity, whereas the plastic viscosity of portland cement mortars decreased, with the greatest difference recorded with the vinyl copolymer.

DISCUSSION

The obtained results reveal hitherto unreported differences in the rheological behavior of alkali-activated slag pastes and mortars, depending on the activator used, as well as in the effect of different HRWRAs and shrinkage-reducing admixtures on rheological properties.

In pastes, when the alkali activator used was waterglass, the yield stress values found for the first cycle up-ramp curve (refer to Fig. 1(a)) were much greater than the down-ramp curve values, resulting in a very large area between the two. This hysteresis is an indication that, in the case of waterglassactivated slag pastes, neither the mixing nor the 2 minutes of treatment at 200 s^sup -1^ before the test cycle sufficed to rupture and separate the flocs formed as a result of the alkali-activation of the slag; rather, such separation continued to take place during the test, as the time of application of shear stress increased. In Cycles 4 and 5, in turn, the area was much smaller than in the earlier cycles, for the up- and downramp curves practically coincide.

Yield stress also decreased over time when waterglassactivated slag pastes were subjected to a prolonged constant shear rate of 200 s^sup -1^. This unexpected behavior of waterglass-activated slag pastes may be explained by the mechanism governing the reaction. Immediately after contact with the waterglass solution, the slag particles are surrounded by a thin layer of primary C-S-H formed by the interaction between the silicate ions in the waterglass solution and the Ca2+ ions in the slag.23 Under these circumstances, the slag particles, attracted to one another by colloidal forces, form flocs. During mixing and treatment before rheological testing, these flocs are partially separated, but the rapid precipitation of massive amounts of primary C-S-H gel continues, generating larger flocs. The result is that greater and greater shear stress is needed to start flow. Once the threshold value is exceeded, however-in this case, 3 minutes into the test-the floc structure breaks down entirely with a significant decline in shear stress.

Support for this notion is shown by the fact that the waterglass- activated slag pastes not only did not harden, but remained fluid in the 30 minutes of this test and also in the 25 minutes of the cyclic tests. This contrasts strikingly with previous studies6,8,24 that report that these pastes may set and harden in the first 40 minutes.

Setting time tests confirm the impact of prolonged shearing on the setting time of waterglass-activated slag pastes. An increase of 2 and 8 hours in the final setting time was observed in waterglass-activated slag pastes when mixing is extended to 10 and 30 minutes, respectively, compared with pastes mixed for 3 minutes (refer to Table 3). Consequently, it may be inferred that an increase in the time that shear stress is applied to waterglass-activated slag pastes prolongs the setting times. This novel finding has not been reported before and has the potential to overcome one of the chief problems in waterglass-activated slag pastes, namely, rapid setting.

The rheological model governing the alkali activation of slag pastes is observed to depend on the nature of the alkaline activator used. In waterglass-activated pastes, the down-ramp curve fits the Herschel-Bulkley model because of incomplete structural breakdown, whereas both NaOH-activated slag and portland cement pastes exhibit Bingham fluid rheology. The common variable in the two rheological models is yield stress, which is a measure of the strength and number of the interactions (flocs) that must be broken by the applied shear before the material flows.

The initial yield stress of waterglass-activated slag pastes is nearly double that for NaOH-activated slag pastes, confirming that the flocs are stronger. As the test advances, however, yield stress declines, as the number of bonds remaining between particles decreases.11

In NaOH-activated slag pastes, only the naphthalene derivative HRWRA of all the admixtures tested enhanced fluidity significantly. This is because the naphthalene admixture is stable and retains its plasticizing properties in the alkaline solution.25 In contrast, melamine, polycarboxylate, and vinyl copolymer admixtures are unstable in high-alkaline media and, as a result, they lose their fluidizing properties. The presence of melamine admixture initially produces a significant reduction of the yield stress in NaOH-activated slag pastes, but fluidity is quickly lost as a consequence of this instability of melamine admixture in highly alkaline media.25,26 It is not clear why the increase of the yield stress in NaOH-activated slag is higher in the presence of melamine admixture compared with the other admixtures, and further work would be necessary to answer this. In addition, in waterglass-activated slag pastes, the admixtures not only fail to improve fluidity, but in two cases-the vinyl copolymer HRWRA and the shrinkage-reducing admixtures- they actually raise the yield stress. Moreover, HRWRAs, except the naphthalene derivative, enhance the fluidity of alkali-activated slag paste only slightly during the first few minutes of the test, again due to their instability in highlyalkaline media.25 Finally, despite its stability in highly alkaline solutions, the shrinkage-reducing admixture did not modify the fluidity of NaOH-activated slag pastes.

All this contrasts with the findings for portland cement pastes, whose yield stress was reduced by up to 90% by all the HRWRAs studied, confirming what many authors have previously reported.9,10,27,28

In mortars, regardless of the activator solution used, the first hysteresis cycle had a larger area than the others, an indication that the flocs were broken down during testing (refer to Fig. 7(a) and (b)). Hysteresis in waterglass-activated mortars was less pronounced than in the equivalent pastes and, in addition, under the constant shear of 3.3 rev/s, there was no initial increase in torque. In contrast, the torque at constant speed gradually declined to a constant value of approximately 0.05 N.m (0.037 lb/ft), suggesting that the presence of the aggregate in mortar and the more intense mixing procedures with respect to the mixing by hand in pastes, increases the shear energy applied to the binder and breaks down the flocs initially forming in the primary C-S-H gel.

In waterglass-activated slag mortars, the down-ramp curve fits the Herschel-Bulkley model (Eq. (2)), whereas both NaOHactivated slag and portland cement mortars fit the Bingham model. This suggests that there is much less structure remaining in the mortar after the completion of the up-ramp than in the equivalent paste. The yield torque in portland cement mortar was double that for both alkali-activated slag mortars (refer to Fig. 8). This is an indication that the interactions between the flocs formed were much stronger in portland cement mortars than in alkali-activated slag mortars.23 Plastic viscosity, in contrast, was twice as high in NaOH-activated slag as in portland cement mortars (refer to Fig. 12). This suggests that although the interfloc interactions were stronger in portland cement mortars, the size and number of flocs were larger in NaOH-activated slag mortars.

Adding 1% of admixture to waterglass-activated slag mortars had little effect on the yield torque, with a small decrease only in the early stages of the test. As before, this is due to the instability of HRWRAs in waterglass.25

All the NaOH-activated slag mortars exhibited Bingham behavior, except those containing the polycarboxylate admixture, which fit the Herschel-Bulkley model. Papo and Piani13 observed similar behavior in portland cement pastes containing polycarboxylate admixtures and suggested that the high concentration of deflocculating agents may have caused particle aggregation instead of repulsion. The stability of the naphthalene-derivative admixture in highly basic media25 was confirmed, with a significantly reduced yield torque. In fact, 1.0% dosage of this admixture is probably an overdosage and less could be used. Once again, while the melamine and vinyl copolymer HRWRAs reduced yield stress in the early stages of the test, this parameter later rose due to instability of both admixtures in highly basic media and the concomitant loss of their fluidizing properties.

CONCLUSIONS

The chief conclusions to be drawn from the present study are as follows:

1. The rheology of alkali-activated slag pastes and mortars depends strongly on the nature of the alkali activator used. When the activator solution is waterglass, they conform to a Herschel-Bulkley model with pronounced structural breakdown, whereas when NaOH is the activator, these pastes and mortars behave like Bingham fluids;

2. The greatest drop in yield stress in alkali-activated slag cements is found when a naphthalene-derivative HRWRA is added to NaOH-activated slag pastes and mortars because of its inherent stability in alkaline media. The other admixtures incorporated in alkali-activated slag pastes and mortars, however, do not significantly modify their rheological parameters; and

3. Increasing the mixing time can dramatically retard the setting of waterglass-activated slag pastes, with initial and final setting times of nearly 3 and 10 hours, respectively, now a possibility. This offers a practical, useful way of controlling the undesirable rapid setting of these materials, which has the potential for facilitating a wide range of technological applications.

ACKNOWLEDGMENTS

Funding for Project MAT 2001-1490 was provided by the Spanish Ministry of Science and Technology (MCyT). M. Palacios worked under a fellowship awarded by the Regional Government of Madrid.