Analysis of Monoethanolamine and Its Oxidative Degradation

Ind. Eng. Chem. Res. 2006, 45, 2437-2451

2437

Analysis of Monoethanolamine and Its Oxidative Degradation Products during CO2 Absorption from Flue Gases: A Comparative Study of GC-MS, HPLC-RID, and CE-DAD Analytical Techniques and Possible Optimum Combinations Teeradet Supap,† Raphael Idem,*,‡ Paitoon Tontiwachwuthikul,‡ and Chintana Saiwan† Petroleum and Petrochemical College, Chulalongkorn UniVersity, Pathumwan, Bangkok, Thailand 10330, and Process Systems Engineering Laboratory, Faculty of Engineering, UniVersity of Regina, Regina, Saskatchewan, Canada S4S 0A2

A comparative study of gas chromatography-mass spectrometry (GC-MS), high-performance liquid chromatography-refractive index detection (HPLC-RID), and capillary electrophoresis-diode array detection (CE-DAD) techniques was carried out for the purpose of analyzing MEA and its degradation products in MEA/H2O/O2 and MEA/H2O/O2/CO2 systems. The experiments were conducted in a 600-mL reactor using an MEA concentration of 5 kmol/m3, an O2 pressure of 250 kPa, a CO2 loading of 0.51 mol of CO2/mol of MEA, and degradation temperatures of 328-393 K. GC-MS using an HP-35MS column (intermediate polarity) performed the best only if analysis of the degradation products was of interest, whereas HP-Innowax (highpolarity column) was best only if analysis of MEA was required. Analyses of the same sample using two different columns (e.g., HP-35MS and HP-Innowax) would be required if both MEA and its degradation products are to be followed. HPLC-RID using a Nucleosil column with phosphate buffer was the best and only technique in which simultaneous analysis of MEA and degradation products was possible. CE-DAD using phosphate and borate electrolytes was able to detect degradation products. Because the results in terms of degradation product distribution, decline of MEA, and role played by CO2 as observed by all techniques were consistent, a combination of these techniques is recommended for confirming MEA oxidative degradation systems. 1. Introduction Emission of carbon dioxide (CO2) is considered to be environmentally unacceptable because of its contribution to enhancing the greenhouse effect. This underscores the significance of CO2 capture from large point sources such as flue gas streams, coal-fired power plants, refineries, cement manufacturing plants, etc., in mitigating CO2 emissions. In the case of lowpressure streams such as flue gases from power plants, absorption with chemical reaction using aqueous alkanolamine solutions is one of the methods available that can reach a desired target for CO2 capture. Various choices of alkanolamines are commercially available, with monoethanolamine (MEA) and diethanolamine (DEA) as examples. One of the advantages of alkanolamines is that, structurally, they contain at least one hydroxyl group, which helps to reduce their vapor pressures but increases their solubilities in aqueous solution. On the other hand, the amino group provides for the necessary alkalinity to absorb CO2.1 However, the disadvantage is that most alkanolamine-based solvents can deteriorate upon contact with oxygen (O2).2 O2 concentrations in the range of 3-12% typical of flue gas streams3,4 are known to induce oxidative degradation of alkanolamines. This undesirable breakdown reduces their CO2 absorption capacity and introduces unwanted degradation products, thus forcing the solutions to be eventually discarded. An operational burden such as corrosion is also induced, as it has been periodically reported that the degradation products can lead to severe corrosion.5,6 Past research has provided useful information on various aspects of the oxidative degradation of alkanolamines, in which * Corresponding author. Fax: (306) 585-4855. E-mail: rapheal.idem@ uregina.ca. † Chulalongkorn University. ‡ University of Regina.

the most attention has been paid to the identification of degradation products7,8 and their effects on corrosion and the formation of heat-stable salts.9-11 The kinetics of the process has also gained considerable interest over the past several years because of its significance for a better understanding of the oxidative degradation process. However, the kinetics formulation is so far limited to measurements of either the concentration decline of the alkanolamine reactant12 or the production rate of a gaseous product (e.g., ammonia).13,14 A more favorable kinetics is one based on a degradation mechanism that incorporates the major degradation products into the kinetic equations. This has not yet been undertaken, even though it is an essential element for the development of an oxidative degradation prevention strategy. Achieving such a strategy requires the development of adequate analytical techniques that must be able to effectively capture a wide variety of the products typically formed within alkanolamine oxidative degradation systems. Gas chromatography (GC) with conventional detectors such as thermal conductivity (TCD) and flame ionization (FID) was one of the earliest methods utilized for analysis of alkanolamines and their degradation products. Its early application, however, was limited to separation of alkanolamines such as MEA, DEA, and TEA (triethanolamine) using different packed columns, in which tedious derivatizations were required prior to the analysis.15,16 Analysis of degradation products by GC was later explored using CO2-induced DEA degradation.17 The use of GC coupled with a more advanced detector such as a mass spectrometer (MS) to analyze degraded DEA solutions was also investigated, but it was limited to only degradation reactions induced by CO2 and carbonyl sulfide (COS).18,19 GC-MS was later applied to the oxidative degradation of MEA for kinetics studies.12 A more recent study used a combination of detectors such as a mass spectrometer (MS), a Fourier transform infrared

10.1021/ie050559d CCC: $33.50 © 2006 American Chemical Society Published on Web 10/15/2005

2438

Ind. Eng. Chem. Res., Vol. 45, No. 8, 2006

(FTIR) absorption spectrophotometer, and an atomic emission detector (AED) with GC, as well as a low-voltage, highresolution mass spectrometry (LVHRMS), combined with inductively coupled plasma-atomic emission spectroscopy (ICP-AES) and ion chromatography (IC) for identification of the degradation products of MEA plant samples.20 Liquid chromatography has also been a choice for alkanolamine degradation analysis. Dating back to 1982, the ionexclusion technique has been capable of separating formic, oxalic, and acetic acids in air-bubbled MEA, DEA, and MDEA (methyldiethanolamine) samples.8 It was later that an ion chromatograph equipped with an anion-exchange column found use in the detection and measurement of carboxylate ions including acetate, formate, and glycolate ions in various alkanolamines that had been in contact with air for some period of time.21 The separations of various inorganic and organic anions in rich and lean MEA and MDEA plant samples have also been carried out using anion-exchange columns of Ionpac AS10 and Ionpac AS9-SC with a combination of Na2B4O7, NaHCO3, and NaOH as the eluents.22 Separations of alkanolamines such as MEA, DEA, TEA, and MDEA, as well as cations, corrosion inhibitors, and piperazine in MDEA solution, were investigated later using a cation-exchange column (e.g., Ionpac-CS10) with various contents of H2SO4 eluents.23 Because the conductivity mode of the detector was used for both techniques, additional ion-suppression systems also served to lower the detector response to the eluents in order to enhance the signal from eluted compounds, thus increasing the cost of analysis. This disadvantage was later overcome by using a direct ion chromatographic technique as shown in the work of Kaminski et al.,24 which made use of the refractive index detector (RID) with an HPLC in cation-exchange mode with a Nucleosil SA column to separate MEA, DEA, and MDEA plant samples and their inorganic cations and some degradation products using a single component of phosphate for the mobilephase eluent. However, this approach was tested only in the desulfurization process and only on wastewater. Analysis of amines and their degradation products in the CO2 capture process has not yet been evaluated. Capillary electrophoresis (CE) also shows benefits such as a reduction in costs and in the effort needed for method development as compared to the HPLC technique. CE has recently been applied to the determination of alkanolamines in water/ethanol extracts of wrapping materials containing volatile corrosion inhibitors.25 Basic and acidic drugs could also be analyzed by this technique using NaH2PO4 and Na2B4O7‚10H2O, respectively, as electrolytes.26,27 Therefore, this technique shows potential for use in the detection of alkanolamines and, perhaps, basic and acidic degradation products being formed in oxidative degradation systems. However, this technique has not yet been applied to the analysis of alkanolamines and their degradation products. It has also been observed that most studies often make use of a single technique to characterize the oxidative degradation of alkanolamines. Although GC, especially GC-MS, has been a powerful and heavily used tool, its analysis can be limited to volatile products. Degradation products of high molecular weights might potentially be left undetected, resulting in incomplete information on their formation pathways. LC might overcome this problem and could be used as a confirmation technique. CE could also be used in the same manner as LC. To adequately gain information on alkanolamine degradation in terms of an alkanolamine decline as well as the distribution of its degradation products in the liquid phase for a formulation

of a mechanism-based kinetics, a combination of analytical techniques is required. A summary showing gas chromatographic, liquid chromatographic, and capillary electrophoresis techniques previously used for analysis of alknolamines and their degradation products is provided in Table 1. This table also shows the present limitations of these techniques. In this study, we are developing improved methods as well as exploring the best combinations of gas chromatographic, liquid chromatographic, and capillary electrophoresis analytical techniques to completely characterize the oxidative degradation of two aqueous monoethanolamine systems, MEA/H2O/O2 and MEA/ H2O/O2/CO2. In the GC-MS technique, three different chromatographic columns of different stationary-phase polarity were evaluated for the separation, detection, identification, and quantification of MEA and its degradation products for both systems. On the other hand, in the HPLC technique, two different columns and two different mobile phases were evaluated for the same purpose. The capillary electrophoresis (CE) technique with selected electrolyte solutions was also explored. Some standards were used to identify degradation products using spiking techniques. The results for all of these techniques are reported in this paper in terms of the compatibility of each analytical technique to the separation of aqueous MEA and its oxidative degradation products, as well as the best combinations of techniques to achieve this objective. 2. Experiments 2.1. Equipment and Chemicals. The oxidative degradation experiments were conducted using an MEA concentration of 5 kmol/m3. The degradation temperature was fixed at 328 or 393 K, representative of absorption-regeneration conditions in a typical flue gas treating process. An O2 pressure of 250 kPa was used to perform accelerated oxidative degradation experiments. This is a precursor for our current work (beyond the scope of the present paper), which will determine the products and mechanism of the actual conditions for O2 in a typical CO2 capture plant. For CO2-loaded experiments, a loading of 0.51 mol of CO2/mol of MEA was employed. The oxidative degradation reactions were carried out using a 600-mL stainless steel batch reactor (model 5523, Parr Instrument Co., Moline, IL). The removable reactor head assembly consisted of a magnetic drive connected to a stainless steel (T316) stirring shaft and two adjustable four-rectangular-blade impellers (diameter of about 1.5 in.) of which one was positioned almost at the bottom of the vessel and the other was about 2.5 in. above the bottom one. The positions of these impellers were held constant throughout the experiments. In addition, the reactor also had a 0-300 psi Bourdon-type pressure gauge; gas inlet, gas release, and liquid sampling valves; a preset safety rupture disk; a J-type thermocouple; a dip tube for gas introduction and sample removal; and a cooling coil for maintaining the process at a constant temperature, irrespective of the temperature rise caused by the exothermic nature of the reaction of MEA and O2, and preventing any temperature overshoot during the experiment that might be caused by furnace heating. The cooling system was regulated by a solenoid valve. The heat was supplied to the reactor by a furnace in which the reactor vessel was inserted and regulated by a temperature controller (model 4836, Parr Instrument Co., Moline, IL), whereas the temperature of the reaction mixture was measured by a J-type thermocouple. The temperature accuracy of the controller was within (0.1%. Research-grade O2 and CO2 were supplied from Praxair (Regina, Saskatchewan, Canada). Concentrated MEA (reagent grade with >99% purity) was purchased from Fisher Scientific, Neopean,

packed

packed

packed and capillary capillary

capillary

ion exclusion

anion exchanger

anion exchanger

cation exchanger

cation exchanger

cation exchanger

capillary

capillary

capillary

16

17

18, 19

20

8

21

22

20

23

24

25

26

27

a

N/A ) not available.

12

packed

column type

15

ref(s)

none

none

none

Nucleosil 100-5 SA

Ionpac CS10/Ionpac CS12A

Ionpac CS14

IonPac AS10/AS9-SC

not provided

N/Aa

5% neopentylglycol succinate 3% OV-1, methylpolysiloxane 2,6-diphenyl-p-phenylene oxide 2,6-diphenyl-p-phenylene oxide/poly(ethylene glycol) cross-linked poly(ethylene glycol) 14% (cyanopropylphenyl)methylpolysiloxane/ nitrotetraphthalic acidmodified poly(ethylene glycol)

stationary phase

operating conditions detector

UV UV

Na2B4O7‚10H2O

Capillary Electrophoresis DAD

PED using conductivity mode RID

not provided PED using conductivity mode conductivity

Liquid Chromatography N/Aa

MS/FTIR/ AED

MS

FID/MS

FID

FID

Gas Chromatography TCD

imidazole/ HIBA/ 18-crown-6 NaH2PO4

KH2PO4

not provided Na2B4O7/ NaHCO3/ NaOH Na2CO3/ NaHCO3 H2SO4

N/Aa

He

He

N2

N2

Ar, N2

He

mobile phase/ electrolyte

acidic drugs

basic drugs

volatile corrosion inhibitors

amines, products, and wastewater in a desulfurization process

inorganic species in MEA used in a flue gas treatment plant alkanolamine, monovalent and divalent cations

O2-induced MEA/DEA/MDEA degradation O2- and O2/CO2-induced degradation of various alkanolamines heat-stable salts in MDEA plant samples

MEA degradation products in samples from a flue gas treatment plant

O2-induced MEA degradation

COS-induced DEA degradation

CO2-induced DEA degradation

alkanolamine analysis

alkanolamine analysis

system studied

not yet applied to alkanolamine and degradation products not yet applied to alkanolamine and degradation products

not yet applied to alkanolamine and degradation products

overlap of products when applied to oxidative degradation systems

need for extra suppression system, increasing costs need for extra suppression system, increasing costs

need for extra suppression system, increasing costs

lack of simutaneous MEA detection

lack of simultaneous MEA detection

caustic-modified samples tested; the technique was not applied to actual degraded samples

peak overlaps

limited to COS-induced degradation

low-efficiency packed column

tedious derivatization

tedious derivatization

limitations/shortcomings

Table 1. Summary of Gas Chromatographic, Liquid Chromatographic, and Capillary Electrophoretic Techniques Previously Used for Analysis of Alkanolamines and Their Degradation Products

Ind. Eng. Chem. Res., Vol. 45, No. 8, 2006 2439

2440


Ontario, Canada. MEA was diluted to the desired concentration using deionized distilled water. Its exact concentration was established by volumetric tritration with a standard solution of 1 kmol/m3 hydrochloric acid (HCl), also from Fisher Scientific, to the endpoint of the methyl orange indicator. 2.2. Typical Experimental Run. 2.2.1. MEA/H2O/O2 Degradation System. About 400 mL of 5.1 kmol/m3 MEA solution was loaded into the reactor, and the reactor head assembly was placed on top and tightly sealed to obtain a leak-free environment. The solution was stirred at 500 rpm while being heated to 328 or 393 K. A few minutes was allowed for the solution to stabilize at the desired temperature. At this point, some vapor pressure due to water was observed as the pressure inside the reactor, and then 250 kPa of O2 was additionally introduced by opening the O2 cylinder set at a predetermined value through the gas inlet valve. O2 was sparged into the solution through the dip tube to enhance the contact area with the solution. By using an equation developed by Rooney and Daniels,28 the initial 250 kPa of O2 pressure was equivalent to a dissolved O2 concentration of 2.45 mol/m3 in the MEA solution. This was the amount of O2 initially dissolved in the MEA solution, causing a drop in O2 pressure inside the reactor of approximately 5 kPa. As a result, an extra 5 kPa of O2 pressure was added to compensate for that dissolved in the solution in order to maintain a constant pressure. According to the ideal gas law, the 5 kPa of O2 pressure was approximately equal to an O2 concentration of 1.50 mol/m3. The amounts of O2 initially added and later compensated were therefore very close. The total pressure of the experiments involving O2 alone was a combination of the water vapor pressure and 250 kPa of O2 pressure (i.e., 250 kPa at 328 K and 450 kPa at 393 K). At predetermined intervals of time, samples of approximately 2.5-mL volume were withdrawn from the reactor through the liquid sampling valve into 5-mL sampling bottles. Extra O2 was quickly introduced into the reactor to compensate for the pressure loss during the sampling process. Cold water was run over the vials containing the samples to quickly cool the sample and quench the degradation reaction. These samples were kept in a refrigerator at 4 °C for less than 1 week to allow sufficient time for GC-MS, HPLCRID, and CE-DAD analyses. Our experiments showed that, under these conditions, further degradation of the samples was avoided. Oxidative degradation of MEA is an exothermic process. Therefore, a cooling water system was used to ensure and maintain the degradation process under isothermal conditions. 2.2.2. MEA/H2O/O2/CO2 Degradation System. Just as in the case of the MEA/H2O/O2 system, 400 mL of 5.03 kmol/m3 MEA solution was transferred into the reactor, and the head assembly was positioned to tightly seal the reactor. Prior to heating, a predetermined value of 250 kPa of CO2 was introduced into the solution through the gas inlet valve by opening the CO2 cylinder for 12 h, after which 4 mL of the solution was removed from the reactor through the liquid sampling valve to check for the CO2 loading by titration against standard solution of 1 kmol/m3 HCl, whereby CO2 was liberated and measured for its quantity by displacement of a NaCl/ NaHCO3/methyl orange mixture. The loading was calculated from the number of moles of CO2 per mole of MEA. The mixture in the reactor was heated to 328 or 393 K, and the CO2 loading was again determined. The total pressure of the reactor at this point was the sum of the pressures of water vapor and nondissolved CO2. Then, 250 kPa of O2, as set at the O2 cylinder, was additionally introduced into the solution through the gas inlet valve. The rest of the procedures were then

conducted following those for MEA/H2O/O2 experiments. As in the case of the O2-alone degradation experiments, the total pressure of the O2/CO2 experiment was the sum of the water vapor pressure and 250 kPa of O2 and CO2 vapor pressure (i.e., 1070 kPa at 393 K). The overall schematic of the oxidative degradation experiments and analysis and a sketch of the internals of the reactor are shown in Figure 1a and b, respectively. 2.3. Analysis of Degradation Products. 2.3.1. Gas Chromatography-Mass Spectrometry (GC-MS) Technique. The GC-MS instrument (model 6890/5073) was obtained from Hewlett-Packard Canada, Ltd., Montreal, Quebec, Canada. Chromatographic capillary columns of different stationary phases and polarities were used as variables for comparison of the separation of MEA and its degradation products. These columns were HP-Innowax (high-polarity), HP-35MS (intermediate-polarity), and HP-5MS (nonpolar) having cross-linked poly(ethylene glycol) and cross-linked 5% and 35% phenylmethylsiloxane, respectively, as stationary phases. All columns had the same dimension of 0.25-µm thickness × 0.25-mm i.d. × 30-m length and were obtained from Hewlett-Packard Canada, Ltd., Montreal, Quebec, Canada. The introduction of sample was done using an autosampler/autoinjector (model 7683, Hewlett-Packard Canada, Ltd., Montreal, Quebec, Canada). The autoinjector volume was reproducible to within 0.3% relative standard deviation (RSD) in terms of area percent of the peaks. Identical GC-MS operating conditions were applied to all columns. The criteria for GC-MS condition setup began at the injector inlet, where the tmperature had to be high enough to completely evaporate the liquid sample. The initial temperature of the oven was set at 100 °C to ensure the absence of condensation after the evaporated sample left the injection chamber. The ramp rate was 7 °C/min lower than that of our previous conditions12 to improve separation of the products. The final temperature was set at 240 °C, which was high enough to provide for complete elution. It was also set to about 10 °C lower than the limiting temperature of the columns used in this work, thus preventing the columns from being damaged. For a typical run, a 1-µL sample was injected at the GC inlet set at 250 °C using a split injection mode with a split ratio of 30:1. The GC oven was initially set at 100 °C and ramped to 240 °C at a rate of 7 °C/min. The temperature was kept at 240 °C for as long as 45 min to ensure complete elution of all degradation products. Ultra-high-purity- (UHP-) grade helium was used as the carrier gas and was regulated at a constant flow rate of 1 mL/min. The GC-MS interface, MS quadrupole, MS source, and EM voltage were kept at 250 °C, 150 °C, 230 °C, and 1858 V, respectively. MS scan mode was used with a mass range from 10 to 300. The products were identified by matching their mass spectra with commercial mass spectra of the National Institute of Standards and Technology (NIST) database (1998 version). Each sample was diluted using deionized distilled water in the ratio of 1:5 prior to the analysis to avoid column overload and improve separation of the components. The samples were analyzed twice to check for reproducibility. MEA measurements were done in terms of concentration calibrated with standard MEA, peak area, and area percent, whereas those of products were based on peak area and the corresponding area percent only and, thus, represented relative concentrations. This methodology was also applied to the analyses that were used for the HPLC and CE techniques. A matching technique that compared the mass spectra of the GC-separated components with the NIST database was used for the initial product identification. Verification of some of


Figure 1. Schematic of MEA oxidative degradation experiments and analysis: (a) Experimental setup and analysis, (b) sketch of the internals of the reactor.

the species was subsequently performed by comparing both the mass spectra and the GC retention time of commercially available pure standards with those of the initially identified components. In addition to MEA, some of the standards used included imidazole, formic acid, N-(2-hydroxyethyl)acetamide, 1-(2-hydroxyethyl)-2-imidazolidinone, and 18-crown-6. Figure 2a-c illustrates mass spectra of imidazole in an MEA/H2O/ O2-degraded sample, imidazole in the NIST database, and imidazole of a pure standard, respectively. The mass spectral pattern of the sample was found to be in excellent agreement with those of the NIST database and the pure standard, thus confirming the presence of imidazole. Figures 3a-c, 4a-c, and 5a-c are shown for the cases of formic acid, N-(2-hydroxyethyl)acetamide, and 1-(2-hydroxyethyl)-2-imidazolidinone, respectively. Figure 6a-c indicates the presence of 18-crown-6, although its sample mass spectrum gave a low-quality match to those of the NIST database and the standard verification. Although not perfectly matched, Figure 6a for this sample contains major peaks similar to those of the NIST database and standard shown in Figure 6b and c, respectively. This results from coelution of 18-crown-6 and other components (not known) from the GC, resulting in a mixed mass pattern. Standard injection was also used to verify that the GC-MS operating conditions did not cause further degradation, so that the products seen were actually a result of MEA oxidative degradation. Chromatograms of standard 4 kmol/m3 MEA and 1 kmol/m3 imidazole are given as examples in Figures 7 and 8, respectively, to confirm that their presence was not a result of thermal degradation under GC-MS conditions. These figures show that only the water (used as a solvent in this study) peak

Figure 2. Mass spectra of imidazole: (a) in a degraded MEA sample, (b) NIST database, (c) standard imidazole.

and those of these compounds were present, confirming that the GC-MS conditions were innocuous to the samples. 2.3.2. High-Performance Liquid Chromatography (HPLC) Technique. The HPLC instrument used for analysis of the liquid samples was equipped with a refractive index detector (RID)

2442


Figure 3. Mass spectra of formic acid: (a) in a degraded MEA sample, (b) NIST database, (c) standard formic acid.

Figure 5. Mass spectra of 1-(2-hydroxyethyl)-2-imidazolidinone: (a) in a degraded MEA sample, (b) NIST database, (c) standard 1-(2-hydroxyethyl)2-imidazolidinone.

Figure 4. Mass spectra of N-(2-hydroxyethyl)acetamide: (a) in a degraded MEA sample, (b) NIST database, (c) standard N-(2-hydroxyethyl)acetamide.

and an on-line degasser (model 1100/G1315B/G1322A, Agilent Technologies Canada, Mississauga, Ontario, Canada). Two types of columns, Nucleosil 100-5 SA containing a strong cationic exchanger of sulfonic acid (Macherey-Nagel, Du¨ren, Germany) of 250-mm length × 4.6-mm i.d. and Shodex YK-421 packed with a weak carboxyl-coated silica exchanger (Showa Denko, Tokyo, Japan) of 125 mm-length × 4.6-mm i.d., were selected. Two types of mobile phases were used. One was 0.05 kmol/m3 potassium dihydrogen phosphate solution (KH2PO4) adjusted to a pH of 2.6 by adding 85% w/w phosphoric acid (H3PO4) to obtain a substantial modification based on the work of Kaminski et al.24 to suit our present system. This is based on the fact that their work was developed for wastewater and amines used in desulfurization processes, whereas our application is specifically

Figure 6. Mass spectra of 18-crown-6: (a) in a degraded MEA sample, (b) NIST database, (c) standard 18-crown-6.

for the CO2 capture process in terms of the analysis of MEA and MEA oxidative degradation products. The other mobile phase had a combination of 0.005 kmol/m3 L-tartaric acid (C4H6O6), 0.001 kmol/m3 2,6-pyridinecarboxylic acid (C7H5NO4), and 0.024 kmol/m3 boric acid (H3BO3). Detection was aimed at MEA and those degradation products having the ability to acquire positive charges under acidic conditions. An autosampler (model G1313A, Agilent Technologies Canada, Mississauga, Ontario, Canada) was used for sample introduction. For a typical analysis, as much as 8 µL of sample was injected in order to ensure visualization of low-concentration products.


Figure 7. Chromatogram of standard 4 kmol/m3 MEA.

Figure 8. Chromatogram of standard 1 kmol/m3 imidazole.

The columns were kept at 30 °C. All analyses were done using a simple isocratic approach in which 100% of a single mobile phase flowing at a rate of 1 mL/min was used throughout the analysis. The RID optical unit was set at a temperature of 30 °C and operated under positive mode. Prior to the analysis, samples were diluted to 1: 40 with nanopure water, followed by filtration using a 0.20-µm nylon membrane filter. All mobile phases were degassed in an ultrasonic bath for at least 3 h and filtered through a 0.20-µm nylon membrane filter prior to use. All chemicals for mobile phases were reagent grade and were purchased from Sigma-Aldrich Canada, Mississauga, Ontario, Canada, except for l-tartaric acid, which was purchased from Fisher Scientific, Neopean, Ontario, Canada. 2.3.3. Capillary Electrophoresis (CE) Technique. A CE instrument equipped with a diode array detector (DAD; model HP 3D CE, Hewlett-Packard Canada, Ltd., Montreal, Quebec, Canada) was employed for detection of MEA and its basic and acidic degradation products. Selection of electrolyte solutions and CE conditions for MEA25 and basic and acidic products26,27 was based on suggestions in the literature, modified to suit the analysis of the present study. A mixture of 0.015 kmol/m3 imidazole (C3H4N2), 0.01 kmol/m3 hydroxyisobutyric acid (HIBA, C4H8O3), and 0.01 kmol/m3 18-crown-6 (C12H24O6) was used for MEA detection. A solution of 0.025 kmol/m3 KH2PO4 that was adjusted to a pH of 2.6 using 85% w/w of H3PO4 was used for the separation of basic degradation products, whereas 0.015 kmol/m3 sodium tetraborate decahydrate (Na2B4O7‚ 10H2O) was used for the separation of acidic degradation products. For all CE analyses, a bare-fused silica capillary column with dimensions of 75-µm i.d. × 805-mm length (720mm effective length, Agilent Technologies Canada, Mississauga, Ontario, Canada) was used. The capillaries were flushed before each analysis with 1 kmol/m3 of NaOH solution and deionized water for 30 min and then flushed with the electrolyte for 45 min. Between runs, the capillaries were flushed with the electrolyte for 5 min, and they were replenished after four runs. The capillaries were kept constant at 29 and 30 °C throughout the analysis for MEA and basic/acidic degradation products, respectively. Sample injections were carried out using hydrodynamic mode by applying 50 mbar pressure to the sample vial for 5 s. A voltage of +18 kV was applied after injection throughout the run. The DAD signal was set to a wavelength of 400 nm (reference at 214 nm) for indirect detection of MEA, whereas a wavelength of 200 nm was used for direct measurement of basic and acidic MEA degradation products. To ensure

Figure 9. GC-MS chromatograms of degraded MEA solution in the MEA/ H2O/O2 system (5.13 kmol/m3 initial MEA concentration, 393 K, 250 kPa O2 pressure, and 220-h reaction time): (a) HP-Innowax column (high polarity), (b) HP-35MS column (intermediate polarity), (c) HP-5MS column (nonpolar).

complete detection of all compounds, analysis times of 20, 35, and 45 min were used for MEA and basic and acidic degradation products, respectively. Samples of 1:4000 and 1:500 dilutions were used for MEA and basic/acidic product analysis. Nylon membrane filters of 0.20 µm were used to filter samples before analysis. The electrolyte solutions were degassed for at least 3 h and also filtered through a 0.20-µm nylon membrane prior to use. All chemicals for CE experiments were reagent grade and were purchased from Sigma-Aldrich Canada, Mississauga, Ontario, Canada, except for NaOH, which was obtained from Hewlett-Packard Canada, Ltd., Montreal, Quebec, Canada. 3. Results and Discussion 3.1. MEA Degradation at 393 K. 3.1.1. GC-MS Analysis. The numbers of observed degradation products from the two systems (i.e., MEA/H2O/O2 and MEA/H2O/O2/CO2) for the three different GC columns having widely different polarities were different. The columns were HP-Innowax (high-polarity), HP-35MS (intermediate-polarity), and HP-5MS (nonpolar). Figure 9a-c shows chromatograms obtained by HP-Innowax, HP-35MS, and HP-5MS, respectively for the MEA/H2O/O2 system, whereas Figure 10a-c represents the respective columns for the MEA/H2O/O2/CO2 system. Although not a primary goal of this study, the detected peaks were labeled, and their identification was attempted, as shown in Table 2. Although some of the commonly found products (e.g., formic and acetic acids, ammonia), as well as additional compounds (i.e., pyrimidine, acetamide, 2-methylaminoethanol, acetaldehyde, and ethanol) did not appear in the chromatograms (except for formic and acetic acids), they were detected, and most of them were verified in this work. These products either were not completely separated and so exhibited coelution with the water peak or MEA peak or existed in low concentrations. The primary products also reacted further with themselves, MEA, or O2, eventually resulting in the stable final degradation products labeled in Figures 9a-c and 10a-c, of which some have been

2444


Table 2. Summary of Degradation Products Suggested by the GC-MS Technique and Previously Reported in the Literature degradation products peak label A B C D E F G H I J K L M N O P Q R T V W X0 X1 X2

% confidence

compound

standard verification

Products Labeled in This Work 1-methylazetidine 58 D,L-homoserine lactone 86 imidazole 80 N-(2-hydroxyethyl)acetamide 83 N-methyl formamide 40 1,3-dioxane 46 2-ethyl-1H-imidazole 25 5-hydrazinocarbonyl-1H-imidazole 53 uracil 35 N-(2-hydroxyethyl)succinimide 72 1-amino-4-methylpiperazine 64 2-pyrolidinone 38 1-methyl-4-imidazole-5-carboxylic acid 50 N-methylene ethanamine 25 5-aminovaleric acid 59 D,L-aspartic acid 72 2-[(2-aminoethyl)amino] ethanol 92 ethylamine 43 4,5-dimethyloxazole 64 18-crown-6 47 ethylurea 38 N-glycylglycine 59 dimethylhydrazone-2-propanone 72 1-(2-hydroxyethyl)-2-imidazolidinone 86 ammonia formic acid acetic acid pyrimidine acetamide 2-methylaminoethanol acetaldehyde ethanol

Additional Products Found in This Workb 30 86 80 86 70 75 65 80

X X

ref(s)

7,a 20 7a

20

X

X

X X X X X

20, 30, 31 7, 13, 21 8, 21 8, 21 29 21

Some of the Previously Reported Products (Not Included in This Work) oxalic acid glycolic acid bicine

7, 21 7, 21 11

a Reported in terms of general amides. b These products did not appear in the chromatogram because of either incomplete coelution or low concentration, except for formic and acetic acids.

consistently reported in the literature.7,8,13,20,21,29-31 In addition, Table 2 also provides the percent identification confidence of the labeled peaks. Some were verified with standards, as indicated earlier. The presence of both imidazole and N-(2hydroxyethyl)acetamide was confirmed. As indicated earlier, 18-crown-6 was also confirmed, even though it had a low percent match, which was caused by incomplete separation from other products, resulting in mixed mass fragmentation patterns. Using a similar explanation, the remaining labeled peaks having low percent confidences, although not verified, might possibly be present in MEA oxidation systems. Certainly, for future work, verification has to be carried out to confirm their presence. For both degradation systems, the peaks were well separated, with good shapes and little tailing for the Innowax and HP35MS columns. On the other hand, the nonpolar column, HP5MS, produced peaks with significant overlap, poor shapes, and some tailing. This implies that most of the MEA degradation products are polar and, thus, were able to match the high- and intermediate-polarity stationary phases. The superior outcomes of the Innowax and HP-35MS columns having high and intermediate polarities, respectively, as compared to the nonpolar HP-5MS column, in terms of clean peak separation could also result in easy relative quantitative analysis of these degradation products as peak areas are mostly used to determine their concentrations. The distribution of degradation products was evaluated as a function of degradation time for the three columns by constructing degradation time curves for the labeled peaks shown in

Figures 9 and 10 for each column and degradation system. Figures 11a-c and 12a-c show plots of the distributions of degradation products against the degradation time corresponding to the peaks in Figures 9 and 10 for the MEA/H2O/O2 and MEA/ H2O/O2/CO2 systems, respectively. Because of the wide differences in the extents of formation of these degradation products, the peak areas were converted to logarithmic values so that the products eluted from the same GC column could be plotted clearly in the same graph. As a result, the y axis represents only arbitrary quantities on a logarithmic scale and, hence, serves only for qualitative evaluation of trends as a function of degradation time. Thus, only the trends of the formation of degradation products are intended for presentation in both Figures 11 and 12. This procedure was also later applied to the analysis of HPLC-RID and CE-DAD data. As illustrated in Figures 11a-c and 12a-c, the numbers of detected degradation products are in the order of HP-35MS (16) > HP-Innowax (12) ) HP-5MS (12) for the MEA/H2O/O2 system and HP35MS (16) > HP-Innowax (14) > HP-5MS (12) for the MEA/ H2O/O2/CO2 system. The numbers of detected intermediate products also observed were in the order HP-35MS (2; i.e., peaks W and P) . HP-Innowax (none) ) HP-5MS (none) for the MEA/H2O/O2 system and HP-35MS (3; i.e., peaks A, B, and N) > HP-Innowax (2; i.e., peaks A and B) ) HP-5MS (2; i.e., peaks A and N) for the MEA/H2O/O2/CO2 system. In both degradation systems, HP-35MS clearly revealed the most information on degradation products as well as intermediates. This is crucial for the development of degradation mechanisms


Figure 10. GC-MS chromatograms of degraded MEA solution in the MEA/ H2O/O2/CO2 system (5.03 kmol/m3 initial MEA concentration, 393 K, 250 kPa O2 pressure, 0.51 mol of CO2/mol of MEA, and 216-h degradation time): (a) HP-Innowax column (high polarity), (b) HP-35MS column (intermediate polarity), (c) HP-5MS column (nonpolar).

and is attributed to the intermediate polarity of its stationary phase being a compromise between the high-polarity and nonpolar phases of HP-Innowax and HP-5MS, respectively. This means that, if most of the products have widely different polarities, the optimum number of products that could be detected in a single column would be obtained by using a column with intermediate polarity (e.g., HP-35MS). Also, most of the products separated with HP-Innowax and HP-5MS could also be separated with HP-35MS. Despite the fact that HP-35MS was superior to the rest of the columns in terms of the total number of degradation products that could be clearly separated, one shortcoming was that the MEA peak separated with an abnormal shape and was also too close to the water peak, thereby causing overlap of the MEA peak with the water peak (note that the MEA peak is not shown in the figure). As a result, quantitative measurement of MEA became difficult. The use of HP-5MS results in a problem similar to that encountered with HP-35MS. In contrast, HP-Innowax separated MEA from water with an acceptable peak shape, even though some tailing was still exhibited, except in Figures 11a and 12a where a decline in quality was observed. The above results show that, in order to quantitatively analyze the oxidative degradation of MEA in the presence or absence of CO2, columns of high polarity and intermediate polarity (e.g., HP-Innowax and HP-35MS, respectively) have to be used in combination for analysis if both the MEA concentration and the degradation product concentrations need to be followed. On the other hand, if only the degradation products are of interest, then a column with intermediate polarity (e.g., HP-35MS) is adequate. A nonpolar column (e.g., HP-5MS) is unsuitable in all cases. Attempts were made to characterize and compare the oxidative degradation of MEA in the presence and in the absence of CO2 by using information extracted from the product distribution

Figure 11. GC-MS degradation product distribution of the MEA/H2O/O2 system at 393 K: (a) HP-Innowax column (high polarity), (b) HP-35MS column (intermediate polarity), (c) HP-5MS column (nonpolar).

obtained from the above results. Figures 11b and Figure 12b were used for this comparison because of the excellent performance of the HP-35MS column as described earlier. In Figure 11b, for the O2-alone environment, apart from two intermediate products (W and P) that disappeared, six other

2446


Figure 13. HPLC-RID chromatograms using the Nucleosil column at 393 K and a mobile phase of: (a) 0.05 kmol/m3 potassium dihydrogen phosphate, MEA/H2O/O2 system; (b) 0.05 kmol/m3 potassium dihydrogen phosphate, MEA/H2O/O2/CO2 system; (c) mixture of 0.005 kmol/m3 L-tartaric acid, 0.001 kmol/m3 2,6-pyridinecarboxylic acid, and 0.024 kmol/m3 boric acid, fresh MEA solution.

Figure 12. GC-MS degradation product distribution of the MEA/H2O/O2/ CO2 system at 393 K: (a) HP-Innowax column (high polarity), (b) HP35MS column (intermediate polarity), (c) HP-5MS column (nonpolar).

products (A, E, G, F, R, and V) seemed to stabilize as final products after reaching certain reaction times (about 300 h of degradation time). The concentrations of the remaining eight products (B, C, D, H, I, L, M, and X2) continued to increase even after 300 h of degradation time. This indicates that the increasing concentrations of these eight products can be attributed to the decline in the concentrations of W and P. On

the other hand, the concentration of MEA still decreased even beyond 300 h of degradation time, showing that it was also responsible for the continued increase of the concentrations of B, C, D, H, I, L, M, and X2. Therefore, these products could possibly be formed by reactions of MEA with O2, MEA with intermediates, O2 with intermediates, and intermediates with intermediates. For the O2/CO2 system, shown in Figure 12b, a total of eight components (D, E, F, I, J, M, O, and P) were sufficiently stable as final products, whereas only five (K, L, Q, X1, and X2) of the compounds continued to increase even after 450 h of degradation time. MEA decreased up to a degradation time of 300 h, beyond which it stabilized with degradation time. These results show that, for the O2/CO2 system, MEA was not responsible for the continued increase of the concentrations of K, L, Q, X1, and X2, so their presence could arise only from O2-intermediates and intermediatesintermediates reactions. The presence of CO2 seemed to induce the formation of stable products and hinder further transformation, thus minimizing the chances of MEA to further react with those products and thereby reducing the extent of degradation. This could also explain why the decrease in MEA concentration with degradation time was less steep in the O2/CO2 degradation system than that in the O2-alone system, as shown in Figures 11a and 12a. 3.1.2. HPLC-RID Analysis. The initial trials were to evaluate the compatibility of the Nucleosil and Shodex YK-421 columns with the two mobile phases used for this study (potassium dihydrogen phosphate and a mixture of l-tartaric acid, 2,6pyridinecarboxylic acid, and boric acid). Degraded MEA/H2O/ O2 and MEA/H2O/O2/CO2 samples as well as fresh aqueous MEA solutions were chosen for the tests. Figure 13a and b


Figure 14. HPLC-RID chromatograms of the MEA/H2O/O2 system at 393 K obtained using Shodex YK-421 and a mobile phase of: (a) 0.05 kmol/ m3 potassium dihydrogen phosphate; (b) mixture of 0.005 kmol/m3 L-tartaric acid, 0.001 kmol/m3 2,6-pyridinecarboxylic acid, and 0.024 kmol/m3 boric acid.

shows the chromatograms for MEA/H2O/O2 and MEA/H2O/ O2/CO2, respectively, obtained on the Nucleosil column with potassium dihydrogen phosphate as the mobile phase. These chromatograms show that the buffer produced an excellent separation of MEA and its degradation products for both systems. The total analysis time was about 35 min. Most products eluted within approximately 10 min after the MEA peak. A mixture of l-tartaric acid, 2,6-pyridinecarboxylic acid, and boric acid was also used with the same column for which, even with fresh solution, MEA was eluted with a poor peak shape and excessive tailing, as shown in Figure 13c. This indicates the incompatibility of the column with the mobile phase. We therefore decided not to carry out any further runs of degraded MEA samples with this arrangement. Figure 14a presents the chromatogram for the MEA/H2O/O2 system produced using Shodex YK-421 with potassium dihydrogen phosphate as the mobile phase. Although the phosphate buffer could separate MEA from most of the products within a short time, the overlap of MEA with its impurity caused an error in MEA quantification. The separation of the products eluting after MEA was also incomplete. A mixture of l-tartaric acid, 2,6-pyridinecarboxylic acid, and boric acid was also run with this same column. The results for this eluent are shown in Figure 14b, which indicates that the separation of degradation products was improved but still suffered from a minor overlap of a couple of compounds. Coelution of MEA with one of the degradation products labeled as product* also occurred. Attempts were also made to improve separation on the Shodex YK-421 column, by reducing the flow rates of both mobile phases from 1 to 0.5 mL/min. It was observed that a decrease in the flow rate seemed to shift all components almost equally to the right without improving separation efficiency. Analysis time also increased, especially for the three-component mobile phase, for which it was almost as long as the runs for the Nucleosil column with phosphate buffer. Although the length of column might affect separation performance, an increase of column length for Shodex YK-421 might not be economical, as the cost of a 125-mm length of Shodex YK-421 is already approximately 4 times that of the Nucleosil column. As a result, the Nucleosil column with a phosphate mobile phase was selected for further tests on the distribution of degradation products with degradation time.

Figure 15. HPLC-RID degradation product distribution at 393 K obtained using the Nucleosil column and phosphate mobile phase: (a) MEA/H2O/ O2, (b) MEA/H2O/O2/CO2.

For the accurate analysis of the distribution of degradation products, evaluation was made only with the well-separated components labeled in Figure 15a and b. These plots show the distributions of degradation products versus degradation time for the MEA/H2O/O2 and MEA/H2O/O2/CO2 systems, respectively. The plots correspond to peaks shown in Figure 13a and b. In Figure 15a, two intermediate products labeled C1 and E1 were clearly observed, as their concentrations increased to maxima before declining thereafter. Two intermediates labeled B2 and G2 were also observed to reach maxima and then decline for the O2/CO2 system, as shown in Figure 15b. Also, only product B1 in the O2-alone environment increased to a maximum and tended to stabilize. On the other hand, in the O2/CO2 reaction system, only two of the products (labeled A2 and C2) tended to stabilize. The remaining detected products for both systems are those whose concentrations still increased even after 450 h. In the O2 system, MEA continued to decrease, showing that MEA and the intermediates contributed to the increasing concentration of products with time. On the other hand, in the O2/CO2 system, MEA stabilized, showing that it was the intermediates that contributed to the increasing concentration of some of the products with degradation time. These outcomes were consistent with those obtained by GC-MS

2448


Figure 17. CE chromatogram of fresh MEA solution using a mixture of imidazole, 18-crown-6, and HIBA as the electrolyte.

Figure 16. Identification using standard imidazole: (a) MEA/H2O/O2, (b) MEA/H2O/O2/CO2.

analysis in which CO2 was observed to influence the formation of stable products. This also suggests the validity of both techniques for degradation analysis of MEA and O2 in the presence or absence of CO2. Although this was not a primary objective, attempts were made to identify the products using the spiking technique. Only three standards (imidazole, 2-methylaminoethanol, and 18crown-6) were used. A typical example of product identification for the MEA/H2O/O2 system using standard imidazole is shown in Figure 16a. Chromatograms of pure standard imidazole, 220h-degraded MEA solution, and 220-h-degraded MEA solution spiked with standard imidazole have been superimposed for ease of comparison. Not only was the standard imidazole eluted at about the same location as peak F1, but the peak also increased in size in the spiked sample. Therefore, peak F1 could be identified as imidazole. Figure 16b shows a similar result for the MEA/H2O/O2/CO2 system in which peak F2 was identified as imidazole. This outcome is consistent with the GC-MS analysis in which imidazole was also detected and identified. Peaks C1 and C2 were matched with standard 2-methylaminoethanol using the same approach. 18-Crown-6 was not detected by the ion-exchange technique because of its inability to acquire a charge. Verification by mass spectrometry is suggested to confirm the presence of these degradation products. 3.1.3. CE-DAD Analysis. Three types of electrolytes were selected mainly to target the separation of MEA, basic degradation products, and acidic degradation products. Figure 17 shows the chromatogram of fresh MEA solution obtained using a mixture of imidazole, 18-crown-6, and HIBA as the electrolyte. Although a well-separated, symmetric MEA peak could be obtained, sample preparation was tedious and time-consuming and, thus, made the method unattractive from our perspective. MEA samples had to be diluted up to 4000 times as compared to 5 and 40 times for the GC and HPLC techniques, respectively. Dilution serves to maintain MEA linearity and also prevents column overload. The dilution causes an inconvenience if used in the CO2 capture process where the MEA concentration can be as high as 5 kmol/m3. Although the technique was quite

Figure 18. Typical CE chromatograms for analysis of basic products of degraded MEA solution after approximately 200 h: (a) MEA/H2O/O2, (b) MEA/H2O/O2/CO2.

simple, high dilution imposes an analytical difficulty that can possibly cause quantitative error. Despite the incompatibility with MEA quantification, CE was capable of detecting some degradation products using KH2PO4 and Na2B4O7‚10H2O as electrolytes. Phosphate solution was employed for the analysis of basic products capable of being protonated at a pH of 2.6. Figure 18a and b shows typical chromatograms of degraded MEA solution after approximately 200 h of degradation time in the O2 and O2/CO2 systems, respectively. The number of detected products in the O2-alone system was approximately twice as high as that in the O2/CO2 system, confirming the higher severity of degradation in the former system. On the other hand, the borate electrolyte was used for the analysis of acidic compounds for which ionization could take place under alkaline conditions. Figure 19a and b shows chromatograms of degraded MEA samples obtained after approximately 400 h of degradation time in the absence and presence of CO2, respectively. MEA/H2O/O2 again generated more acidic products than MEA/H2O/O2/CO2. Trends from Figure 20a and b were matched with corresponding peaks in Figure 18a and b to check the product distribution. Approximately, the concentrations of 40% of the total basic products detected in the MEA/H2O/O2/CO2 system continued to increase as a function of time up to 150 h of degradation time but then leveled off. In the case of the MEA/ H2O/O2 system, the concentrations of only 23% behaved in a similar way. A similar trend was also exhibited in the case of acidic products, as shown in Figure 21a and b, plotted


Figure 19. Typical CE chromatograms for analysis of acidic products of degraded MEA solution after approximately 400 h: (a) MEA/H2O/O2, (b) MEA/H2O/O2/CO2.

Figure 21. CE-DAD acidic degradation product distribution: (a) MEA/ H2O/O2, (b) MEA/H2O/O2/CO2.

Figure 20. CE-DAD basic degradation product distribution: (a) MEA/ H2O/O2, (b) MEA/H2O/O2/CO2.

corresponding to the labeled peaks in Figure 19a and b. Again, these results were consistent with those of the GC and HPLC

techniques, for which the presence of CO2 was found to result in a higher number of stable compounds, thus limiting further MEA degradation. Identification of products was done similarly to the case for the HPLC-RID analysis. Imidazole could be identified for the MEA/H2O/O2 system and is labeled as peak B3 in Figure 18a. MEA/H2O/O2/CO2 also produced imidazole, but its concentration was too small to be recognized by the technique used, so it is not labeled in Figure 18b. However, its peak could be observed to be located as the second peak from the left of peak A4. 3.2. MEA Degradation at 328 K. The degradation of the MEA/H2O/O2 system was conducted at the absorption temperature of 328 K and was also evaluated by GC-MS and HPLCRID. In terms of the degradation products, a product labeled “I” in a previous GC-MS analysis at 393 K was detected only after 500 h of degradation time using the GC-MS technique. On the other hand, no degradation product was observed using HPLC-RID technique. The quality of the MEA peak as a function of degradation time was also determined and is plotted in Figure 22. To evaluate the HPLC-RID and GC-MS techniques for data consistency, the MEA measurement made using the GC-MS technique with the HP-Innowax column is also included in the same figure as the HPLC-RID results. No significant change in MEA concentration was observed by the two techniques. These results confirm the much lower level of

2450


analysis time for component separation was in the order of GCMS < CE-DAD for basic products < HPLC-RID < CE-DAD for acidic products. In terms of sample and mobile phase preparation, GC-MS was the least tedious technique, requiring the lowest dilution not requiring filtration. Preparation of a mobile phase was not needed. On the other hand, CE required the highest dilution, which was as high as 4000 times for MEA analysis and 500 times for analysis of basic and acidic degradation products, as compared to only 40 times for HPLC-RID. Prior to the analysis, both the HPLC and CE analytical techniques needed sample filtration and multistep mobile phase preparation such as pH adjustment and degassing. A few advantages of the CE-DAD technique could be identified. The column cost of CE-DAD was approximately $60, whereas the GC and HPLC columns were about 12 and 6 times more expensive, respectively. If compared to the HPLC-RID technique, CE-DAD consumed much less sample and mobile phase, which could lead to a higher cost saving. Even though each technique showed advantages and disadvantages with respect to each other, a common ground still existed. For MEA oxidative degradation in the absence or presence of CO2 characterized by GC-MS, HPLC-RID, and CEDAD, similar trends in regard to product distribution, decline of MEA, and role played by CO2 were observed. This consistency indicates the validity of the three techniques, showing their capabilities as analytical tools for characterizing MEA oxidative degradation. In addition, each of the techniques could be used to confirm the outcomes from the other techniques.

Figure 22. MEA concentration as a function of degradation time at 328 K measured using the GC-MS and HPLC-RID techniques.

degradation in the absorber region of the CO2 capture plant. Thus, temperature plays a significant role and can be used to determine whether O2-induced degradation will be significant in the absorber. 3.3. Comparison of GC-MS, HPLC-RID, and CE-DAD. The best conditions for each technique were picked for comparison. Table 3 presents a summary of this comparison of the three techniques for analysis of MEA oxidative degradation. GC-MS provided the most sensitive detection by detecting the greatest number of products (16 components) for both systems studied. Perhaps, this can be attributed to the fact that GC-MS required no component manipulation during the analysis. For GC-MS, vaporization is the major characteristic needed of the sample. Thus, the components in the sample are likely to enter the analytical unit as originally present in the liquid samples. Unlike the GC-MS technique, in the HPLC and CE techniques, the components must be converted into detectable forms either by protonation or by ionization at a certain pH prior to the analysis. Therefore, compounds to be detected by these techniques were limited to those favoring protonation or ionization. HPLC-RID, however, was capable of capturing the most degradation intermediates, calculated to be 29% for both degradation systems, whereas CE-DAD performed equally for basic intermediate products and GC-MS performed equally for all intermediates. Acidic products analyzed by CE-DAD showed no reaction intermediates for both degradation systems. HPLCRID was the only technique capable of simultaneously detecting MEA and MEA degradation products within a single analysis. GC-MS and CE-DAD needed an additional column and an additional electrolyte, respectively, for MEA detection. The

4. Industrial Applications Applications of the analytical techniques used in this study can be envisioned for industry. Information extracted from GCMS could guide the analysis of the CO2 capture process for solvent quality in terms of chemical stability by selection of appropriate GC columns to match the required analysis (i.e., MEA, degradation products, MEA + degradation products). HPLC-RID with the tested conditions could also be applied because it showed the capability of detecting MEA and some degradation products in a single analysis. CE-DAD, although it required heavy sample dilution, was simple enough for analysis of the degradation products. The GC-MS, HPLC-RID, and CE-DAD techniques, if combined, could be used to verify results obtained from one another. This could be advantageous compared to the existing analytical techniques used in real applications, for which a single technique is often used.

Table 3. Comparative Details of GC-MS, HPLC-RID, and CE-DAD Techniques for Analysis of MEA Oxidative Degradation GC-MS parameter compared column product type mobile phase degradation products total number of products detected intermediates revealed MEA detectionb sample preparation dilution level filtration mobile phase preparation analysis time column costc

system Aa

system Ba

HP-35MS neutral helium

HPLC-RID system Aa

system Ba

Nucleosil 100-5 SA protonated KH2PO4, pH 2.6

CE-DAD system Aa

system Ba

system Aa

system Ba

standard bare fused silica basic (protonated) acidic (ionized) KH2PO4, pH 2.6 Na2B4O7‚10H2O

16 13% no

16 19% no

7 29% yes

7 29% yes

12 17% no

7 14% no

8 0 no

3 0 no

1 in 5 no

1 in 5 no

1 in 40 yes

1 in 40 yes

1 in 500 yes

1 in 500 yes

1 in 500 yes

1 in 500 yes

no 20 min ∼$700

yes 35 min ∼$360

yes 25 min ∼$60

yes 45 min ∼$60

a Degradation system A ) MEA/H O/O ; degradation system B ) MEA/H O/O /CO . b MEA detection by GC-MS could be done on HP-Innowax. MEA 2 2 2 2 2 could also be detected by CE-DAD with mobile phase of imidazole, 18-crown-6, and HIBA. c Amounts are in Canadian dollars.


5. Conclusions (1) A GC column of intermediate polarity (i.e., HP-35MS) performed the best for analysis of degradation products in the MEA/H2O/O2 and MEA/H2O/O2/CO2 degradation systems in terms of component separation and numbers of total products and intermediates detected. This column was found to be adequate if analysis of only the degradation products was required. A high-polarity column (HP-Innowax) had to be used in combination if information on the MEA decline was needed. The nonpolar HP-5MS column was unsuitable in all cases. (2) The HPLC-RID technique using a combination of the cation-exchanger Nucleosil 100-5 SA column and a 0.05 kmol/ m3 KH2PO4 mobile phase was superior to the rest of the systems studied. MEA and its degradation products could be simultaneously analyzed in this single column in a single run. (3) Because of the high concentration of MEA used in the CO2 capture process, CE-DAD using an electrolyte consisting of a mixture of imidazole, 18-crown-6, and HIBA that required heavy dilution was not suitable for MEA analysis. However, some basic and acidic products could be analyzed through the use of KH2PO4 and Na2B4O7‚10H2O electrolytes, respectively. (4) In addition to GC-MS, the spiking technique using standards could be used as a method for identifying the degradation products separated by HPLC-RID and CE-DAD. Imidazole was detected and identified by all techniques, thus confirming its presence. (5) GC-MS was found to be the most sensitive technique, detecting the greatest number of products within the least time. In addition, it required the least work in terms of sample preparation. (6) The presence of CO2 was found to induce more stable products. Therefore, further degradation by reaction with MEA was reduced. Thus, the rate of MEA degradation was seen to be lower than that in the system including O2 alone. (7) Consistency was obtained in all techniques in terms of product distribution, decline of MEA, and role played by CO2, thus indicating their validity. Confirmation of the results could be accomplished through the use of a combination of GC-MS, HPLC-RID, and CE-DAD. Acknowledgment The authors acknowledge the financial support provided by The Royal Golden Jubilee Ph.D. program (Grant PHD/0034/ 2546); the Natural Science and Engineering Research Council of Canada (NSERC); and CANMET Energy Research Centre, Ottawa, Canada. Literature Cited (1) Kohl, A. L.; Reisenfeld, F. C. Gas Purification, 4th ed.; Gulf Publishing Co.: Houston, TX, 1997. (2) Hendricks, C. Carbon Dioxide RemoVal from Coal-Fired Power Plants; Kluwer Academic Publishers: Dordrecht, The Netherlands, 1994. (3) Astarita, G.; Savage, D. W.; Bisio, A. Gas Treating with Chemical SolVents; John Wiley & Sons: New York, 1983. (4) Chakma, A.; Mehrotra, A. K.; Nielson, B. Comparison of Chemical Solvents for Migrating CO2 Emissions from Coal-Fired Power Plants. Heat RecoVery Syst. CHP 1995, 15 (2), 231. (5) Blake, R. J. How Acid-Gas Treating Processes Compare. Oil Gas J. 1967, (January), 105. (6) Rooney, P. C.; Bacon, T.; Dupart, M. S. Part 2: Effect of Heat Stable Salts on MDEA Solution Corrosivity. Hydrocarbon Process. 1997, (April), 65. (7) Hofmeyer, B. G.; Scholten, H. G.; Lloyd, W. G. Contamination and Corrosion in Monoethanolamine Gas Treating Solutions; Internal Report No. 722; The Dow Chemical Company: Midland, MI, 1965.

(8) Blanc, C.; Grall, M.; Demarais, G. The Part Played by Degradation Products in the Corrosion of Gas Sweetening Plants Using DEA and MDEA. In Proceedings of the 32nd Annual Laurance Reid Gas Conditioning Conference; Oklahoma University: Normal, OK, 1982. (9) Rooney, P. C.; Dupart, M. S. Corrosion in Alkanolamine Plants: Causes and Minimization. In Corrosion 2000: NACE International Annual Conference and Exposition; Orlando, FL, 2000; Paper 494. (10) Veldman, R. R. Alkanolamine Solution Corrosion Mechanisms and Inhibition from Heat Stable Salts and CO2. In Corrosion 2000: NACE International Annual Conference and Exposition; Orlando, FL, 2000; Paper 496. (11) Howard, M.; Sargent, A. Operating Experiences at Duke Energy Field Services Wilcox Plant with Oxygen Contamination and Amine Degradation. In Proceedings of the 51st Annual Laurance Reid Gas Conditioning Conference; Oklahoma University: Normal, OK, 2001. (12) Supap, T.; Idem, R.; Veawab, A.; Aroonwilas, A.; Tontiwachwuthikul, P.; Chakma, A.; Kybett, B. D. Kinetics of the Oxidative Degradation of Aqueous Monoethanolamine in Flue Gas Treating Unit. Ind. Eng. Chem. Res. 2001, 40, 3445. (13) Chi, S.; Rochelle, G. T. Oxidative Degradation of Monoethanolamine. Ind. Eng. Chem. Res. 2002, 41, 4178. (14) Goff, G. S.; Rochelle, G. T. Monoethanolamine Degradation: O2 Mass Transfer Effects under CO2 Capture Conditions. Ind. Eng. Chem. Res. 2004, 43, 6400. (15) Brydia, L. E.; Persinger, H. E. Quantitative Gas Chromatographic Determination of Ethanolamines as Trifluoroacetyl Derivatives. Anal. Chem. 1967, 39, 1318. (16) Piekos, R.; Kobylczyk, K.; Grzybowski, J. Quantitative Gas Chromatographic Determination of Ethanolamine as Trimethylsilyl Derivatives. Anal. Chem. 1975, 47, 1157. (17) Kennard, M. L.; Meisen, A. Gas Chromatographic Technique for Analyzing Partially Degraded Diethanolamine Solutions. J. Chromatogr. 1983, 267, 373. (18) Dawodu, O. F.; Meisen, A. Gas Chromatographic Analysis of Alkanolamine Solutions Tenax-GC Using Capillary and Packed Columns. J. Chromatogr. 1993, 629, 297. (19) Dawodu, O. F.; Meisen, A. Identification of Degradation Products Resulting from Carbonyl Sulphide-Induced Degradation of Diethanolamine. J. Chromatogr. 1991, 587, 237. (20) Strazisar, B. R.; Anderson, R. R.; White, C. M. Degradation Pathways for Monoethanolamine in a CO2 Capture Facility. Energy Fuels 2003, 17, 1034. (21) Rooney, P. C.; Dupart, M. S.; Bacon, T. R. Oxygen’s Role in Alkanolamine Degradation. Hydrocarbon Process. 1998, (July), 109. (22) Kadnar, R.; Rieder, J. Determination of Anion in Amine Solutions for Sour Gas Treatment. J. Chromatogr. A 1995, 706, 339. (23) Kadnar, R. Determination of Amines used in the Oil and Gas Industry (Upstream Section) by Ion Chromatography. J. Chromatogr. A 1999, 850, 289. (24) Kaminski, M.; Jastrzebski, D.; Przyjazny, A.; Kartanowicz, R. Determination of the Amount of Wash Amines and Ammonium Ion in Desulfurization Products of Process Gases and Results of Related Studies. J. Chromatogr. A 2002, 947, 217. (25) Pereira, E. A.; Tavares, M. F. M. Determination of Volatile Corrosion Inhibitors by Capillary Electrophoresis. J. Chromatogr. A 2004, 1051, 303. (26) Altria, K. D.; Frake, P.; Gill, I.; Hadgett, T.; Kelly, M. A.; Rudd, D. R. Validated Capillary Electrophoresis Method for the Assay of a Range of Basic Drugs. J. Pharm. Biomed. Anal. 1995, 13, 951. (27) Altria, K. D.; Bryant, S. M.; Hadgett, T. A. Validated Capillary Electrophoresis Method for the Analysis of a Range of Acidic Drugs and Excipients. J. Pharm. Biomed. Anal. 1997, 15, 1091. (28) Rooney, P. C.; Daniels, D. D. Oxygen Solubility in Various Alkanolamine/ Water Mixtures. Pet. Technol. Q. 1998, 3 (1), 97. (29) Bello, A.; Idem, R. O. Pathways for the Formation of Products of the Oxidative Degradation of CO2 Loaded Concentrated Aqueous Monoethanolamine Solutions During CO2 Absorption from Flue Gases. Ind. Eng. Chem. Res. 2005, 44, 945. (30) Polderman, L. D.; Dillon, C. P.; Steele, A. B. Why MEA Solution Breaks Down in Gas Treating Service. Oil Gas J. 1956, 54, 180. (31) Talzi, V. P.; Ignashin, S. V. NMR Study of Decomposition of Monoethanolamine under Conditions of Industrial Gas Treatment. Russ. J. Appl. Chem. 2002, 75 (1), 80.

ReceiVed for reView May 13, 2005 ReVised manuscript receiVed September 9, 2005 Accepted September 13, 2005 IE050559D

Analysis of Monoethanolamine and Its Oxidative Degradation

Recommend Documents