Blast Furnace Slag Weathering Study

BLAST FURNACE SLAG WEATHERING STUDY Battagin, Arnaldo Forti1 and Pecchio, Marcelo2 1 2

Brazilian Portland Cement Association, São Paulo, Brazil. E-mail: [email protected] Brazilian Portland Cement Association, São Paulo, Brazil. E-mail: [email protected]

ABSTRACT Blast-furnace slag as an addition to Portland cement is largely used in Brazil. In spite of the high addition content (up to 70 mass %), the cement industry cannot absorb the total slag production, leading to storage of millions of tons in open air. Slag is thus exposed to weathering for years. This paper presents the study carried out on three long-stored slag samples compared to two new ones, aiming at checking their aging degree and influence on their reactivity and consequently on the mechanical performance of the corresponding blended cement. Old slag samples showed lower hydration heat liberation and significant - in some cases reaching 30% to 40% - loss in performance, followed by changes in setting times. 1. INTRODUCTION About 7.3 million tons of blast furnace slag were produced in Brazil in 2000 [1] and around 6.0 million tons were formed by granulated blast furnace slag (GBFS). Granulation is obtained by quenching the molten slag into glass using high pressure water jets. The rapid cooling prevents crystallization producing a granular glassy material that, after grinding has good hydraulic properties making it a Portland clinker substitute or extender. GBFS in slag cement or blended Portland slag cement has been largely used in Brazil for more than 50 years [2] but the cement industry cannot absorb the whole slag production. Although consumption in 2000 was about 4.9 million tons there was an excess 1.1 million tons that was not used. Due to this, around 12 million tons of GBFS are stockpiled mainly in cement plants. Many publications have dealt with the use of fresh-made GBFS as a cementitious material but less work has been done about the use of weathered slag. In fact, there is only one known contribution of Hiroshima and Igarashi [3] that studied the effect of weathering of GBFS and concluded that there was a loss of 28-day compressive strength and an increase of setting times of cements containing “old” slag. On the other hand, Brazilian cement manufacturers have reported an improvement in slag grindability. It is generally agreed that slag reactivity is due to both inherent characteristics and external factors. Inherent characteristics are related to operational conditions of blast furnace, temperature and viscosity of molten slag, granulation process and installation, etc. External factors are characteristics that are imposed to slag by means of handling, storage and grinding. For practical purposes and quality control we can summarize as slag inherent characteristics their chemical composition and their glass content.

External factors determining slag reactivity are mainly slag fineness, alkali concentration of reacting system, temperature during early phases of hydration process and weathering degree. This paper reports results from a laboratory study on the weathering effects on slag reactivity using chemical analysis, microscopy examination, conduction calorimeter, accelerated compressive strength tests by NaOH activation and cement compressive strength test with slag partial replacement.

2. EXPERIMENTAL DETAILS AND RESULTS 2.1. Chemical Composition and Glass Content Five slag samples were selected: UV1, UV2 and UN produced by USIMINAS, a Brazilian steel plant located in Minas Gerais State, VRV and VRN produced by Companhia Siderurgica Nacional (CSN) located in Volta Redonda, Rio de Janeiro. UV1, UV2 and VRV were stockpiled in unground form for more than 10 years and UN and VRN have just been produced. Chemical compositions are given in Table 1. Table 1 – Chemical composition % in mass UV1 UV2 UN 42.54 42.28 41.91 34.68 35.10 35.10 12.23 12.08 12.54 0.96 0.84 0.53 2.40 2.44 2.65 6.18 6.38 6.39 0.37 0.40 0.39 0.11 0.10 0.11 0.52 0.40 0.37 99.99 100.02 99.99 1.76 1.73 1.73 1.23 1.20 1.19

Oxide CaO SiO2 Al2O3 Fe2O3 SO3 MgO K2O Na2O TiO2 Total IB1(*) IB2(**) (*)IB1 –

I=

CaO + MgO + Al2O3 >1 SiO2

VRV 42.50 35.65 11.32 1.14 2.45 6.09 0.52 0.113 0.19 99.973 1.68 1.19

VRN 42.47 34.31 11.42 3.26 2.58 5.08 0.28 0.12 0.46 99.98 1.72 1.24

(**)IB2 - i = CaO/SiO2

Results show that slags are basic and comply with the requirements of Brazilian Standards. There are no significant differences among slag chemical compositions. Although expressed as SO3, sulphur is mostly present as sulphide, especially for UN and VRN slag samples. Sulphide oxidation is supposedly more expressive in old slags (UV1, UV2 and VRV). The glass content was determined by means of optical polarizing microscope according to the method proposed by McMaster [4]. Results are given in Table 2.

Table 2 – Slag glass content Slag

%

UV1

95.2

UV2

93.6

UN

97.9

VRV

92.9

VRN

86.9

The slag melt has a high content of thermal energy. If this energy is fully dissipated by slow cooling a stable crystalline and non-hydraulic slag is formed. On the other hand, if quick cooling is used, a glassy slag with latent hydraulic properties is obtained. It is generally agreed that slag should have a high glass content to develop cementitious properties. As shown in Table 2, samples present a high glass content, most of them higher than 90%. Only VRN exhibited glass content below 90% but it seems that small amounts of crystalline material in vitreous slag will not have very detrimental effects. Demoulian and collaborators [5] have even reported that crystal nuclei have an advantageous effect. 2.2. Loss on Ignition (LOI), Insoluble Residue (IR) and Scanning Electron Microscopy In order to evaluate the effect of storage conditions on slag LOI and IR were determined in all samples. LOI value could detect a certain stage of hydration/carbonation because it is known that slags are formed after solidification of high temperature molten material and do not present any LOI. However, vitreous slags are soluble in HCl solution and any IR value detected could be related to contaminations of slag during storage. Table 3 presents the results of LOI and IR. Table 3 – LOI and IR values Sample

Insoluble Residue (%)

Loss of Ignition (%)

UV1

0.70

2.30

UV2

0.25

0.90

UN

0.25

-0.50

VRV

0.49

2.50

VRN

0.25

-1.70

(Obs.) Negative sign indicates a gain of mass due to an oxidation process Table 3 shows that “old” slags presented higher values of LOI and IR compared to new ones. Figures 1 and 2, obtained by sample observation under scanning electron microscope show that weathered slag exhibits grains with a coating surface of hydration/carbonation and probably sulphate products. It is believed that the hydrated layer is thin but tends to increase the higher the time of storage at open air environment.

Figure 1 - Slag grains with coating surface of hydration/carbonation and probably sulphate products

Figure 2 – Slag grains with coating surface of hydration/carbonation and probably sulphate products 2.3. Hydraulic Activity of Slag by Reaction with Alkali: Compressive Strength Tests and Heat of Hydration Accelerated strength tests by using sodium hydroxide solution as mixing water were employed. Slags were ground to a fineness of (4000 ± 150)cm2/g. Cylindrical test specimens of 50x100mm, in proportion 1:3 (slag: sand), water cured at 23ºC, s/alk sol.=0.5, according to an adaptation of Brazilian Standard NBR 7215 [6] were carried out. Alkali solution (20%) was prepared by dissolving 200g of sodium hydroxide in 1 liter of distilled water and cooled to room temperature. Average strength results at 48h are given in Table 4.

Table 4 – Slag hydraulic activity with NaOH Hydraulic Density Blaine surface activity with Samples (g.cm-3) area (cm2.g-1) NaOH (MPa) UV1 2.79 3950 7.4 UV2 2.84 4120 12.2 UN 2.91 3980 15.0 VRV 2.80 4040 8.6 VRN 2.94 4030 12.8 A conduction calorimeter was used to measure the evolution of heat of hydration. Slag pastes with slag/alkaline solution equal to 0.4 and curing temperature 23ºC were adopted. Results of total heat liberation up to 72 hours are shown in Figure 3. 120 UV1 100

UN

Hydration Heat (J/g)

VRN VRV

80

UV2 60

40

20

0 0

10

20

30

40

50

60

70

80

Time (h)

Figure 3 – Total heat of hydration liberation From Figure 3 we conclude that the total heat liberated by “new” slag from a single source is higher than that of the corresponding weathered slag. Accordingly, the same conclusion can be deduced from results of hydraulic activity with NaOH shown in Table 4. Weathered slags both from USIMINAS (UV1 and UV2) and Companhia Siderurgica Nacional (VRV) presented lower values of hydraulic activity with NaOH compared to the new slags from the same origin. Brazilian former studies [7] have shown that typical compressive strengths for good quality slags are expected in range of 9 to 11MPa. We can deduce that “new” slags comply with these values while weathered slags presented values below of these limits. It should be noted, however, that slag UV2 presented a good behavior perhaps due to its higher fineness.

2.4. Cement Compressive Strength Test and Setting Time In order to evaluate the mechanical performance behavior, each slag sample after ground to a fineness (4000±150)cm2/g was mixed with an industrial ordinary Portland cement with Blaine fineness 4400cm2/g. The proportions of mixes were 0, 20%, 30%, 40%, 50% and 60% of slag replacement. Test specimens were prepared and tested on 7- and 28-day compressive strength in accordance with NBR 7215. Results obtained are given in Table 5 and Figures 4 and 5.

Slag UV1 content 7 28 (%) days days 0

-

-

Table 5 - Compressive Strength (MPa) UV2 UN VRV VRN 7 28 7 28 7 28 7 28 days days days days days days days days

Reference 7 28 days days

-

-

-

-

-

-

-

-

40.3

47.1

20

34.9 44.6

38.2

46.2

40.8

51.0

37.4

44.6

41.1

50.7

-

-

30

33.4 43.1

35.8

46.4

40.5

53.4

36.0

44.9

38.3

50.1

-

-

40

29.6 40.8

33.9

44.9

41.3

54.3

37.0

42.3

37.2

52.3

-

-

50

25.3 37.2

29.9

42.1

35.8

51.9

29.4

40.7

34.4

51.4

-

--

60

21.4 34.0

25.2

39.9

33.6

50.6

25.1

36.6

29.3

49.2

-

-

7 Day - Compressive Strength (MPa)

45

40

35

UV1

30

UV2 UN VRV

25

VRN 20 0

10

20

30

40

50

Slag content (%)

Figure 4 – Slag contents versus 7-day compressive strength

60

28 Day - Compressive Strength (MPa)

60

55

50

45 UV1 UV2

40

UN VRV

35

VRN 30 0

10

20

30

40

50

60

Slag Content (%)

Figure 5 – Slag contents versus 28-day compressive strength Figure 4 shows the 7-day compressive strength of cement mortars versus slag replacement from 20% to 60%. It is showed that the higher the slag replacement, the higher the strength loss for both weathered and new slags. However, it seems that strength loss is higher for weathered slags compared to “new” slags. A different behavior was found for 28-day compressive strength, Figure 5. Accordingly, a strength gain was obtained for mixes containing new slags. For 40% slag replacement strength gains are 15% and 10%, compared to reference for UN and VRN, respectively. By comparing mortar strength of mixes with same slag replacement and from a single source, results showed that strength losses are higher for weathered slags. For instance, for 60% slag replacement, UV1 presented a loss of 36% compared to UN at 7 days. Table 6 clearly shows that raising the slag replacement there is an increase of setting time for both weathered and new slag. By comparing weathering and new slags from a same source, there is a slight decrease of initial setting time and a shortening of final setting time for mixtures containing weathered slag that reached up to 85 minutes. Table 6 shows the results of initial and final setting times Table 6 - Setting Time (min) UV1 UV2 Slag content (%) Initial Final Initial Final

UN Initial

VRV

VRN

Final Initial

Final Initial

Final

20

155

235

165

235

175

245

175

235

155

235

30

185

245

165

235

200

240

145

205

215

245

40

170

260

175

245

170

240

175

225

185

275

50

185

275

195

265

205

265

155

215

235

295

60 195 3. CONCLUSIONS

275

145

245

205

275

170

220

225

305

From the results of tests carried out in this investigation the following conclusions may be drawn. • • • •

•

•

•

Slag weathering is linked to hydration/carbonation/oxidation process that causes a loss of slag reactivity; By comparing weathered and new slags from a single source tests showed a decrease of heat of hydration liberation as well as a decrease of alkali accelerated strength in weathered slag; LOI and IR tests, for its simplicity, could be used for quality control of slag weathered degree; For 7-day compressive strength, test of mixes with slag replacement from 20% to 60% showed that the higher the slag replacement, the higher the strength loss for both weathered and news slags compared to the reference. However, it seems that strength loss is higher for weathered slags; A different behavior was found for 28-day compressive strength tests. In spite of mixes containing weathered slags have exhibited a strength loss of for all slag replacement, mixes containing new slags presented a strength gain of 10% to 15% with replacement from 30% to 50%; For some slag replacement, mixes presented strength loss of up to 40% by comparing new and weathered slags from a single source. Finally, by raising the slag replacement, there is an increase of setting time for both weathered and news slags. By comparing weathered and new slags from a single source there was a slight decrease of initial setting time and a shortening of final setting time for mixes containing weathered slag that reached up to 85 minutes.

4. REFERENCES [1] Annual Report 2000 – National Institute of Siderurgy [2] Battagin,A.F. et all – Contribution to the knowledge of Blast Furnace Slag Cement Properties. Technical Report nº 90. Brazilian Cement Association,1987. (in Portuguese) [3] Hiroshima, A; Igarashi,T. Effect by Weathering of Granulated Blast Furnace Slag Powder on the Quality of Portland Blast Furnace Slag Cement, CAJ Review, 1983, p. 65-66. [4] Hooton,D.R. and Emery, J.J. Glass content determination and strength development predictors for vitrified Blast Furnace Slag. In Fly ash, silica fume, slag&others mineral by-products in concrete. ACI SP-79, vol. II, pag 943-962. [5] Demoulian, E.;Vernet,C.;Hawthorn,F.;Gourdin,P. Détermination de la teneur en laitier dans le ciments par dissolucion seletive. In. Int. Congr. Chemistry of Cement,7th . Paris, 1980. Proceedings. v II, p 151-156.

[6] NBR 7215, Determination on Cement Compressive Strength, Brazilian Association of Technical Standards, 1996. (in Portuguese) [7] Cincotto, M.A. and Battagin, A.F. – Characteristics of blast furnace slag and its use as a cimentitious material and as concrete aggregate. Technical Research Institute Bulletin 65, 1992 p.18 (in Portuguese) [8] Demolian, E et alii – Influence de la composition chimique et de la texture des laitiers sur leur hidraulicité. In: Int. Congress on Chemistry of Cement, 7th, Paris, 1980 V4, p17-20.