An ATR-FTIR Study on the Effect of Molecular Structural Variations on the CO2 Absorption Characteristics of Heterocyclic Amines, Part II

This paper reports on an ATR-FTIR spectroscopic investigation of the CO2 absorption characteristics of a series of heterocyclic diamines: hexahydropyrimidine (HHPY), 2-methyl and 2,2-dimethylhexahydropyrimidine (MHHPY and DMHHPY), hexahydropyridazine (HHPZ), piperazine (PZ) and 2,5- and 2,6-dimethylpiperazine (2,6-DMPZ and 2,5-DMPZ). By using in situ ATR-FTIR the structure–activity relationship of the reaction between heterocyclic diamines and CO2 is probed. PZ forms a hydrolysis-resistant carbamate derivative, while HHPY forms a more labile carbamate species with increased susceptibility to hydrolysis, particularly at higher CO2 loadings (>0.5 mol CO2/mol amine). HHPY exhibits similar reactivity toward CO2 to PZ, but with improved aqueous solubility. The α-methyl-substituted MHHPY favours HCO3− formation, but MHHPY exhibits comparable CO2 absorption capacity to conventional amines MEA and DEA. MHHPY show improved reactivity compared to the conventional α-methyl- substituted primary amine 2-amino-2-methyl-1-propanol. DMHHPY is representative of blended amine systems, and its reactivity highlights the advantages of such systems. HHPZ is relatively unreactive towards CO2. The CO2 absorption capacity CA (mol CO2/mol amine) and initial rates of absorption RIA (mol CO2/mol amine min−1) for each reactive diamine are determined: PZ: CA=0.92, RIA=0.045; 2,6-DMPZ: CA=0.86, RIA=0.025; 2,5-DMPZ: CA=0.88, RIA=0.018; HHPY: CA=0.85, RIA=0.032; MHHPY: CA=0.86, RIA=0.018; DMHHPY: CA=1.1, RIA=0.032; and HHPZ: no reaction. Calculations at the B3LYP/6-31+G** and MP2/6-31+G** calculations show that the substitution patterns of the heterocyclic diamines affect carbamate stability, which influences hydrolysis rates.


Introduction
The dominant sources of anthropogenic CO 2 emission are fossil-fuel combustion and industrial processes. [1,2] Sequestration of this CO 2 is now a major target for reduction of atmospheric CO 2 levels. While it is clear that any significant longterm reduction in greenhouse gas emissions must involve changing our approach to energy production and consumption, technologies are required to reduce levels in the short term. Coal-fired power stations are the largest point-source emitters of CO 2 in Australia and worldwide. [3] The prospect of integrating post-combustion CO 2 capture (PCC) technology in both existing and new coal-fired power stations offers the potential to lower CO 2 emissions in the face of existing and predicted growth in the number of coal-fired power stations. [4] Currently, aqueous amine-based PCC is viewed as the most promising and near-ready technology for the reduction of CO 2 emissions from coal-fired power stations. PCC involves separating CO 2 from a flue gas stream by chemical absorption and rereleasing CO 2 from the absorbent by heating in a two-step process for subsequent storage or industrial use. PCC is industrially proven with absorbents such as aqueous monoethanolamine (MEA), used for decades for CO 2 removal from gas streams in small-scale commercial processes such as ammonia production and natural-gas processing. [5,6] Despite being an established technology, deployment of current industry-standard technology (30 wt % aqueous MEA) on a large scale applies a considerable efficiency penalty to the power generation process. Regeneration of PCC absorbent is energy-intensive [7] and will result in up to 25 % reduction in the net efficiency of a coal-fired power plant. [8,9] Clearly, the absorption/regeneration characteristics of the amine-based PCC absorbent will influence the economic feasibility of this technology. One approach to reducing the energy requirements and cost of the PCC process is the development of more cost effective and better performing amines. There is considerable scope to develop absorbents that show higher CO 2 absorption, lower regeneration costs and greater chemical stability, particularly in the face of an increasing move towards demonstration-scale PCC plants.
The CO 2 absorption/desorption by aqueous amine-based absorbents has been, and continues to be, extensively studied in the search for improvements in PCC efficiency. The CO 2 absorption/desorption process is shown schematically in Figure 1. Typically, CO 2 reacts with aqueous amines to generate carba-This paper reports on an ATR-FTIR spectroscopic investigation of the CO 2 absorption characteristics of a series of heterocyclic diamines: hexahydropyrimidine (HHPY), 2-methyl and 2,2-dimethylhexahydropyrimidine (MHHPY and DMHHPY), hexahydropyridazine (HHPZ), piperazine (PZ) and 2,5-and 2,6-dimethylpiperazine (2,6-DMPZ and 2,5-DMPZ). By using in situ ATR-FTIR the structure-activity relationship of the reaction between heterocyclic diamines and CO 2 is probed. PZ forms a hydrolysis-resistant carbamate derivative, while HHPY forms a more labile carbamate species with increased susceptibility to hydrolysis, particularly at higher CO 2 loadings (> 0.5 mol CO 2 /mol amine). HHPY exhibits similar reactivity toward CO 2 to PZ, but with improved aqueous solubility. The a-methyl-substituted MHHPY favours HCO 3 À formation, but MHHPY exhibits comparable CO 2 absorption capacity to conventional amines MEA and DEA. MHHPY show improved reactivity compared to the conventional a-methylsubstituted primary amine 2-amino-2methyl-1-propanol. DMHHPY is representative of blended amine systems, and its reactivity highlights the advantages of such systems. HHPZ is relatively unreactive towards CO 2 . The CO 2 absorption capacity C A (mol CO 2 /mol amine) and initial rates of absorption R IA (mol CO 2 /mol amine min À1 ) for each reactive diamine are determined: PZ:  (2) and a protonated amine. The amine substitution pattern affects the products produced. Primary amines such as MEA and secondary amines like diethanolamine (DEA) and piperazine (PZ) react with CO 2 under aqueous conditions to form a carbamate derivative R 1 R 2 NCOO À ( Figure 1). The 2:1 reaction stoichiometry restricts CO 2 absorption capacity of primary and secondary amines to a theoretical upper limit of 0.5 mol CO 2 /mol amine. However, it has been demonstrated that the carbamate can be hydrolysed at high CO 2 loadings (> 0.5 mol CO 2 /mol amine) to produce bicarbonate and regenerate a free amine, [10] which allows for slightly improved absorption capacities ( Figure 1). Given the chemical stability of carbamates formed from 18 and 28 amines, hydrolysis does not occur at an industrially relevant rate, [11] with the exception of heterocyclic secondary amines such as piperidine, which have been demonstrated to form a labile carbamate that is readily hydrolysed. [10] Tertiary amines (R 1 R 2 R 3 N) cannot react directly with CO 2 to form carbamates. [12][13][14] Tertiary amines are believed to act as catalysts facilitating the hydrolysis reaction between CO 2 and OH À to form bicarbonate. [12,14,15] This 38-amine pathway is kinetically and thermodynamically less favourable than carbamate formation. [16] Bicarbonate formation is advantageous in consuming only one molecule of amine per molecule of CO 2 , allowing for increased CO 2 absorption capacities. CO 2 absorption by aqueous amines is a reversible process, and the degree of reversibility is amine-dependent. Amines that form stable carbamates exhibit faster reaction rates, but a larger input of energy is required for absorbent regeneration. Conversely, amines that form more bicarbonate than carbamate exhibit slower reaction rates and require less energy for regeneration. Recent technological advances have allowed for the convenient and rapid analysis of these chemical species to be carried out in situ during the PCC absorption/desorption cycle by attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy. In particular, the ability of ATR-FTIR to distinguish carbamate from bicarbonate formation accelerates the screening of potential PCC amines. [10,17] We recently reported the application of ATR-FTIR in a model PCC absorbent system with substituted piperidines. [10] Herein we report on the in situ CO 2 absorption characteristics of a series of heterocyclic diamines ( Figure 2): piperazine (PZ), 2,6dimethyl-and 2,5-dimethylpiperazine (2,6-DMPZ and 2,5-DMPZ), hexahydropyrimidine (HHPY), 2-methylhexahydropyrimidine (MHHPY), 2,2-dimethylhexahydropyrimidine (DMHHPY) and hexahydropyridazine (HHPZ).

Infrared Spectral Analysis
The effect of structural variations on the CO 2 absorption characteristics of the heterocyclic diamines shown in Figure 2 were assessed in relation to the IR-identifiable products, that is, carbamate versus bicarbonate absorbance; CO 2 absorption capacity, defined as moles of CO 2 absorbed per mole of amine in solution (mol CO 2 /mol amine); and the initial rate of CO 2 absorption (mol CO 2 /mol amine min À1 ). Full details of our experimental approach is given in the Experimental Section and also in our previous work. [10] Each experiment was conducted until equilibrium was established and a maximum CO 2 loading achieved. This was amine-dependent but typically required 45- Figure 1. Reaction mechanism leading to carbamate (1) formation for the reaction of CO 2 with primary (R 1 R 2 NH, where R 1 or R 2 = H) and secondary (R 1 R 2 NH) amines (top), [10] which occurs in the absorption step of the PCC process (bottom). 90 mins. Calculations were also performed to investigate the electronic/steric effects of the structural variations on the amine-carbamate derivatives.

Piperazine (PZ)
Our investigations commenced with parent heterocyclic diamine PZ ( Figure 2). As can be seen from the partial (1750-950 cm À1 ) FTIR spectrum collected during a typical CO 2 absorption experiment with an aqueous PZ solution (1.5 mol L À1 ), five major FTIR peaks evolve during CO 2 absorption ( Figure 3). The carbamate (NCOO À ) derivatives of heterocyclic monoamines have been identified as giving rise to several strong absorbance bands in the 1600-1260 cm À1 region, including the asymmetric (n asCOO À , 1600-1500 cm À1 ) and symmetric (n sCOO À , 1450-1350 cm À1 ) vibrations of the COO À moiety and the NÀCOO À stretching vibration (n NÀCOO À , 1300-1260 cm À1 ) of the NCOO À derivative. [10] The protonated amine (NH 2 + ) generated on absorption of CO 2 was found to give rise to an absorbance band in the 1479-1474 cm À1 region (NH 2 + bending mode). In an amine/CO 2 /H 2 O system, the bicarbonate (HCO 3 À ) species was identified as giving rise to a broad peak in the 1360-1354 cm À1 region (n sCÀO ). [10,17] Assignment was based on the spectral data acquired for 1-methylpiperidine (tertiary amine)/ CO 2 /H 2 O and 2-amino-2-methyl-1-propanol (AMP)/CO 2 /H 2 O systems. It is known that the absorption of CO 2 by aqueous AMP (a-dimethyl-substituted MEA derivative) leads to the formation of mostly HCO 3 À with no significant NCOO À formation. [14,18,19] Herein these peaks can be related to the vibrational modes of the potential ionic reaction products, including PZ-carbamate (PZ-COO À ), protonated PZ (PZ-H + ) and bicarbonate (HCO 3 À ). PZ, being a secondary diamine, should react with CO 2 in solution to form NCOO À , predominately in the form of a protonated PZ-COO À derivative ( + H 2 NR 1 R 2 NCOO À ). [20,21] One amine moiety acts as the absorption site for CO 2 , and the other as a proton acceptor. PZ has also been reported to form the dicarbamate ( À OOC-PZ-COO À ), which was detected by 1 H and 13 C NMR spectroscopy, at CO 2 loadings of 0.2-0.8 mol CO 2 /mol amine. [20,21] The FTIR spectra for the PZ/CO 2 /H 2 O system closely resembles that we previously reported for the piperidine/CO 2 /H 2 O system, differing only in slight shifts in key IR stretching frequencies. [10] At low levels of absorbed CO 2 the PZ/CO 2 /H 2 O system exhibits the n asCOO À (1524 cm À1 ), n sCOO À (1432 cm À1 ) and n NÀCOO À (1276 cm À1 and 1294 cm À1 ) of the PZ-COO À derivative and the NH 2 + vibration of PZ-H + (1470 cm À1 ). These peaks shift to 1546, 1425 and 1289 cm À1 , respectively, with increasing CO 2 absorption levels ( Figure 3). As anticipated, a near-linear relationship between cumulative CO 2 absorption and IR peak intensity is observed for the spectral peaks assigned to n asCOO À , n sCOO À and PZ-H + . Increased peak absorbance is concomitant with the rate of NCOO À formation at the reaction onset plateauing as a maximum CO 2 loading of 0.92 mol CO 2 /mol amine is approached (Figure 4). This near-linear relationship differs from that observed for the n NÀCOO À bands at 1276 and 1294 cm À1 . From the data presented in Figures 3 and 4 the primary n NÀCOO À absorbance emerged at 1276 cm À1 and was the dominant peak, but only at CO 2 /mol amine loadings of 0.4-0.5 mol CO 2 /mol amine. At amine loadings greater than 0.5 CO 2 /mol amine the n NÀCOO À absorbance decreases correspondingly with a sharp increase in intensity of the absorbance band at 1294 cm À1 and a frequency shift to 1289 cm À1 . This trend in n NÀCOO À peak absorbance in the 1294-1276 cm À1 region is attributed to formation of À OOC-PZ-COO À . The IR absorbance of PZ/CO 2 /H 2 O in this region differs from those of all other heterocyclic amine and diamine systems thus far reported, and the remaining subset of secondary heterocyclic diamines analysed in this study (see below) displays only a single n NÀCOO À absorbance band in the 1283-1272 cm À1 region.
As the IR stretching frequencies of PZ-dicarbamate had not been previously reported we turned to computational approaches to facilitate the assignment of key vibrational modes of PZ-COO À , in particular n NÀCOO À . Calculations were performed at the B3LYP/6-31 + G** and MP2/6-31 + G** levels (gas phase, Spartan '08). [22] The B3LYP/6-31 + G** calculations assigned PZ-n NÀCOO À to a single band at 1282 cm À1 (no scaling), while MP2/6-31 + G**  positioned this band at 1284 cm À1 (no scaling), similar in shape, but not as broad, as that which initially emerges at 1276 cm À1 in Figure 3. For the À OOC-PZ-COO À species B3LYP/ 6-31 + G** gave two sharp n NÀCOO À absorbances at 1297-1266 and 1348-1345 cm À1 , which correlated well with MP2/6-31 + G** calculated positions of 1302-1274 cm À1 and 1364-1355 cm À1 . These values correlate well with the experimentally observed peaks at values of 1266, 1276 and 1294 cm À1 , with the latter two shifting to 1289 cm À1 with CO 2 absorption. The B3LYP/6-31 + G** and MP2/6-31 + G** calculations confirm our peak assignments for the PZ-CO 2 carbamate absorption species above. The evolution of a weak broad absorbance band in the 1360-1350 cm À1 region of the PZ/CO 2 /H 2 O IR spectral profile ( Figure 3) was assigned to n sCÀO of HCO 3 À . This absorbance band was far less prominent than that we observed for the piperidine/CO 2 /H 2 O system. Additionally, this absorbance in the PZ system does not follow the trend observed with the corresponding piperidine system; that is, the depletion of n asCOO À , n sCOO À and n NÀCOO À absorbance bands with concomitant increase in the HCO 3 À absorbance band for CO 2 loadings greater than 0.5 mol CO 2 /mol amine. [10] The increase in HCO 3 À absorbance in the piperidine system is attributable to hydrolysis of the initially formed carbamate, which strongly suggests, consistent with the IR data presented herein, that the PZ system forms a hydrolysis-resistant carbamate.
Due to steric congestion arising from the two a-CH 3 moieties, initial CO 2 absorption most likely occurred at the less hindered and more nucleophilic amine moiety, resulting in NCOO À formation. This reduced nucleophilicity and hence reactivity towards CO 2 hindered dicarbamate formation, correlating with the observation of a single n NÀCOO À peak in the IR spectrum. The reduced prevalence of dicarbamate formation resulted in increased hydrolysis and HCO 3 À , as evidenced by rapid growth of the n sCÀO band at 1354 cm À1 (Figure 5 a, b). The a-dimethylamine moiety acted catalytically, in a manner analogous to that reported for sterically hindered amines, to accelerate formation of HCO 3 À . [10] The subtle structural variations between 2,6-DMPZ and 2,5-DMPZ resulted in a significant change in the IR profile. In the case of the 2,5-DMPZ/CO 2 /H 2 O system the most dominant feature is HCO 3 À absorbance, as evidenced by the intense peak in Figure 5. a) Partial IR spectral profile collected for an aqueous solution of 2,6-DMPZ (1.5 mol L À1 ) as CO 2 is absorbed to a maximum loading of 0.86 mol CO 2 /mol amine. b) Relationship between the cumulative CO 2 absorption and IR absorbance for 2,6-DMPZ. the 1400-1300 cm À1 region. There is little evidence to support formation of a stable carbamate ( Figure 6). [10,17]

Hexahydropyrimidine (HHPY)
The HHPY/CO 2 /H 2 O system displayed some similarity with the PZ/CO 2 /H 2 O system in terms of signal positioning but with evidently weaker signals, due in part to the lower concentration of amine (HHPY) available for this study. Notwithstanding this, Figure 7 shows evolution of n asCOO À at 1570-1520 cm À1 , n sCOO À at 1427 cm À1 and n NÀCOO À at 1293 cm À1 of HHPY-COO À ; the NH 2 + bending mode of HHPY-H + at 1479 cm À1 ; and HCO 3 À absorbance at 1354 cm À1 . The HCO 3 À absorbance band was more prominent than that observed for the PZ/CO 2 /H 2 O system, that is, HHPY forms a more labile NCOO À derivative that is more susceptible to hydrolysis.

2-Methylhexahydropyrimidine (MHHPY)
The IR spectrum of the MHHPY/CO 2 /H 2 O system is dominated by the broad HCO 3 À absorbance band in the 1400-1300 cm À1 region ( Figure 8), which is characteristic of a-substituted amines such as AMP. [10,17] MHHPY is the methyl-substituted an-alogue of HHPY. Given the intensity of the HCO 3 À band in Figure 8, MHHPY was readily hydrolysed under the study conditions, with HCO 3 À formation dominating on absorption of CO 2 .

2,2-Dimethylhexahydropyrimidine (DMHHPY)
Given the structural similarity between DMHHPY and MHHPY, we anticipated predominant HCO 3 À formation on CO 2 absorption by DMHHPY. However the IR profile obtained for the DMHHPY/CO 2 /H 2 O system (Figure 9 a) was significantly different to that obtained with MHHPY ( Figure 8) and the HHPY and PZ systems (Figures 7 and 3, respectively). Here DMHHPY is acting more in keeping with a blended-amine PPC absorbent system. Close examination of the in-house synthesized DMHHPY revealed the presence of unconverted 1,3-diaminopropane (DAP) which had been unavoidably carried forward to the final product. Hence, the contamination of DMHHPY with DAP explains the observed blended-system-like IR profile (Figure 9 a). Re-examination of the IR spectrum of the DMHHPY/CO 2 /H 2 O system identified the NH bending mode of 1,3-diaminopopane at 1602 cm À1 . While 1,3-diaminopropane [approximately 37 %, 1.91 g ( 1 H NMR)] was the minor component within the   To allow potential deconvolution of the DAP and DMHHPY signals in the original DMHHPY/CO 2 /H 2 O IR profile, data were collected separately for a DAP/CO 2 /H 2 O system at a DAP concentration of 0.6 mol L À1 (Figure 9 b). It is apparent that the original DMHHPY system is dominated by the reactivity of DAP (cf. Figure 9 a, b). Both systems show evolution of n asCOO À (1565 and 1568 cm À1 , respectively), n sCOO À (1440 cm À1 ) and n NÀCOO À (1328 and 1330 cm À1 , respectively) of the DAP-COO À derivative and the NH 3 + bending mode of protonated DAP (1492 cm À1 ). For the blended DMHHPY/CO 2 /H 2 O system, weaker absorbance bands were also observed to emerge at 1385 and 1370-1350 cm À1 at CO 2 loadings above 1.0 mol CO 2 /mol amine. These new peaks are consistent with NCOO À hydrolysis and HCO 3 À formation. Carbamate hydrolysis was not observed for the pure DAP/CO 2 /H 2 O system. In the initial DMHHPY/CO 2 /H 2 O system (Figure 9 a) the DAP-COO À absorbance bands dominate the IR spectrum. However in the DAP/CO 2 /H 2 O systems the carbamate absorbances are considerably weaker, the initial effect of which was thought to be that the DAP concentration on the DMHHPY system appears to be significantly higher than the 0.6 mol L À1 evident in Figure 9 b. However, a similar difference in intensity between the carbamate absorbance bands of a blended AMP (2.4 mol L À1 )/PZ (0.6 mol L À1 ) system ( Figure 10, further described below) versus an unblended PZ (0.6 mol L À1 ) system was also observed ( Figure 11). Carbamate absorbance in the unblended PZ system was found to be considerably weaker than that observed for the AMP/PZ blended system, despite equivalent PZ concentrations (6 mol L À1 ). Based on the percentage concentrations determined by 1 H NMR spectroscopy the ratio of DMHHPY to DAP in the blended system was 0.95/ 0.85 mol L À1 (total concentration 1.8 mol L À1 ).
For comparative purposes an AMP/PZ blended amine system (2.4 mol L À1 /0.6 mol L À1 , respectively) was also investigated. Similarly to the blended DMHHPY absorbent, the AMP/ PZ blend consists of an amine that forms predominately HCO 3 À on absorption of CO 2 (AMP, major constituent) and an amine that forms predominately NCOO À (PZ, minor constituent). A similar trend in IR absorbance was observed in the spectral data collected for the AMP/PZ/CO 2 /H 2 O system ( Figure 10) to that described above for the blended DMHHPY/ CO 2 /H 2 O system. Despite PZ being the minor constituent of the amine blend, the PZ-carbamate absorbance bands dominated the IR spectral profile. Figure 10 shows the evolution of n ascoo -(1533 cm À1 ), n scoo -(1421 cm À1 ) and n NÀCOO À (1276 cm À1 and shifts to 1263 cm À1 ) of the PZ-COO À derivative and the NH 2 + vibration of PZ-H + (1471 cm À1 ). HCO 3 À absorbance was seen to emerge in the 1386-1330 cm À1 region after a CO 2 loading of about 0.5 mol CO 2 /mol amine. For comparison Figure 11 a and b present the IR spectral profile for an unblended AMP/CO 2 /H 2 O system (3 mol L À1 ) and PZ/CO 2 /H 2 O system (0.6 mol L À1 ), respectively. The HCO 3 À absorbance band was more prominent in the IR spectra of the AMP/PZ/CO 2 /H 2 O system compared to that of the DMHHPY/CO 2 /H 2 O system. This was most likely due to the difference in amine concentrations, with the AMP/PZ blend having a total concentration of 3 molL À1 and the DMHHPY/1,3-diaminopropane blend a total concentration of 1.5-1.8 mol L À1 . Figure 10. Partial IR spectral profile of an aqueous solution of an AMP/PZ blend (2.4/0.6 mol L À1 , respectively) as CO 2 is absorbed to a maximum loading of 1.00 mol CO 2 /mol amine. Figure 11. Partial IR spectral profile of an aqueous solution of a) unblended AMP (3 mol L À1 ) and b) unblended PZ (0.6 mol L À1 ) as CO 2 is absorbed to a maximum loading of 0.84 and 0.92 mol CO 2 /mol amine, respectively.

Hexahydropyridazine
HHPZ absorbed no CO 2 during a typical CO 2 absorption/FTIR experiment. HHPZ is a hydrazine derivative that is reported to have a pK a value of 7.9, [23] which is much lower than that of PZ (9.73), [24] HHPY (9.75) [25] or 2,5-DMPZ (9.66). [26] The low basicity of HHPZ compared to the other diamines (pK a > 9.5) would significantly reduce the reactivity of the amine towards CO 2 .

Absorption Capacity and Absorption Rate
Having established the ability of our diamines to absorb CO 2 , the initial absorption rate R IA and absorption capacity C A were determined. The R IA value was measured by a thermal gravimetric analysis (TGA) method, and C A was measured simultaneously with the IR spectral data (see Experimental Section). These data are presented in Table 1. For comparison, the reactivity of conventional absorbents MEA, DEA and AMP (a-dimethyl-substituted MEA) are also included.
The current industry-standard PCC amine MEA returned C A = 0.56 mol CO 2 /mol amine and R IA = 0.027 mol CO 2 /mol amine min À1 . From the data amassed for PZ, 2,6-DMPZ, HHPY and DMHHPY in Table 1, superior C A and R IA values were observed for all these diamines relative to MEA. Superior C A values were also observed for 2,5-DMPZ and MHHPY, but with lower R IA values. HHPZ did not react with CO 2 , and DMHHPY was blended with DAP. Diamine C A values ranged from 0.85 (HHPY) to 0.92 (PZ) mol CO 2 /mol amine and R IA values from 0.018 (2,5-DMPZ) to 0.045 (PZ) mol CO 2 /mol amine min À1 . While in absolute terms PZ was the standout pure diamine with the highest C A (0.92 mol CO 2 /mol amine) and R IA (0.045 mol CO 2 /mol amine min À1 ), C A and R IA are not the sole factors to be considered in determining the most efficient PCC diamine absorbent; aqueous solubility and stability of the carbamate also play a role. HPPY displays higher water solubility than PZ, on the basis of observations when preparing 1.5 mol L À1 amine solutions. HHPY was readily soluble at this concentration, as opposed to PZ, which required heating and stirring for dissolution. Additionally, HHPY displays high C A (0.85 mol CO 2 /mol amine) and R IA (0.032 mol CO 2 /mol amine min À1 ; Table 1) and showed clear evidence of formation of a hydrolysis-susceptible carbamate (see above and Figure 7).
The PZ analogues 2,6-DMPZ and 2,5-DMPZ displayed lower R IA values of 0.025 and 0.018 mol CO 2 /mol amine min À1 , respectively. As these analogues differ only in the number of methyl substituents (PZ has none) and their positioning, these data suggested that introduction of methyl moieties had an adverse effect on the initial rate of CO 2 absorption. 2,6-DMPZ has both methyl groups a to a single NH group, while 2,5-DMPZ has one methyl group a to each NH group. The measured R IA values indicate that the effect of addition of a-methyl groups is cumulative, with R IA dropping from 0.032 (PZ) to 0.025 (2,6-DMPZ) to 0.018 mol CO 2 /mol amine min À1 (2,5-DMPZ). Concurrent with the reduction in R IA was an increased prevalence towards HCO 3 À formation for 2,5-DMPZ (see above and Figure 6). The reactivity of 2,6-DMPZ towards CO 2 was found to be similar to that of MEA, with the exception of a higher C A value. The propensity of 2,5-DMPZ for HCO 3 À formation was similar to that observed with MHHPY (see above and Figure 8), which was reflected in the almost identical C A (0.88 and 0.86 mol CO 2 /mol amine respectively) and R IA (0.018 and 0.018 mol CO 2 /mol amine min À1 , respectively) values obtained for these amines. The reactivity of 2,6-DMPZ towards CO 2 was found to be similar to that of MEA, with the exception of a higher C A value. 2,6-DMPZ was found to form predominantly carbamate on absorption of CO 2 ( Figure 5), similar to HHPY (Figure 7). While the propensity for carbamate hydrolysis and subsequent HCO 3 À formation is also similar to that observed with HHPY (cf. Figures 5 and 7), which was reflected in the almost identical C A (0.88 and 0.85 mol CO 2 /mol amine, respectively) values, 2,6-DMPZ returned a lower R IA value (0.025 and 0.032 mol CO 2 /mol amine min À1 , respectively). Despite forming predominately HCO 3 À , the initial absorption rates obtained for both MHHPY and 2,5-DMPZ were much higher than that obtained for AMP and comparable to those of MEA and DEA (Table 1). Of the diamines examined, DMHHPY exhibited the highest C A (1.1 mol CO 2 /mol amine) value and an R IA value higher than that of MHHPY and comparable to that of HHPY. This is an artefact of the serendipitous blending with DAP, which contributes significantly to the observed CO 2 absorption capacity. Aqueous DAP has C A = 0.95 mol CO 2 /mol amine. The bicarbonate-forming DMHHPY further promotes CO 2 absorption, resulting in C A > 1.0 mol CO 2 /mol amine. The 18 amino groups of DAP contribute towards DMHHPY's increased initial reaction rate compared to MHHPY.

Effect of Structural Variations on Carbamate Structures
The effect of diamine structural variation on the ability to form stable carbamates was examined at the B3LYP/6-31 + G** and MP2/6-31 + G** levels of theory. Geometry optimisations were initially performed on PZ, 2,6-DMPZ, 2,5-DMPZ, HHPY, MHHPY, DMHHPY and HHPZ. Table 2 lists selected atomic properties including electrostatic potential (ESP) partial charges on the amino nitrogen atoms and the exposed area on the nitrogen atoms for these diamines. The trends in results obtained at the two levels of theory were found to be in good agreement with one another. Diamines can react with CO 2 in aqueous solution to form three possible forms of carbamate species: amine carbamate (HNR 1 R 2 NCOO À ), protonated amine carbamate ( + H-HNR 1 R 2 NCOO À ) and dicarbamate. Of these three forms, + HHNR 1 R 2 NCOO À was expected to be the main reaction product, with one amino group acting as the binding site for CO 2 and the other as a proton acceptor. Geometry optimisations were next performed for the protonated amine-carbamate ( + H-HNR 1 R 2 NCOO À ) and amine-carbamate (HNR 1 R 2 NCOO À ) species. For the lowest-energy conformer of each + HHNR 1 R 2 NCOO À and HNR 1 R 2 NCOO À derivative, Table 3 provides the calculated NÀCOO À bond length (r NÀC []), r C1ÀO1 /r C1À O2 [] ( Figure 12) and ESP partial negative charges on both oxygen atoms as a measure of charge delocalisation. As anticipated, methyl substitution significantly increased the partial negative charges on the amino groups of 2,6-DMPZ (N1 À0.718; N2 À0.752), 2,5-DMPZ (N1, N2 À0.746), MHHPY (N1 À0.851; N2 À0.728) and DMHHPY (N1 À0.959; N2 À0.957) relative to PZ (N1, N2 À0.584) and reduced the exposed area     (Figure 3 and 4) demonstrated that PZ forms a hydrolysis-resistant NCOO À derivative, and thus the optimised geometries of H + -PZ-COO À and PZ-COO À were used as the baseline against which the remaining amine NCOO À derivatives were compared. For both H + -PZ-COO À and PZ-COO À resonance stabilisation of the carboxylate moiety is evident, with calculations showing identical charges on O1 and O2 as well as displaying identical r CÀO values. The r CÀO values of 1.247 and 1.257 reveal partial double-bond character for H + -PZ-COO À and PZ-COO À , respectively ( Table 3). The NÀCOO À bond length of both PZ-COO À species was of single-bond character [standard single-bond r NÀC in PZ is 1.466 (B3LYP) and 1.465 (MP2)]. H + -2,6-DMPZ-COO À and 2,6-DMPZ-COO À also exhibit this stable resonance structure, allbeit with slightly shorter NÀ COO À bond lengths. These findings are in keeping with our IR studies on 2,6-DMPZ and PZ ( Figure 5), which gave very similar outcomes, with the exception of the emergence of a small HCO 3 À absorbance band. The protonated and unprotonated 2,6-DMPZ-COO À forms of isomer 2 (Figure 12), the minor reaction component, contributed to HCO 3 À formation.
H + -HHPY-COO À displays lower levels of charge delocalisation across the two oxygen atoms. Changes in charge distribution and bond length were noted with r CÀO1 = 1.227 and r CÀO2 = 1.277 , which were mirrored in the change in electron density at O1 (À0.641) and O2 (À0.735) (B3LYP). The shift in electron distribution was much less pronounced in the HHPY-COO À species with an r CÀO1 of 1.257 and a r CÀO2 of 1.259 which was mirrored in the change in electron density at O1 (À0.762) and O2 (À0.767) (B3LYP). The NÀCOO À bond lengths of H + -HHPY-COO À (1.507 ) and HHPY-COO À (1.475 ) are similar to that of the PZ-COO À species (1.471 ). These findings are in keeping with our IR studies (Figure 7), in which HHPY was identified as forming a more labile NCOO À derivative that is more susceptible to hydrolysis than that of PZ. The lowest energy conformer obtained for H + -HHPY-COO À , as opposed to HHPY-COO À species, exhibited intramolecular hydrogen bonding between the COO À group and the NH 2 + moiety. The low-energy conformers of H + -MHHPY-COO À and H + -HHPZ-COO À were also found to exhibit the same intramolecular hydrogen bonding. The resonance structure of the H + -MHHPY-COO À was less delocalised with r CÀO1 = 1.211 and r CÀO2 = 1.354 , which was mirrored in the changes in electron density at O1 (À0.650) and O2 (À0.772) (B3LYP). The NÀCOO À bond length of H + -MHHPY-COO À was found to be much shorter (1.417 ). The shift in electron distribution of the carboxylate resonance structure was again much less pronounced in the MHHPY-COO À species compared to the H + -MHHPY-COO À species. MHHPY-COO À has a shorter NÀCOO À bond length (1.463 ) than PZ-COO À (1.471 ) and HHPY-COO À (1.475 ). MHHPY forms predominantly HCO 3 À on absorption of CO 2 , as does 2,5-DMPZ. Both species display shorter NÀCOO À bonds and lower levels of resonance stabilisation. In our IR studies DMHHPY was found to be representative of a blended amine system. Nonetheless, optimised geometries of the H + -DMHHPY-COO À and DMHHPY-COO À were still analysed, and both exhibited reduced resonance in the carboxylate moiety with r CÀO1 = 1.235 and r CÀO2 = 1.244 , which was mirrored in the changes in electron density at O1 (À0.653) and O2 (À0.699), and with r CÀO1 = 1.252 and r CÀO2 = 1.261 , which was mirrored in the changes in electron density at O1 (À0.758) and O2 (À0.819) (B3LYP), respectively. Given the structural similarity of DMHHPY and MHHPY, DMHHPY was expected to form predominantly HCO 3 À on absorption of CO 2 .
Experimentally HHPZ was unreactive towards CO 2 . Calculations revealed H + -HHPZ-COO À to have a significantly shorter NÀCOO À bond length of 1.390 (significant double-bond character), as well as the largest displacement in electron distribution of the carboxylate resonance structure with r CÀO1 = 1.215 and r CÀO2 = 1.350 , which was mirrored in the changes in electron density at O1 (À0.575) and O2 (À0.660) (B3LYP). This was much less pronounced in the HHPZ-COO À species; nonetheless, in a diamine system it is typical for one amino group to act as binding site for CO 2 while the other is protonated. These data support our experimental observations.

Conclusions
A series of heterocyclic diamines (PZ, 2,6-DMPZ, 2,5-DMPZ, HHPY, MHHPY, DMHHPY and HHPZ) have been evaluated as potential PCC absorbents by in situ ATR-FTIR spectroscopy. Of these diamines, PZ displayed both the highest CO 2 absorption capacity (C A = 0.92 mol CO 2 /mol amine) and highest initial absorption rate (R IA = 0.045 mol CO 2 /mol amine min À1 ). These values represent a significant enhancement over currently used amines such as MEA. PZ forms a hydrolysis-resistant carbamate, as well as a dicarbamate. This behaviour is unique to PZ. Hydrolysis of the carbamate derivative of HHPY was observable in the IR spectra collected during CO 2 absorption. HHPY displayed similar CO 2 absorption characteristics to PZ, but with a higher propensity for HCO 3 À formation. The introduction of a-methyl substituents increased the propensity towards carbamate hydrolysis and HCO 3 À formation. Additionally a-methyl substitution decreased R IA , with PZ analogues 2,6-DMPZ and 2,5-DMPZ displaying lower R IA values of 0.025 and 0.018 mol CO 2 /mol amine min À1 , respectively. Increasing the number of methyl groups a to the NH group also increases the rate of HCO 3 À formation. Despite forming predominately HCO 3 À , the R IA of MHHPY (0.018 mol CO 2 /mol amine min À1 ) was much higher than that of the corresponding a-dimethyl substituted 18 amine AMP (R IA = 0.006 mol CO 2 /mol amine min À1 ) and comparable with that of the industrially relevant MEA and DEA. The serendipitously blended DAP/DMHHPY exhibited the highest C A (1.1 mol CO 2 /mol amine) and excellent R IA (0.032 mol CO 2 /mol amine min À1 ). HHPZ was found to be relatively unreactive towards CO 2 . In all instances our calculations at the B3LYP/6-31 + G** and MP2/6-31 + G** levels of theory supported our experimental observation. Microscale Thermogravimetric Analysis (TGA): A Setaram Labsys TG-DTA/DSC thermogravimetric analyser (TGA) was used in isothermal mode at 40 8C to analyse aqueous CO 2 /amine reactivity on a microscale (100 mL). 1.5 mol L À1 aqueous amine solutions were exposed to a gas stream of 15 vol % CO 2 (> 99.9 % purity, BOC Australia) in N 2 at atmospheric pressure. A gas flow rate of 30 mL min À1 was used for all experiments.
To determine total CO 2 uptake two separate TGA experiments were performed for each amine. The first experimental run determined the mass loss due to evaporation when the test solution was exposed to a 100 % N 2 gas stream. The second experimental run determined the mass increase of the test solution when exposed to CO 2 (15 vol % CO 2 in N 2 gas stream) over the same length of time. Each experiment was performed on a fresh 100 mL aliquot of the test solution in a 100 mL alumina crucible (Setaram). From the data collected an absorption curve was then calculated for each amine by subtracting the mass at time t of the evaporation run from the mass at time t of the absorption run. Initial absorption rates could then be calculated by using linear regression to determine the slope of the absorption curve. Absorption/FTIR Experiments: The absorption reactor apparatus used to analyse aqueous CO 2 /amine reactivity has been described previously. [10] Briefly, a gas stream of 13 vol % CO 2 (> 99.9 % purity, BOC Australia) in N 2 with a flow rate of 1.8 L min À1 was bubbled through a 1.5 mol L À1 aqueous amine solution in a glass reactor vessel, maintained at 40 8C by a temperature-controlled water bath (Techne). The CO 2 content of both the gas inflow and outflow was measured by using a Horiba VA 3000 CO 2 analyser. The difference between the CO 2 concentration of the reactor gas inflow and gas outflow was used determine the amount of CO 2 absorbed by the amine solution (mol CO 2 /mol amine). Each experiment was run until the measured CO 2 concentration in the outflow returned to the original percentage value, that is, equilibrium was reached. A typical run lasted between 45 and 90 mins. Solution volumes of 30 mL were used for all experiments. For the duration of each absorption experiment an ATR diamondtipped IR probe, coupled via a mirrored K6 conduit to an ic10 FTIR spectrometer (Mettler-Toledo), was immersed in the aqueous amine solution. In situ IR measurements were obtained simultaneously with the CO 2 absorption measurements, with the FTIR spectrometer set to continuously collect spectra for the duration of the absorption experiment over the spectral range of 4000-650 cm À1 . Each spectrum was recorded as the average of 256 scans over a sampling interval of fifteen seconds with a resolution of 4 cm À1 . The amines investigated were synthesised according to the procedures detailed above, with the exception of commercially available PZ, 2,6-DMPZ and 2,5-DMPZ (Sigma-Aldrich). Computational Details: The computational software package Spartan '08 was used to calculate and compare optimised geometries (gas phase) of the heterocyclic diamines and their carbamate deriv-atives. [22] First, molecular mechanics calculations using the MMFF94 force field and Monte Carlo search algorithm were used to obtain a set of low-energy conformers for each amine and carbamate molecule. Each subset of low-energy conformers were then re-submitted as a geometry optimisation at the B3LYP/6-31 + G** and MP2/6-31 + G** levels to obtain an equilibrium geometry corresponding to an energy minimum (characterized by a gradient < 0.001). Vibrational analysis was performed for all optimised geometries to ensure that they correspond to local minima, that is, there are no imaginary frequencies.