Article pubs.acs.org/est
Influence of Amine Structural Characteristics on N‑Nitrosamine Formation Potential Relevant to Postcombustion CO2 Capture Systems Ning Dai† and William A. Mitch‡,* †
Department of Chemical and Environmental Engineering, Yale University, New Haven, Connecticut 06520, United States Department of Civil and Environmental Engineering, Stanford University, Stanford, California 94305-4020, United States
‡
S Supporting Information *
ABSTRACT: Concerns have arisen for the possible contamination of air or drinking water supplies downwind of amine-based CO2 capture facilities by potentially carcinogenic N-nitrosamines formed from reactions between flue gas NOx and amine solvents. This study evaluated the influence of amine structure on the potential to form total N-nitrosamines within the absorber and washwater units of a laboratory-scale CO2 capture reactor, and in the solvent after a pressure-cooker treatment as a mimic of desorber conditions. Among 16 amines representing 3 amine classes (alkanolamines, straightchain and cyclic diamines, and amino acids), the order of the amine was the primary determinant of total N-nitrosamine formation in the absorber unit, with total N-nitrosamine formation in the order: secondary amines ≈ tertiary amines ≫ primary amines. Similar results were observed upon pressurecooker treatment, due to reactions between nitrite and amines at high temperature. For secondary and tertiary amines, total N-nitrosamine formation under these desorber-like conditions appeared to be more important than in the absorber, but for primary amines, significant formation of total N-nitrosamines was only observed in the absorber. For diamines and amino acids, total N-nitrosamine accumulation rates in washwaters were lowest for primary amines. For alkanolamines, however, total N-nitrosamine accumulation in the washwater was similar regardless of alkanolamine order, due to the combined effects of amine reactivity toward nitrosation and amine volatility. While total N-nitrosamine accumulation rates in washwaters were generally 1−2 orders of magnitude lower than in the absorber, they were comparable to absorber rates for several primary amines. Decarboxylation of the amino acid sarcosine resulted in the accumulation of significant concentrations of N-nitrosodimethylamine and N-nitrodimethylamine in the washwater.
■
(e.g., eq 2 for triethanolamine (TEA)).4 The CO2-rich amine solution then proceeds to a “desorber” column, where high temperatures reverse the absorption reaction, regenerating the amine and releasing the CO2. The CO2 is compressed and routed to underground storage for sequestration. Countercurrent water scrubbing units (washwater units) generally are employed to remove amines and other potential contaminants from flue gases prior to atmospheric release.
INTRODUCTION Amine-based carbon capture has emerged as one of the most promising technologies for implementing postcombustion CO2 capture in the near future. Used by Statoil since 1996 to capture CO2 from North Sea natural gas fields precombustion,1,2 it remains one of the few technologies for which there is full-scale experience.3 The Norwegian government, in concert with Statoil, is planning to construct one of the first postcombustion CO2 capture facilities at a combined heat and power plant at the Mongstad refinery in Norway, and amine-based capture technologies are major contenders for this facility. In an amine-based postcombustion CO2 capture system, flue gases are passed through an “absorber” column where a countercurrent amine-containing aqueous solvent absorbs CO2. Primary and secondary amines react with CO2 to form a covalent carbamate intermediate (e.g., eq 1 for monoethanolamine (MEA)), while tertiary amines and some sterically hindered primary and secondary amines behave as bases to solubilize CO2 as HCO3− © 2013 American Chemical Society
Received: Revised: Accepted: Published: 13175
August 8, 2013 October 16, 2013 October 18, 2013 October 18, 2013 dx.doi.org/10.1021/es4035396 | Environ. Sci. Technol. 2013, 47, 13175−13183
Environmental Science & Technology
Article
desorber unit; 1-nitrosopiperazine, the nitrosation product of piperazine, constituted only 62% of the total N-nitrosamine signal.19 In 2.5 M amine solvents purged with 25 ppmv each of NO and NO2, the total N-nitrosamine accumulation rates for a tertiary amine (MDEA) and two primary amines (AMP and MEA) were 1−2 orders of magnitude lower than that for a secondary amine (PZ), but were measurable.19 Reactions between amines and nitrite accumulating in the solvent under the high temperature conditions in the desorber may also form N-nitrosamines. For example, when an MEA-based solvent collected from a pilot system lacking a desorber unit was autoclaved to mimic desorber conditions, the concentration of N-nitrosodiethanolamine increased from 5.3 μM to 190 μM.20 Additionally, when 5 M PZ was heated at 100 °C in the presence of 2.9 M CO2 and nitrite, complete nitrite conversion to 1-nitrosopiperazine was reported.21 Because the washwater unit generally represents the last treatment unit prior to atmospheric release, there are concerns regarding the potential for N-nitrosamines in the washwater to be stripped into the exhaust gas. In previous work, we measured total N-nitrosamine concentrations in washwaters collected from a pilot facility lacking a desorber unit of 0.73 μM from a MEA-based solvent, and 59 μM from a mixed PZ/AMP-based solvent.19 Moreover, we demonstrated that washwaters can potentially serve as a source of N-nitrosamines when residual NOx reacts with amines accumulating within the washwater.22 Addressing the potential release of these compounds in the exhaust gas, the Norwegian Climate and Pollution Agency has granted CO2 Technology Centre Mongstad a discharge permit for its operations limiting the sum of all N-nitrosamines and N-nitramines to 0.3 ng/m3 in downwind airsheds, and 4 ng/L in downwind water supplies.23 Unfortunately, there is no total N-nitramine method currently available. However, the comparison of N-nitrosomorpholine and N-nitromorpholine formation rates within a 5 M morpholine solvent purged with NOx indicated that N-nitrosamine formation exceeded N-nitramine formation regardless of the relative concentrations of NO and NO2.19 The purpose of this study was to systematically vary solvent amine structures to compare their potential to form total N-nitrosamines under conditions relevant to carbon capture. Three classes of compounds being considered for carbon capture were evaluated: alkanolamines, straight-chain and cyclic diamines, and amino acids (see Figure 1 for amine structures and abbreviations). Within these classes, the amines were selected for study based upon systematic variations in amine structural characteristics, rather than for commercial relevance. A primary goal was to evaluate the influence of the order of the amine. Total N-nitrosamine accumulation rates were evaluated in the solvent and washwater of a laboratory-scale reactor without a desorber unit, and amine accumulation rates were monitored in the washwaters. Additionally, total N-nitrosamine formation after treatment of the solvent at 120 °C in a pressure-cooker was used to estimate the importance of N-nitrosamine formation within desorber units. Our results indicated that the amine order was of primary importance for determining N-nitrosamine accumulation rates. Because primary amines demonstrated the lowest potential to form N-nitrosamines, additional experiments systematically varied primary amine structures to evaluate the effects of the following: (1) steric hindrance of the amino group in primary alkanolamines, (2) the alkyl chain length of straightchain primary diamines and amino acids, and (3) substituent functional groups of primary amines.
Altering the structure of the solvent amine is a major focus for improving the performance of amine-based CO2 capture systems. An ideal solvent would feature high CO2 absorption capacity, fast absorption kinetics, low heat requirement for regeneration, and high chemical stability over extended absorption−desorption cycles. Primary amines, including the benchmark solvent MEA, exhibit fast CO2 absorption kinetics,5 but low CO2 absorption capacities, and high heat requirements for regeneration in the desorber unit,6 and are susceptible to degradation by CO 2 and O 2 . 7,8 Tertiary amines (e.g., methyldiethanolamine (MDEA)), on the contrary, have high CO2 absorption capacities,6,9 low heat requirements for regeneration,9 and high chemical stabilities,7,8 but feature low CO2 absorption kinetics.6 In comparison, secondary amines and sterically hindered primary amines (e.g., 2-amino-2-methyl-1propanol (AMP)) exhibit a good balance between fast CO2 absorption kinetics and high capacities. They also feature low regeneration heat requirements,6,9 but are susceptible to degradation.7,8 An emerging concern for the application of amine-based technologies to postcombustion CO2 capture is the formation of potentially carcinogenic N-nitrosamines and N-nitramines10−12 from reactions between flue gas NOx and solvent amines.13 NO and NO2 react to form the nitrosating agent N2O3 (eq 3), while NO2 can dimerize to form N2O4 (eq 4), which has nitrating and nitrosating tautomers.13 Extensive previous biomedical research on N-nitrosamine formation from the reaction between nitrite and amines under the acidic conditions of the stomach has suggested that N2O3 is the responsible nitrosating agent.14 Under these conditions, secondary amines, which are able to form stable N-nitrosamines directly,14 are the most potent precursors. In comparison, primary15,16 and tertiary17 amines are less prone to forming N-nitrosamines, because they require initial transformation to nitrosatable secondary amine precursors (see Schemes SI-1 and SI-2 of the Supporting Information, SI). NO• + NO2• ↔ N2O3
(3)
2NO2• ↔ N2O4
(4)
However, the influence of the amine structure on the formation potential of N-nitrosamines and N-nitramines has not been systematically evaluated under conditions relevant to carbon capture. A recent study applying 0.8% NO and 5% O2 to amines in a batch system demonstrated that the secondary amines piperazine (PZ), diethanolamine (DEA), and 2piperidinemethanol formed their associated specific N-nitrosamines at measurable levels; but no specific N-nitrosamines were observed during treatment of the tertiary amine methyldiethanolamine (MDEA), or the primary amines AMP and MEA.18 Primary and tertiary amines can undergo degradation due to high temperature and reactions with flue gas components, thereby being converted to secondary amines that serve as precursors for N-nitrosamines. The complexity of these pathways renders it difficult to predict the specific N-nitrosamine products. Using a recently developed total N-nitrosamine assay, we detected 11 mM of total N-nitrosamines in a solvent constituted from a mixture of a hindered primary amine (AMP, 25%) and a secondary amine (PZ, 15%) from a pilot system without a 13176
dx.doi.org/10.1021/es4035396 | Environ. Sci. Technol. 2013, 47, 13175−13183
Environmental Science & Technology
Article
Figure 1. Structures of model solvent amines.
■
MATERIALS AND METHODS Sources and purities of chemicals are provided in the SI. Laboratory-Scale CO2 Capture System. This system consists of an absorber column and a washwater column, both of which are 5 cm diameter × 40 cm length, containing 6 mm borosilicate glass Raschig rings. A 2.5 M amine solution (200 mL, “solvent”) was circulated through the absorber unit at 15 mL/min in a countercurrent fashion. The absorber column and solvent reservoir were maintained at 48 °C during reactor operation. The amine solutions were self-buffered, but amino acid solutions were pH adjusted with sodium hydroxide to 1 pH unit above the pKa of the amino group. For piperazine (PZ), 2.5 M exceeds its water solubility at room temperature; the PZ solvent was therefore prepared by directly dissolving the solids into deionized water preheated to approximately 50 °C. Deionized water (500 mL) was circulated through the washwater unit at 30 mL/min at room temperature in a countercurrent fashion. A synthetic flue gas
containing N2, CO2, and O2 was purged through a prewetting column containing deionized water at 48 °C, resulting in a mixture of 75% N2, 6% CO2, 7% O2, and 12% H2O. NO and NO2 (25 ppmv each) were added to the flue gas between the prewetting column and absorber column. Note that these NOx conditions may not replicate any particular NOx conditions in an authentic capture plant. These equimolar conditions were selected to compare solvent amines under conditions maximizing nitrosation potential. Due to the lack of desorber unit, the system was pre-equilibrated with CO2 without NOx for 1 h, after which CO2 was turned off and NOx was turned on simultaneously. The volumetric flow rate of CO2 was made up by N2. The CO2 loading of the solvent was measured after the 1-h loading period (Table SI-1 of the SI). Solvent and washwater samples were collected at various time points after NOx addition and analyzed for nitrite and total N-nitrosamine concentrations. Because the volume of the solvent and washwater could change over the 13177
dx.doi.org/10.1021/es4035396 | Environ. Sci. Technol. 2013, 47, 13175−13183
Environmental Science & Technology
Article
the experimental time periods (e.g., Figure SI-2 of the SI for solvent MEA). The order of nitrite accumulation rates in the absorber for different amines was primary < secondary < tertiary, but the variation was less than a factor of 10 across the amines (Figure 2B).
course of the experiment (e.g., due to volatilization/accumulation of amines and water vapor), the volumes changes in solvent and washwater reservoirs were monitored to allow the calculation of nitrite and total N-nitrosamine mass accumulation rates. To simulate desorber conditions, 5 mL solvent samples retrieved from the absorber were subject to 120 °C treatment in a pressure cooker at 15 psi for 10 h in duplicate. Nitrite and total N-nitrosamines were measured before and after pressure-cooker treatment. Sample Analysis. Nitrite was measured by the N-(1naphthyl)ethylenediamine dihydrochloride colorimetric method.24 Concentrations of parent amines in the washwater were measured by cation chromatography (Dionex, ICS-1000), with a methanesulfonic acid eluent of 5 mM for alkanolamines, 20 mM for primary and secondary diamines, or 35 mM for tertiary diamines at a flow rate of 1 mL/min. Primary amino acids were derivatized using the AccQ-Fluor reagent kit (Waters), and analyzed using LC-MS with gradient elution with acetonitrile and 20 mM ammonium formate: 5% acetonitrile for 3 min, ramping to 25% acetonitrile over 2 min and holding for 7 min. Electrospray ionization parameters were as follows: nebulizer gas pressure 35 psi, drying gas pressure 10 psi, temperature 350 °C, and capillary voltage 80 V. Full scan mass spectrometry was used with a RF loading of 85%. The total N-nitrosamines analysis was described previously;19 a summary is provided in the SI. Error bars represent standard deviations from replicate experiments (n = 2−3). CO2 loadings of solvent amines were measured according to the procedure of Weiland and Trass.25 Briefly, each of two aliquots of 0.5 mL solvent samples were diluted to 10 mL with deionized water. One of the 10 mL aliquots was heated in a water bath to 90 °C for 20 min in the presence of 0.38 M barium chloride. After 20 min, the solution was immediately cooled down to 4 °C, and filtered through 0.2-μM syringe filters. The filtrate and the other 10 mL aliquot were titrated with 0.5 N hydrochloric acid to pH 3. The CO2 loading was calculated using eq 5. CO2 loading(mol CO2 /mol amine) =
N2O3 + H 2O → 2NO2− + 2H+
(6)
N2O4 + H 2O → NO2− + NO3− + 2H+
(7)
Solvent samples retrieved from the absorber were treated at 120 °C in a pressure-cooker for 10 h as an initial estimate of the importance of N-nitrosamine formation in desorber units. Recent research has suggested that N-nitrosamine formation under desorber-like conditions is affected by nitrite concentration and the CO2 loading of amines.21 As amine structural differences affect their percent CO2 loading under absorber conditions (Table SI-1 of the SI), it would be difficult to load all 16 amines to the same percentage. Instead, we opted to collect absorber samples after a set time for introduction to the pressurecooker. This more operational approach incorporated the different CO2 affinities of amines with different structures and their effects on nitrite accumulation in the absorber solvents. For 13 of the 16 amines, samples were collected from the absorber after 4 h of absorber operation. For the remaining three amines (tertiary amines TEA, tMEdA, and dMPZ), samples were retrieved at the end of the absorber experimental runs (7−9.4 h). Although the latter samples may feature higher nitrite concentrations (by a factor of ∼2), and slightly lower CO2 loading than their 4 h equivalents, the effects of these discrepancies on total N-nitrosamine formation during the pressure-cooker treatment were anticipated to be less than an order of magnitude, smaller than the order of magnitude differences observed between amine categories (see below). Although these results bear more uncertainty due to the difference in sample collection time, data from these samples were also included in Figure 2C,D for comparison. Recent research demonstrated that the N-nitrosamine formation rate under such desorber-like conditions in a batch system declines over time as nitrite is consumed.21 Therefore, rather than presenting the formation and loss rates, Figure 2C,D present the increase in total N-nitrosamine mass and decrease in nitrite mass, respectively, after 10 h pressure-cooker treatment, in comparison to the mass present in the original sample from the absorber unit. The change in total N-nitrosamine levels before and after pressure-cooker treatment was insignificant for primary amines EdA and Gly, but 1.4 ± 0.2 μmole for MEA. In comparison, the increase for secondary amine-based solvents was 2−3 orders of magnitude higher than for primary amines, and 50−300% higher than for the corresponding tertiary amines. In addition, the increase for secondary amines was comparable across the different classes of amines. Figure 2D provides the loss of nitrite during 10 h of pressure-cooker treatment for different solvent amines. Although the nitrite mass loss for all solvents was within an order of magnitude, the loss of nitrite was lowest for primary amines, and similar for secondary and tertiary amines, except for piperazine (PZ), which exhibited the greatest loss of nitrite. The accumulation of total N-nitrosamines in washwaters represents an important environmental concern due to the potential for these compounds to be stripped into the exhaust gas. The total N-nitrosamine accumulation rates in washwaters were 1−2 orders of magnitude lower than in the absorber unit (Figure 2E). The total N-nitrosamine accumulation rate in the
(V2 − V1) × 0.5 C × 0.5 (5)
where V1 and V2 are the volumes of 0.5 N hydrochloric acid needed to titrate the 10 mL solution with and without barium chloride treatment, respectively, and C is the initial amine concentration in the samples.
■
RESULTS Order of the Amino Group in Alkanolamines, Diamines, and Amino Acids. In the absorber, the amount of total N-nitrosamines in the solvent accumulated linearly over time (e.g., Figure SI-1 of the SI for solvent MEA) upon exposure to NOx after accounting for volume changes during reactor operation. Figure 2A compares the total N-nitrosamine accumulation rates in primary, secondary, and tertiary amine solvents. Across the different classes of amines, secondary amines generally exhibited the highest N-nitrosamine accumulation rates. Tertiary amines formed N-nitrosamines at comparable rates (e.g., tMEdA) or up to 5 times lower (e.g., TEA and MDEA) when compared to the corresponding secondary amines, while primary amines exhibited accumulation rates nearly 1−2 orders of magnitude lower. Nitrite is one of the products of NOx hydrolysis (eqs 6 and 7),26 and its mass accumulation rates within the absorber unit were constant over 13178
dx.doi.org/10.1021/es4035396 | Environ. Sci. Technol. 2013, 47, 13175−13183
Environmental Science & Technology
Article
Figure 2. (A) Total N-nitrosamine and (B) nitrite formation rates in the absorber; (C) total N-nitrosamine formation and (D) nitrite loss during pressure-cooker treatment; and (E) total N-nitrosamine and (F) parent solvent amine accumulation rates in washwaters for primary, secondary, and tertiary amines. Error bars indicate standard deviations for experimental replicates (n = 2−3). NS = not significant. NM = not measured.
washwaters associated with diamines and amino acids followed a similar pattern as observed in the absorber (i.e., secondary ≈ tertiary ≫ primary amines). However, for alkanolamines, the total N-nitrosamine accumulation rates in the washwaters were similar regardless of the order of the amine. Previously, we observed that N-nitrosamines can form in situ in the washwater unit from the reaction between residual NOx and the amines accumulating in the washwater.22 Therefore, the accumulation of solvent amines in washwater was also monitored (Figure 2F). Because preliminary experiments indicated that amines accumulated linearly in the washwater (Figure SI-3 of the SI), amine accumulation rates were calculated using the amine concentrations in the washwater samples collected at the final time point. While primary and secondary diamines accumulated in the washwater at similar rates, the two tertiary diamines, tMEdA and dMPZ, exhibited approximately 4 orders of magnitude higher accumulation rates than the other diamines. For alkanolamines, the accumulation rates of solvent amines in the washwater decreased as the amine order increased. For most amines, only the parent amine was detected at significant concentrations; however, DEA and MEA were detected in the washwater associated with TEA at concentrations approximately 4 times higher than TEA. MEA was also detected in the washwater
of the other tertiary alkanolamine MDEA. In addition, dimethylamine was detected in washwaters associated with SAR, exhibiting an accumulation rate of 0.33 (±0.06) nmol/s. Although SAR was not measured, its accumulation rate is anticipated to be low, similar to those of β-Ala (
