Waste Management 30 (2010) 132–139

Contents lists available at ScienceDirect

Waste Management

journal homepage: www.elsevier .com/ locate/wasman

Accelerated carbonation of steel slags in a landfill cover construction

S. Diener a,*, L. Andreas a, I. Herrmann a, H. Ecke b, A. Lagerkvist a

a Division of Waste Science and Technology, Luleå University of Technology, 971 87 Luleå, Swedenb Civil and Materials Engineering, Vattenfall Research and Development AB, SE-814 26 Älvkarleby, Sweden

a r t i c l e i n f o a b s t r a c t

Article history:Accepted 23 August 2009

0956-053X/$ - see front matter � 2009 Elsevier Ltd.doi:10.1016/j.wasman.2009.08.007

* Corresponding author. Tel.: +46 920 49 1702; faxE-mail address: Silvia.Diener@ltu.se (S. Diener).

Steel slags from high-alloyed tool steel production were used in a full scale cover construction of a muni-cipal solid waste (MSW) landfill. In order to study the long-term stability of the steel slags within the finalcover, a laboratory experiment was performed. The effect on the ageing process, due to i.e. carbonation,exerted by five different factors resembling both the material characteristics and the environmental con-ditions is investigated. Leaching behaviour, acid neutralization capacity and mineralogy (evaluated bymeans of X-ray diffraction, XRD, and thermogravimetry/differential thermal analysis, TG/DTA) are testedafter different periods of ageing under different conditions.

Samples aged for 3 and 10 months were evaluated in this paper. Multivariate data analysis was used fordata evaluation. The results indicate that among the investigated factors, ageing time and carbon dioxidecontent of the atmosphere were able to exert the most relevant effect. However, further investigationsare required in order to clarify the role of the temperature.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Large amounts of materials are necessary to build a final coverof a landfill. Such a cover, consisting of vegetation layer, protectionlayer, drainage layer, liner layer and second drainage layer, is nor-mally several meters high. Both, environmental properties andcosts are in focus when it comes to choose the appropriate materi-als for building a final cover. In the optimum case, also availablealternative materials such as sewage sludge, ash, excavated mate-rial, and industrial by-products arising in the area are consideredand tested. The reuse of industrial by-products occurring in largeamounts such as steel slags as well as the chance to take advantageof their positive chemical and physical properties (for examplealkaline nature, low hydraulic conductivity and strength) gaveincentives for using steel slags as construction material for theliner of the presented field test.

Two types of steel slags have been used as secondary construc-tion material in the liner layer of a full scale cover construction ona municipal solid waste (MSW) landfill in Hagfors, Sweden. Sincelandfill covers are expected to function for centuries, the long-termstability of the cover components is of special interest. Mineraltransformations take place due to non-equilibrium conditions be-tween the materials and their environment; they can affect boththe properties of the materials and their interaction with theenvironment.

Under the conditions occurring at a landfill site, carbonation isexpected to be an important weathering process due to the high

All rights reserved.

: +46 920 49 28.

CO2 content in MSW landfill gas. According to the findings of Huij-gen et al. (2004, 2005), the rate-determining step (below optimalconversion temperature, i.e. 200 �C) of aqueous carbonation isthe Ca diffusion from the slag surface through the slag matrix. Itwas shown by Huijgen et al. (2005) with scanning electron micros-copy measurements that Ca leaching leads to a Ca-depleted silicaterim around a Ca-silicate core. The carbonated rim is assumed tohinder further Ca diffusion and therewith reducing the carbonationrate. Particle size and reaction temperature were the two main fac-tors determining the reaction rate. Calcite formation due to car-bonation is limited by the amount of available CaO (Fernandez-Bertos et al., 2004) and Ca(OH)2, the temperature, CO2 concentra-tion and relative humidity in the pore structure (Saetta et al.,1993). In addition to Ca, also Mg, Sr and Ba can form carbonatesduring aqueous carbonation of steel slags (Huijgen and Comans,2006). If carbonates are formed, the concentration of that alkalineearth metal in the leachate decreases. Carbonation of Mg(OH)2

(brucite) occurs less readily than for Ca(OH)2 (portlandite) as bru-cite does not dissolve in water as much as portlandite does (Gold-ring and Juckes, 1997); this is an important limiting factor, sincedissolution of potentially reactive elements is an essential pre-req-uisite for carbonation to occur.

In order to study the effect of five different factors on leaching,acid neutralization capacity (ANC) and mineralogy (X-ray diffrac-tion, XRD and thermogravimetry/differential thermal analysis,TG/DTA) of aged steel slags over time, a laboratory experimentwas designed. The aim of this work was to identify possiblechanges in the steel slags after accelerated ageing and to rankthe impact of the investigated factors; carbon dioxide content, rel-ative humidity, ageing time, temperature, and water quality.

S. Diener et al. / Waste Management 30 (2010) 132–139 133

2. Material and methods

Steel slags from high-alloyed tool steel production have beenused in the liner of a landfill cover of a field test in Sweden. Partic-ularly, Electric arc furnace slag (EAFS) and ladle furnace slag (LFS)have been used. The steel slags have been sieved <19 mm beforethey arrived at the laboratory. Before the steel slags were sievedagain, 70% of EAFS particles ranged between 1 and 19 mm, whilethe LFS was finely grained with more than 75% of the particlesbeing smaller than 0.5 mm in size. The steel slags had a total solidscontent of 99 wt.%. Steel slag specimens were built correspondingto the composition of the liner layer in the field. A mixture of 50%EAFS (<8 mm) and 50% LFS (<19 mm) was used. The steel slag mix-ture consists of about 36% CaO, 19% SiO2, 15% MgO, 16% Al2O3 withonly a minor content of Fe oxide as shown in Fig. 1. The composi-tion of Portland cement is shown additionally as a reference.

About 10 wt.% of water was added to the mixture and thencompacted in small cylinders with a proctor device. After weighingand marking, the specimens were placed into small containersaccording to the different atmospheric conditions.

54 Sample specimens (height 40 mm), including replicates,were prepared according to the following procedure: material mix-ing, water addition (10 wt.%) and mixing, compaction into cylindri-cal shapes with 50 mm diameter and curing under controlledconditions selected according to the design. Further informationsuch as the construction of the landfill cover, the materials grainsize distributions, the curing and compaction properties as wellas the hydraulic conductivity of the mixture were reported in aprevious study (Andreas et al., 2005).

Accelerated carbonation and ageing of steel slag was studied ina laboratory experiment using a reduced factorial design where thefive factors time, CO2 content, relative humidity, temperature andquality of the water added during sample preparation were inves-tigated (Table 1). For each factor combination, the samples wereplaced in small plastic containers with the corresponding temper-ature, relative humidity and CO2 content. The CO2 level was set byforcing the flow of either atmospheric air, a mixture of N2/CO2 (80/

Fig. 1. Main oxides in the investigated steel slag mixture compared to Portlandcement type II (Dobrowolski, 1998).

Table 1Factors investigated in the experiment.

Factor Low

Relative humidity 30%Carbon dioxide Air (0.038% vol.)Temperature 5 �CTime 3 monthsWater quality for building samples Distilled water

20) or pure CO2 in the containers with a pump. The containers withthe samples were placed either in a fridge, a heating cabinet or aclimate room for the control of the temperature at 5, 60 and30 �C, respectively. The humidity content in the boxes was ad-justed by either drying or humidifying the flowing gas with a bottlecontaining silica gel or water. The humidity and the temperaturewere frequently measured and recorded with USB data loggers(Measurement Computing Corp.).

Analyses were performed according to Table 1 starting with theso called ‘‘zero-samples” aged for a period of 3 months. The 3months-aged samples have been analysed for ANC, TG/DTA, XRD,and leached using a 24 h standard leaching test according to SS-EN 12457-4 (SIS, 2003) with a liquid–solid ratio of 10 (L/S 10).The 10 months-aged samples have undergone the same analyses,except the TG/DTA measurements. For XRD and TG/DTA analyses,particles from the surface of the sample specimens were removedbefore the specimens were crushed for further analyses. The sur-face particles were ground for 20 s in a vibratory disc mill afterdrying at 105 �C. The XRD instrument used was a Siemens D5000using Cu Ka radiation. Analyses were performed at Bragg-angles(2H) from 10 to 90 �, with 0.02 � per step and 4 s per step.

The weight loss between 100 and 1000 �C was investigatedusing thermogravimetry (TG) while differences in temperature be-tween the sample and a reference (Al2O3) were recorded with dif-ferential thermal analysis (DTA). TG/DTA analyses includedquadrupole mass spectrometry (QMS) for CO2 analysis. It was per-formed with a Netzsch STA 409 in an argon atmosphere with a con-stant flow of 100 ml Argon/min. The heating rate was 10 �C/minbetween 100 and 1000 �C. Four steel slag samples (aged for 3months) were analysed with TG/DTA and QMS. The sample weightwas 21.0 ± 0.91 mg.

For the leaching tests (<9 mm) and the ANC titrations, the bulkof each aged specimen was crushed and sieved. Additionally, thesamples for the ANC analyses were grinded after crushing, driedat 105 �C and then sieved <125 lm. The ANC test was carried outin duplicates. About 1 g of sample was titrated in a suspension withdistilled water at L/S 110. ANC was tested according to the stan-dard SS-EN ISO 9963-1 (SIS, 1996) using a TitraLab system (Radi-ometer Analytical SA, Lyon, France) with an ABU 901 autoburetteand the TIM900 titration manager, both controlled by the com-puter software TimTalk9 version 2.1 (LabSoft, Radiometer Analyti-cal SA). The pH was measured continuously using pH glasselectrodes with Red Rod Technology (Radiometer Analytical SA).The titration was conducted in two steps: from the natural pH ofthe suspension to pH 8.3 (ANC 8.3) and from pH 8.3 to pH 4.5(ANC 4.5). The experiment was terminated once the final pH wasstable for 5 min. The total titration time was between 10.5 and19 h.

The results of the analyses were evaluated with multivariatedata analysis (Software SIMCA, Umetrics AB). A model was createdaccording to principal component analysis (PCA) and partial leastsquares (PLS) method. A PCA of the ANC data was performed topresent a multivariate data table in a low-dimensional plane whichgives an overview of the data. Such an overview can reveal groupsamong data points, trends and outliers. Another aim is to uncoverthe relationships between observations (samples) and variables

Middle High

65% 100%20% vol. 100% vol.30 �C 60 �C10, 22 months 30 months– Leachate (from a column experiment)

134 S. Diener et al. / Waste Management 30 (2010) 132–139

(measured data). Observations that are close to each other in thescore plot are similar and vice versa. The same is valid for the load-ing plot where variables positioned close to each other are posi-tively correlated and vice versa. The distance to the origin ofeach plot is also important. The most important variables are foundin the periphery of the loading plot; the least influential are situ-ated close to the origin (Umetrics AB, 2006).

The results of a PCA and a PLS are illustrated in two comple-mentary, superimposable plots: the Score plot containing the

Fig. 2. ANC of 3 and 10 mo

Fig. 3. Evaluation of ANC data of 3 and 10 months-aged samples with a PCA model.Distribution of the samples along x-axis according to their ANC 8.3 and ANC 4.5, respectivhigher ANC 8.3 have a lower ANC 4.5, and vice versa. ANC 8.3 and ANC Start pH are po

observations and the Loading plot showing the variables. The loca-tion of the observations and variables gives information on howthey correlate. The variables in the loading plot inform about themagnitude of the correlation (small, large) and the manner (posi-tive, negative) in which the variables contribute to the scores.

The Score plot shows also the Hotelling T2 ellipse (at 0.95 con-fidence level). An observation located outside the T2 ellipse is be-yond the 95% confidence region of the model and can beinterpreted as strong outlier.

nths-aged steel slags.

Score plot (left) and loading plot (right) of first and second principal component.ely. Those two responses are negatively correlated which means that samples with asitively correlated.

Table 2Leachate composition (L/S 10) of 3 and 10 months-aged steel slags.

0.038% CO2 100% CO2 0.038% CO2 100% CO2

30% rh 30% rh 100% rh 100% rh5 �C 60 �C 5 �C 60 �C3 months 3 months 3 months 3 months

Mean(n = 3)

Mean(n = 3)

Mean(n = 3)

Mean(n = 3)

pH – 11.7 11.0 11.7 11.2Ca mg/kg 3034.0 1226.0 2893.2 1624.1Al mg/kg 2410.1 822.8 2252.1 1197.2Na mg/kg 57.22 18.83 65.31 23.26K mg/kg 17.34b 12.71b 42.27 15.99b

S mg/kg 11.79b 149.57 13.16 133.57Si mg/kg 0.76 6.66 0.90 2.14Mg mg/kg 0.64b 1.05b 0.62a 0.74b

Ba mg/kg 0.512 0.176 0.367 0.301Fe mg/kg 0.42a 0.40b 0.42a 0.40b

Cr mg/kg 0.068 0.062 0.069 0.084Sr mg/kg 3.433 1.492 3.189 0.954V mg/kg 0.046 0.090b 0.047 0.090

0.038% CO2 100% CO2 0.038% CO2 100% CO2 20% CO2

30% rh 30% rh 100% rh 100% rh 65% rh60 �C 5 �C 60 �C 5 �C 30 �C10 months 10 months 10 months 10 months 10 months

Mean(n = 3)

Mean(n = 3)

Mean(n = 3)

Mean(n = 3)

Mean(n = 2)

pH – 11.8 10.8 11.8 9.9 11.6Ca mg/kg 2444.7 1351.9 1933.6 196.9 2258.9

S. Diener et al. / Waste Management 30 (2010) 132–139 135

3. Results

3.1. ANC of 3 and 10 months-aged steel slags

The results of the first titration step to pH 8.3 and the secondtitration step to pH 4.5 varied significantly between the samplesaged for 3 and 10 months (Fig. 2). The ANC 8.3 value was lowerfor the specimens aged in a CO2-enriched atmosphere, whereasthe ANC 4.5 was higher.

The factor CO2 co-varied with the factors relative humidity andwater quality. However, the latter two did not show an impact onANC. It seems that only the factor CO2 had an influence on the ANCof the steel slag samples. The total of ANC 4.5 and ANC 8.3 was onaverage between 11.8 and 13.8 mmol/g and did not change signif-icantly over time or with ageing conditions.

A PCA model including the ANC results is presented in Fig. 3.Most samples exposed to 100% CO2 during ageing are situated onthe left side of the score plot (Fig. 3). Correspondingly, the variableANC 4.5 is located on the left side of the loading plot. This meansthat the samples treated with high CO2 by trend had a higherANC 4.5 and a lower ANC 8.3. The variable ANC total is positionedin between ANC 4.5 and ANC 8.3 since it presents the sum of both.The x-axis (first principal component) explains about 58% of thedata variation with the above mentioned correlations. The two firstprincipal components PC1 and PC2 (x- and y-axes) together explain91% of the data variation within the model.

Al mg/kg 1539.7 1336.9 879.5 137.0 1923.3Na mg/kg 21.58 6.95 32.53 7.25 5.41K mg/kg 5.01a 5.01a 19.71 11.89 6.75S mg/kg 303.59 211.13 137.23 120.82 505.37Si mg/kg 2.20 0.47a 3.08 13.29 1.11Mg mg/kg 0.90a 17.29b 0.90a 254.06 0.90a

Ba mg/kg 0.284 0.359 0.016 0.031 0.372Fe mg/kg 0.20a 0.15a 0.15a 0.04a 0.20a

Cr mg/kg 0.036 0.022a 0.023 0.007 0.030a

Sr mg/kg n.a. n.a. n.a. n.a. n.a.V mg/kg n.a. n.a. n.a. n.a. n.a.

n.a. not analysed.a All values below detection limit,b At least one value below detection limit.

3.2. Leaching of 3 and 10 months-aged steel slags

The pH values measured in the eluate after 24 h were up to twopH units lower for the samples aged most intensively (Table 2). Be-sides a reduced pH, the exposure to high levels of CO2 togetherwith a high temperature led to an increased leaching of Mo, Si,Mg and V and to a reduced leaching of Ca, Ba and Al after 3 monthsageing. But only Ca, Si and Mg showed a similar behaviour after10 months ageing. Na and K leaching also seemed to decrease un-der the influence of CO2 and high temperature (after 3 months)while S leaching increased under those conditions. Again, K and Sleaching were less dependent on the CO2 content after 10 monthsageing. Possibly, the factor relative humidity became relevantthere. Cr phases were rather stable in the steel slags and therewithCr leaching was low.

The leaching test results are also evaluated with a PLS model ofwhich the first two principal components are presented in Fig. 4.The score plot contains 26 observations (samples) while the corre-sponding loading plot gives information about the influence of 26variables (chemical elements) that are included. The two plots rep-resent about 54% of the data variation. According to Fig. 4, the var-iable time had the largest influence on the leaching data as it issituated the furthest from the origin, followed by CO2 and temper-ature. This suggests that the x-axis represents the variation overtime which is also seen by the grouping in the score plot. The ele-ments Mg, Si and V are found close to the variable CO2 in the load-ing plot which suggests a positive correlation, while elements asCa, Sr and Ba are found on the opposite side of the plot. This con-firms the negative correlation between CO2 content and those ele-ments as expected from the literature. It is not possible to evaluatesome elements such as As, Cd, Co, Cu, Fe, Hg, Mn, Ni, Pb and Zn fur-ther as the majority of those results were below the detection limitafter 10 months ageing.

3.3. XRD

EAF and LF slags have many minerals in common, but someminerals are specific to each type of steel slag. The samples con-

tained a mixture of the mineral phases originating from thesetwo slag types. Two XRD examples are presented in Fig. 5. Themineralogy of the steel slag samples is complex and contains anumber of possible phases. Various calcium silicates such asakermanite (Ca2MgSi2O7), bredigite (Ca7Mg(SiO4)4), gehlenite(Ca2Al(AlSi)O7), monticellite (CaMgSiO4) and merwinite (Ca3Mg(SiO4)2) as well as mayenite (Ca12Al14O33) were consistent withthe sample pattern. The intensities of specific peaks differ be-tween samples A and B. For example, the peak corresponding topericlase (MgO), one of the main phases, was much smaller inthe sample aged for 10 months. A decrease in one calcium silicatepeak (c-Ca2SiO4) was found. After 10 months’ ageing, the peakscorresponding to quartz (SiO2) and calcite (CaCO3) increased inintensity (sample B). Some peaks corresponding to merwinite in-creased in a similar fashion. The detailed chemistry of the spinelphase could not be analysed using XRD. Additional mineral phasescould be present.

3.4. TG of 3 months-aged steel slags

Figs. 6 and 7 present the results of the thermal analysis (TG/DTA) of four differently treated samples. In addition to the TG data,the derivative thermogravimetric (DTG) curves and the CO2 analy-ses are illustrated.

The aged steel slag samples appear to behave similar for tem-peratures up to about 550 �C. A small difference between ageing

Fig. 4. Evaluation of the leaching behaviour of 3 and 10 months-aged samples with a PLS model. Score plot (left) and loading plot (right) of first and second principalcomponent. The observations (score plot) are combined acc. to the experimental conditions, i.e. each sign represents several samples (except no. 25 and 26). The loading plotshows responses (squares) and factors (dots).

136 S. Diener et al. / Waste Management 30 (2010) 132–139

in air (at low temperature) and ageing in CO2 (at high tempera-ture) is discernable between 100 and 200 �C. At these tempera-tures the samples stored in air underwent a weight reduction ofalmost 2 wt.% (Fig. 6), whereas the weight loss of other sampleswas negligible. This indicates the loss of free moisture. A similarloss occurred for samples C and D at higher temperatures, proba-bly from hydroxides. Ageing includes hydration leading to the for-mation of, for example Ca(OH)2 and Mg(OH)2, which may becarbonated subsequently to form CaCO3 and MgCO3. Two endo-thermic reactions were found in the DTA curve between 350and 650 �C. Those endothermic reactions were largest for sampleB (aged at high humidity). Probably, this corresponds to thedecomposition of hydrates where chemically bonded water isreleased.

At about 500 �C, CO2 levels in the gas phase started to rise(Fig. 7). Only minor weight changes occurred at this temperature.The decomposition of calcium carbonate occurs in an argon atmo-sphere at 680–760 �C. The CO2 levels measured in the gas phasewere highest between 680 and 730 �C. The greatest weight loss,at about 720 �C, coincided with this release of CO2 and was there-fore most probably caused by the decomposition of carbonates.The lowest CO2 concentration in the gas was measured for sampleB, followed by A, C and D. Notably, the strongest peak in the regionof hydrate dissociation (430 �C) and also the lowest CO2 concentra-tion was observed for sample B. Hydration is the first reaction tooccur during ageing of steel slags.

The greatest weight loss was observed for sample D at around10 wt.% between 100 and 740 �C and the final weight was2–4.4 wt.% lower than for the other samples. Weight losses forsamples A and C were 7–8 wt.% and 5.6 wt.% for sample B. Thecurve shapes for samples C and D are similar. Looking at the effectof the factor carbon dioxide, the weight loss was largest for thesamples stored in 100% CO2.

4. Discussion

4.1. ANC of 3 and 10 months-aged steel slags

The lower ANC 8.3 for the steel slag samples exposed to pureCO2 (Fig. 2) is assumed to be caused by carbonate formation andbuffering reactions of carbonates. As the samples aged with pureCO2 were exposed to higher levels of the weak acid CO2, the formedcarbonates began to dissociate and the pH decreased. Silicate min-erals reacting with OH- ions could have also contributed to adecreasing pH.

Buffering reactions involving for example hydrogen bicarbonatemight be the cause for the higher ANC 4.5 of the samples aged with100% CO2. Another possibility could be that the steel slags dis-solved to a greater extent during the second step of the ANC anal-ysis (ANC 4.5) leading to more material available to react and,hence, a higher ANC. The evaluation of further acid neutralizationcapacity measurements (below pH 4.5) may further specify thesensitivity of steel slags towards decreasing pH and possible fur-ther buffering reactions.

4.2. Leaching of 3 and 10 months-aged steel slags

The PLS model including the L/S 10 leaching test suggests thatthe main factor influencing the leaching after ageing was the timefollowed by both CO2 and temperature. The factors water qualityand relative humidity did not have a significant impact on the re-sults according to the model. For the factor relative humidity noimpact was seen so far because the humidity values were not sta-ble at the beginning of the experiment.

A reduced solubility of metals such as Ca, Ba and Sr was ex-pected because of a lower pH after carbonation. Decreasing Ba lev-

Fig. 5. XRD of two steel slag samples after ageing at: (a) 0.038% CO2, 30% relative humidity, 5 �C, distilled water, 3 months, and (b) 100% CO2, 100% relative humidity, 5 �C,leachate, 10 months.

Fig. 6. Weight loss (TG) and heat flux (DTA) of four steel slag samples aged at: (A) 0.038% CO2, 30% relative humidity, 5 �C, distilled water, 3 months; (B) 0.038% CO2, 100%relative humidity, 5 �C, leachate, 3 months; (C) 100% CO2, 30% relative humidity, 60 �C, distilled water, 3 months; (D) 100% CO2, 100% relative humidity, 60 �C, leachate,3 months.

S. Diener et al. / Waste Management 30 (2010) 132–139 137

Fig. 7. Derivative of weight loss (DTG) and CO2 content in the gas stream of four steel slag samples aged at: (A) 0.038% CO2, 30% relative humidity, 5 �C, distilled water,3 months; (B) 0.038% CO2, 100% relative humidity, 5 �C, leachate, 3 months; (C) 100% CO2, 30% relative humidity, 60 �C, distilled water, 3 months; (D) 100% CO2, 100% relativehumidity, 60 �C, leachate, 3 months.

138 S. Diener et al. / Waste Management 30 (2010) 132–139

els in the leachate after carbonation can be caused due to BaCO3 ordue to solid solutions with calcite. Solid solutions with calcite werereported for Sr as well (Walton et al., 1997). However, Ba leachingafter 10 months ageing was more reduced when samples were ex-posed to high humidity levels than when exposed to low humidity(independent from CO2 level). Mg being an alkaline earth metalwas expected to decrease due to carbonation similar to Ca. How-ever, this was not observed (Table 2). Possibly, the CO2 pressurewas too low to achieve significant carbonation of Mg minerals(Huijgen et al., 2005).

Another reason could be inefficient dissolution of the Mg min-erals and therewith a low conversion into carbonates. At the sametime higher metal leaching of carbonated steelmaking slag wasfound for Si, Mg and V. This can be attributed to changes in themicrostructure due to carbonation. When Ca during carbonationdiffuses to the steel slag surface, Si–Mg minerals such as akerma-nite and monticellite deplete in Ca and are exposed to weatheringthemselves. Huijgen and Comans (2006) have also been foundincreasing V levels in the leachate of carbonated samples agreeingwith the presented results, possibly due to the dissolution of Ca-vanadate. But it can be caused by hydration of free lime as well,which increases the pH and therewith reduces dissolution of Vcontaining calcium silicates (Drissen, 2006).

Comparing the leached amounts with the solid steel slag, espe-cially S (27%), Na (12%) and K (4%) leaching was higher than forother elements (mostly below 2%). Higher Na and K values are typ-ical for weathering of silicate minerals, which were found in steelslags in different forms (see XRD results). Due to the overall leach-ate concentrations, the environmental impact caused by leachingfrom a steel slag liner is assessed as low.

4.3. XRD

The XRD analyses could reveal a number of minerals as dical-cium silicates, periclase and mayenite in the steel slags. Howeveronly calcite formed due to carbonation could be identified after10 months ageing time. One reason for this is the complexity ofthe materials. The large number of diffraction peaks (numerousminerals possibly present) and varying intensities in between thesamples aged under same conditions and for the same time, a highbackground in the diffractogram and small peak shifts due to solidsolution effects cause difficulties in detecting all phases. Further-

more, XRD analysis allows a detection of a phase if its amount islarger than 4% of the sample. However, the changes in peak inten-sity reveal that further mineralogical transformations are possibleand decreased peak intensities for MgO and calcium silicate due tocarbonation of steel slag were also reported by Johnson et al.(2003). Since the specimens have been compacted, which reducedthe sample surface exposed to the atmospheric conditions, theevaluation of more samples aged for a longer time is necessaryto be able to detect further mineral changes and allocate them tothe respective ageing conditions.

4.4. TG/DTA of 3 months-aged steel slags

The TG/DTA measurements and gas analyses for CO2 provedthat ageing with high CO2 and over a longer time lead to the high-est carbonate formation. However, also a clear difference betweenthe samples stored at 5 �C (A and B) and at 60 �C (C and D) was ob-served for the DTG curve in Fig. 7 in the region between 100 and300 �C where the dissociation of hydrates is expected. Presumably,the factor temperature influences steel slag hydration as well.However, H2O was not included in the gas analysis.

5. Conclusions

Changes of the steel slag mixture due to ageing could be proved.Ageing at high CO2 content decreased the pH about two units andthe ANC 8.3 about 50%. The ANC 4.5 increased, which indicatesbuffering reactions, possibly of newly formed carbonates. The totalANC did not change significantly after 10 months.

Leaching from aged steel slags was low with concentrations be-low detection limits for most trace elements (As, Cd, Co, Cu, Hg,Mn, Ni, Pb, Zn) after 10 months ageing. According to the presentedPLS model, the leaching behaviour was mostly influenced by age-ing time, CO2 content of the atmosphere and temperature. Leach-ing of Ca and Sr decreased while leaching of Mg, Si and Vincreased due to exposure to high CO2 levels. The environmentalimpact of a liner built of steel slags due to leaching is assessed aslow.

Mineral phases contained in the steel slags were mainly calciumsilicates such as dicalcium silicate (Ca2SiO4), merwinite (Ca3Mg(SiO4)2), monticellite (CaMgSiO4) as well as mayenite (Ca12Al14O33),periclase (MgO), and a spinel phase. Mineralogical changes due to

S. Diener et al. / Waste Management 30 (2010) 132–139 139

ageing observed using X-ray diffraction included calcite formation.Other mineralogical changes such as changes in the silicate structureare possible as well, but could not be verified yet.

Thermogravimetry, differential thermal and CO2 analyses couldprove the formation of carbonates during ageing as well. Hence,carbonate formation is expected to occur around a liner of a landfillcover consisting of steel slag. Buffering due to carbonates in theliner material will prevent a fast dissolution of the liner materialfurther ensuring its chemical stability.

The factors ageing time and CO2 showed the biggest impact onsteel slag properties under the conditions that were investigatedhere. The impact of temperature, relative humidity and water qual-ity on the observed changes of sample characteristics could not beclarified yet; no clear trends were observed by now. The evaluationof the remaining samples (22 and 30 months ageing time) is ex-pected to increase the understanding of the processes occurringduring ageing of steel slags under landfill cover conditions and tofurther clarify the impact of the investigated factors.

References

Andreas, L., Herrmann, I., Lidstrom-Larsson, M., Lagerkvist, A., 2005. Physicalproperties of steel slag to be reused in a landfill cover. Sardinia ‘́05. In: Cossu,Stegmann (Ed.), 10th International Waste Management and LandfillSymposium, October 3–7, 2005. CISA, Environ. Sanitary Engineering Centre,Italy.

Dobrowolski, J.A., 1998. Concrete Construction Handbook, fourth ed. Mc Graw-HillInc., New York. p. 1.15.

Drissen, P., 2006. Binding of trace elements in steel slags. In: Proceedings of the FifthEuropean Slag Conference (Euroslag), Luxembourg.

Fernandez-Bertos, M., Simons, S.J.R., Hills, C.D., Carey, P.J., 2004. A review ofaccelerated carbonation technology in the treatment of cement-based materialsand sequestration of CO2. Journal of Hazardous Materials 112 (3), 193–205.

Goldring, D.C., Juckes, J.M., 1997. Petrology and stability of steel slags. Ironmakingand Steelmaking 24 (6), 447–456.

Huijgen, W., Witkamp, G.-J., Comans, R., 2004. Mineral CO2 sequestration in alkalinesolid residues. In: The Seventh International Conference on Greenhouse GasControl Technologies (GHGT), Vancouver, Canada.

Huijgen, W.J.J., Witkamp, G.-J., Comans, R.N.J., 2005. Mineral CO2 sequestration bysteel slag carbonation. Environmental Science and Technology 39 (24), 9676–9682.

Huijgen, W.J.J., Comans, R.N.J., 2006. Carbonation of steel slag for CO2 sequestration:leaching of products and reaction mechanisms. Environmental Science andTechnology 40 (8), 2790–2796.

Johnson, D.C., Mac Leod, C.L., Hills, C.D., 2003. Acid neutralisation capacity ofaccelerated carbonated stainless steel slag. Environmental Technology 24 (5),545–551.

Saetta, A.V., Schrefler, B.A., Vitaliani, R.V., 1993. The carbonation of concrete and themechanism of moisture, heat and carbon dioxide flow through porousmaterials. Cement and Concrete Research 23 (4), 761–772.

SIS, 1996. Water quality – Determination of alkalinity – Part 1: Determination oftotal and composite alkalinity. SS-EN ISO 9963-1. Swedish Standards Institute(SIS), Stockholm.

SIS, 2003. Characterization of waste – Leaching – Compliance test for leaching ofgranular waste materials and sludges, Part 4. SS-EN 12457-4. SwedishStandards Institute (SIS), Stockholm.

Umetrics AB, 2006. Multi- and Megavariate Data Analysis: Part I Basic Principlesand Applications, second revised ed., Sweden, p. 33.

Walton, J.C., Bin-Shafique, S., Smith, R., Guitierrez, N., Targuin, A., 1997. Role ofcarbonation in transient leaching of cementitious wasteforms. EnvironmentalScience and Technology 31 (8), 2345–2349.