The interaction of flotation reagents with reacting surfaces. A study on the flotation of cement

Chemical Engineering Science. Vol. 40, No. Printed in Great Britain.

II. pp. 2141-2148.

1985. 0

00@~2509/85 S3.00+0.00 1985. Pergamon Press Ltd.

THE INTERACTION OF FLOTATION REAGENTS WITH REACTING SURFACES. A STUDY ON THE FLOTATION OF CEMENT E. NAGELE University GesamthochschuleKassel, FB 14, MBnchebergstr.7, D-3500 Kassel. F.R.G. (Received

12 Juty 1984)

Abstract-Fresh concrete can be separated into cement and aggregate by a single-steptlotation process in aqueous suspension. The results of an experimental investigation comprising zeta-potential measurements, SIMS analyses and flotation testsare discussed.The topochemicalreactions,which permit the adsorption of collector on cement, ace consideredin detail, because cement is a material that reacts vigorously with water in aqueous suspension. The cement Rotation process can be used to determine the composition of fresh or hardened concrete.

Table 1. Data of the cements

1. INTRODUCTION

Flotation is widely used in mineral processing technology. The mechanisms of the flotation of sulphideand oxide-type minerals are quite well understood, although there still exist unsolved problems due to the large number of parameters afFecting any flotation process [l. 21. However, no information exists about the flotation behaviour of materials having a reacting surface. This problem is encountered in the flotation of materials like cement. It has been shown that cement can be floated easily, using alkyl sulphates as collectors [3]. In fresh concrete the ratio of the specific surfaces of cement and aggregate is roughly inverse to their mass or volume ratio. In addition, in a normal concrete mix only the cement is able to react with water. This reactivity, together with the surface ratio of cement and aggregate, led to the conclusion that selective separation of cement and aggregate may be achieved by means of flotation. For an optimum design of this process, knowledge of the basic mechanisms governing the adsorption of collector on cement is essential. Due to the reaction of the cement with water, the situation on the surface of cement is very complicated. The aim of this paper is to show that the flotation process can be used to perform quantitative separation of cement and aggregate. The reagents suited for cement flotation are presented and the adsorption mechanism and other interesting properties of this rather special flotation system are discussed. 2. EXPERIMENTAL

2.1. Materials All experiments described in this paper were performed with a German portland cement PZ 35 F and a German blast furnace slag cement HOZ 35L, both with a specified strength after 28 days of 35 N/mm2 [4]. The compositions of the cements are given in Table 1.

PZ 35 F (a) Composition (% by weight) Loss on ignition 1000°C Insoluble residue CaO SiOa Al,% Fe&b so3 MgO

Nat0 R,D Mn203 s-

(b) Specific surface (d/kg) cording to Ref. [4]

HOZ

35 L

1.62 0.73 63.20 18.40 3.94 3.91 3.08 2.40 0.29 1.61 0.39 -

0.70 0.80 48.30 28.30 13.10 1.50 1.70 3.50 0.36 0.71 0.44 0.60

315

398

ac-

Mortars with a maximum grain size of the aggregate of 1 mm were used instead of concrete in the flotation experiments [3]. The mortars were made according to German standards [4] with the above cements and siliceous aggregate from the Rhine. The water/cement ratio was 0.5 throughout the tests. 2.2. Zeta-potential measurements The electrokinetic mobility of cement particles was measured using a Zeta-meter-microelectrophoresis apparatus, and the zeta-potentials were calculated according to Ref. [S]. The zeta-potentials of both cements were measured in distilled water for up to 30 min after mixing. The results of these measurements are presented and discussed in Section 4.2. Due to the high electrical conductivity of the suspension, it was not possible to perform such measurements with the flotation reagents added. 2.3. Secondary ion mass spectrometry (SIMS) SIMS measurements were performed at DornierSystems GtnbH, Friedrichshafen, F.R.G., using a

2141

E. NAGELE

2142

Balxers SIMS-spectrometer. Both cements were hydrated in water for 10 min and depth profiles of various elements were recorded down to 100 nm from the particle surface for these hydrated samples. The mass spectra obtained were evaluated using a wellestablished empirical model, because a complete theory of the SIMS process does not yet exist [6].

2(2CaO.

2(3CaO - SiOz) + 6HsO 3Ca0 - A120, 4CaO.

2.4. Chemical analyses The effect of some flotation reagents on the reaction between cement and water was studied by preparing an aqueous suspension of the cements with a water/cement ratio of 2.0. This suspension was stirred vigorously in a glass beaker and the reagents under consideration were added 1 min after mixing cement and water. Samples were then taken from the suspension 5, 10, 15, 30, 45, 60 and 120 min after mixing cement and water. The samples were filtered through a Btichner funnel under suction and the clear filtrate was analysed for Ca2 + and OH - ions by titration with 0.1 N EDTA and 0.1 N HCl, respectively.

2.5. Flotation tests The flotation tests were performed in a commercially available laboratory flotation machine, type MN 934, manufactured by KHD Humbold-WEDAG, F.R.G. The machine was equipped with a 4 1 PMMA vessel covered with a perforated plate. A typical flotation experiment was performed as follows: the vessel was filled with 3 1 of tap water and the sample consisting of 300 g of mortar as described above was added. The sample was suspended by stirring vigorously. After 1 mitt of stirring, the reagents were added and allowed to react for 1 min. Then the air supply was opened and the flotation started. The froth was removed continuously and collected in a porcelain dish. At the end of the test, the cement floated was dried at 105°C and subsequently weighed. The results were calculated as per cent recovery of the cement.

3. SURVEY

OF CEMENT

CHEMISTRY

WITH

RESPECT

TO FLOTATION

It is useful to summa rize here some general aspects of the chemistry of cement. This may help the reader through the rather special topics of the following sections. Portland cement consists of portland cement clinker and about 5 per cent by weight of a calcium sulphate. Portland cement clinker is a mixture of tricalcium about 65% by mass), dicalcium silicate (3CaO*Si02, stlicate (2CaO. SiOz, about 15 O%by mass), tricalcium about 10 % by mass) and aluminate (3Ca0.Alz0,, tetraealcium aluminateferrite (4Ca0.A1203.FeZOj, about 5 % by mass). The overall hydration reaction of Cement with water is given by the following scheme [8]:

SiO*) + 4HsO + 3CaO. 2SiOt . 3Hz0 + ‘XOH),

+ Ca(OH),

(2)

+ 12Hz0 --, 4CaO - AlaOj - 12H20

(3)

Al,OJ - Fe,O, + 2Ca(OH), 3CaO. A1303 ‘3CaO.

3(3CaO - Al,O,)

(1)

-, 3CaO. 2SiOo * 3H,O + Ca(OH)s

+ 10HzO + Fez03. 6HsO

(4)

+ 3(CaSOI _2H10) + 26H,O + 9CaO. 3Al,O, .3CaS04. 32H20.

(5)

The behaviour of this system of simultaneous and interacting chemical reactions provides the technical basis of concrete manufacture and hence has been studied extensively. For recent reviews see, for example, Refs [7] and [S]. It was shown that about OS-2 y0 of the clinker reacts during the first 5 min of hydration [8]. Then a comparatively long period of chemical inactivity follows, called the “induction period”. During this period, which may last up to 6 h, the concrete remains workable. Then the reaction starts again, leading to the final hydration products, which are calcium silicate hydrates, calcium aluminate hydrates and calcium hydroxide. The induction period is of special importance for the cement flotation. If we look at the outer surface of a clinker particle during the first period of the hydration reaction, we note that the composition of the surface changes from pure silicates to silicate hydrates with calcium hydroxide and finally to sulphoaluminate with calcium hydroxide. The composition of the outer layers then remains rather unchanged during the induction period. Hence, fresh concrete is an aqueous suspension of high pH value containing the following particles: calcium sulphate, portland cement clinker covered with layers of sulpho-aluminates and aggregate. The aggregate can consist of a large variety of minerals and for simplification will be treated in this work as pure quartz (Si02). Blast furnace slag cement is a mixture of portland cement and amorphous blast furnace slag. This slag also reacts according to reaction (5) and hence all the purely qualitative considerations in this section are also valid for blast furnace slag cement. There is only one important difference: the slag contains no and hence both the pH value and the Ca(OH)z Ca(OHh concentraticms in the suspension and in the surface layers are significantly lower than those with portland cement. 4. RESULTS

AND

DlSCUSSION

4.1. SZMS measurements In Fig. 1 the depth profiles of Mg, Ca and SO, after 10 min of hydration are shown for the portland cement. Similar results are obtained for the blast furnace slag cement. From these depth profiles the depth of hydration can be calculated directly, because

Interaction of tlotation reagents with reacting surfaces

0.1

1.0 D,r+ance

100.0 nm

*).O

from

surface

the

Fig. 1. Depth profiles of Mg, Ca and SOL ions for portland cement hydrated for 10 min.

the compositions of the primary hydration products and the unhydrated cement differ significantly [6]. The degrees of hydration calculated from these data according to Ref. [6] agree well with the data obtained by other methods [S]. For portland cement the hydration depth after 10 min of hydration is about 15 nm, whereas for blast furnace slag cement 7 nm has been measured. Hence, after 10 min of hydration only a thin layer of about 2040 atomic layers on a cement particle is in a hydrated state. After 10 min of hydration the outer layers of both cements also contain large quantities of S04containing phases, whereas the alkali ions have been dissolved in the surrounding water. These results agree well with the zeta-potential measurements and they elucidate the composition and structure of the outer shell of a reacting cement particle in water [6]. 4.2. Zeta-potential measurements In Fig. 2 the zeta-potentials of the portland cement and of the blast furnace slag cement in aqueous suspension are plotted against the hydration time. The zeta-potential of the portland cement is positive due to

2143

the adsorption of Ca’+ ions from dissolved calcium sulphate and calcium hydroxide on the surface of the cement particles. The blast furnace slag cement, however, shows a negative zeta-potential, which may be attributed to SO;or OH- ions on the surface of the particles. Thus the composition of th/e surface must be different for portland cement and blast furnace slag cement. As the hydration proceeds, the zeta-potentials of both cements go to zero because the hydrating cements release ions into the surrounding solution, thus increasing its ionic strength. Hence the absolute values of the zeta-potentials decrease with increasing hydration time and converge towards zero. These results are in good agreement with data reported in the literature [9, 10-J. Hinterthan measured the effect of various ions on the zeta-potential of pure clinker minerals in aqueous suspension [l l]. He found that all clinker minerals have positive zeta-potentials at pH values above 11.0, i.e. in NaOH or KOH solutions with an ionic strength above 10T3 N. Thus, if the pH value of the flotation pulp is increased above 11 by adding NaOH or KOH, all clinker minerals should have the same flotation properties, which is essential for quantitative flotation of cement. In addition, the calcium sulphate particles also have a high positive zeta-potential at these high pH values and thus can be floated together with the clinker particles [12]. So, zeta-potential measurements provide the basis for the design of an effective flotation reagent system for cement flotation. 4.3. Collectors for cement flotation (4.3.1) General considerations. Having studied the surface chemistry of hydrating cement particles, the next step is to design a proper set of reagents for the flotation of cement. According to the results of the preceding sections, the collectors must adsorb on cement particles, i.e. on

Zeta.Potenkal

imV1 10

._ Fig. 2. Zeta-potential

7

3

30

10

me

[nml

of portland cement (PZ) and blast furnace slag cement in water as a function of hydration time.

2144

E. NAGELE

the hydrated outer layers of the clinker particles and the calcium sulphate particles, respectively_ Furthermore, there should be no adsorption of collector on the aggregate particles. As the clinker particles react chemically with the surrounding water in the flotation pulp, this reaction must be used for the separation of cement particles from the aggregate particles. From general physicochemical arguments it is clear that the aggregate surface will also be hydrated in an aqueous suspension. But in general this hydration is confined to the weak adsorption of some layers of water. This adsorption can be treated as being in thermodynamic equilibrium with the surrounding liquid at any time of the hydration reaction. In contrast, the cement surface is far from being in equilibrium with its environment. There are reactions both at the surface [mostly the incorporation of sulphate ions and the binding of water according to reaction (S)] and at the boundary between hydrated material and unhydrated clinker. Now, if the collector is able to participate in this reaction, selective flotation of cement from the aggregate is possible. The collector then becomes part of the new surface layers by a chemical reaction, which may be described as an incorporation of the polar group of the collector into the product layers on the surface. Furthermore, as the surface compositions of quite different cements are very similar, though not necessarily the same, as has been mentioned above, such collectors should be applicable for all technical cements based on silicate hydration. The chemisorption of the collector to a reacting surface is the main difference between cement flotation and conventional flotation processes with oxide minerals, where a reaction of the surface normally leads to desorption of the collector, which in this case is adsorbed physically in the Stern layer [2, 13). Furthermore, the chemisorption of the collector leads to quantitative flotation of the cement and to selectivity of the separation of cement and aggregate that is far better than usually found in flotation processes with oxide minerals. Hence the process can be applied to the analysis of fresh concrete.

(4.3.2) Flotation tests. From the arguments stated in the previous sections it follows immediately that cement flotation should be performed in a highly alkaline medium, where all cements show positive zetapotentials. Therefore, all subsequent flotation experiments were performed in 0.02N NaOH solution unless stated otherewise and only anionic collectors were used. The mortars were mixed 5 min prior to flotation. For the first series of tests, C,,-surf&&ants with difl’erent types of end-groups were selected. The collector concentration was kept constant at 1500 ppm with respect to the cement. The results of these tests are shown in Table 2. The carboxylate is precipitated by the Ca2+ ions in

Table 2. Recovery of cement with various collectors

Collector Sodium laurylsulphate (SLS) Sodium laurylsulphonate Sodium laurylsulphate .5 PEO

Di-sodium laurylsulphosuccinate

Laurylsulphosuccinamide Dodecylphosphate disodium salt Dodecylphosphate .5 PEO disodium salt

CI,H2&OONa

Recovery of cement ( % hy weight) Portland Blast furnace cement

slag cement

100 2

100

90 90

loo 100

92

100

94

96

86

0

95

0

the pulp as the insoluble Ca’+ salt and hence cannot act as a collector for cement. All other anionic surfactants tested can act as collectors for cement. The best results were obtained with sodium laurylsulphate (SLS) and sodium laurylsulphonate. The phosphates showed good recoveries but had a bad froth structure and a poor froth volume, which made the flotation difficult. The higher recoveries obtained with the blast furnace slag cement in these experiments than with portland cement are due to the lower particle size of this cement (see also Table 1 for specific surface)_ The non-floating material from the portland cement are the particles with diameters about 1OOpm. The flotation machine available cannot generate air bubbles large enough to goat these large particles. X-Ray diffraction analysis showed that all of the calcium sulphate had been floated. 4.4. Efect of collector concentration Cement flotation requires surprisingly low collector concentrations. The concentrations involved correspond to values known for the flotation of other oxide minerals [2, 13, 141. The effect of collector concentration was studied using SLS as the collector. The concentrations were varied between 750 and 3000 ppm with respect to the cement. Since both the amount of cement in the samples and the pulp volume were kept constant throughout the tests, these concentrations correspond to 0.85 x 10e4 and 3.4 x 10d4 mol/l of collector, respectively. The results of these tests are given in Fig. 3, where the amount of cement floated is plotted against the square root of the collector concentration. For the portland cement there is no flotation below a critical value, which was estimated from additional experiments to be about 600ppm. Above this value, the recovery increases rapidly and then remains nearly constant for collector concentrations between 1500 and 3000 ppm. The recovery of the blast furnace slag cement increases proportionally to the square root of the collector concentration in the concentration interval under consideration.

Interaction of flotation reagents with reacting surfaces

x 0

Blast s1a.g

2145

furnace

cement Portland cement collectaconcentration r-at scale

square 3000

Fig. 3. Effect of colkctor

concentration

on the recovery of portland cement.

These results can be interpreted in terms of different adsorption behaviour of the collector on both cements. The reason for this behaviour is the different structures of the hydration products on portland cement and blast furnace slag cement. On portland cement there is a comparatively thick layer of hydration products (at least 2040 monolayers) and hence the collector is fixed quite well by incorporation of the SO0 group into the hydration products of the outermost layers. This binding is not significantly disturbed by the hydration of unhydrated clinker 40 layers deeper. The binding energy is strong enough to prevent the collector molecule from escaping from the surface. If the concentration of collector in the pulp is low, the surface cannot be covered to a sufficient degree by the collector and hence there is no flotation. If there is enough collector present in the pulp to make the surface hydrophobic, the collector will be bound to the surface and flotation will occur. If the surface is su!Iiciently hydrophobic, any further increase in collector concentration wiI1 increase the recovery only because of secondary effects, as, for example, the modscation of the bubble size distribution, etc., but not because more collector is adsorbed on the cement. Hence the recovery does not depend much on the collector concentration if it is maintained above a certain critical value. The adsorption of &Rector on portland cement can thus be described by a Langmuirtype adsorption isotherm [lS]. These considerations confirm the data presented in Refs [lo] and [14] for calcium silicates and aluminium silicates, respectively. The thickness of the product layers around the particles of blast furnace slag cement is much less and hence the hydration reaction occurring in the deeper layers significantly affects the collector adsorption by lowering the bond energy between collector and surface. The lower bond energy between collector and cement leads to a considerable exchange of collector molecules between the cement surface and the pulp. The rate of this exchange, however, is determined significantly by the bulk concentration of collector. The square-root concentration dependence therefore

I wm1

cement and blast furnace slag

indicates that collector adsorption on blast furnace slag cement is a diffusion-controlled process. This simple model will surely need some refinement in the future. But it is capable of describing the overall effects correctly and is sufficient for practical use. A detailed study of the adsorption mechanism of collectors on reacting surfaces is far beyond the scope of this paper. But as a result of the present investigation, it can be stated that the adsorption energy of collector on cement depends on the reactivity of the cement with respect to water. 4.5. The eflect of the hydrophobic

groups in the collector

molecule

To study the effect of the hydrocarbon chain length of the collector, additional experiments were performed as described in Section 2.5, using sulphate collectors with different lengths of the hydrocarbon chain. The hydrocarbon chain length was varied between C4 and C1s and only pure substances were investigated except for a sulphated coconut fat. The results of these experiments are presented in Table 3 and may bc interpreted as follows. Four C-

Table 3. Recovery of cements with sulphatecollectors with different lengths of the hydrocarbon chain nc (collector concentration2000 ppm) Recovery of cement ( % by weight) Collector Butylsulphate sodium salt Nonylsulphate sodium salt Dodecylsulphate

sodium salt

Hexadecylsulphate sodium salt Sulphated

coconut fat

nc

Portland cement

Blast furnace. slag cement

4

0

0

9

72.3

0

12

95

100

16

20

20

12-18

100

100

E. NAGELE

2146

atoms in the hydrophobic part of the collector are not enough to make the surface sufficiently hydrophobic for flotation. Hence butylsulphate cannot act as a collector for cement. Nonylsulphate can float portland cement but not blast furnace slag cement. Dodecylsulphate (SLS) is capable of floating both portland cement and blast furnace slag cement. Pure hexadecylsulphate is nearly insoluble in water and hence provides a very low eff‘ective concentration of collector. Therefore the recovery of cement is low and hexadecylsulphate is not a good collector for cement. If a C,2-C18 mixture such as the sulphated coconut fat is used, the solubility of the C1,X,8 fraction is increased significantly by the shorter components of the mixture and hence the longer hydrocarbon chains are capable of repelling the cement particles. So, these mixtures are excellent collectors for cement. With respect to the effects of the hydrocarbon chain length, the cement flotation system behaves as a normal flotation system of oxide minerals [2]4.6. The eflect of NaOH and SLS on the hydration of ce?nent In Fig. 4 the ratio of OH- and Ca2+ ions liberated from the portland cement is shown for distilled water, 0.02 N NaOH and a mixture of 0.02 N NaOH/1.7 x 10-4 N sodium laurylsulphate. In distilled water the portland cement releases more OH--ions than corresponding to Ca(OH),. The reason for the positive zeta-potential of this cement in water is therefore an excess of Ca2+ ions on the surface of the hydration products_ This excess originates from a stronger dissolution of OHions compared to Ca2+ ions, which can be understood by simple equilibrium thermodynamics considering the concentrations of the ions present. The solution surrounding the cement particles contains Ca2+ ions from the sulphate salts, thus limiting the dissolution rate of Ca 2+ but not those of other ions such as OH-, Na+, K+, etc. If a surfactant such as SLS is added to the cement suspended in NaOH solution, the surface properties are changed significantly. The

,o ‘D

z

1

I f T

dissolution rates of Ca2+ and OH- are much lower when SLS is present. Now it takes the system 30 min to reach a OHjCa ratio of 2. After 60 min the surfactant has lost its effect, probably because it has been incorporated in the hydration products growing on the cement surface. The significant effect of SLS on the hydration reaction in the first 30 min must be interpreted in terms of a stopped hydration reaction which is due to the hydrophobic character of the cement surface after SLS addition. Thus SLS is able to suppress the hydration of cement for a comparatively long period of time. This is a very important effect because it allows the cement recovered by flotation to be dried after filtration without the danger of subsequent hydration.

4.7. The effect of NaOH in cement flotation For siliceous materials, H+ and OH- are potentialions determining ions [2, 5, 93. Furthermore, OHafikct the reactions at the cement surface and hence also affect the flotation behaviour. In concrete a pH value of 12.3 [saturated Ca(OH)3 solution] is convenient. pH values above 12.3 result in precipitation of Ca(OH)2 on the surfaces of both the cement and the aggregate particles and therefore must be avoided in cement flotation. So the optimum PI-I-range for cement flotation is between 11.0 and 12.3 and alkali hydroxides are activators for cement. In Fig. 5 the effect of NaOH additions to the flotation pulp is shown for both cements. Because PZ35 F liberates more OHions, its recovery depends less on the concentration of OHions added than does the recovery of the blast furnace slag cement. It is interesting to note that cement flotation can be performed even in 0.1 N NaOH solution, but then due to the Ca(OH)2 precipitation, fine aggregate is floated too. On the other hand, high pH values result in an enhanced flotability for all kinds of cement and other pozzolanic material. So, there is a possibility of separating different kinds of cement and other poz-

x Portland

cement

without

l

Portland

cement

wth

NaOH

NaflH

0

Portland

cement

wth

NaDH

and

SLS

i 5

10 15

30

Fig. 4. Ratio of OH-/Ca

45

60

120

lime

Iminl

‘+ ions for portland cement in various solutions.

Interaction of flotation reagents with reacting surfaces

concentratm

of

2147

activator

Fig. 5. Effect of activator concentration on the recovery of cement (SLS as collector).

zolanic materials by means of flotation using adequate pH values in the pulp. 4.8. Time-dependent eficts The most striking effect in cement flotation is the independence of cement recovery on the hydration time. This enables us to analyse fresh concrete at any stage of the working process and makes the flotation method a valuable tool in this field. However, the cement does react with water in a fresh concrete mix, forming hydration products. This reaction is expected to disturb the adsorption of collector, at least when the hydration restarts after the induction period. To study this effect, flotation tests were performed on cement and cement paste hydrated up to 28 days. The paste was crushed in a ball mill in order to pass a 0.063 mm sieve prior to flotation The results obtained for the pure cements and pastes are summarized in Fig. 6, where the recovery ofcement is shown as a function of the hydration time of the cement paste. The values at zero hydration time are the results obtained from the flotation of the dry mixes of cement and aggregate. The recovery of cement increases during the first 5 min after the addition of mixing water. This is due to

the strong reaction in the initial period, where, after the formation of primary silicate hydration, mainly tricalcium aluminate and sulphates react to give ettringite [reaction (5)]. Also, large quantities of calcium hydroxide are formed. Thus the chemisorption of the collector is disturbed and the recovery of cement is smaller than at times higher than about 5 mitt, when the cement is in the induction period. Because this reaction slows down with an increasing number of layers of primary hydrates, there is better adsorption of collector with increasing hydration time. During the next few hours, i.e. during the induction period, the recovery of cement is constant and the theory described above applies. The beginning of the hydration of the silicate phases, the final gel formation, is then characterized by a sudden decrease in the recovery of cement. This decrease is not large, but nevertheless significant. When the formation of the final hydrates begins, the chemical reactions restart on a large scale, thus again hindering the adsorption of collector. This results in the observed decrease of the cement recovery. When a certain amount of the final gel structure has been formed at the end of the setting period, the amount of cement floated again becomes constant with time. In

1

3

7

14

28

days

Fig. 6. Recovery of portland cement (PZ) and blast furnace. slag cement (HOZ) from mortars as a function of hydration time (SLS as collector).

E. NAGELE

2148

this state of hydration the cement particles are linked together by needles of CSH gel, which also covers their surface completely. Further hydration proceeds far in the inner regions of the cement particles, leaving the surfaces unchanged. Hence the collector can be adsorbed on the CSH gel and the flotation is unaffected by the hydration reaction. During the induction period, the collector is incorporated in the products of reaction (5) and not in the CSH gel, which is formed only at the later stages of the hydration. Therefore, we can conclude that the adsorption mechanism has changed from chemisorption to physisorption. This change can be observed by the sign&ant decrease in cement recovery. Hence flotation also provides an insight into the mechanisms of cement hydration and the properties of the hydration products [16]. 5. SUMMARY Cement can be floated using sulphate or sulphonate collectors in an alkaline pulp. The collector is adsorbed on the first reaction products of the cement hydration by a chemisorption process. The adsorption energy is different for the different types of cement and depends on the reactivity of the cement with respect to water. The effects of collector concentration and collector structure are the same as those encountered in normal oxide mineral flotation systems, where the collector acts via physisorption in the Stem layer. The chemisorption of collector leads to high selectivity of the cement flotation and permits quantitative flotation of the cement and determination of the cement content of fresh and hardened concrete by means of flotation. Acknowledgement-The author wishes to express his appreciation to Professor P. Ney, University of K6ln, for the zeta-potential measurements and to Dr. W. Gerhard, Domier Systems. Friedrichshafen, for his assistance in the SIMS measurements.

REFERENCES [l]

Rogers J., Principles of sulphide mineral flotation. Froth Flotation, 50th Anniversary Volume, Am. Inst. Min. Petr. Engrs, New York 1962. [2] Aplan F. F. and Fuerstenau D. W.. Principles of nonFroth Flotation, metallic mineral flotation. 50th Anniversary Volume, Am. Inst. Min. Petr. Engrs, New York, 1962. [3] Niigele E. and Hilsdorf H. K., A new method for cement content determination of fresh concrete. Cement Concrete Res. 1980 10 23-34. Hochofenund c41 DIN 1164, Portland-Eisenportland TraBzement. Aqube November 1978. und Flotierbarkeit van c51 Nev P.. Zeta-Potentiale h&eralLn. Springer, Berlin. C61 Gerhard W. and N8gele E., The hydration of cement studied by secondary ion mass spectrometry. Cement Concrete Res. 1983 13 849-860. c71 Young F. and Slcalny J., Proc. 7th Inc. Symp. Chem. Cement, Paris, 1980. PI Lecher. F. W.. Richartz W. and Sprung S., Erstarren von Cement. Zement-Kalk-Gips 1976 29 435. PI Suzuki K., Nichikawa T., Kate K., Hayashi H. and Ito S., Approach by zeta-potential-measbrement on the surface charge of hydrating CJS. Cement Concrete Res. 1981 11 759-764.

cm

Stein. H. N., Surface charges on calcium silicatesand calcium silicate hydrates. J. COIL Inz. Sci. 1968 Z? 203-213.

Cl13 Hinterthan

CM Cl31 Cl41 Cl51 Cl61

0.. Zeta-Potentiale als Indikator fiir Reaktionen der Zementminerale mit Wasser und wiissrigen Lasungen. Diplomarbeit, Math.-natw. Fak., Univ. Kiiln, 1974. Keck W. E. and Jasberg P., A study on the flotation protxrties of -sum. Min. Technol. 1937 I. 1. ?wzh&ert Hz- _ Aufbereitung fester mi&czlischer Rohstofi, VEB Verlag fiir Grundstoffindustrie. Lei~zia. 2nd 4%. -1978. Smolik T. J., Hamman R. W. and Fuerstenau D. W., Surface characteristics and flotation behaviour of aluminosilicates. Trans. AIME. pp. 367-375, Dec. 1966. Osipow L., Surface Chemistry. Reinhold, New York, 1962. Nggele E. and Knittel T., A note on the hydration of cement. Cement Concrete Res. 1983 13 141-145.