Comparison of alkaline industrial wastes for aqueous mineral carbon sequestration through a parallel reactivity study

Waste Management xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Waste Management

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

Comparison of alkaline industrial wastes for aqueous mineral carbonsequestration through a parallel reactivity study

http://dx.doi.org/10.1016/j.wasman.2014.03.0090956-053X/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +1 4436556602.E-mail address: [email protected] (C.W. Noack).

Please cite this article in press as: Noack, C.W., et al. Comparison of alkaline industrial wastes for aqueous mineral carbon sequestration through areactivity study. Waste Management (2014), http://dx.doi.org/10.1016/j.wasman.2014.03.009

Clinton W. Noack a,⇑, David A. Dzombak a, David V. Nakles a, Steven B. Hawthorne b, Loreal V. Heebink b,Neal Dando c, Michael Gershenzon c, Rajat S. Ghosh c

a Department of Civil and Environmental Engineering, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA 15213, United Statesb Energy and Environmental Research Center, University of North Dakota, 15 North 23rd Street, Stop 9018, Grand Forks, ND 58202, United Statesc Alcoa Technical Center, 100 Technical Drive, New Kensington, PA 15068, United States

a r t i c l e i n f o

Article history:Received 5 December 2013Accepted 18 March 2014Available online xxxx

Keywords:Carbon sequestrationAlkaline industrial wastesMineral carbonationChemical equilibrium modeling

a b s t r a c t

Thirty-one alkaline industrial wastes from a wide range of industrial processes were acquired andscreened for application in an aqueous carbon sequestration process. The wastes were evaluated for theirpotential to leach polyvalent cations and base species. Following mixing with a simple sodium bicarbon-ate solution, chemistries of the aqueous and solid phases were analyzed. Experimental results indicatedthat the most reactive materials were capable of sequestering between 77% and 93% of the available car-bon under experimental conditions in four hours. These materials – cement kiln dust, spray dryer absor-ber ash, and circulating dry scrubber ash – are thus good candidates for detailed, process-orientedstudies. Chemical equilibrium modeling indicated that amorphous calcium carbonate is likely responsiblefor the observed sequestration. High variability and low reactive fractions render many other materialsless attractive for further pursuit without considering preprocessing or activation techniques.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Precipitation of stable carbonate minerals provides a means forcapturing and sequestering carbon dioxide (CO2) in aqueous scrub-ber solutions. Alkaline industrial wastes are potential sources ofpolyvalent cations, especially calcium and magnesium, for reactionwith aqueous carbonate and precipitation of carbonate minerals. Inaddition to the carbon sequestration that is achieved, this processprovides an opportunity for beneficial reuse of abundant industrialwastes.

Interest in aqueous mineral carbonation as a large-scale optionfor carbon sequestration arose from observations of natural silicateweathering and abundance of thermodynamically suitable rawmaterials (Seifritz, 1990). Owing to their ubiquity, these minerals(chiefly magnesium silicates such as olivine and serpentine) con-tinue to be the focus of research for myriad process schemes, treat-ment additives, and condition optimizations (Zevenhoven et al.,2011). Similarly, industrial waste materials with abundant alkalinecontents such as fly ash, iron and steel slag, and alumina refiningwastes have been studied as alternatives to mined ores in direct,

gas–solid carbonation schemes as well as aqueous carbonationschemes, the focus of this work (Bobicki et al., 2012).

Variability in the mineralogies of the cation-source materialsalong with the dynamic process conditions under which they aregenerated complicates comparison of results between independentstudies. This issue is exacerbated by the wide variety of experi-mental designs and analytical tools used by researchers to investi-gate the carbon sequestration capacity of these materials. Forexample Montes-Hernandez et al. (2008) determined a carbonsequestration capacity of 26 g CO2 per kg of fly ash in a batch sys-tem while Back et al. (2008) found a capacity of 230 g CO2 per kg offly ash in a flow-through design. Hence, it is useful and necessaryto examine a variety of materials under identical conditions to gaininsight into their relative behaviors.

Aqueous CO2 capture technologies described in the literaturecommonly examine the use of the ubiquitous enzyme carbonicanhydrase (CA) to overcome slow CO2 hydration kinetics (Bondet al., 2001; Favre et al., 2009; da Costa Ores et al., 2012; Vinobaet al., 2010). Studies that have examined CA systems typically usedbrines, either natural or synthetic, as the polyvalent cation source,and have inadequately assessed the potential role for alkalineindustrial wastes (Liu et al., 2005; Favre et al., 2009; Rawlins,2008). Alternatively, use of sodium hydroxide (NaOH) to captureCO2 from ambient air has been proposed, with CO2 sequestered

parallel

http://dx.doi.org/10.1016/j.wasman.2014.03.009

mailto:[email protected]


http://www.sciencedirect.com/science/journal/0956053X

http://www.elsevier.com/locate/wasman


2 C.W. Noack et al. / Waste Management xxx (2014) xxx–xxx

via causticization and calcination prior to geologic storage (Zeman,2007; Stolaroff et al., 2008). For this study, sodium bicarbonate isused to mimic the chemistry of a solution that might result frombuffered CO2 capture with CA or from NaOH absorption.

This work examined directly comparable reactivities of a widerange of alkaline industrial wastes by exposure to a simple sodiumbicarbonate solution with a goal of providing guidance to future,process-oriented studies of these materials. The overall goal ofthe study was to investigate relative efficiencies of alkaline indus-trial waste reaction with aqueous carbonate. The specific objec-tives were to: estimate the relative potential aqueous carbonsequestration capacities of various alkaline industrial wastes; esti-mate the reaction time necessary to reach short-term pseudo-equi-librium; and determine the most promising high-volume industrialwastes for use in an aqueous carbon sequestration process.

2. Materials and methods

2.1. Industrial residuals

The types of materials investigated in this study were selectedbased on reported mineral compositions and related prior work.Thirty-one samples representing several distinct industrial opera-tions were acquired (see Table 1). The test samples included: coalcombustion fly ash (FA), spray dryer absorber ash (SDA), circulat-ing dry scrubber ash (CDS), cement kiln dust (CKD), blast- or basicoxygen-furnace slag (FS), electric arc furnace dust (EAFD), and wetflue gas desulfurization gypsum (GYP). Additionally, two bench-mark samples that were previously studied (Dilmore et al., 2009)– fly ash and spray drier absorber ash – were obtained from the Na-tional Energy Technology Laboratory (NETL) of the U.S. Department

Table 1Matrix of samples acquired for study. Sample types are: coal combustion fly ash (FA),spray dryer absorber ash (SDA), circulating dry scrubber ash (CDS), cement kiln dust(CKD), blast- or basic oxygen-furnace slag (FS), electric arc furnace dust (EAFD), andwet flue gas desulfurization gypsum (GYP). Boiler and fuel types specified by sampleproviders.

Sample ID Sample type Boiler type Fuel type

FA-1 FA Pulverized coal (PC) LigniteFA-3 FA Cyclone LigniteFA-5 FA PC LigniteFA-4 FA Cyclone LigniteFA-7 FA Gassifier LigniteFA-8 FA Spreader stoker LigniteFA-6 FA Fluidized bed LigniteFA-11 FA Cyclone SubbituminousFA-2 FA PC SubbituminousFA-9 FA PC SubbituminousFA-10 FA PC SubbituminousFA-DOE FAFA-12 FA PC S/B BlendSDA-3 SDA PC LigniteSDA-2 SDA Cyclone LigniteSDA-5 SDA PC SubbituminousSDA-7 SDA PC SubbituminousSDA-1 SDA PC SubbituminousSDA-4 SDA PC SubbituminousSDA-DOE SDASDA-6 SDA BituminousCDS-1 CDS IndustrialCDS-2 CDS BituminousCKD-1 CKD N/A N/ACKD-2 CKD N/A N/ACKD-3 CKD N/A N/AFS-1 BOFS N/A N/AFS-2 BFS N/A N/AEAFD-1 EAFD N/A N/AEAFD-2 EAFD N/A N/AGYP-1 GYP

Please cite this article in press as: Noack, C.W., et al. Comparison of alkaline indreactivity study. Waste Management (2014), http://dx.doi.org/10.1016/j.wasm

of Energy. All SDA and CDS samples came from coal-fired powerplant operations without fly ash pre-collection and thus representa blend of combustion fly ash and desulfurization solids. SamplesFA-1, FA-2, FA-3, FA-5, FA-9, FA-10, SDA-1, SDA-2, SDA-3, andSDA-4 were part of the EERC coal combustion product (CCP)sample bank and had been analyzed as part of a previous study.No discernable, systematic differences were observed betweenthe previously analyzed and freshly collected samples.

2.2. Preliminary solid composition characterization

The bulk chemical composition of all samples was determinedusing X-ray fluorescence (XRF) spectroscopy. Two XRF techniques(fused pellet and pressed pellet) were employed for the analysisof the samples (ASTM, 2004; ASTM, 2011). The EERC CCP samplebank samples were analyzed as fused pellets while all othersamples were analyzed as pressed pellets.

2.3. Leaching methodology for screening and detailed experiments

2.3.1. Leaching protocolPrior to testing, the samples were ground, if necessary, so that

80% of the sample mass passed through a 1 mm screen. Twentygrams of sample were mixed with 120 mL of 0.5 M NaHCO3 solu-tion at 350–500 RPM by a magnetic stir bar in a 500 mL glass bea-ker. The sodium bicarbonate solution was prepared using distilledwater. Similar tests were conducted with distilled water (specificconductivity <20 lmhos/cm) as the extraction agent. Measure-ments of pH were taken of the stirred slurry throughout the exper-iment. pH meters were calibrated using standard buffer solutionsat pH 7, pH 10, and pH 12.

After the contact period, the slurry was allowed to settle briefly.The supernatant was decanted, centrifuged, and then filtered at0.45 lm. Finally, five grams of the filtrate was titrated to an end-point of pH 4.5 (and an intermediate endpoint of pH 8.3) with1.0 N HCl to determine alkalinity. Each filtered supernatant was ti-trated in duplicate.

2.3.2. Screening criteriaScreening tests were conducted on the entire suite of solids for

24 h in each leach solution with procedural duplicates in 0.5 MNaHCO3. Samples were ranked according to reactivity and seques-tration capacity. These screening tests were used to focus subse-quent, detailed testing and analysis on the best-performingsamples. Reactivity was assessed based on the time evolution ofslurry pH while sequestration capacity was calculated using finalpH and alkalinity (Section 2.6). Screening-stage leachates andresidual solids received no additional analysis.

Based on reactivity and calculated carbon sequestration, themost reactive solids were selected for more detailed investigation.To identify the subset of samples for in-depth testing and model-ing, consideration was given to identifying the most promising car-bon sequestration candidates while also ensuring that samplesfrom each of the material types were included, with highly similarsamples being excluded to avoid redundancy.

2.3.3. Detailed leaching experimentsLeach testing on the set of best-performing samples was con-

ducted in distilled water for four hours and in 0.5 M NaHCO3 forone and four hours. As with the screening tests, pH was measuredand the slurry was separated at the end of the test period with afiltered supernatant sample titrated to determine alkalinity. Addi-tionally, supernatants and solids were reserved for further analysiswith the solids being dried at 42–48 �C to preserve any hydratedminerals which may have precipitated.

ustrial wastes for aqueous mineral carbon sequestration through a parallelan.2014.03.009


C.W. Noack et al. / Waste Management xxx (2014) xxx–xxx 3

2.4. Solution chemistry analysis

Aliquots of filtered supernatant samples were analyzed byinductively coupled plasma-atomic emission spectroscopy (ICP-AES) according to EPA method 6010b (EPA, 1996) for cations andby ion chromatography (IC) according to EPA method 300 (EPA,1993) for anions.

2.5. Solids analysis

Pre- and post-test solids from the 4-h bicarbonate leach testwere analyzed for mineral phases and for total carbon. X-raydiffraction (XRD) was used for mineral identification with diffrac-tion patterns collected on a Bruker AXS D8 Advance XRD systemand interpreted using the PDF-2 database (ICDD, 2007). Mineralphases identified with XRD were quantified by Rietveld quantita-tive phase analysis using whole-pattern, general least-squaresrefinement.

A Carbon–Hydrogen–Nitrogen (CHN) technique was used todetermine the total carbon in the as-received samples and thedried samples following the 4-h bicarbonate screening test. Theanalyses were performed on an Exeter Analytical CE 440 ElementalAnalyzer. Total carbon in the as-received samples was also ana-lyzed using a LECO Total Carbon Analyzer for comparison, howeverpost-test solids were not evaluated by this method.

2.6. Sequestration capacity calculations

Sequestration capacity was estimated using two approaches:(1) alkalinity of filtered supernatants measured by acid titration,and (2) changes in total carbon of the solids measured by CHNanalysis. The alkalinity-based approach allowed for estimation ofsequestration capacity from screening data for all 31 test solids(where solid phase carbon content was not measured) and aidedin the selection of the samples for more detailed analysis. Compar-isons of the total carbon measurements on solids before and afterreaction allowed direct assessment of the sequestration of carbon-ate by precipitation.

Alkalinity in these systems was initially assumed to be carbon-ate dominated (Eq. (1); i.e. ignoring hydroxylated metals, silicicacids, phosphates, and sulfides). A subscript, alk, has been appliedto the carbonate species to indicate that these values include aque-ous complexes, such as NaCO�3 , which contribute to alkalinity butwere not directly measured. Complexes of carbonate are knownto account for large fractions of total carbonate, however their con-tributions to alkalinity are captured during titration (Stumm andMorgan, 1996). Aqueous carbonate speciation was calculated usingthe measured pH values, alkalinity, second carbonic acid dissocia-tion constant (Ka ¼ 10�10:3; Stumm and Morgan, 1996), and ionactivity coefficients (c) calculated from the Davies Equation foran ionic strength of 0.5 M (Eq. (2)). Changes in ionic strength,due to mineral dissolution and precipitation, and activity coeffi-cients are expected to be minimal, thus motivating the use of0.5 M. Combining Eqs. (1) and (2) allows for calculation of total dis-solved inorganic carbon.

Alk ðeq=LÞ ¼ �½Hþ� þ ½OH�� þ ½HCO�3 �alk þ 2 � ½CO2�3 �alk ð1Þ

½CO2�3 �alk ðMÞ ¼

Ka � c1 � ½HCO�3 �alk

c2 � fHþg

ð2Þ

Carbon sequestration, occurring over a test period of time s,was estimated by a mass balance on total dissolved carbon species(P½C� ¼ ½HCO�3 �alk þ ½CO2�

3 �alk; as estimated from titration data) inthe system before and after reaction with 0.5 M NaHCO3 (Eq.(3)). Carbon sequestration capacity was calculated based on testconditions of 120 mL solution and 20 g of solid.


DCalk ðMÞ ¼X½C�t¼0 �

X½C�t¼s ð3Þ

A solid-phase carbon mass balance was used to estimate carbonsequestration for the samples chosen for additional testing.Changes in percent carbon, determined by CHN analysis beforeand after reaction with 0.5 M NaHCO3, were converted to changesin carbon mass based on the measured sample mass before andafter exposure to 0.5 M NaHCO3 for time s (Eq. (4)). This changein solid phase carbon (converted to a molar value) and the initialtest sample mass were used to yield a sequestration capacity inthe same units employed in the alkalinity estimation method.

DCsolid ðg CÞ ¼ ð%Ct¼s �mt¼sÞ � ð%Ct¼0 �mt¼0Þ ð4Þ

As this study was intended to assess the sequestration potentialof a wide variety of samples and identify best-performing materialclasses, the absolute values of sequestration capacity defined byEqs. (3) and (4) were meant to serve only as selection criteriaand are thus not directly comparable to published literature re-sults. Under the test conditions (120 mL of 0.5 M NaHCO3 mixedwith 20 g of solid), the maximum potential sequestration capacitywas 3 mol CO2 per kg dry solid assuming no imbibition of atmo-spheric CO2 during mixing. Further optimization of both absolutesequestration and sequestration efficiency metrics for implemen-tation in a full-scale system would be possible with further testing.

2.7. Chemical equilibrium modeling

Chemical equilibrium modeling of the data generated in theseexperiments was performed to calculate the change in dissolvedcarbon mass and to hypothesize the mineral phases controllingthe observed composition through examination of saturation indi-ces (SI) using PHREEQC (Parkhurst and Appelo, 1999). Post-reac-tion solution chemistries were evaluated to determine thespeciation of ions, the critical aqueous interactions governing thisspeciation, and the contribution of non-carbonate species to alka-linity for carbon sequestration estimates. For all models measuredpH, alkalinity, ICP data, and IC data (excluding CO2�

3 ) were used tocalculate total dissolved inorganic carbon. Examination of solubil-ity controlling minerals was conducted according to methodolo-gies set forth by Khaitan et al. (2009) and Meima and Comans(1997). Solubility products of mineral phases not present in thethermodynamic database associated with PHREEQC were obtainedfrom studies reported in the literature and included manually.

3. Results and discussion

Herein we discuss the results of sample characterization, leach-ing tests, mineralogical analysis, and chemical equilibrium model-ing, focusing on trends within material classes and the implicationsfor carbon sequestration and waste management. Raw data for allindividual samples and all analyses have been included in support-ing information.

3.1. Preliminary sample characterization

The primary polyvalent cations present in the study materialswere expected to be Ca and Mg. Fig. 1 compares the mass fractionsof these metals in the different sample types acquired for thisstudy along with Al and Fe. As seen in Fig. 1, the materials obtainedfor this study generally contained a high fraction of Ca, which hasbeen shown to be more effective than Mg for sequestration reac-tions at ambient conditions (Back et al., 2008). The material classeswith the highest Ca content were FGD ashes (SDA and CDS), ce-ment kiln dusts, and iron/steel furnace slags. The EAFD samplescontained the least Ca. Solubilized Fe and Al are expected to havean important effect on the solution chemistry as discussed in



Aluminum

Calcium

Iron

Magnesium

0

20

40

0

20

40

0

20

40

0

20

40

CDS

CKD

EAFD FA FS

GYP

SDA

Mass

fraction

(%)

Fig. 1. Sample mass fractions of aluminum, calcium, iron, and magnesium,determined by XRF, in material types included in this study. Points in black arethose selected for further testing, while those in gray were excluded. Material typesare: coal combustion fly ash (FA), spray dryer absorber ash (SDA), circulating dryscrubber ash (CDS), cement kiln dust (CKD), blast- or basic oxygen-furnace slag (FS),and electric arc furnace dust (EAFD).


Section 3.3.2. Solid phase Fe was highest in the iron and steel resid-uals while Al was enriched in the fly ashes.

3.2. Screening tests

Time-series pH data provided information about the acid–basereaction kinetics of the solids. Fig. 2 presents the pH vs. time re-sults for two CKD samples in the leach solutions. Asymptoticbehavior of the pH vs. time data was observed for the solids, rap-idly in most cases, and suggested pseudo-equilibrium had beenreached in the system. Plateauing of reaction extent during carbon-ation of waste materials has been attributed to surface passivationby precipitates leading to slower, diffusion controlled reaction

0.0042 0.08 1 2 4 7 24

8

9

10

11

12

13

pH

Time (hrs)

CKD−1 / BC

CKD−1 / DW

CKD−2 / BC

CKD−2 / DW

Sample / Leachant

Fig. 2. Time-series slurry pH test results for two cement kiln dust (CKD) samples indistilled water (DW) and 0.5 M NaHCO3 (BC) solutions at 6:1 liquid to solid ratios.Values with 0.5 M NaHCO3 as the extraction solution represent averages fromduplicate tests. Initial (t = 0) pH values have been excluded from the logarithmictime scale. For all tests presented here the initial pH of the 0.5 M NaHCO3 solutionwas between 7.9 and 8.3.


kinetics (Huntzinger et al., 2009). This behavior was apparent forsample CKD-2 with a transition from rapid, log-linear pH changeto slow pH change in 0.5 M NaHCO3 at approximately 1 h of reac-tion. The samples that most rapidly reached pseudo-equilibriumlikely exhibited the most rapid short-term solids dissolution andcould potentially require a shorter retention time for sequestrationreactions. Time-series pH data are provided for all samples inFig. S1.

Based on Fig. 2, CKD-1 would not be expected to serve as a suit-able sequestrant without pretreatment as it liberates minimal basein both leach solutions. Within the applicable range, the increasedionic strength of the leach solution would increase mineral solubil-ities by decreasing ion activities (Stumm and Morgan, 1996). Thiseffect likely explains why an increase in pH, potentially corre-sponding to alkaline mineral dissolution, was observed duringreaction with 0.5 M NaHCO3 but not with distilled water forCKD-1, however it is not possible to assess this hypothesis withthe current data. By contrast, CKD-2 reacted rapidly to reach highpH in both solvents indicating that in the 0.5 M NaHCO3 sufficientbase had been liberated to overcome the pH buffering of the leachsolution.

During reaction with 0.5 M NaHCO3 solution, 9 of the 31 sam-ples achieved 24-h pH values greater than 10.3. This is an impor-tant threshold for carbon sequestration since above this pHcalcite solubility is thermodynamically minimized (Stumm andMorgan, 1996). All material classes investigated except for EAFDare represented by these nine samples. For the best performing sol-ids a high, stable pH was achieved within four hours of mixing.

Alkalinity measurements were used as a proxy for dissolvedinorganic carbon and used to estimate relative aqueous carbonsequestration capacity through mass balance calculations. Fig. 3presents the estimation of sequestration capacity as a function ofpH at 24 h for the 31 solids evaluated. Sequestration capacityshows clear trends within sample types as a function of pH. Thisis likely due to similar mineral phase compositions within the dis-tinct sample types. The general trend of increasing sequestrationcapacity with pH can be explained by the solubility behavior of cal-cite (limited above pH 10.3) and the acceleration of precipitationkinetics with mineral supersaturation (Spanos and Koutsoukos,1998).

The optimal solids within this experimental context plot in thetop right corner of Fig. 3. These samples created a highly basicsolution and precipitated much of the initial carbonate and likely

0

1

2

3

8 9 10 11 12 13

pH

Calk

(molC

O2

/kg

solid

)

Material classCDS

CKD

EAFD

FA

FS

GYP

SDA

Fig. 3. Apparent aqueous carbon sequestration capacity, calculated from alkalinitydata, vs. final measured slurry pH for 24-h exposure of 20 g solids to 120 mL 0.5 MNaHCO3. Markers represent the average Hþ activity and sequestration capacity;vertical error bars denote � sample standard deviation from 4 titrations; horizontalerror bars mark the range of pH measured in duplicate leach tests. Points in blackare those selected for further testing, while those in gray were excluded. Sampletypes are: coal combustion fly ash (FA), spray dryer absorber ash (SDA), circulatingdry scrubber ash (CDS), cement kiln dust (CKD), blast- or basic oxygen-furnace slag(FS), and electric arc furnace dust (EAFD).



12

12.5

13

pH

CDS + SDA

CKD

Material class


contain significant fractions of lime, CaO or portlandite, Ca(OH)2.Many desulfurization residuals (i.e. CDS, SDA, and GYP) seques-tered significant carbon without increasing pH above 10 (upper leftof Fig. 3). These samples likely represent materials where the reac-tive CaO, used for dry desulfurization, was more effectively con-verted to CaSO3 or CaSO4 minerals. According to the criteriadescribed in Section 2.3, the subset of samples chosen for furtherinvestigation were: CDS-1, CDS-2, CKD-2, EAFD-1, FA-6, FA-7, FA-10, FA-11, FA-12, FS-1, SDA-DOE, SDA-1, and SDA-6.

−0.02 0 0.02 0.04 0.06 0.08 0.1 0.12 0.1411

11.5

Σ [SBT] − Σ [SAT] (eq/kg)

EAFD

FA

FS

Fig. 5. Observed dependence of measured slurry pH on acid–base balance for solidsin contact with distilled water for 4 h. Acid–base balance defined as the differencebetween strong base tracers (SBT) and strong acid tracers (SAT). Base tracer specieswere Ca, Na, and K. Acid tracer species were Cl�; SO2�

4 ; NO�3 , and Br� . Materialclasses are: coal combustion fly ash (FA), spray dryer absorber ash (SDA), circulatingdry scrubber ash (CDS), cement kiln dust (CKD), blast- or basic oxygen-furnace slag(FS), and electric arc furnace dust (EAFD).

3.3. Alkalinity leaching and carbonate precipitation tests

3.3.1. Distilled water leach test solution chemistryTests conducted in distilled water served to examine mineral

solubility and to determine the dominant controls of solution pH.The test solids yielded significant amounts of Ca upon exposureto distilled water along with other alkaline cations such as Naand K. A summary of the leachate chemistry is given in Fig. 4.The material types that released the most Ca were CKD, CDS, andSDA. Magnesium was non-detectable (<0.2 mg/L) in all distilledwater leach tests. As in the 24-h exposure study, the leachates be-came highly basic with an average pH near 12 across all materialclasses.

The acid–base balance in a water sample is estimated from thesum of strong acid tracer anion equivalents and strong base tracercation equivalents (Stumm and Morgan, 1996). This is equivalentto the difference between the sum of dissolved mono- and divalentcations, except for Hþ, and the sum of dissolved anions, excludingOH� and the carbonate species. Fig. 5 presents the relationship

Fig. 4. Average supernatant dissolved component concentrations determined byICP-AES and slurry pH for the material types of the reduced sample set. All elementconcentrations are on the left y-axis in units of charge equivalents while pH is aloneon the right y-axis. Panels correspond to different leach solutions – distilled water(DW) and 0.5 M NaHCO3 (BC) – and the test duration. Numbers in parentheses arenumber of samples of each material class. Material classes are: coal combustion flyash (FA), spray dryer absorber ash (SDA), circulating dry scrubber ash (CDS), cementkiln dust (CKD), blast- or basic oxygen-furnace slag (FS), and electric arc furnacedust (EAFD).


between pH and the acid–base balance tracers in the distilledwater leach tests and demonstrates a positive relationship with ex-cess strong base tracers. This relationship confirms that significantamounts of the alkaline cations were both present and accessibleby solution in an oxide or hydroxide form, which is critical forthe success of sequestration reactions. However, as Fig. 5 indicates,variance of these species does not fully explain the variance in pHof the solutions. Some inconsistency was expected as cations werequantified by ICP-AES and thus represent total dissolved concen-trations while the anions were quantified by IC which did not in-clude any aqueous complexes of the anions.

3.3.2. Sodium bicarbonate leach test solution chemistryThe metals chemistry in the supernatants was important in

evaluating the reactions occurring during leaching tests, as metalions can inhibit calcium carbonate formation, a targeted reactionfor CO2 sequestration. The solution chemistry from the 1- and 4-h 0.5 M NaHCO3 leach tests showed increased concentrations ofiron and magnesium from the distilled water leach tests in whichneither were detected for any sample (Fig. 4).

Inorganic ions can strongly retard calcite crystal growth rates.Ferric iron can completely inhibit calcite growth at concentrationsorders of magnitude lower than calcium and carbonate by adsorp-tion of ions or colloidal solids blocking growth sites (Katz et al.,1993). Any iron leached will oxidize rapidly at high pH (Sungand Morgan, 1980), suggesting that ferric iron dominated the sol-uble iron species. It has been shown that incorporation of otherions such as Naþ; Mg2þ; or SO2�

4 in the calcite crystal lattice en-hance solubility and slow kinetics (Akin and Lagerwerff, 1965;He and Morse, 1993). Moreover, calcite growth inhibitors havebeen show to behave additively Matty and Tomson (1988). Sincethe studied leachates contained other growth inhibitors – such asmagnesium, aluminum, and phosphate (Fig. 4) – the presence ofthese inhibitors may explain, in part, the observed supersaturationof calcite calculated from equilibrium modeling of the supernatantdata (Section 3.4).

3.3.3. Solid mineral phase chemistryBased on previous XRD analyses of similar materials, a signifi-

cant amount of the total calcium was expected to exist as an oxideor hydroxide for these material classes (Dilmore et al., 2009;Huijgen et al., 2005; Huntzinger et al., 2009; Mattigod et al.,1990). Neither the FA samples nor the iron and steel residuals



(A)

(B)

Fig. 7. (A) Comparison of solid-phase (y-axis) and aqueous-phase (x-axis) seques-tration capacity estimates for 4-h reaction of 20 g solid with 120 mL 0.5 M NaHCO3.Abscissa values were calculated through a mass balance on dissolved carbonateestimated from titration data assuming carbonate dominated alkalinity as in Eq. (1)(filled markers) or from PHREEQC model results (open markers), which account fornon-carbonate alkalinity. (B) Final combined (aqueous and solid phase) carbonmass vs. initial combined carbon mass. Note that in (B) the values have been log-transformed. For both subfigures the dashed line represents 1:1 correlation.Material classes are: coal combustion fly ash (FA), spray dryer absorber ash(SDA), circulating dry scrubber ash (CDS), cement kiln dust (CKD), blast- or basicoxygen-furnace slag (FS), and electric arc furnace dust (EAFD).


(EAFD and FS) met this expectation, with the crystalline calciumcontent of these materials occurring primarily as mixed metal sil-icates as determined in the XRD analyses. Other common calciumminerals include various sulfate/sulfite polymorphs. The changesin major calcium and carbonate minerals detected by XRD forpre- and post-exposure solids subjected to four hour leaching with0.5 M NaHCO3 are estimated in Fig. 6.

Of the 13 samples, only four (FA-6, SDA-1, SDA-6, and CDS-2)were found to have gained a substantial calcite mineral phase fol-lowing reaction with 0.5 M NaHCO3. However, for all samples sig-nificant amorphous phases were noted, though not identified orquantified, which could include calcium or magnesium carbonates.More commonly, various sodium carbonate polymorphs were ob-served in the post-exposure samples, including natrite (Na2CO3),natron (Na2CO3 * 10H2O), and trona (NaHCO3 * Na2CO3 * 2H2O). Ifthese minerals were precipitating during reaction, the sequestra-tion process does not accomplish its goals. However, chemicalequilibrium modeling, as discussed in Section 3.4, was used todetermine that these minerals precipitated during the evaporationof pore water while drying and not during reaction.

3.3.4. Carbon sequestration measurementsIn the 1- and 4-h tests, the trends in sequestration capacity

(calculated from titration data) were similar to those observed ininitial screening (i.e., Fig. 3) with the same material types seques-tering high amounts of carbon (CKD, CDS, and SDA). Further, thesesamples also reacted most rapidly, reaching at least 85% of their24-h capacity in only 4 h with most reaching more than 95%. Theseresults support the pseudo-equilibrium assumption made based onpH-time profiles during screening. In addition to having lowerabsolute sequestration capacities, fly ashes and furnace slags re-acted slowly, failing to surpass 70% of their 24-h capacity in fourof six samples.

Measurements of total carbon in the solid phase before andafter reaction with 0.5 M NaHCO3 were used to calculate the massof carbonate sequestered as a solid. While a 1:1 correlation was ex-pected between the aqueous and solid phase carbon sequestrationestimates there was poor agreement of the two estimation meth-ods (Fig. 7A). This observed disparity may have arisen from exper-imental artifacts or analytical uncertainty. However, analysis of

Fig. 6. Calcium and carbonate mineral mass change following 4-h reaction with0.5 M NaHCO3. Mass change of mineral S is calculated byDmS ¼ ð%St¼s �mt¼sÞ � ð%St¼0 �mt¼0Þ. The total amorphous fraction of the totalmass was assumed to remain constant during reaction. Non-detectable phases wereassumed to be zero. Solid lines represent the limit of quantification based on threetimes the instrument detection limit. Material classes are: coal combustion fly ash(FA), spray dryer absorber ash (SDA), circulating dry scrubber ash (CDS), cementkiln dust (CKD), blast- or basic oxygen-furnace slag (FS), and electric arc furnacedust (EAFD).


sources of error in the two methods (see supporting information)revealed that error in the aqueous phase estimates is more easilyreconciled and corrected. For this reason, the alkalinity method,with modifications discussed subsequently to account for non-car-bonate alkalinity, was considered to be more representative of ac-tual sequestration capacity.

Despite poor agreement between sequestration capacity calcu-lations, the carbon mass balance on the total system was closedwithin experimental and analytical uncertainty for almost all sam-ples after accounting for non-carbonate alkalinity (9 of 13 samples< 15% relative error; Fig. 7B). Excess carbon outside of experimen-tal and analytical error was observed in two samples and could beexplained by dissolution of atmospheric CO2 but the amount isvery small, amounting to no more than 1% of the initial carbon insolution. This appears to confirm that our initial assumption, thatatmospheric CO2 imbibition is negligible. The mechanism of carbonloss is unclear, however systems with low initial, solid-phase car-bon (where the largest discrepancies are observed) are more sensi-tive to analytical errors, while procedural errors during leach testsand alkalinity titration are constant for all systems. This furthervalidates the choice to rely on aqueous-phase, model-adjustedsequestration capacities.

3.4. Chemical equilibrium modeling

Aqueous speciation and mineral saturation indices were calcu-lated by PHREEQC for leach test supernatants using experimental




data. Chemical equilibrium modeling enabled calculation ofsequestration capacity accounting for non-carbonate alkalinityand ionic strength. Saturation indices of carbonate minerals inthe supernatants of the bicarbonate leach tests gave some indica-tion of the mineral phases governing the observed sequestration.

3.4.1. Model calculation of sequestration capacityFor the 1-h bicarbonate leach test, calculation of sequestration

capacity using model results for total dissolved inorganic carbonresulted in increased estimates for 11 of the 13 samples. For the4-h bicarbonate leach test, modeling results yielded increasedsequestration capacity estimates for all 13 samples, with an aver-age increase of 38% for the entire sample group. Fig. 7A shows boththe original alkalinity estimation method and updated model esti-mate for the 4-h leach test. As model estimates of carbon seques-tration capacity increased, the disparity between solid andaqueous phase estimations also increased for samples where thesolid-phase metric was less than the aqueous-phase estimate.However, method agreement was improved for samples wherethe solid-phase estimate exceeded the aqueous-phase estimate.

For three samples, FA-11 in distilled water and SDA-1 and FA-6in bicarbonate for 4 h, the measured total aluminum concentration(by ICP-AES) and pH would theoretically yield an alkalinity greaterthan what was measured. This elevation of alkalinity is caused by a4.0 equivalent per mole contribution of the species AlðOHÞ�4 , whichbecomes the dominant dissolved aluminum phase at pH above 7(Stumm and Morgan, 1996). A discussion of how this was correctedfor equilibrium modeling is included in the supportinginformation.

3.4.2. Model calculation of mineral saturationSaturation indices (SI) for important minerals were calculated

using Eq. (5), where IAP is the ion activity product and Ksp is thesolubility product of the mineral of interest.

SI ¼ IAPKsp

ð5Þ

A summary of log SI for select carbonate minerals is presentedin Fig. 8. In the distilled water supernatant speciation, all sampleswere supersaturated with calcite with log SI values ranging from1.00 to 3.96. In the 1- and 4-h bicarbonate leach tests, the calcitelog SI values ranged from 1.15–3.40 and 0.61–3.43 respectively.The cases where the log SI was very high (P1.00) may be a resultof leached calcite inhibitors mentioned previously. Several studiesof alkaline wastes in water show significant supersaturation of cal-cite even after long exposure times, likely due to the leaching ofinhibitors (Bhattacharyya et al., 2011; Duchesne and Reardon,1998; Huijgen and Comans, 2006; Roy and Griffin, 1984).

Calcite am CaCO3 Natron Trona

8

4

0

1 hr 4 hr 1 hr 4 hr 1 hr 4 hr 1 hr 4 hr

Reaction time

log

10S

I

Fig. 8. Saturation indices (SI) for potential carbonate minerals calculated with Eq.(5) based on PHREEQC modeling results of supernatant solution chemistry datafrom 0.5 M NaHCO3 leach tests. Solid lines represent the range of SI for potentialcontrolling mineral phases (Meima and Comans, 1997).


Meima and Comans (1997) considered any minerals where thelog SI approached zero to be potential controlling phases. For cal-cite, only 4-h bicarbonate leach samples FA-6 and SDA-1 met thiscriterion with log SI values of 0.61 and 0.72 respectively; thesewere also two of the four minerals where an appreciable increasein solid-phase calcite was observed by XRD. For amorphous CaCO3,after the 1-h bicarbonate leach test all samples except for SDA-DOEand CDS-1 were in the log SI range �1 to 1, while after 4 h only thelog SI for sample CKD-2 exceeded 1. Amorphous CaCO3 is a meta-stable predecessor of calcite which will naturally undergo dissolu-tion and recrystallization to maximize thermodynamic stability(Brecevic and Nielsen, 1989). In all samples carbonate was in ex-cess of calcium, meaning that there would be little additionalsequestration capacity gained if the aqueous systems were to cometo calcite equilibrium.

Experimental results indicated that sodium carbonate poly-morphs made up the majority of crystalline carbonate phaseswhich increased significantly during reaction (Fig. 6). Based onthe modeling it would appear unlikely that either natron or tronaprecipitated during reaction. Rather, given that the calculated SIvalues for both minerals in both bicarbonate leach tests were�1.98 or lower and that anhydrous sodium carbonate polymorphs(e.g. Na2CO3) have higher solubility (Monnin and Schott, 1984),these minerals probably formed as a result of evaporation duringsample drying. Moreover, a simplified evaporation model usingrelevant supernatant data (Na, Ca, alkalinity, and pH) from sampleCDS-1, which exhibited a large increase in natron, showed that na-tron precipitation occurs only after 93% evaporation (Fig. S2).

4. Conclusions

This work has demonstrated the feasibility of using select alka-line industrial wastes as sources of divalent cations in an aqueouscarbon sequestration scheme. Experimental results indicate themost reactive materials were able to sequester between 2.3 and2.8 mol CO2 per kg dry solid, or 77% and 93% of the initial aqueouscarbonate respectively, in a 4-h reaction under the test conditions.These materials were cement kiln dust, spray dryer absorber ash,and circulating dry scrubber ash, all of which are materials withhigh fractions of reactive calcium content. In addition to sequester-ing significant amounts of carbonate, these solids liberated exten-sive base and caustic alkalinity which is desired for an aqueous CO2

sequestration process. However, screening of a number of samplesin each material class indicated that there was high variability insequestration performance within each class, likely due to operat-ing conditions within the process generating the waste. Additionaltesting is needed to optimize both the sequestration capacity andefficiency.

Less reactive materials, such as fly ash as well as iron and steelresiduals, may still be viable options because of their abundantalkaline cation content, but their use would be enhanced by addi-tional preprocessing or activation. However, these less desirablesequestrant materials were effectively neutralized indicating thatcarbonate exposure may be an interesting remedy for these alka-line wastes.

Two methods were used to estimate carbon sequestrationcapacity: evaluation of changes in solution chemistry and evalua-tion of changes in carbon content of the solids, The use of solidphase changes was found to be problematical because low-carbonsamples experienced changes in carbon content near the instru-ment detection limit. Chemical equilibrium modeling of solutionchemistry for estimation of sequestration capacity allowed for con-sideration of non-carbonate alkalinity in the supernatant. Carbonsequestration estimates by the solution phase analysis approachwere less affected by analytical error in the solution phase




measurements than were carbon sequestration estimates from so-lid phase carbon analyses. Supernatant chemical modeling indi-cated that the controlling carbonate mineral phase in aqueousleaching of the alkaline waste materials was primarily amorphousCaCO3, which evolves over time to form more stable calcite.

Associated content

Solid phase characterization data (XRF, CHN, and XRD), titrationdata (pH and alkalinity), and solution chemistry data (ICP-AES andIC) for all samples and experiments are provided in the supportinginformation. Additional discussion of discrepancies between aque-ous and solid-phase sequestration capacity estimates as well as thecorrection of equilibrium models for aluminum are also included inthe supporting information.

Acknowledgements

This work was funded by the U.S. Department of Energy grantFOA-0000015 to Alcoa, and by Alcoa. The authors would like tothank the following individuals for assistance with sample acquisi-tion: Jason Laumb of EERC, Yee Soong of NETL, and various indus-trial partners of both organizations. Finally, the authorsacknowledge the contributions of Dr. Richard Lunt, who providedvaluable advice at the beginning of this study but who passed awaybefore the study was completed. The authors also recognize Asso-ciate Editor Kaimin Shih and three anonymous reviewers, whosecomments significantly improved this manuscript.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.wasman.2014.03.009.

References

Akin, G., Lagerwerff, J., 1965. Calcium carbonate equilibrium in solutions open tothe air II. Enhanced solubility of CaCO3 in the presence of Mg2þ and SO2�

4 .Geochim. Cosmochim. Acta 29, 353–360.

ASTM, 2004. Standard test methods for sampling and testing fly ash or naturalpozzolans for use in portland-cement concrete.

ASTM, 2011. Standard test method for ash in the analysis sample of coal and cokefrom coal.

Back, M., Kuehn, M., Stanjek, H., Peiffer, S., 2008. Reactivity of alkaline lignite fly ashtowards carbon dioxide in water. Environ. Sci. Technol. 42, 4520–4526.

Bhattacharyya, P., Reddy, K.J., Attili, V., 2011. Solubility and fractionation ofdifferent metals in fly ash of powder river basin coal. Water Air Soil Poll. 220 (1–4), 327–337.

Bobicki, E.R., Liu, Q., Xu, Z., Zeng, H., 2012. Carbon capture and storage using alkalineindustrial wastes. Prog. Energy Combust. Sci. 38 (2), 302–320.

Bond, G.M., Stringer, J., Brandvold, D., Simsek, F.A., Medina, M.-G., Egeland, G., 2001.Development of integrated system for biomimetic carbon dioxide sequestrationusing enzyme carbonic anhydrase. Energy Fuels 15, 309–316.

Brecevic, L., Nielsen, A.E., 1989. Solubility of amorphous calcium carbonate. J. Cryst.Growth 98, 504–510.

da Costa Ores, J., Sala, L., Cerveira, G.P., Kalil, S.J., 2012. Purification of carbonicanhydrase from bovine erythrocytes and its application in the enzymic captureof carbon dioxide. Chemosphere 88 (2), 255–259.


Dilmore, R., Howard, B., Soong, Y., Griffith, C., Hedges, S., DeGalbo, A., Morreale, B.,Baltrus, J., Allen, D., Fu, J., 2009. Sequestration of carbon dioxide in mixtures ofcaustic byproduct and saline waste water. Environ. Eng. Sci. 26 (8), 1325–1333.

Duchesne, J., Reardon, E., 1998. Determining controls on element concentrations incement kiln dust leachate. Waste Manage. 18, 339–350.

EPA, 1993. Determination of inorganic anions by ion chromatography.EPA, 1996. Inductively coupled plasma-atomic emission spectrometry.Favre, N., Christ, M.L., Pierre, A.C., 2009. Biocatalytic capture of CO2 with carbonic

anhydrase and its transformation to solid carbonate. J. Mol. Catal. B: Enzym. 60(3), 163–170.

He, S., Morse, J.W., 1993. The carbonic acid system and calcite solubility in aqueousNa–K–Ca–Mg–Cl–SO4 solutions from 0 to 90 �C. Geochim. Cosmochim. Acta 57(3533–3554).

Huijgen, W.J., Comans, R.N., 2006. Carbonation of steel slag for carbon dioxidesequestration: leaching of products and reaction mechanisms. Environ. Sci.Technol. 40, 2790–2796.

Huijgen, W.J., Witkamp, G.-J., Comans, R.N., 2005. Mineral carbon dioxidesequestration by steel slag carbonation. Environ. Sci. Technol. 39, 9676–9682.

Huntzinger, D., Dierke, J., Kawatra, S., Eisele, T., Sutter, L., 2009. Carbon dioxidesequestration in cement kiln dust through mineral carbonation. Environ. Sci.Technol. 43, 1986–1992.

ICDD, 2007. Pdf-2, database sets 1–45. International Centre for Diffraction Data.Katz, J., Reick, M., Herzog, R., Parsiegla, K., 1993. Calcite growth inhibition by iron.

Langmuir 9, 1423–1430.Khaitan, S., Dzombak, D.A., Lowry, G.V., 2009. Chemistry of the acid neutralization

capacity of bauxite residue. Environ. Eng. Sci. 26 (5), 873–882.Liu, N., Bond, G.M., Abel, A., McPherson, B.J., Stringer, J., 2005. Biomimetic

sequestration of CO2 in carbonate form: role of produced waters and otherbrines. Fuel Process. Technol. 86, 1615–1625.

Mattigod, S., Rai, D., Eary, L., Ainsworth, C., 1990. Geochemical factors controllingthe mobilization of inorganic constituents from fossil fuel combustion residues:I. Review of major elements. J. Environ. Qual. 19 (2), 188–201.

Matty, J., Tomson, M., 1988. Effect of multiple precipitation inhibitors on calciumcarbonate nucleation. Appl. Geochem. 3, 549–556.

Meima, J.A., Comans, R.N.J., 1997. Geochemical modeling of weathering reactions inmunicipal solid waste incinerator bottom ash. Environ. Sci. Technol. 31 (5),1269–1276.

Monnin, C., Schott, J., 1984. Determination of the solubility products of sodiumcarbonate minerals and an application to trona deposition in Lake Magadi(Kenya). Geochim. Cosmochim. Acta 48 (3), 571–581.

Montes-Hernandez, G., Pérez-López, R., Renard, F., Nieto, J., Charlet, L., 2008. Mineralsequestration of carbon dioxide by aqueous carbonation of coal combustion fly-ash. J. Hazard. Mater. 161 (2-3), 1347–1354.

Parkhurst, D.L., Appelo, C., 1999. PHREEQC (Version 2): a computer program forspeciation, batch-reaction, one-dimensional transport, and inverse geochemicalcalculations. US Geological Survey Denver.

Rawlins, C., 2008. Geological sequestration of carbon dioxide by hydrous carbonateformation in steelmaking slag. Ph.D. thesis, Missouri University of Science andTechnology.

Roy, W., Griffin, R., 1984. Illinois basin coal fly ashes. 2. Equilibria relationships andqualitative modeling of ash–water reactions. Environ. Sci. Technol. 18, 739–742.

Seifritz, 1990. Carbon dioxide disposal by means of silicates. Nature 345, 486.Spanos, N., Koutsoukos, P.G., 1998. Kinetics of precipitation of calcium carbonate in

alkaline ph at constant supersaturation. spontaneous and seeded growth. J.Phys. Chem. B 102 (34), 6679–6684.

Stolaroff, J.K., Keith, D.W., Lowry, G.V., 2008. Carbon dioxide capture fromatmospheric air using sodium hydroxide spray. Environ. Sci. Technol. 42 (8),2728–2735.

Stumm, W., Morgan, J., 1996. Aquatic Chemistry, third ed. John Wiley & Sons, Inc.,New York.

Sung, W., Morgan, J., 1980. Kinetics and product of ferrous iron oxygenation inaqueous systems. Environ. Sci. Technol. 14 (5), 561–568.

Vinoba, M., Kim, D.H., Lim, K.S., Jeong, S.K., Lee, S.W., Alagar, M., 2010. Biomimeticsequestration of CO2 and reformation to CaCO3 using bovine carbonicanhydrase immobilized on sba-15. Energy Fuels 25 (1), 438–445.

Zeman, F., 2007. Energy and material balance of CO2 capture from ambient air.Environ. Sci. Technol. 41 (21), 7558–7563.

Zevenhoven, R., Fagerlund, J., Songok, J.K., 2011. CO2 mineral sequestration:developments toward large-scale application. Greenhouse Gases Sci. Technol.1 (1), 48–57.




http://refhub.elsevier.com/S0956-053X(14)00100-7/h0040



























































































Documents

Comparison of alkaline industrial wastes for aqueous mineral carbon sequestration through a parallel reactivity study