A Gas Chromatography Method for Simultaneous Quantification of Inorganic Gases and Light Hydrocarbons Generated in Thermochemical Processes

This paper reports a method for simultaneous determination of H2, O2, N2, CO, CO2, CH4, C2H2, C2H4, C2H6, C3H4 (propadiene and propyne), C3H6, C3H8 and C4H10 (n-butane and iso-butane) by gas chromatography using thermal conductivity and flame ionization detectors. A single porous layer open tubular column (0.53 mm internal diameter × 30 m length × 30 μm thick) was applied and no catalytic converter was needed to convert CO and CO2 into CH4 to enable identification by a flame ionization detector. The most appropriate chromatographic conditions were defined for the method and it was validated according to the recommendations of the National Health Surveillance Agency and the National Institute of Metrology, Standardization and Industrial Quality. Chromatographic conditions defined for the target gases presented satisfactory linearity (r > 0.99), and limits of detection ranged between 0.0916 and 2.75 ppm. High accuracy (98-101%) obtained for the gas chromatography/thermal conductivity detector/flame ionization detector method associated to low relative standard deviation (< 2%) confirmed its applicability in routine quantification of target gases formed during the pyrolysis of municipal refuse-derived fuel.


Introduction
Thermochemical processes (pyrolysis and gasification) constitute alternatives to minimize and convert refusederived fuel (RDF) from municipal solid waste (MSW, also known as municipal refuse-derived fuel (MRDF)) into energy. Inorganic gaseous compounds (CO, CO 2 , H 2 , N 2 , O 2 , etc.) and light hydrocarbons (C 1 -C 6 ) are generated as products of these processes. 1,2 Some of these gases are combustible (CO, H 2 and C 1 -C 6 ) and may be used for energy production, thus adding value to the application of thermochemical processes.
The composition and proportion of gases formed during thermochemical processes vary according to the type of waste, reactor, and operational conditions, such as temperature and heating rate. [3][4][5] As some of these gases may be used as alternative energy sources, it is critical to develop methodologies to enable simultaneous characterization and quantification of all gases formed during thermochemical treatment. Table 1 summarizes chromatographic conditions applied in standard methods (ASTM D3612-02, methods A and C) 6 and by Supelco 7 to quantify inorganic gases (H 2 , O 2 , N 2 , CO and CO 2 ) and light hydrocarbons (C 1 -C 4 ) by thermal conductivity (TCD) and flame ionization (FID) detectors, respectively. 6,7 ASTM D3612-02 (method A) requires two columns connected in series (a molecular sieve and a Poparak N columns) to separate and identify the inorganic gases and light hydrocarbons. Besides, a catalytic converter (methanizer) is needed to convert CO and CO 2 into CH 4 for detection by FID under acceptable sensitivity using argon as carrier gas. Other limitations of this method are: (i) light hydrocarbons propane and propylene are not separated under the furnished conditions; (ii) C 3 H 4 (propadiene and propyne) are not targeted by this method, and (iii) it enables the identification of butane only (it is not clear if is n-or iso-butane). 6 The second standard test method is ASTM D3612-02 (method C) which enables the analysis of these target gases by also employing two columns connected in series (a molecular sieve and a porous layer open tubular (PLOT) column (Carboxen-1006)), a methanizer and argon as a carrier gas. However, light hydrocarbons (C 3 H 4 (propyne and propadiene), C 3 H 6 and C 4 H 10 (n-and iso-butane)) are not evaluated in this method. 6 On the other hand, a method using a single column (Carboxen-1010 PLOT or Carboxen-1006 PLOT) and argon or helium as carrier gases were proposed by Supelco. 7 The proposed method applies Carboxen-1010 PLOT column, argon as carrier gas, FID and TCD detectors and a methanizer, yet no hydrocarbons containing 3 or 4 carbon atoms were evaluated. Although it is possible to analyze C 3 H 4 (propyne), C 3 H 6 , C 3 H 8 and C 4 H 10 (n-butane) by the method proposed by Supelco, light hydrocarbons such as C 3 H 4 (propadiene) and C 4 H 10 (iso-butane) as well as inorganic gases (H 2 and O 2 ) were not evaluated by this method. Furthermore, the method does not present a complete and effective separation of CO and N 2 analytes. In addition, no details regarding method validation such as: linear range, linearity, repeatability (intra-day and interday studies), limits of detection (LOD) and quantification (LOQ) were presented in Supelco studies. 7 These evidences demonstrate that it is critical to perform more studies involving the use of Carboxen-1010 PLOT, which is a more efficient column for the separation of inorganic gases and light hydrocarbons up to 3 carbon atoms. The use of helium rather than argon as carrier gas must also be evaluated as it shows better performance for TCD due to higher thermal conductivity and response factor. 8 The combination of these two factors may lead to the development of a single method for separation, identification and quantification of inorganic gases and light hydrocarbons.
This work proposes a new chromatographic method which covers a broader scope of analytes when compared to ASTM D3612-02 (methods A and B) 6 and Supelco 7 methods. The new method was developed and validated aiming the simultaneous quantification of inorganic gases (CO, CO 2 , H 2 , N 2 , O 2 ) and light hydrocarbons (C 1 -C 3 and C 4 H 10 (n-butane and iso-butane)) by using a single column (Carboxen-1010 PLOT) helium as carrier gas, and detection via TCD and FID without the need for a methanizer. Besides, the procedure was also applied for the identification of inorganic gases and light hydrocarbons generated during the pyrolysis of real MRDF.

MRDF sample
MRDF (with 15 wt.% moisture content) was produced in an industrial solid waste processing line (SWPL) as detailed previously by Infiesta et al. 9 by using MSW generated in the city of Boa Esperança, Minas Gerais, Brazil. MSW is pretreated by mechanical processes such as selection, crushing and drying in this SWPL. The mass balance of the SWPL (4148 kWh), lower heating values (LHV) of the MSW (9.3 MJ kg −1 ) and MRDF (15.8 MJ kg −1 ), as well as average characterization of the MRDF produced from MSW were all determined in the previous study. 9

Chromatographic conditions
Chromatographic analyses were performed by using a gas chromatographer (GC) (Shimadzu GC-2014, Kyoto, Japan) equipped with TCD and FID detectors, respectively, which were operated in series. Data were processed using the GC-Solution software. A Carboxen 1010 PLOT column (0.53 mm internal diameter × 30 m long × 30 µm thick) was used as stationary phase.

Method validation
Analytical curves and linearity Method linearity was evaluated by developing calibration curves with data obtained from the injection of 5-10 different concentrations (ranged from 0.0916 to 274 ppm, and prepared in triplicate) of each analyte. [10][11][12][13] Tedlar bags of polyprolyene (1 L, CEL Scientific Corporation, Cerritos, USA) were used to transfer the sample of each isolated analyte from the cylinders to the atmospheric pressure. Dilution of gases was performed by adding argon gas to each analyte directly by using a suitable microsyringe (fixed needle, Teflon tip and capacity of 1000 µL) for collecting gas samples (Hamilton Gastight 1001, Nevada, USA).
The concentration of each analyte was calculated considering the volume of gas in the temperature of 0 °C to facilitate the comparison with the results obtained with ASTM D3612-02 (methods A and C). 6

Selectivity
Method selectivity was calculated considering the resolution (R s ) between the different target compounds by using retention times, and base width of the peak for each compound (equation 1): (1) where t rA : retention time of compound A; t rB : retention time of compound B; wA: base width of peak A; wB: base width of peak B.

LOD and LOQ
LOD and LOQ were calculated for each target compound by using the signal-to-noise ratio method (LOD = 3:1 and LOQ = 10:1, signal-to-noise ratio, respectively). 6,14,15 Precision Both intra-day and inter-day precision were assessed for the mixture of different concentrations of analytes prepared in triplicates in three different concentrations. Three separate bags were prepared with the mixture for the three evaluated concentrations (low, medium and high), and the linear range obtained for each compound was checked as shown in Table S1 (Supplementary Information (SI) section). Subsequently, the mixture of each bag was injected only once for each concentration under analysis.
For the evaluation of intra-day precision (repeatability), samples were injected in the GC/TCD/FID four times within intervals of 2 h (1, 3, 5 and 7 h). Inter-day precision (reproducibility) was evaluated by injecting sample in over 5 different days (1, 3, 7, 15 and 30 days). The relative standard deviations (RSD, in percentage) were determined as according to data obtained during these runs.

Method application
A laboratory scale pyrolysis reactor (50 mL) ( Figure S1, SI section) was used for the production of synthesis gases from MRDF. Initially, 20.1 g of MRDF were inserted into the reactor. The reactor was heated externally by using an electrical resistance coupled to a temperature controller (up to 900 ºC). 16,17 Condensable gases generated during the pyrolysis process were retained in the condenser, and non-condensable gases were collected in the combustion cylinder. After a pressure of 8 bar was reached, generated gases were extracted from the combustion cylinder ( Figure S1) by using Tedlar bags, and kept at rest for 15 min to reach room temperature and pressure. Then, sample were injected in the GC/TCD/FID.

Accuracy
Method accuracy was evaluated by assessing analyte recovery in the synthesis gas generated in the pyrolysis of real MRDF. Three samples prepared in triplicates, were fortified by adding different concentrations of the analytes (low, medium and high) within the linear range obtained for each one (Table S1). The determined concentration of fortified samples was divided by the theoretical concentration of the fortified samples to assess recovery (equation 2): (2) where C1: experimental concentration of analyte in the fortified sample; C2: theoretical concentration of analyte in the fortified sample.

Results and Discussion
Evaluation of chromatographic conditions and method validation Table 2 presents all chromatographic conditions tested in this study. The most appropriate conditions of operation were selected as according to signal intensity associated to the detection and selectivity of target compounds. Figure 1 shows the chromatographic profile of the analyte mixture under the best chromatographic conditions. R s values presented in Table 3 were calculated by using equation 1. R s values greater than 1.5 were obtained for the following analytes: H 2 /O 2 , N 2 /CO, CO/CO 2 , CH 4 /C 2 H 2 , C 2 H 2 /C 2 H 4 , C 2 H 4 /C 2 H 6 , C 2 H 6 /C 3 H 4 (propyne), C 3 H 8 /iso-C 4 H 10 and iso-C 4 H 10 /n-C 4 H 10 , thus indicating a separation of 100% between the peaks of each of these analytes. 8 R s values between 1.18-1.38 were obtained for O 2 /N 2 , C 3 H 4 (propadiene)/C 3 H 4 (propyne), C 3 H 4 (propyne)/ C 3 H 6 and C 3 H 6 /C 3 H 8 , indicating an overlap of only 2% between peaks. 8 These results demonstrate the appropriate selectivity of the proposed GC/TCD/FID method.
Method linearity (represented by the correlation coefficient, r) is shown in Table 4, and was determined by using calibration curves. High linearity was obtained for all analytes (r values > 0.99) and comply with requirements of the National Health Surveillance Agency (ANVISA) 14  LOD values determined by the signal-to-noise ratio were compared with those reported for ASTM D3612-02 (methods A and C) 6 (Table 4). Lower LOD values were obtained for the present study when compared to ASTM D3612-02 (method A). Depending on the analyte, the proposed method enables the detection of concentrations ranging from 546 times lower for N 2 and O 2 to 34 times lower for CO and CO 2 , 11 times lower for CH 4  not possible to compare the LOQ values obtained in the proposed method with values obtained by ASTM D3612-02 since no LOQ values are presented for methods A and C. Besides, as previously described in the Introduction section, the ASTM D3612-02 (method C) requires the use of two columns connected in series (a molecular sieve and a Carboxen-1006 PLOT columns) and a methanizer to convert CO and CO 2 to CH 4 for acceptable sensitivity by using argon as carrier gas via detection by FID. In addition, light hydrocarbons (C 3 H 4 (propyne and propadiene), C 3 H 6 and C 4 H 10 (n-and iso-butane)) are not within the scope of method C. 6 It can also be observed in this work that H 2 presented the higher LOD value when compared to the other inorganic gases and light hydrocarbons (Table 4). This can be justified by the proximity between heat capacity values pertaining to helium and hydrogen, thus generating a reduced difference on TCD signal. RSD values obtained for target gases for inter (between 0.31 and 1.3%) and intra-day reproducibility tests (between 0.76 and 2.0%) (Tables S2 and S3, SI section) were lower than 2%, while higher RSD values (between 3 and 13%) were reported for ASTM D3612-02 (method C). These results show low variability between measurements obtained for each replicate made within a day or in different days, which guarantees the reliability of results obtained by the application of the proposed method.
Application of the method under the best chromatographic conditions for determination of syngas characterization during pyrolysis of MRDF Figure 2 shows the chromatogram obtained from the syngas generated during the pyrolysis of MRDF.
The LHV (25.5 ± 1.7 MJ Nm −3 ) of the synthesis gas is equivalent to values reported for oily sludge (23.5 ± 4.3 MJ Nm −3 ) 19 (Table 5). On the other hand, the present LHV obtained for the synthesis gas via pyrolysis of MRDF is higher than the LHV obtained for pyrolysis of sewage sludge (9.5 ± 0.3 MJ Nm −3 ), 10 rice straw (11.6 ± 0.2 MJ Nm −3 ), 11 leather-tannery waste (6.0 ± 6.0 MJ Nm −3 ), 12 and horse manure biowaste (13.9 ± 1.8 MJ Nm −3 ). 13 Gasification process applied to the same matrix (MRDF), also resulted in synthesis gases which presented an inferior LHV (between 5.5 and 17.0 ± 4.7 MJ Nm −3 ) (Table 5). [20][21][22] On the basis of these  results, the increased LHV obtained for the synthesis gas analyzed by pyrolysis of MRDF in this study is justified by the high concentration of hydrocarbons (ΣC1-C4 = 52.1% v/v) and absence of H 2 in the sample (Table 5). Finally, the accuracy of the GC/TCD/FID method was evaluated by recovery tests performed before and after spike of samples containing known concentrations of target gases. Recovery values ranged from 98 to 101% (Table 6) and are in accordance with recommendations made by INMETRO (between 98 and 102%). 15 In addition, these results indicate the absence of matrix interference. Hence, the proposed chromatographic method may be considered as adequate for the accurate measurement of each analyte in the method.

Conclusions
A GC/TCD/FID method was developed and validated for the simultaneous quantification of inorganic gases and light hydrocarbons by gas chromatography using a single Carboxen 1010 PLOT column. The proposed method complies with standards recommended by ANVISA and INMETRO. As the proposed method was successfully applied for characterization of the synthesis gas generated during the pyrolysis of real MRDF, it is useful for the identification and quantification of combustible gases generated during thermal processes applied as waste treatment alternatives and which may be explored as energy source. Therefore, the present work supports the use of GC/TCD/FID as a straightforward solution for routine quantification of inorganic gases and  light hydrocarbons generated in thermochemical treatment processes using different matrices.