CO2 Uptake of Carbonation-Cured Cement Blended with Ground Volcanic Ash

Accelerated carbonation curing (ACC) as well as partial replacement of cement with natural minerals are examples of many previous approaches, which aimed to produce cementitious products with better properties and environmental amicabilities. In this regard, the present study investigates CO2 uptake of carbonation-cured cement blended with ground Saudi Arabian volcanic ash (VA). Paste samples with cement replacement of 20%, 30%, 40%, and 50% by mass were prepared and carbonation-cured after initial curing of 24 h. A compressive strength test, X-ray diffractometry (XRD), and thermogravimetry were performed. Although pozzolanic reaction of VA hardly occurred, unlike other pozzolana in blended cement, the results revealed that incorporation of VA as a supplementary cementitious material significantly enhanced the compressive strength and diffusion of CO2 in the matrix. This increased the CO2 uptake capacity of cement, reducing the net CO2 emission upon carbonation curing.


Introduction
Increasing global CO 2 emission has been posing threats to the Earth's atmospheric environments. It was reported that the annual CO 2 concentration at the Earth's surface in 2017 reached 405 ppm, which is approximately quadruple to that in the early 1960s [1]. The cement industry and the production of Portland cement (PC) embody 5-8% of global CO 2 emissions, which is one of the major greenhouse gases contributing to global warming [2,3]. Numerous attempts have been identified as potential means to reduce the CO 2 associated with the production of cement, such as (1) alternative clinkers, including reactive belite-rich PC and belite-ye'elimite-ferrite cement, which have less limestone input and lower sintering temperature [4,5], alkali-activated binders derived from alkaline activation of aluminosilicate precursors sourced from industrial by-products or natural pozzolans [6][7][8][9], and phosphate-derived hydraulic binders, including both calcium and magnesium phosphate cement [10]; (2) PC clinker substitution with supplementary cementitious materials (SCMs), which are industrially sourced or naturally abundant [11,12]; and (3) accelerated carbonation curing (ACC), in which CO 2 generated from various industries are collected and are utilized [13][14][15]. As the removal of emitted CO 2 is a serious concern, numerous attempts have been made to stabilize gaseous CO 2 by carbon sequestration under the theme of reusing CO 2 with ACC technologies [16,17].
ACC technology for concrete production has received significant attention because it does not only facilitate storing a large amount of CO 2 , but also has a general tendency to improve the mechanical  3 3.27 SO 3 2.13 L.O.I. 1 3.27 1 Loss-on-ignition.
Materials 2018, 11, x FOR PEER REVIEW 3 of 12 Table 1. The VA obtained from Harrat Rahat, western region of Saudi Arabia, southern part of Al-Madina city, was used to partially substitute PC to achieve binary binding materials. The VA was ball-milled for an hour to achieve the particle size distribution, as shown in Figure 1. Note that the median powder value of ball-milled VA was 19.84 μm. The chemical compositions of the VA used in this study can be found in Table 2. The mineralogical constituent of the VA mostly consisted of plagioclase feldspar (andesine (Al1.488Ca0.491Na0.499O8Si2.506, #PDF 01-079-1148)) with an amorphous hump in the region of 20-30° 2θ ( Figure 2). It was calculated from the quantitative analysis of X-ray diffractometry (XRD, Malvern Panalytical, Malvern, UK) pattern that 65.1% of VA is composed of the amorphous glassy phase.  1 3.27 1 Loss-on-ignition.   The samples were made with a constant water-to-powder ratio of 0.4 (mass ratio). PC was substituted with VA by 20%, 30%, 40%, and 50% by mass of the cement. The mixture proportion and designated ID used in this study are summarized in Table 3. Powder (PC and VA) and tap water were mixed for 5 min under a laboratory condition (25 • C) and poured into cubical molds with dimensions of 50 × 50 × 50 mm thereafter. The samples were kept in a sealed condition for 24 h using plastic wraps before the samples were placed in an accelerated carbonation chamber. It should be noted that the samples for chemical analyses were crushed to pass through a 3 mm sieve before carbonation to achieve uniform carbonation. The samples were carbonation-cured at a 5% atmospheric CO 2 concentration, 25 • C, and 60% R.H., until 28 days. Similarly, reference samples were cured for an identical period, in a sealed-condition using plastic wraps (i.e., free from carbonation), and placed in a thermo-hygrostat chamber (25 • C).  The samples were made with a constant water-to-powder ratio of 0.4 (mass ratio). PC was substituted with VA by 20%, 30%, 40%, and 50% by mass of the cement. The mixture proportion and designated ID used in this study are summarized in Table 3. Powder (PC and VA) and tap water were mixed for 5 min under a laboratory condition (25 °C) and poured into cubical molds with dimensions of 50 × 50 × 50 mm thereafter. The samples were kept in a sealed condition for 24 h using plastic wraps before the samples were placed in an accelerated carbonation chamber. It should be noted that the samples for chemical analyses were crushed to pass through a 3 mm sieve before carbonation to achieve uniform carbonation. The samples were carbonation-cured at a 5% atmospheric CO2 concentration, 25 °C, and 60% R.H., until 28 days. Similarly, reference samples were cured for an identical period, in a sealed-condition using plastic wraps (i.e., free from carbonation), and placed in a thermohygrostat chamber (25 °C).

Test Methods
The compressive strength and CO2 uptake of carbonation-cured PC blended with VA were evaluated by means of compressive strength tests, XRD, and thermogravimetry/derivative thermogravimetry (TG/DTG). The samples were immersed in acetone at 7 and 28 days to arrest further evolution of   1 The following denotation was used to identify samples: "C" and "V" indicate Portland cement (PC) and volcanic ash (VA)., respectively, while the numbers following "C" and "V" indicate their mass ratio. 2 Powder denotes the summation of the amount of cement and volcanic ash.

Test Methods
The compressive strength and CO 2 uptake of carbonation-cured PC blended with VA were evaluated by means of compressive strength tests, XRD, and thermogravimetry/derivative thermogravimetry (TG/DTG). The samples were immersed in acetone at 7 and 28 days to arrest further evolution of hydrates. XRD and TG were conducted for the samples obtained at 7 and 28 days of curing, while other tests were performed at 28 days only.
The compressive strength was measured using a 3000 kN compression testing machine with a constant loading speed of 0.02 mm/s. The compressive strength was averaged from three replicas. An empyrean instrument with CuKα radiation at 40 kV and 30 mA was adapted for high-resolution XRD analysis. Samples were scanned in a range of 5-65 • 2θ at a rate of 1 • /min. The TG/DTG was performed using a TGA/DSC1/1600 LF instrument manufactured by Mettler-toledo (Columbus, OH, USA). The mass change was recorded under a constant heating rate of 10 • C/min in N 2 .

Compressive Strength
The compressive strength of VA-blended PC is depicted in Figure 3. The compressive strength of uncarbonated samples varied depending on the amount of substituted VA. An increase in the compressive strength was most noticeable for the samples with 20% substitution with VA. Meanwhile, the samples incorporating more than 30% of VA showed compressive strength lower than that of the control specimen (i.e., C10).
compressive strength was most noticeable for the samples with 20% substitution with VA. Meanwhile, the samples incorporating more than 30% of VA showed compressive strength lower than that of the control specimen (i.e., C10).
Upon carbonation curing, most samples showed a significant increase in the compressive strength. Compared to uncarbonated samples, carbonation lead to an increase in strength that was highest for C8V2, and the strength decreased as the content of VA in the sample increased. In particular, carbonation curing of the samples blended with 50% VA was observed to reduce its strength. Overall, the compressive strength of carbonation-cured samples exhibited a similar trend with that of uncarbonated samples, showing a decrease as the amount of VA increased. It should be mentioned that the strength increases in the normally cured C8V2 and C7V3 samples compared to the C10 sample is possibly attributed to the filler effect of VA [40]. Also note that the strength increase in the carbonation-cured samples is mainly associated with the increased hydration reaction of PC promoted by carbonation curing. However, the decrease in the compressive strength of these samples is possibly attributed to the dilution effect, which occurs when the incorporated mineral admixture significantly influences the water-to-cement ratio [41]. Therefore, a long term compressive strength test is required to assess the effect of the pozzolanic reaction degree on the compressive strength.

XRD
The XRD patterns of VA-blended PC are shown in Figure 4. The uncarbonated samples showed peaks corresponding to the presence of portlandite (Ca(OH)2, PDF #00-044-1481), C-S-H (Ca1.5SiO3.5H2O, PDF #00-033-0306), and larnite (Ca2SiO4, PDF #01-077-2010). The presence of ettringite (Ca6Al2(SO4)3(OH)12·26H2O, #PDF 00-013-0350) was commonly observed in the uncarbonated samples with several peaks attributed to andesine, which correspond to the unreacted fractions of raw volcanic ash. The formation of AFm phases, i.e., hemicarbonate (Ca4Al2O7(CO2)0.5·12H2O, PDF #00-036-0129) and monocarbonate (Ca4Al2O7(CO2)·11H2O, PDF #00-036-0377)), was observed in the VA-blended samples. The formation of these phases is possibly attributed to the use of partially carbonated VA given that Ca-, Mg-, and Fe-carbonates can be precipitated in natural basaltic VA due to weathering carbonation [42]. It should be noted that the peak associated with portlandite was observed even in the samples blended with 50% VA at 28 days. Note that portlandite is found to be fully consumed in Upon carbonation curing, most samples showed a significant increase in the compressive strength. Compared to uncarbonated samples, carbonation lead to an increase in strength that was highest for C8V2, and the strength decreased as the content of VA in the sample increased. In particular, carbonation curing of the samples blended with 50% VA was observed to reduce its strength. Overall, the compressive strength of carbonation-cured samples exhibited a similar trend with that of uncarbonated samples, showing a decrease as the amount of VA increased. It should be mentioned that the strength increases in the normally cured C8V2 and C7V3 samples compared to the C10 sample is possibly attributed to the filler effect of VA [40]. Also note that the strength increase in the carbonation-cured samples is mainly associated with the increased hydration reaction of PC promoted by carbonation curing. However, the decrease in the compressive strength of these samples is possibly attributed to the dilution effect, which occurs when the incorporated mineral admixture significantly influences the water-to-cement ratio [41]. Therefore, a long term compressive strength test is required to assess the effect of the pozzolanic reaction degree on the compressive strength.
Note that portlandite is found to be fully consumed in PC blended with 20% fly ash or silica fume according to the thermodynamic calculation [11], indicating that the pozzolanic reaction of VA was relatively low. addition, peaks corresponding to the carbonation products of VA can be identified from the presence of siderite (FeCO3, PDF #00-029-0696) and dolomite (CaMg(CO3)2, #PDF 00-036-0426). At 7 days of carbonation curing, the C10 and C8V2 samples showed a peak corresponding to portlandite. The intensity of the peak corresponding to portlandite was found to decrease as the amount of VA substitution increased. This shows that carbonation had fully occurred in the samples blended with 50% VA. At 28 days of carbonation curing, the presence of ettringite and portlandite was not observed in the XRD patterns of VA-blended samples except for C10. In addition, peaks corresponding to the presence of hemicarbonate and monocarbonate disappeared at 28 days. The intensity of the peaks corresponding to andesine was persistent at 28 days, indicating that this phase contained in the raw VA remains unaffected by carbonation.   The carbonation-cured samples showed peaks associated with the presence of calcium carbonate polymorphs, such as calcite (CaCO 3 , PDF #01-072-1937) and vaterite (CaCO 3 , #PDF 00-001-1033). In addition, peaks corresponding to the carbonation products of VA can be identified from the presence of siderite (FeCO 3 , PDF #00-029-0696) and dolomite (CaMg(CO 3 ) 2 , #PDF 00-036-0426). At 7 days of carbonation curing, the C10 and C8V2 samples showed a peak corresponding to portlandite. The intensity of the peak corresponding to portlandite was found to decrease as the amount of VA substitution increased. This shows that carbonation had fully occurred in the samples blended with 50% VA. At 28 days of carbonation curing, the presence of ettringite and portlandite was not observed in the XRD patterns of VA-blended samples except for C10. In addition, peaks corresponding to the presence of hemicarbonate and monocarbonate disappeared at 28 days. The intensity of the peaks corresponding to andesine was persistent at 28 days, indicating that this phase contained in the raw VA remains unaffected by carbonation.

TG/DTG
The TG/DTG curves of VA-blended PC are shown in Figure 5. The uncarbonated samples showed a weight loss peak at 80-100 • C and a shoulder at 120-140 • C due to the dehydration of chemically bound water from C-S-H [43] and ettringite [44]. The VA-blended samples showed a lesser weight loss at this region, implying that the amount of hydration products (C-S-H and ettringite) was much less. The weight loss at 400-450 • C is attributed to the dehydroxylation of portlandite [45], and was observed in all samples at seven and 28 days regardless of the amount of VA substitution. This is in fair agreement with the XRD results, showing that the pozzolanic reactivity of the VA was low. The weight loss at 500-720 • C can be attributed to the decarbonation of calcium carbonate [46] and unreacted clinker [47], and its intensity reduced with the amount of cement replacement.

TG/DTG
The TG/DTG curves of VA-blended PC are shown in Figure 5. The uncarbonated samples showed a weight loss peak at 80-100 °C and a shoulder at 120-140 °C due to the dehydration of chemically bound water from C-S-H [43] and ettringite [44]. The VA-blended samples showed a lesser weight loss at this region, implying that the amount of hydration products (C-S-H and ettringite) was much less. The weight loss at 400-450 °C is attributed to the dehydroxylation of portlandite [45], and was observed in all samples at seven and 28 days regardless of the amount of VA substitution. This is in fair agreement with the XRD results, showing that the pozzolanic reactivity of the VA was low. The weight loss at 500-720 °C can be attributed to the decarbonation of calcium carbonate [46] and unreacted clinker [47], and its intensity reduced with the amount of cement replacement.
The carbonation-cured samples showed a weight loss at regions corresponding to the dehydration of chemically bound water from hydration products [43,44] and dehydroxylation of portlandite [45], which is similar with the uncarbonated samples, but with reduced intensity. Meanwhile, a weight loss at 500-720 °C was observed in all samples at seven and 28 days, which is attributed to the decarbonation of calcium carbonate [46], mainly calcite, as observed by the XRD analysis. It is worth noting the carbonation-cured reference sample (i.e., without VA) showed a weight loss due to dehydroxylation of portlandite, while portlandite was found to have been fully consumed in the VA-blended samples at 28 days of carbonation curing due to their lower PC content, resulting in a rapid carbonation. The relative weight loss attributed to dehydration of C-S-H and ettringite, dehydroxylation of portlandite, and decarbonation of calcium carbonate of the samples at 28 days as calculated by TG is summarized in Table 4.  The CO2 uptake capacity per unit weight of cement in the total blend at 28 days of carbonation curing, relative to the unit weight of PC, is summarized in Table 5, which is calculated by the following Equation (1) [14]: The carbonation-cured samples showed a weight loss at regions corresponding to the dehydration of chemically bound water from hydration products [43,44] and dehydroxylation of portlandite [45], which is similar with the uncarbonated samples, but with reduced intensity. Meanwhile, a weight loss at 500-720 • C was observed in all samples at seven and 28 days, which is attributed to the decarbonation of calcium carbonate [46], mainly calcite, as observed by the XRD analysis. It is worth noting the carbonation-cured reference sample (i.e., without VA) showed a weight loss due to dehydroxylation of portlandite, while portlandite was found to have been fully consumed in the VA-blended samples at 28 days of carbonation curing due to their lower PC content, resulting in a rapid carbonation. The relative weight loss attributed to dehydration of C-S-H and ettringite, dehydroxylation of portlandite, and decarbonation of calcium carbonate of the samples at 28 days as calculated by TG is summarized in Table 4. Table 4. Relative weight loss (%) attributed to the decomposition of reaction products at 28 days.

Curing Condition
Specimen ID The CO 2 uptake capacity per unit weight of cement in the total blend at 28 days of carbonation curing, relative to the unit weight of PC, is summarized in Table 5, which is calculated by the following Equation (1) [14]: CO 2 uptake capacity (%) = M CO2 /M PC × 100 (1) Here, M CO2 is the mass of CO 2 sequestrated by PC during the carbonation curing, and is calculated by computing the difference between the weight loss at 500-720 • C of carbonation-cured and uncarbonated samples. M PC is the unit mass ratio of the PC used (i.e., 0.71, 0.57, 0.5, 0.43, and 0.36 for C10, C8V2, C7V3, C6V4, and C5V5 samples, respectively). The CO 2 uptake capacity of the reference sample was 12.3%, showing a similar level of CO 2 uptake (13.5%) observed in a previous study [14]. The relationship between the CO 2 uptake capacity of VA-blended PC and VA substitution is correlated in Figure 6. The CO 2 uptake capacity of VA-blended PC was found to generally increase with the amount of VA substitution.

Discussion
The obtained results can be discussed in terms of two main roles of VA blended in PC, namely hydration and carbonation. AFm phases, such as hemicarbonate and monocarbonate, were observed in the VA-blended samples, which can be attributed to the incorporation of SCM rich in Al, as typically found in fly ash-blended cement [11]. The DTG results showed that a significant amount of portlandite remained unconsumed in the samples blended with a high amount of VA (i.e., 50%), indicating that the dilution effect occurred in the samples and that 28 days was insufficient for the VA to exhibit pozzolanic reaction. This is also supported by the fact that the weight loss due to the dehydration of hydration products in the DTG curve was relatively lower in these samples.
Considering the fact that the CO2 uptake capacity of cementitious materials heavily depends on their Ca content, which effectively facilitates as a CO2 sink by precipitation of CaCO3 polymorphs [13], one may assume that incorporation of SCMs with a low content of Ca may lead to a lower CO2 uptake in the cementitious material. The amount of CO2 in the samples after 28 days of carbonation curing ranged

Discussion
The obtained results can be discussed in terms of two main roles of VA blended in PC, namely hydration and carbonation. AFm phases, such as hemicarbonate and monocarbonate, were observed in the VA-blended samples, which can be attributed to the incorporation of SCM rich in Al, as typically found in fly ash-blended cement [11]. The DTG results showed that a significant amount of portlandite remained unconsumed in the samples blended with a high amount of VA (i.e., 50%), indicating that the dilution effect occurred in the samples and that 28 days was insufficient for the VA to exhibit pozzolanic reaction. This is also supported by the fact that the weight loss due to the dehydration of hydration products in the DTG curve was relatively lower in these samples.
Considering the fact that the CO 2 uptake capacity of cementitious materials heavily depends on their Ca content, which effectively facilitates as a CO 2 sink by precipitation of CaCO 3 polymorphs [13], one may assume that incorporation of SCMs with a low content of Ca may lead to a lower CO 2 uptake in the cementitious material. The amount of CO 2 in the samples after 28 days of carbonation curing ranged from 6.9-11.0%, showing a large deviation. However, the CO 2 uptake capacity relative to the unit mass ratio of PC approximately showed a linear increase with the amount of VA substitution, meaning that the CO 2 uptake efficiency can be vastly improved by VA substitution. Furthermore, the relative amount of stored CO 2 (fourth column of Table 5), without consideration of unit cement in total blend, also demonstrated that incorporated VA increases CO 2 diffusivity even in samples with less PC content. This may be attributed to the presence of CaO, Fe 2 O 3 , and MgO components in VA, which amounts to 23.87 wt.%, precipitating carbonates in addition to the carbonation of PC. The traces of these carbonates were observed in the XRD patterns of carbonated samples, which are overlapped with the peaks corresponding to calcite. However, the presence of various carbonates is supported by the fact that the peak intensities corresponding to these carbonates increased as the amount of incorporated VA increased. VA incorporation in PC can therefore be described to promote the diffusion of CO 2 into a cementitious matrix, thereby increasing the rate of carbonation, as suggested by the higher compressive strength being reached by VA-blended samples (Figure 3). In light of the industrial practice of carbonation curing for concrete production, VA incorporation is effective to lower the overall CO 2 footprint by reducing the use of PC and enhancing the CO 2 uptake efficiency of the cement used. In particular, these advantageous aspects are considered to be very useful when VA blended PC is used in the production of precast members.

Conclusions
The present study investigated the CO 2 uptake capacity of carbonation-cured cement blended with ground VA. Samples at various blend ratios were manufactured and carbonation-cured for 28 days. The results obtained in this study showed that carbonation curing of VA incorporated cement exhibited a significant enhancement in the compressive strength, thus enabling a reduction in the CO 2 footprint. The following conclusions were drawn from this work: (1) A significant enhancement in the compressive strength by carbonation curing was observed in the VA-blended cement except for the samples blended with more than 40% VA, which were found to be fully carbonated.
(2) The CO 2 uptake capacity of VA-blended samples relative to the unit weight of PC showed a linear increase as the amount of VA substitution increased due to the partial carbonation of VA.
(3) VA incorporation of cement can significantly reduce the CO 2 footprint throughout its production and manufacturing, and is capable of improving CO 2 diffusivity.
(4) The utilization of VA as a SCM has a slight effect in improving the compressive strength of PC, although the dilution effect somewhat hindered the strength development of the samples blended with VA. Nevertheless, combining VA substitution with ACC resulted in a significant improvement in the compressive strength and CO 2 uptake capacity.