Online High Throughput Measurements for Fast ... - ACS Publications

Subscriber access provided by Kaohsiung Medical University

Article

Online High Throughput Measurements for Fast Catalytic Reactions Using Time-Division Multiplexing Gas Chromatography Marco Robert Wunsch, Rudolf Lehnig, Christiane Janke, and Oliver Trapp Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.8b01805 • Publication Date (Web): 12 Jul 2018 Downloaded from http://pubs.acs.org on July 24, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 9 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Marco R. Wunsch†, Rudolf Lehnig†, Christiane Janke† and Oliver Trapp*,§ † §

BASF SE, Carl-Bosch-Str. 38, 67056 Ludwigshafen, Germany Department Chemie, Ludwig-Maximilians-Universität München, Butenandtstr. 5-13, Munich, Germany

ABSTRACT: Developing new catalysts is crucial for optimization of chemical processes. Thus, advanced analytical methods are required to determine the catalytic performance of new catalysts accurately. Usually, gas chromatographic methods are employed to analyze quantitatively the product distribution of volatile compounds generated by a specific catalyst. However, the characterization of rapidly changing catalysts, e.g. due to deactivation, still poses an analytical challenge because gas chromatographic methods are too slow for monitoring the change of the complex product spectra. Here, we developed a gas chromatographic technique based on the concept of multiplexing gas chromatography (mpGC) for fast and comprehensive analysis of the product stream from a catalytic testing unit. This technique is applied for the study of the catalytic reaction of Methanol-To-Olefins (MTO) conversion. For this method, the time distance between two measurements are chosen so that the chromatograms but not the peaks themselves are superimposed. In this way, stacked chromatograms are generated in which the components from successively injected samples elute baseline separated next to each other from the column. The peaks from different samples are interlaced and for this reason the method is referred to as time-division multiplexing gas chromatography (td-mpGC). The peaks are analyzed by direct peak integration not requiring a Hadamard transformation for deconvolution of the raw data as usual for many mpGC applications. Therefore, the sample can be injected equidistantly. The integrated peaks have to be allocated to the correct retention times. The time distance between two measurements for studying the reaction and regeneration cycles of MTO catalysts is 4.3 min and 38 s, respectively. Column switching techniques such as backflush and heart-cut are introduced as general tools for multiplexing gas chromatography.

Catalysis is a key technology to promote a clean environment, the efficient use of energy, health and good life quality.1 Also, the change of oil-based raw materials to renewable ones creates the need to develop new catalysts. For understanding the mechanisms of a catalytic reaction and for the design and development of a catalytic process, the change of the product spectrum over the lifetime of a catalyst needs to be investigated. For the profitability analysis of a chemical process, often the exact knowledge of the occurrence of different isomers in the product stream is crucial. Therefore, a comprehensive analytical method is needed for the analysis of the products of a catalytic reaction. Especially in the case of rapidly changing catalyst performance, the analytical method has to allow for a high sampling frequency. In most cases, only chromatographic methods allow for a quantitative analysis of complex product spectra from catalyzed reactions.2 Sample intervals in the range of up to one hour are often the case for standard chromatographic methods. We are presenting a new approach for an online high throughput measurement technique based on the concept of multiplexing gas chromatography (mpGC) for the catalytic characterization of Methanol-To-Olefin (MTO) catalysts3. This new mpGC technique combines the possibility to analyze quantitatively a broad spectrum of different products with a high sampling frequency. These two features allow to monitor the rapid changes of the product composition for the reaction

and regeneration cycles of the MTO reaction. Note that MTO reaction cycles in fixed bed reactors over SAPO-34-like catalyst can be rather short as the conversion and selectivity patterns change within a few hours due to coke formation.2,4 However, with standard gas-chromatography, consecutive samples of the product spectra of those catalysts are generally analyzed every 30 to 60 minutes, which is usually insufficient for an exact characterization of the catalytic properties.5-8 One reason for these long sampling intervals is that the water in the product stream is adsorbed on the alumina columns which are needed for efficient separation of all C1-C5 isomers.9 This typically causes peak shifting. Therefore, these columns either have to be protected from contamination by water using a different precolumn or the water has to be desorbed from the column using a temperature program. Both methods are highly time consuming. For this reason, the increase of the sample throughput is very difficult with classical gas chromatographic methods, e.g. steep temperature ramps and high carrier gas pressure. For fully automated data analysis as needed for process analytics, highly reproducible retention times are required. Therefore, isothermal methods are preferred. However, the online high throughput measurement based on the concept of mpGC using column switching techniques–as presented in this paper–allows for a sample interval in the range of few minutes. In a classical mpGC measurement, the sample is injected according to the pattern of a n-bit (m = 2n – 1 elements) pseudo random binary sequence (PRBS).10 A “1” in the PRBS

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

stands for an injection and a “0” stands for no injection. The time duration for one PRBS is m x Δt, where Δt is the time interval between two elements of the PRBS. For a mpGC measurement, the chromatographic runtime t is longer than Δt. Therefore, a superimposed (convoluted) chromatogram is recorded. As suggested by Smit11 mpGC has been used for many applications to enhance the signal to noise ratio (SNR) by making use of the multiplex advantage.12-22 However, the original idea was to use mpGC as a tool for process analytics.10 For this application, a process stream is continuously sampled according to an endlessly repeated PRBS and a deconvoluted chromatogram containing the averaged information of the components which elute within the time duration of the last PRBS is calculated after every time interval Δt.23 However, the calculation can begin at the earliest after the elution of the last component from the last injection of the first PRBS.23 For a process stream with a constant composition, the convoluted chromatogram can be described by the forward Hadamard transformation given in eq 1.24 [𝐒] ∙ [

deconvoluted convoluted ]=[ ] chromatogram chromatogram

(1)

Where S (m x m matrix) is the convolution matrix derived from the PRBS.24 The deconvoluted chromatogram can be obtained by multiplying the inverse convolution matrix with the convoluted chromatogram according to the inverse Hadamard transformation given in eq 2. [𝐒]−1 ∙ [

convoluted deconvoluted ]=[ ] chromatogram chromatogram

(2)

Changes in the composition of a process stream introduce non-linearities during linear superposition of the single chromatograms.25 As a result so called correlation noise26 (lowfrequency noise and ghost peaks, respectively) emerges in the deconvoluted chromatogram.25 The standard deviation of the correlation noise grows linearly with the mean squared deviation of the signal change due to changing concentrations.27 Therefore, the method will not be suitable if the change of the composition of a process stream is too high over the time duration of one PRBS. Concentration changes of any of the components in a sample might introduce correlation noise that is on the same scale as the peaks of low-concentration components.25,28 Kaljurand et al. calculated that for a narrow peak (peak width less than 1 % of the time duration of one PRBS) whose concentration changes about 50 % over the time duration of a PRBS with 512 injections, the standard deviation of the correlation noise is about 5 % of the chromatographic peak height.29 A second peak in this chromatogram with a concentration of 5 % compared to the first peak is in this case already in the same scale as the correlation noise. Multiplexing GC was used to study the thermal decomposition of polymers.29-32 The chromatograms in this case represent an averaged chromatogram of the peaks which elute within the time duration of the last PRBS. However, for a true highthroughput measurement it is expected that the composition of a sample can be analyzed exactly at the time of each injection.

Such a method (htMPGC – high throughput multiplexing gas chromatography) has been developed for the offline analysis of different chemically similar samples.33-37 The samples are injected with a special injector design according to a PRBS into the GC.38 Peak positions as well as peak widths are determined in the averaged chromatogram (deconvolution of the raw data with eq 2). These peak parameters are then used for fitting the convoluted raw data to obtain the single chromatograms of each injection. To guarantee an acceptable peak to correlation noise ratio in the deconvoluted overview (averaged) chromatogram and a good fit of the raw data, the same samples are injected multiple times. The more the composition of the samples differ the more often the same samples are injected.33,34 However, this technique cannot be applied to process analysis for continuously rapidly changing process streams because every sample has an unique composition and cannot be injected repeatedly. This is in particular problematic for fast catalytic processes and in homogeneous catalysis, where the sample shows continuous catalytic activity. To overcome this hurdle, the complexity of the convoluted chromatogram has to be reduced by using longer time intervals Δt and column switching techniques. The time interval Δt has to be chosen so that every peak in a superimposed chromatogram can be recognized at least by standard peak fitting algorithms or standard peak integration algorithms as implemented in chromatographic data systems. Thus, no deconvolution of the raw data with Hadamard transformation as usual for mpGC is needed. Therefore, the sample can be injected equidistantly. The time distance between two measurement points is equal to the time interval Δt for equidistant injections. The methods presented in this paper for the characterization of MTO catalysts even yield chromatograms with baseline separated peaks. Only the chromatograms and not the peaks themselves are superimposed. The peaks of the components from successively injected samples are interlaced continuously in these chromatograms. This is called a time division multiplexing process39 and for this reason the method is referred to as time-division multiplexing gas chromatography (td-mpGC). However, for this high throughput measurement technique all potential components in the sample have to be known for method development to make sure that a full peak superposition does not take place. For process analytical measurements, this is usually the case as the reactions products are known and their exact concentrations at a certain point of time is of interest. Here, column switching techniques are used to limit the number of components from the sample reaching the detector. Column switching techniques such as backflush and heart-cut have been developed as general tools for multiplexing gas chromatography and are reported in this paper. Since the introduction of mpGC in 1967, several very fast gas injection systems which define the sample volume via gas volume flow and time have been designed and tested for mpGC. Such an injection can take place by controlling the time for opening and closing of solenoid valves10,11,16,25,30 or with fluidic switching elements like a bistable fluidic amplifier40 or a deans switch41,42. According to Hagen–Poiseuille’s law, a laminar volume flow along a pipe depends also on the viscosity of the sample. In cases where concentrations and therefore also the viscosity of the sample are changing, injecting reproducible sample volumes via controlling the duration of a volume flow is not possible. Therefore, the sample vol-


Page 2 of 9

Page 3 of 9 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


ume is defined via a sample loop, an approach which has already been used for mpGC.43,44 This technique is also common in liquid chromatography (LC) and online GC.

Reference gas for calibration and application development was purchased from Linde (Munich, Germany). The reference gas contains 1,3-butadiene, 1-butene, n-butane, cis-2-butene, ethane, ethylene, isobutane, isobutylene, methane, propane, propylene and trans-2-butene in an Argon matrix. The concentrations are 0.963, 0.978, 0.988, 0.984, 0.996, 1.02, 0.995, 0.983, 0.994, 0.995, 0.995 and 0.973 Vol%, respectively. CO (2.0) was bought as pure gas from Linde (Munich, Germany) and CO2 of analytical grade (4.5) from Basi (Rastatt, Germany). Dimethyl ether (3.0) was purchased from Air Liquide (Düsseldorf, Germany). Methanol (>99.9 %) and Isopentane (>99%) was supplied by Sigma Aldrich (Steinheim, Germany). Pressurized air was used for diluting gases and to vaporize liquids for the development of the applications. The mpGC experiments were carried out with a dual channel gas chromatograph (model 7890B by Agilent) equipped with a flame ionization detector (FID) and a thermal conductivity detector (TCD). For studying the regeneration cycle of MTO catalysts, O2, N2, CO, and CO2 have to be separated This analysis was conducted on two pre-columns (type: HP-PLOT/Q+PT from Agilent with a length of 15 m, an inner diameter of 0.53 mm, and a film thickness of 40 µm) and a commercially available set of two parallel columns as main column (type: “select permanent gases/CO2” from Agilent) with the TCD as detector (isothermal and isobaric at 70 °C and 250 kPa with a 1:5 split ratio) (see Figure 1 for the arrangement of the columns). For studying the reaction cycle of MTO catalysts, additionally methanol, water, and dimethyl ether have to be separated from O2, N2, CO, and CO2. This measurement was also conducted with this set-up of separation columns (in this case: isothermal and isobaric at 86 °C and 180 kPa with a 1:10 split ratio). Furthermore, for studying the reaction cycle of MTO catalysts, 1,3-butadiene, 1-butene, n-butane, cis-2-butene, ethane, ethylene, isobutane, isobutylene, methane, propane, propylene, and trans-2-butene have to be separated. This analysis was conducted on two pre-columns (type: GS-GasPro from Agilent with a length of 5 m, and an inner diameter of 0.32 mm) and one main column (type: “Select Al2O3 MAPD” from Agilent with a length of 50 m, an inner diameter of 0.53 mm, and a film thickness of 10 µm) with the FID as detector (isothermal and isobaric at 86 °C and 150 kPa with a 1:10 split ratio) (see Figure 1 for the arrangement of the columns). The sample loop has a volume of 100 µL for the TCD channel and a volume of 10 µL for the FID channel. For the injections, 6-port diaphragm valves are used (model: MDVG-6-16HT-0-0 from AFP, Quebec, Canada). These types of valves were also used for sample block and bleed right before the injection takes place. 10-port diaphragm valves (model: 724DV223110 from VICI, Schenkon, Switzerland) are used for backflush. The valves for injection, block and bleed as well as back-flushing were pneumatically controlled by the internal solenoid valves of the GC. The switching times for the solenoid valves were defined in the time event table in the software ChemStation (by Agilent) used to control the GC.

Multiplexed chromatograms were acquired with the software ChemStation (by Agilent) at a data acquisition rate of 50 Hz for the FID and of 5 Hz for the TCD, respectively. The software ChemStation was also used for peak integration. The retention times and peak areas of all peaks were exported into an ASCII file. The exported data were the basis for fully automatic assignment of the peaks to a component from a certain injection with a MATLAB script. Calibration and conversion of peak area to concentration was also performed with a MATLAB script. The MTO reaction was performed in a fixed bed reactor equipped with a continuous flow system. The catalyst was a Mg-modified Chabazite. 6 ml of silicon carbide (1.6-2.0 mm) were placed into the reactor. The reaction temperature and the pressure were set to 450 °C and 150 kPa. The inlet composition of the gas was MeOH/N2 = 21/79 Vol% and the gas hourly space velocity was set to 2534 nLgas/h*Lcat-1. The regeneration experiment was performed with a spent MTO catalyst which contained significant amounts of coke. The temperature and the pressure were adjusted to 600 °C and 50 kPa during the regeneration. The inlet composition of the gas was set to O2/N2 = 8/92 Vol%.

For process analytical measurements, column switching techniques such as back-flush or heart-cut are regularly used to separate the target components from those with much shorter or much longer retention times.45 When using such techniques for mpGC, it has to be considered that superposition of chromatograms has to take place on the main column since parts of the sample are either held back on the pre-column (back-flush) or are vented to the exhaust (heart-cut). In Figure 1 a scheme is depicted showing how multiplexing gas chromatography is combined with back-flushing a system of two pre-columns to prevent compounds with long retention times from reaching the main column. pre-column I

aux EPC 2 S/SL-Inj.

1

10

3

9

4

8 5

6

main column

detector

7

exhaust

pre-column II

Fig. 1: Set-up of Multiplexing gas chromatography back-flush technique: The sample is injected with a split/ splitless injector (S/SL-Inj.) which is connected via a capillary to port 3 of a 2 way 10-port valve (the straight lines represent the off-position and the dotted lines represent the on-position of the valve). The sample flows either through pre-column I or II depending on whether the valve is switched to the off or on position, respectively. Pre-column I and pre-column II are of the same type and of the



2∙𝑡𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟 𝑛

(3)

Where n is the number of pre-columns and ttransfer is the time for transferring the components of interest from one of the pre-columns to the main column. The same time is needed for back-flushing this pre-column. Therefore, ttransfer has to be accounted twice when determining Δtmin. The pre-column system can be parallelized by connecting even more than two pre-columns to one main column to decrease Δtmin. To guarantee a complete back-flush within ttransfer, the inverse gas flow should to be larger compared to the forward gas flow. For measurements requiring a heart cut, a 2-way 6-Port valve is added between port 8 of the 10-port valve and the main column for the set-up depicted in Figure 1. With such an arrangement of valves, the high boiling components are backflushed and the low boiling components are vented to the exhaust. Multiplexing GC back-flush technique as well as heart-cut technique can be used in general for mpGC measurements applied for both improving the signal to noise ratio or–as in this case–for increasing the sample throughput.

A total of three td-mpGC methods have been developed for the measurements shown in this paper. For studying the reaction cycle of the MTO catalysts, two td-mpGC methods have been set-up, designated as method I and II. They run simultaneously on two separate channels of one GC using a FID and a TCD as detector, respectively. For studying the regeneration cycle of the catalysts, a third td-mpGC method has been developed (method III). It runs on the same GC using the channel with the TCD as detector. With the td-mpGC method I (channel with FID), methane, ethane, ethylene, propane, pro-

Intensity / 25 µV

∆𝑡𝑚𝑖𝑛 =

sum peak:

𝑡 O2, N2, CO 𝑡 O2

100 0

CO2

CO

0

a.)

𝑡

N2

200

1

2

3

Time / min Intensity / 25 µV

The sample is injected with a split/ splitless injector (S/SL-Inj.) into the carrier gas flow. The gas is conducted via the ports 3 and 2 of a 2-way 10-port valve switched to the off-position (straight lines in Figure 1) over pre-column I onwards to the main column. Port 10 is connected to an auxiliary electronic pressure controller (aux EPC) operated independently from the EPC that controls the forward carrier gas flow over the pre-columns and the main column. While the sample flows over pre-column I, pre-column II is back-flushed with the gas stream controlled by the auxiliary EPC. As soon as all target components have left pre-column I and are transferred to the main column, the 10-port valve is switched to the on-position (dotted lines in Figure 1). This leads to an inverse gas flow over pre-column I to the exhaust (back-flush) as controlled by the aux EPC and a forward gas flow over pre-column II to the main column. The next sample is injected and flows via the ports 3 and 4 over pre-column II onwards to the main column. If both pre-columns are of the same type and of the same length, a constant forward gas flow through the main column will be provided. The shortest possible time interval Δtmin for the column switching system shown in Figure 1 is given in eq 3.

pylene, isobutane, n-butane, trans-2-butene, 1-butene, isobutylene, cis-2-butene, isopentane, and 1,3-butadiene are analyzed. As shown in a separate measurement, hydrocarbons with five or more C-atoms are not produced in significant amounts (> 0.5 Vol%) for the MTO catalysts tested in these experiments. However, to avoid any interference of trace amounts of those hydrocarbons they are back-flushed together with H2O, methanol, and dimethyl ether which all elute after 1,3-butadiene on the pre-column (type: GS-GasPro) using the column switching set-up shown in Figure 1. With the td-mpGC method II (channel with TCD), O2, N2, CO, CO2, H2O, methanol, and dimethyl ether are analyzed. All components which elute after methanol on the pre-column (type: HPPLOT/Q+PT) are back-flushed, again using the column switching set-up shown in Figure 1. Methane, ethane, ethylene, propane, and propylene elute before methanol on this pre-column and are therefore also detected. They are already measured with the td-mpGC method I using the FID instead of the TCD. Thus, the detection of these components with method I has a higher sensitivity than with method II. Therefore, these components which do not have to be measured with method II may overlay with themselves but not with the target components. However, such interfering substances reduce the available space in the chromatogram for detection of the target compounds. With the td-mpGC method III (channel with TCD), O2, N2, CO, and CO2 are analyzed for studying the regeneration cycles of the MTO catalysts. Compared to the td-mpGC method II, all components which elute after CO2 on the pre-column (type: HP-PLOT/Q+PT) are back-flushed by defining a shorter transfer time ttransfer. In Figure 2, four chromatograms are shown to illustrate the concept of td-mpGC.

𝑡

𝑡

200

4

5

𝑡 b.)

𝑡

100 0

0

1

2

3


same length. Port 10 is connected to an auxiliary pressure controller (aux EPC) which provides the carrier gas for back-flush. Switching the 10-port valve causes an inversion of the gas flow on both pre-columns while the gas flow on the main column remains constant.

𝑡

200

4

5

𝑡

c.)

𝑡

100

0

0

1

2


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 9

𝑡

200

3

𝑡

4

𝑡

𝑡

5

𝑡 d.)

𝑡

100 0

0

1

Time / min


2

∆𝑡

3

∆𝑡

∆𝑡

4

∆𝑡

∆𝑡

5

Page 5 of 9

Fig. 2: Measured as well as simulated chromatograms of continuously injected samples of the same composition with various cycle times. (a) Measured chromatogram with a cycle time of t+tgap which is standard for conventional online gas chromatography (b) Simulated chromatogram with a cycle time t: The following sample is injected right after the elution of all components from the previous sample. (c) Simulated chromatogram with a cycle time t-t0: Injection of the following sample before elution of all components from the previous sample, so that the first peak of the following chromatogram elutes right after the last peak of the previous sample. (d) Measured chromatogram with a cycle time Δt: Chromatograms are stacked with a time interval Δt in such a way that peaks do not overlap.

chromatogram is shown that has been measured with the cycle time being even further reduced compared to the method MISER in fast injection mode as illustrated in Figure 2c. The chromatograms shown in Figure 2d are stacked by injecting the samples with a sampling interval Δt which is smaller than t-t0 while making sure that components from different samples elute in between the components of other samples so that peaks do not overlap. A similar method was published for non-continuous measurements with few injections of samples of little complexity and was referred to as MISER in stacked injection mode.50,51 The chromatogram depicted in Figure 2d has been measured with the td-mpGC method III with a time interval Δt of 38 s. This corresponds to a nearly five-times shorter cycle time compared to the cycle time of the single chromatograms of 2.8 min as shown in Figure 2a. For a continuous interlacing of peaks from different samples as shown in Figure 2d, an individual timeslot ΔtRet. within every time interval Δt is reserved for every component that could possibly be in the sample. Therefore, this injection scheme is called time-division multiplexing gas chromatography (td-mpGC).39 The relative reduction of the sample interval for a td-mpGC online high throughput measurement compared to a conventional GC measurement can be calculated according to eq 4.

Figure 2a shows two chromatograms measured with a sample gas containing O2, N2, CO, and CO2. For the first chromatogram (retention time from 0 to 1.8 min), the gas has been guided through pre-column I and for the second chromatogram (retention time from 2.8 to 4.6 min) through pre-column II (see Figure 1). Both chromatograms show the same peak pattern, however with different intensity since the split ratio varies slightly depending on the pre-column and therefore also the amount of sample gas. The first three peaks in both chromatograms are assigned to O2, N2, and CO (in the order of increasing retention times). The fourth peak is a sum peak of O2, N2, and CO which is observed since the main column is a set of two parallel columns (type: select permanent gases/CO2 from Agilent). The fifth peak in both chromatograms belongs to CO2. The chromatographic run time is referred to as time t. There will be a time gap tgap after the chromatographic run is finished until the following sample is injected if both samples are measured one after the other as in the present case. Depending on the type of GC used for the analysis, tgap is typically in the order of 30 to 60 s (here: 60 s). The total cycle time is therefore t+tgap. Figures 2b and 2 c show chromatograms simulated by using the single chromatograms presented in Figure 2a. The simulated chromatogram in Figure 2b will be obtained if the subsequent sample is injected right after all components from the previous injection have left the separation column, thereby omitting tgap. The cycle time is therefore reduced to the chromatographic runtime t. In Figure 2c the simulated chromatograms are shown which will be obtained if the subsequent sample is injected before all components from the previous injection have left the separation column so that the first peak of the following chromatogram elutes right after the last peak of the previous injection. The cycle time for such a case is t-t0, where t0 indicates the time when the elution of the first component begins. This chromatographic method is already published as MISER (multiple injections in a single experimental run) in fast injection mode.46-49 In Figure 2 d the

[

𝑐𝑜𝑛𝑣𝑒𝑛𝑡𝑖𝑜𝑛 𝑙 𝐺𝐶 𝑚𝑢𝑙𝑡𝑖 𝑙𝑒𝑥𝑖𝑛 𝐺𝐶

]=

𝑡+𝑡𝑔𝑎𝑝

(4)

∆𝑡

As required for process analytical measurements, the td-mpGC methods have to be set-up for a continuous operation. The prerequisite to determine a suitable time interval for a td-mpGC method is that the number of potential components that elute for the given chromatographic method as well as their retention times and peak widths are known. Let A and B be the time where the elution of one of the peaks starts and ends, respectively. For continuously stacked chromatograms without peak overlay the sampling interval Δt has to be chosen in a way that the timeslot ΔtRet. defined as 𝛥𝑡𝑅𝑒𝑡. = [𝐴 𝑚𝑜𝑑 (𝛥𝑡); 𝐵 𝑚𝑜𝑑 (𝛥𝑡)]

(5)

is different for each peak. Accordingly, a large number of peaks within a chromatogram does not limit the possibility of stacking but requires a longer time interval Δt. The chromatogram depicted in Figure 3 shows a representative section of the chromatogram measured with the td-mpGC method I for analyzing all alkanes and alkenes with up to 4 C atoms as well as isopentane. 16 14 12 10 8 6 raw data injection 5 injection 4 injection 3 injection 2 injection 1

Intensity / pA

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


2

3

4

5

6

8 7 Time / min

9

10

11

12


13


Fig. 3: td-mpGC measurement of methane, ethane, ethylene, propane, propylene, isobutane, n-butane, trans-2-butene, 1-butene, isobutylene, cis-2-butene, isopentane, and 1,3-butadiene. Every injected sample contains all of these components. The black line is a section of the raw data acquired by the FID. The peaks, which elute in this time range, originate from 5 different injections at different times. The peaks belonging to these five injections are depicted in different colors in the foreground.

The black line in the background of Figure 3 shows a section of the continuously recorded data. The peaks that are observed originate from 5 different injections which are depicted in front of the black line in different colors. The red peaks, injection 3, correspond to a full chromatogram of a single injection. The elution order is methane, ethane, ethylene, propane, propylene, isobutane, n-butane, trans-2butene, 1-butene, isobutylene, cis-2-butene, isopentane, and 1,3-butadiene. Injection 4 (green peaks) and injection 5 (brown peaks) are injected one and two time intervals Δt later, respectively, compared to injection 3. Injection 1 (orange peaks) and injection 2 (blue peaks) are injected one and two time intervals Δt earlier, respectively, compared to injection 3. For injections 1, 2, 4, and 5, not all peaks belonging to each injection are visible in Figure 3. The time interval Δt for this measurement method is 4.1 min.

The three td-mpGC measurement methods I, II, and III were used for the characterization of MTO catalysts in a catalytic testing unit. The td-mpGC method III for the analysis of O2, N2, CO, and CO2 (see Figure 2 d) was utilized for studying the regeneration cycles of those catalysts. In organic reactions on zeolite catalysts such as the catalytic MTO process, coke is formed on the catalyst. This reduces both the activity and the selectivity of the catalyst.52,53 Therefore, zeolite catalysts have to be regenerated oxidatively. The exact determination of the end of the regeneration process is an important information for the design of the industrial process. As depicted in Figure 4 oxygen is no h through the catalyst bed in the reactor. The time distance Δt between two measurement points is 38 s for the td-mpGC method III. 100 90 concentration / vol.-%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 9

8

6 4

2 0

5.7

5.9 6.1 time on stream / h N2

CO2

6.3 O2

This time interval Δt allows for a detailed characterization of the regeneration cycles. Although reactant and product stream from the catalytic testing unit are measured alternately leading to an increase of the sampling interval to 76 s for both reactant and product stream. Therefore, the time resolution of the data depicted in Figure 4 is 76 s since only the concentrations of N2, O2, and CO2 in the product stream of the catalytic testing unit are depicted. CO is below the detection limit in the time range from 5.7 h to 6.3 h. Therefore, this compound is not shown although the set-up has been calibrated for detection of CO. MTO catalysts commonly deactivate due to coke formation within a few hours after start of the reaction.2,4 In Figure 5 the methanol conversion and the carbon selectivities of the products from a reaction cycle of an MTO catalyst are presented. The difference between the concentration of methanol in the reactant and product stream is interpreted as the amount of methanol which is converted during the catalytic reaction. The methanol conversion is given as the ratio (in %) of converted methanol to the amount of methanol in the reactant stream. The concentrations of all components detected in the product stream were converted into the carbon selectivity (C-Sel.) according to eq 6. [ 𝑥𝑖 ] ∙ 𝑘 𝑖

(𝐶-𝑆𝑒𝑙. )𝑖 = ∑𝑛

𝑖=1 [ 𝑥𝑖 ] ∙ 𝑘𝑖

(6)

The concentration of each component i is xi and the number of C-atoms in its molecular structure is ki. In the first 1.5 h of the reaction over 95 % of the methanol is converted into alkanes and alkenes with one to four C-atoms. The rest (5%) is converted into CO and CO2. During this time range the catalyst has a C-selectivity of over 60 % for both ethylene and propylene. After 1.5 h from the start of the reaction, the methanol conversion drops and at the same time the conversion of methanol to dimethyl ether increases. This indicates that the catalyst starts to deactivate. The time resolution of the data depicted in Figure 5 for the analysis of the product stream of the catalytic testing unit is 4.3 min. For calculation of the carbon selectivity, the concentrations of all components measured with the FID and the TCD channel (td-mpGC method I and II, respectively) are needed. Since both methods run with different time intervals Δt (method I: 4.1 min and method II: 4.3 min) the depicted carbon selectivity in Figure 5 has to be calculated with the longer time interval Δt of 4.3 min. Nevertheless, such a time interval Δt allows for observation of fast composition changes during the reaction cycle of the MTO catalysts.

Fig. 4: Regeneration cycle of MTO catalysts. The concentrations of N2, O2, and CO2 coming out of the reactor are shown in the time range from 5.7 to 6.3 h after start of the regeneration cycle. This corresponds to the time interval when the O2 concentration increases and therefore the regeneration ends. The time distance between two measurement points is Δt = 76 s.


80

80

60

60

40

40

20

20

0

0.5 X(methanol) CO2 CO cis-2-butene

1.0 1.5 2.0 time on stream / h trans-2-butene iso-butylene 1-butene propylene

2.5

ethylene n-butane propane ethane

0

methane dimethyl ether

Fig. 5: Reaction cycle of a MTO catalyst. The fraction of each product is given as C-selectivity (left y-axis). The methanol conversion is plotted on the right y-axis. After 1.5 h the methanol conversion and the C-selectivity of all products drop and the methanol to dimethyl ether conversion increases. This indicates deactivation of the catalyst. The time distance between two measurement points is Δt = 4.3 min.

The accuracy and cycle time of a conventional online-GC method and the newly developed td-mpGC method have been compared. For this experiment, the product stream of the catalytic testing unit is guided first through the sampling valves of the GC used for the td-mpGC methods and afterwards to the sampling valves of a conventional online-GC system. The results of these experiments are depicted in Figure 6. The concentrations measured with both methods are converted into the carbon selectivity as described above. The carbon selectivity of ethylene, propylene, methane, olefins (ethylene, propylene, trans-2-butene, 1-butene, isobutylene, cis-2-butene, 1,3-butadiene), alkanes without methane (ethane, propane, isobutane, n-butane, isopentane), dimethyl ether, and of the sum of CO and CO2 are depicted. The conversion of methanol is also shown. There is no significant difference in the quantitative results of both methods. The td-mpGC data (dots) clearly show that for a time span of 1.5 h after start of the reaction a selectivity for olefins of over 80 % is maintained before the catalyst starts to deactivate. This onset of the deactivation of the catalyst cannot be seen from the data measured with conventional GC (crosses). The time resolution is increased with td-mpGC (dots) about a factor of 10 in comparison to the conventional GC method (crosses). The conventional GC method uses only one pre-column and is heated slowly during every measurement to detect also aromatic and higher hydrocarbons. Furthermore, the measurement method of the

conventional GC is not fully optimized for a short analysis time.

100

100

80

80

60

60

40

40

20

20

0

0 0.5

1.0 1.5 2.0 time on stream / h

methanol conversion / %

100

methanol conversion / %

100

C-selectivity / %

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


C-selectivity / %

Page 7 of 9

2.5

X(methanol)

olefins

ethylene

propylene

CO + CO2

methane

alkanes w/o CH4

dimethyl ether

Fig. 6: Comparison of the results obtained with td-mpGC (dots) with the values measured with conventional online GC (crosses) for the analysis of the product stream in the reaction cycle of a MTO catalyst. The time distance between two measurement points for td-mpGC is Δt = 4.3 min and for conventional online GC Δt = 50 min.

The new technique of time-division multiplexing gas chromatography (td-mpGC) has been applied for high throughput measurements of the product stream of a reaction over a methanol-to-olefin (MTO) catalyst. It should be noted that also higher boiling fractions such as aromatic and higher hydrocarbons can be analyzed with short cycle times using the proposed Multiplexing GC heart-cut technique with appropriate separation columns. The gain in information through such a high throughput measurement is crucial for an efficient development of catalysts for which the performance changes rapidly on a timescale of minutes. For the method td-mpGC, peaks belonging to different sample injections are interlaced while maintaining baseline separation of all target components. The samples are injected equidistantly, in contrast to classical mpGC methods using pseudo random binary sequences for defining an injection scheme. For td-mpGC, the peaks are integrated with standard GC software. In the current experiments, the valve switching events for multiple injections were programmed as entries in the time event table of the software ChemStation (by Agilent). Controlling the time of the start of the GC exactly would allow for running a td-mpGC experiment by defining the valve switching events for a single injection as a method that is continuously restarted. The instrument would be restarted at intervals of Δt. The increase of efficiency provided by td-mpGC is especially high for chromatographic separations marked by a long run time for separation of only a few components.


Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

There is an increased risk of interferences for the presented method. If unknown components are formed in the process stream their peaks could overlap with those of the target substances. Therefore, care must be taken that only a defined spectrum of components reaches the detector. Multiplexing GC backflush technique as well as heart-cut technique are presented as possible methods to define the spectrum of interest from a sample. This column switching techniques can be applied in general for mpGC. They are also especially useful to suppress correlation noise by cutting out high concentrated components if the signal to noise ratio should be increased by Hadamard transformation based mpGC.

*E-mail: [email protected] ¶

M.R.W. and R.L. contributed equally to this work.

The authors declare no competing financial interest.

This work has been funded by BASF SE.

(1) Centi, G.; Perathoner, S. Catal. Today 2008, 138, 69. (2) Chen, D.; Moljord, K.; Holmen, A. Microporous Mesoporous Mater. 2012, 164, 239. (3) Rostami, R. B.; Lemraski, A. S.; Ghavipour, M.; Behbahani, R. M.; Shahraki, B. H.; Hamule, T. Chem. Eng. Res. Des. 2016, 106, 347. (4) Olsbye, U.; Svelle, S.; Bjørgen, M.; Beato, P.; Janssens, T. V.; Joensen, F.; Bordiga, S.; Lillerud, K. P. Angew. Chem. Int. Ed. 2012, 51, 5810. (5) Travalloni, L.; Gomes, A. C.; Gaspar, A. B.; da Silva, M. A. Catal. Today 2008, 133, 406. (6) Li, J.; Qi, Y.; Xu, L.; Liu, G.; Meng, S.; Li, B.; Li, M.; Liu, Z. Catal. Commun. 2008, 9, 2515. (7) Nishiyama, N.; Kawaguchi, M.; Hirota, Y.; Van Vu, D.; Egashira, Y.; Ueyama, K. Applied Catalysis A: General 2009, 362, 193. (8) Ilias, S.; Khare, R.; Malek, A.; Bhan, A. J. Catal. 2013, 303, 135. (9) Scott, C. G. Journal of the Institute of Petroleum 1959, 118. (10) Izawa, K.; Furuta, K.; Fujiwara, T.; Suyama, N. Ind. Chim. Belg. 1967, 32, 223. (11) Smit, H. C. Chromatographia 1970, 3, 515. (12) Wunsch, M. R.; Lehnig, R.; Trapp, O. Anal. Chem. 2017, 89, 4038. (13) Siegle, A. F.; Trapp, O. Anal. Chem. 2014, 86, 10828. (14) Siegle, A. F.; Trapp, O. Anal. Chem. 2015, 87, 11932. (15) Siegle, A. F.; Trapp, O. J. Chromatogr. A 2016, 1448, 93. (16) César, W.; Flourens, F.; Kaiser, C.; Sutour, C.; Angelescu, D. E. Anal. Chem. 2015, 87, 5620. (17) Lin, C.-H.; Kaneta, T.; Chen, H.-M.; Chen, W.-X.; Chang, H.W.; Liu, J.-T. Anal. Chem. 2008, 80, 5755.

(18) Cheng, C. C.; Chang, H. W.; Uchimura, T.; Imasaka, T.; Kaneta, T.; Lin, C. H. J. Sep. Sci. 2010, 33, 626. (19) Cheng, Y.-K.; Lin, C.-H.; Kaneta, T.; Imasaka, T. J. Chromatogr. A 2010, 1217, 5274. (20) Fan, Z.; Lin, C.-H.; Chang, H.-W.; Kaneta, T.; Lin, C.-H. J. Chromatogr. A 2010, 1217, 755. (21) Cheng, Y.-K.; Lin, C.-H.; Kuo, S.; Yang, J.; Hsiung, S.-Y.; Wang, J.-L. J. Chromatogr. A 2012, 1220, 143. (22) Fan, G.-T.; Yang, C.-L.; Lin, C.-H.; Chen, C.-C.; Shih, C.-H. Talanta 2014, 120, 386. (23) Davies, W. D. T. Instrument Practice 1968, 22, 213. (24) Kaljurand, M.; Küllik, E. Chromatographia 1978, 11, 328. (25) Godfrey, K. R.; Devenish, M. Meas. Control 1969, 2, 228. (26) Annino, R.; Bullock, L. E. Anal. Chem. 1973, 45, 1221. (27) Kaljurand, M.; Küllik, E. J. Chromatogr. 1979, 186, 145. (28) Koel, M.; Kaljurand, M. Crit. Rev. Anal. Chem. 1996, 26, 149. (29) Küllik, E.; Kaljurand, M. Anal. Chim. Acta 1986, 181, 51. (30) Urbas, E.; Kaljurand, M.; Kūllik, E. Journal of Analytical and Applied Pyrolysis 1980, 1, 213. (31) Kaljurand, M.; Elomaa, M.; Plit, L. Anal. Chim. Acta 1991, 248, 271. (32) Kaljurand, M.; Smit, H. C. Chromatographia 1994, 39, 210. (33) Trapp, O. Angew. Chem. Int. Ed. 2007, 46, 5609. (34) Trapp, O. J. Chromatogr. A 2008, 1184, 160. (35) Trapp, O. J. Chromatogr. A 2010, 1217, 6640. (36) Siegle, A. F.; Trapp, O. Chem. Ing. Tech. 2014, 86, 1044. (37) Siegle, A. F.; Trapp, O. J. Sep. Sci. 2015, 38, 3839. (38) Trapp, O. United States Patent, US 8297135 B2, 2012. (39) West, J.; Dean, T.; Andrews, J. Network+ guide to networks; Cengage Learning, 2015. (40) Annino, R.; Gonnord, M. F.; Guiochon, G. Anal. Chem. 1979, 51, 379. (41) Koel, M.; Kaljurand, M.; Küllik, E. Anal. Chim. Acta 1987, 199, 197. (42) Kaljurand, M.; Koel, M.; Küllik, E. Anal. Chim. Acta 1990, 239, 317. (43) Villalanti, D. C.; Burke, M. F.; Phillips, J. B. Anal. Chem. 1979, 51, 2222. (44) Valentin, J. R.; Hall, K. W.; Becker, J. F. J. Chromatogr. 1990, 518, 199. (45) Kessler, R. W. Prozessanalytik: Strategien und Fallbeispiele aus der industriellen Praxis; Wiley, 2012. (46) Welch, C. J.; Gong, X.; Schafer, W.; Pratt, E. C.; Brkovic, T.; Pirzada, Z.; Cuff, J. F.; Kosjek, B. Tetrahedron: Asymmetry 2010, 21, 1674. (47) Papp, R.; Andersson, U.; Cantin, L.-D. J. Pharm. Biomed. Anal. 2013, 77, 94. (48) Welch, C. J.; Regalado, E. L.; Welch, E. C.; Eckert, I. M.; Kraml, C. Analytical Methods 2014, 6, 857. (49) Zawatzky, K.; Biba, M.; Regalado, E. L.; Welch, C. J. J. Chromatogr. A 2016, 1429, 374. (50) Schafer, W.; Wang, H.; Welch, C. J. J. Sep. Sci. 2016, 39, 2978. (51) Vistuba, J. P.; Piovezan, M.; Pizzolatti, M. G.; Rebelo, A. M.; Azevedo, M. S.; Vitali, L.; Costa, A. C. O.; Micke, G. A. J. Chromatogr. A 2013, 1274, 159. (52) Liang, J.; Li, H.; Zhao, S.; Guo, W.; Wang, R.; Ying, M. Applied Catalysis A: General 1990, 64, 31. (53) Marchi, A. J.; Froment, G. F. Applied Catalysis A: General 1991, 71, 139.

Insert Table of Contents artwork here


Page 8 of 9

Page 9 of 9

multiplexing GC

conventional GC 80

80

C-selectivity / %

C-selectivity / %

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


60 40

20 0

60 40 20

0 0.5

1.0

1.5

2.0

time on stream / h ethylene methane olefins propylene

0.5

1.0

1.5

2.0

time on stream / h Alkanes w/o CH4 CO + CO2


9

Online High Throughput Measurements for Fast ... - ACS Publications

Recommend Documents