Chemical Engineering Science 64 (2009) 187 -- 191

Contents lists available at ScienceDirect

Chemical Engineering Science

journal homepage: w w w . e l s e v i e r . c o m / l o c a t e / c e s

Reactivity of CaO derived from nano-sized CaCO3 particles through multiple CO2
capture-and-release cycles
Nicholas H. Florin, Andrew T. Harris ∗
Laboratory for Sustainable Technology, School of Chemical and Biomolecular Engineering, University of Sydney, Sydney, NSW, 2006, Australia

A R T I C L E I N F O A B S T R A C T

Article history: The carbonation characteristics of pure CaO derived from nano-sized CaCO3 were investigated as part
Received 9 August 2008 of a multi-cycle performance study which showed potential for exploiting the properties of nano-sized
Received in revised form 8 October 2008 CaO sorbents in a continuous CO2 capture-and-release process. To help understand the approach to the
Accepted 12 October 2008
decay asymptote, which is established through multiple capture-and-release cycles, a qualitative model
Available online 1 November 2008
was proposed. The rate of approach and residual conversion defined by the decay asymptote represents
Keywords: the establishment of an equilibrium between the pore volume and surface area loss during thermal
Calcium oxide sintering; and the pore volume and surface area regeneration as a consequence of a solid-state diffusion
CO2 capture mechanism, and the subsequent release of CO2 in the next calcination cycle. This qualitative explanation
CO2 sorbent is valid for all CaO derived CO2 sorbents.
Nanostructure © 2008 Elsevier Ltd. All rights reserved.
Adsorption
Chemical processes
Environment
Particulate processes

1. Introduction the maximum conversion achieved during carbonation of a fixed du-

ration decreases with cycle number as it is limited by the rate of
The integration of a continuous Ca-looping scheme, exploiting the conversion during the (slow) diffusion-controlled phase. Bhatia and
reversible reaction between CaO and CO2 (Eq. (1)), offers potential Perlmutter (1983) first suggested that the diffusion-controlled phase
for reducing the cost of CO2 capture in a future clean energy system may involve a solid-state mass transfer mechanism. This is evidenced
(e.g., Abanades et al., 2007). However, this potential is limited by by the strong influence of impurities in the product layer on the
the observed decay in CO2 capture capacity of CaO through multiple rate of conversion during the diffusion-controlled phase (Florin and
capture and release cycles (Curran et al., 1966; Barker, 1974a; Silaban Harris, 2008b). In contrast to the CO2 capture reaction, the CO2 re-
and Harrison, 1995; Abanades and Alvarez, 2003). lease reaction is much faster and always goes to completion. In prac-
tise this reaction is carried out at temperatures  850 ◦ C in order
CaO(solid) + CO2 (gas) ⇔ CaCO3 (solid)
to produce a near pure stream of CO2 .
◦
H873.15 = −171.2 kJ mol−1 (for carbonation) (1) Strategies for enhancing multi-cycle performance include steam
The reversible reaction between CO2 and CaO has been well stud- hydration and thermal treatments (e.g., Manovic and Anthony, 2007,
ied for a range of applications, from enhanced H2 synthesis to en- 2008; Lysikov et al., 2007). Tailored CaO-based sorbents, whereby the
ergy storage (Curran et al., 1966; Barker, 1974a; Han and Harrison, active material is supported on high surface area solids, such as alu-
1994; Florin and Harris, 2008a). The CO2 capture reaction is charac- mina (Al2 O3 ) or mayonite (Ca12 Al14 O33 ) have also been proposed
terised by an initial rapid kinetically-controlled phase, which tran- (e.g., Li et al., 2006; Feng et al., 2006). In order to be competitive with
sitions abruptly to a slower diffusion-controlled phase; the abrupt the alternative strategy of simply using large quantities of limestone,
transition is well explained by the formation of a critical product tailored sorbent materials must achieve a significantly enhanced av-
layer thickness (Alvarez and Abanades, 2005). The proportion of the erage conversion through a much larger number of cycles to off-
maximum conversion attributable to the reaction-controlled phase set the economic penalties associated with their manufacture. For
diminishes with the number of capture-and-release cycles due to example, Zhen-Shan et al. (2008) estimated that CaO/Ca12 Al14 O33
a change in the particle morphology resulting from sintering. Thus, (75/25 wt%) was approximately on par with dolomite in terms of the
cost of electricity and CO2 mitigation for coal combustion with CO2
removal from the flue gas. Their cost analysis assumed an average
conversion of 27% with the tailored sorbent, compared to only 14%
∗
Corresponding author. Tel.: +61 2 9351 2926; fax: +61 2 9351 2854. with the dolomite. Despite this result, there is clear scope for further
E-mail address: aharris@chem.eng.usyd.edu.au (A.T. Harris). improvement.

0009-2509/$ - see front matter © 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2008.10.021
188 N.H. Florin, A.T. Harris / Chemical Engineering Science 64 (2009) 187 -- 191

In this paper we investigate the reaction characteristics of nano-

sized sorbent particles for application in a CO2 capture-and-release
process, in particular the unique carbonation characteristics of pure
CaO derived from nano-sized CaCO3 .While acknowledging that this
application will likely involve a fluidised bed of some description,
and thus some form of pelletisation or particle support will be re-
quired for near term applications, there are techniques available to
satisfactorily fluidise primary nanoparticles (e.g., Harris, 2008) which
are likely to become commercially viable with further research.
We also critique the novel pore-skeleton model proposed by
Manovic and Anthony (2008). This critical review is given in the
context of our experimental results and leads to the development
of an alternative model—which does not necessitate differentiation
between the mass transfer phenomena responsible for the rate of
deactivation, and the mass transfer phenomena responsible for pro-
ducing a sorbent morphology that defines the extent of the residual
conversion.

2. Experimental methods and sorbent characterisation

All experiments used nano-sized CaCO3 obtained from

Nanomaterials Technology Pty Ltd (Singapore). High resolution
scanning electron microscopy (SEM) was used to identify morpho-
logical properties and confirm the mean particle size reported by
the manufacturer. Nanoparticles were suspended in resin and SEM
images of the fresh fracture surfaces were obtained. The image in
Fig. 1a shows individual CaCO3 nanoparticles embedded within the
resin material; the apparent size of the individual particles is con-
sistent with the manufacturer's measurement of 40 nm. The image
in Fig. 1b shows the fresh fracture surface of a sample which has
undergone 100 carbonation/calcination cycles and confirms the ten-
dency of the CaCO3 particle to agglomerate; this is consistent with
the mechanistic model of the interrelated mass transfer processes
described in detail below.
Surface area and pore volume analysis was inferred from N2 ad-
sorption measurements. Fig. 2 shows the pore size distributions for
nano-sized CaCO3 and CaO prepared at 700 ◦ C under a N2 atmo-
Fig. 1. (a) SEM image of a fresh fracture surface for virgin CaCO3 material showing
sphere in a tube furnace. The pore size distributions were deter- a large collection of particles with a mean size of 40 nm; (b) SEM image of a fresh
mined from desorption data using the BJH method. The pore volume facture surface for CaCO3 material after undergoing 100 carbonation/calcination
of the nano-sized CaCO3 particles (circles) is predominantly in the cycles.
range 10–150 nm. The pore size distribution shown for the calcine
(squares) shows an increase in pore volume associated with pores
< 10 nm as well as larger pores with diameters greater than about 3. Results and discussion
30 nm. The total increase in the pore volume associated with the
conversion of CaCO3 to CaO based on the N2 adsorption data was The reactivity of CaO derived from nano-sized CaCO3 was first
0.03 cm3 /g. It is likely that the increase in very small pores is due to investigated by Barker (1974b) more than three decades ago in
CO2 evolution and the increase in pore volume associated with the the context of a high-temperature energy storage scheme. Barker
larger pores is due to restructuring of the nanoparticle agglomerate. demonstrated 93% conversion of CaO to CaCO3 , sustained through
All reactivity experiments used a thermogravimetric analyser (TA 30 capture-and-release cycles, with the carbonation reaction span-
Instruments Q500). In a typical experiment ∼5 mg of CaCO3 nanopar- ning 24 h. While this rate of reaction is not practicable for CO2
ticles were evenly distributed within a platinum sample holder and capture-and-release applications, the result demonstrates the re-
heated at 20 ◦ C/min to a calcination temperature of 850 ◦ C, under N2 . versibility of Eq. (1). We repeated Barker's experiment, choosing
The sample was held at 850 ◦ C for 10 min before the temperature was reaction temperatures more representative of the proposed CO2
reduced to 650 ◦ C for carbonation, which was initiated by switch- capture process. Fig. 3 shows the reactivity of CaO derived from
ing the gas composition to 15% CO2 in N2 . The purge gas flowrate nano-sized CaCO3 particles through five cycles, with the carbona-
was held constant at 80 mL/min. Samples were cycled through up tion reaction spanning 24 h. The weight gain corresponding to the
to 100 calcination/carbonation reactions. The use of a TGA is advan- CO2 capture reaction, up to 94 wt%, reflects the complete conver-
tageous for reactivity studies for several reasons, in particular the sion of CaO to CaCO3 . The shortfall from 100 wt% is due to the
ease of switching between the calcination and carbonation regimes presence of physically bound water, released from the sample dur-
across multiple cycles. However, we note that calcination under N2 , ing the first calcination step. This was confirmed by recording the
regardless of the temperature, is not truly representative of a CO2 mass spectra of the evolved gas species during decomposition. Com-
capture process, because in a real system higher temperatures and plete conversion is achieved through five carbonation and calcina-
CO2 partial pressures are required to produce a pure stream of CO2 . tions cycles, confirming the reversibility of Eq. (1), in agreement with
Similarly a TGA system is not able to simulate the particle attrition Barker's earlier result. Thus, it may be concluded that there is no
phenomena which occurs within fluidised beds. morphological impediment to complete conversion of CaO derived
 N.H. Florin, A.T. Harris / Chemical Engineering Science 64 (2009) 187 -- 191 189

0.35
100

0.3

90
Desorption Dv(logd) (cm3/g)

0.25

0.2
80

Weight (%)
0.15

70
0.1
CaO
CaCO3
0.05 60

0
1 10 100 1000 50
Pore diameter (nm)
0 1000 2000 3000 4000 5000 6000
Fig. 2. Pore size distribution determined for nano-sized CaCO3 and CaO derived
Time (min)
from nano-sized CaCO3 . CaO was calcined in a tube furnace at 700 ◦ C under N2 .

0.9
100
0.8

0.7 Cycle 1
90
0.6

0.5
80
X (-)
Weight (%)

0.4

0.3 Cycle 100

70

0.2

60 0.1

0
0 2 4 6 8 10 12 14 16 18 20
50 Time (min)
0 20 40 60 80 100 120 140
Time (hr) Fig. 4. (a) Weight change associated with 100 CO2 capture-and-release cycles; (b)
Conversion (X) in the first carbonation reaction cycle.
Fig. 3. Weight change associated with CO2 capture-and-release; 24 h carbonation at
650 ◦ C with 15% CO2 in N2 and calcination at 850 ◦ C in N2 for 10 min.

from nano-sized CaCO3 particles when sufficient time is permitted the rapid, kinetically-controlled, reaction phase. The transition from
for carbonation. the rapid to slow reaction phase commences after only ∼2 min. The
When the carbonation reaction proceeded for only 20 min we ob- high residual conversion capacity of about 20% is higher than previ-
served the characteristic decay in conversion with cycle number and ously reported (e.g., Lysikov et al., 2007; Manovic and Anthony, 2008)
a decay asymptote was approached with a residual conversion of demonstrating the potential in using nano-sized sorbent materials.
∼20% (Fig. 4). Fig. 4a shows the conversion through 100 CO2 capture- The superior conversion capacity is likely related to the unique mor-
and release cycles, represented in terms of wt%. The conversion (X) phology which evolves through multiple CO2 capture-and-release
profiles for cycle 1 and 100 are shown in Fig. 4b. In cycle 1 a max- cycles using nano-sized particles (Fig. 1b). The exceptionally high
imum conversion of 89% was achieved after 20 min compared with surface area of the nano-sized sorbent particles means that there
only 21% following the 100th cycle. The higher conversion in the first are many points available for mass transfer, which results in a resid-
cycle is due substantially to the extent of conversion achieved during ual morphology with a higher conversion capacity. The particle size
190 N.H. Florin, A.T. Harris / Chemical Engineering Science 64 (2009) 187 -- 191

100 0.9

no presintering
0.8

90
0.7

0.6
80
presintered at 900 °C, 12 hr
Weight (%)

0.5

X (–)
0.4
70

0.3

60 0.2

0.1 presintered at 1100 °C, 12 hr

50
0
0 1500 3000 4500 6000 7500
0 10 20 30 40 50 60 70 80 90 100
Time (min)
Cycle (–)
Fig. 5. Weight change associated with 100 CO2 capture-and-release cycles; the 51st
CO2 capture cycle spanned 24 h. Fig. 6. The effect of thermal pre-sintering on the rate of approach to the decay
asymptote.

distribution and particle shape are also likely to influence the resid- N2 for up to 24 h. Manovic and Anthony described this behaviour
ual morphology. as “self-reactivation” and explained it in terms of a pore-skeleton
The conversion profile corresponding to the 100th cycle indicates model which differentiates between mass transfer associated with
a delay in the start of the CO2 capture reaction, a greatly diminished the carbonation/calcinations reactions, and ion diffusion in the CaO
kinetically-controlled reaction phase and a reduced rate of reaction crystal structure with exposure to relatively high temperatures. Ac-
during the diffusion-controlled phase. We attribute the delay in the cordingly, the former mechanism is responsible for morphological
commencement of the carbonation reaction for the 100th cycle to changes, i.e., the reduction in small-pore volume through multiple
changes in the sorbent morphology which impede gas–solid contact CO2 capture-and-release cycles; and the ion diffusion in the unre-
during the initiation of the rapid-reaction phase, although it is not acted CaO affects the crystal structure resulting in a hard “internal-
possible to be sure that the delay is not, in part, attributable to skeleton”. During thermal treatment the ion diffusion mechanism
experimental variability. We have observed minor variation in the is dominant and this stabilises the internal-skeleton, which pro-
initial rate of carbonation for a range of different CaO-based sorbents tects against severe reduction in the sorbent reactive surface area—a
at different cycles numbers. proposition consistent with improvements in sorbent performance
The rate of approach to the decay asymptote can be delayed by that they observed through 30 reaction cycles.
extending the duration of the carbonation reaction. This is demon- In our experiments, pre-sintering at 900 ◦ C resulted in a sig-
strated in Fig. 5, which shows the effect of extending the carbona- nificant drop in conversion in the first CO2 -capture reaction fol-
tion reaction in the 51st reaction cycle. Carbonation was extended lowed by some enhancement through ∼10 cycles. Similar to the
from 20 min to 24 h resulting in an increase in the conversion from findings of Manovic and Anthony (2008), the rate of decay was de-
20% in the 50th cycle to 35% in the 51st cycle. Although the conver- layed, but eventually we observed the approach to the same residual
sion in the subsequent cycles (20 min) gradually receded back down conversion of 20% after 70–80 cycles. For pre-treatment at 1100 ◦ C
to the decay asymptote, the extended carbonation cycle had a re- the heavily sintered sorbent showed very low conversion in the
generative effect on the sorbent conversion. This can be explained first CO2 -capture reaction, however the reactivity was regenerated
by an increase in the pore volume and reactive surface area; more through ∼20 cycles and the asymptote reached after ∼30 cycles.
CO2 was captured in the 51st cycle than was released in the pre- We contend that the decay asymptote identifies an equilibrium
ceding calcination step, and thus additional pore volume and re- between the loss of surface area and pore volume due to sinter-
active surface area is generated when the larger amount of CO2 is ing under the calcination conditions, and an oppositional increase in
released. surface area and pore volume due to mass transfer associated with
We also investigated the influence of thermal sintering on the solid-state diffusion during carbonation, which evolves with the sub-
rate of approach to the decay asymptote by pre-sintering CaCO3 sequent release of the CO2 . The latter mechanism, while occurring
nano-particles at 900 and 1100 ◦ C for 12 h in air. Fig. 6 compares the at a lower temperature, is enhanced by the presence of CO2 (which
conversion trajectories of the pre-sintered sorbent particles, through is known to enhance the mass transfer mechanism). This model can
up to 100 CO2 capture-and-release cycles. Pre-sintering resulted in explain the regenerative behaviour but does not necessitate making
quite interesting behaviour whereby the extent of conversion in- a distinction between an external skeleton and an internal skele-
creased through a small number of the initial reaction cycles. This ton. Furthermore, we assume that the mass transfer mechanisms
behaviour was observed by Lysikov et al. (2007) and Manovic and associated with thermal sintering and the solid-state ion diffusion
Anthony (2008) with natural limestones, which they exposed to during the diffusion-controlled carbonation reaction are analogous.
thermal pre-treatment at temperatures from 800 to 1200 ◦ C under This qualitative model is supported both by our results in a TGA
 N.H. Florin, A.T. Harris / Chemical Engineering Science 64 (2009) 187 -- 191 191

and those of Fennell et al. (2007) using a fluidised bed, which show References
how extending the time for which sorbent particles are exposed to
both the carbonation and calcination conditions effects the approach Abanades, J.C., Alvarez, D., 2003. Conversion limits in the reaction of CO2 with lime.
Energy and Fuels 17, 308–315.
to the decay asymptote. This behaviour—which we have observed Abanades, J.C., Grasa, G., Alonso, M., Rodriguez, N., Anthony, E.J., Romeo, L.M., 2007.
with pure CaO derived from nano-sized particles—is particularly pro- Cost structure of a postcombustion CO2 capture system using CaO. Environmental
nounced due to the susceptibility of these high surface area par- Science and Technology 41, 5523–5527.
Alvarez, D., Abanades, J.C., 2005. Determination of the critical product layer thickness
ticles to sintering with their numerous contact points to facilitate in the reaction of CaO and CO2 . Industrial and Engineering Chemistry Research
mass transfer (Fig. 1b). The very ordered nature of the agglomera- 44, 5608–5615.
tion shown in Fig. 1b is quite striking and it appears that the high Barker, R., 1974a. The reversibility of the reaction CaCO3 →CaO+CO2 . Journal of
Applied Chemistry and Biotechnology 23, 733–742.
surface area of the nano-sized particles and the narrow particle size
Barker, R., 1974b. The reactivity of calcium oxide towards carbon dioxide and its
distribution, which is apparent from the SEM image of the fresh par- use for energy storage. Journal of Applied Chemistry and Biotechnology 24,
ticles (Fig. 1a), may facilitate the mass transfer process resulting in 221–227.
Bhatia, S.K., Perlmutter, D.D., 1983. Effect of the product layer on the kinetics of the
the residual morphology. In the more realistic case when the sorbent
CO2 -lime reaction. American Institute of Chemical Engineers 29, 79–86.
particles are supported, the agglomeration behaviour would be less Borgwardt, R.H., 1989. Calcium oxide sintering in atmospheres containing water and
significant and more controllable. This work points to the advantage carbon dioxide. Industrial and Engineering Chemistry Research 28, 493–500.
of tailoring sorbent supports. Further implicit in this understand- Curran, G.P., Fink, C.E., Gorin, E., 1966. CO2 acceptor gasification process. In: Schora,
F.C. (Ed.), Fuel Gasification. American Chemical Society, Washington, DC, p. 141.
ing of the decay behaviour is recognition that any factor which may Feng, B., Liu, W., Li, X., An, H., 2006. Overcoming the problem of loss in capacity of
influence solid-state mass transfer will affect the rate of approach calcium oxide in CO2 capture. Energy and Fuels 20, 2417–2420.
and establishment of the decay asymptote, including: the presence Fennell, P.S., Pacciani, R., Dennis, J.S., Davidson, J.F., Hayhurst, A.N., 2007. The effects
of repeated cycles of calcination and carbonation on a variety of different
of impurities in natural limestones, and the effect of H2 O and CO2 limestones, as measured in a hot fluidised bed of sand. Energy and Fuels 21,
(Borgwardt, 1989), temperature, and time of exposure to both calci- 2072–2081.
nation and carbonation conditions. Florin, N.H., Harris, A.T., 2008a. Review of enhanced hydrogen production from
biomass with in situ carbon dioxide capture using calcium dioxide sorbents.
Chemical Engineering Science 63, 287–316.
Florin, N.H., Harris, A.T., 2008b. Development of an experimental protocol for
4. Conclusions screening CaO sorbents for CO2 capture in biomass gasifiers. Energy and Fuels
22, 2734–2742.
Han, C., Harrison, D., 1994. Simultaneous shift reaction and carbon dioxide separation
The carbonation characteristics of pure CaO derived from nano- for the direct production of hydrogen. Chemical Engineering Science 49,
sized CaCO3 were investigated as part of a multi-cycle performance 5875–5883.
study. The decay asymptote, which is established through multi- Harris, A.T., 2008. On the vibration assisted fluidisation of silica nanoparticles.
International Journal of Nanotechnology 5, 179–194.
ple capture-and-release cycles, was explained through a qualitative Manovic, V., Anthony, E.J., 2007. Steam reactivation of spent CaO-based sorbent
model where the rate of approach and residual conversion defined for multiple CO2 capture cycles. Environmental Science and Technology 41,
by the decay asymptote represents the establishment of an equilib- 1420–1425.
Manovic, V., Anthony, E.J., 2008. Thermal activation of CaO-based sorbent and
rium between the pore volume and surface area loss during thermal self-reactivation during CO2 capture looping cycles. Environmental Science and
sintering; and the pore volume and surface area regeneration as a Technology 42, 4170–4174.
consequence of a solid-state diffusion mechanism, and the subse- Li, Z.-S., Cai, N.-S., Haung, Y.-Y., 2006. Effect of preparation temperature on the cyclic
CO2 capture and multiple carbonation–calcination cycles for a new Ca-based
quent release of CO2 in the next calcination cycle. This qualitative
CO2 sorbent. Industrial and Engineering Chemistry Research 45, 1911–1917.
explanation is valid for all CaO derived CO2 sorbents. Lysikov, A.I., Salanov, A.N., Okunev, A.G., 2007. Change of CO2 carrying capacity of CaO
in isothermal recarbonation–decomposition cycles. Industrial and Engineering
Chemistry Research 46, 4633–4638.
Silaban, A., Harrison, D.P., 1995. High temperature capture of carbon dioxide:
Acknowledgement characteristics of the reversible reaction between CaO(s) and CO2 (g). Chemical
Engineering Communications 137, 177–190.
The authors are grateful for the financial support of the Australian Zhen-shan, L., Ning-sheng, C., Croiset, E., 2008. Process Analysis of CO2 Capture from
Flue Gas Using Carbonation/Calcination Cycles. AICHE J 54 (7), 1912–1925.
Research Council through DP0666488.