Extending the scope of gas chromatography - mass spectrometry by means of supersonic molecular beams

(1)

MSc Chemistry

Analytical Sciences

Master Thesis

Extending the scope of gas chromatography – mass spectrometry by

means of supersonic molecular beams

by

Mimi den Uijl

10410465

May 2018

48 EC

September 2017 – May 2018

Supervisor/Examiner:

Examiner:

W.J.L. Genuit, Dr.

P.J. Schoenmakers, Prof. Dr. Ir.

(2)

Abstract

Combined gas chromatography mass spectrometry (GC-MS) is an essential analytical technique that plays a key role in a broad range of disciplines. The power lies in the identification and quantification of volatile and semi-volatile organic compounds in complex mixture. Standard electron ionization (EI) and the automated identification with extensive libraries makes this technique indispensable. However, GC-MS suffers from two major Achilles-heels: it is only applicable to a range of thermally stable and volatile compounds and often the mass spectra have a reduced or missing molecular ion. In addition to GC-EI-MS, chemical ionization (CI) is used as ionization technique, which is a softer method and tends to create more protonated quasi-molecular ions. The downside is that there is no library for CI and that CI is not effective for every component. Using a supersonic molecular beam (SMB) could provide a solution to the current GC-MS restrictions. Through an adiabatic expansion, the analytes cool before they enter the ion source. This creates a spectrum, which is similar to that in standard GC-EI-MS, but with enhanced molecular ion intensity. In addition, the pumping capacity of the SMB system is much higher than that of a standard GC-MS vacuum system, so it is possible to use short columns and high flow rates for the analysis of heavy and thermally labile compounds. Applicability of the GC-SMB-MS and its improvement compared to standard GC-EI-MS is

demonstrated in standard mixtures, high mass alkanes, thermally unstable peroxides and pyrolysis products from various solid organic materials.

(3)

1. Introduction

The combination of gas chromatography with mass spectrometry (GC-MS) is an essential analytical technique and plays a key role in a broad range of disciplines. The power of GC-MS lies in the identification and quantification of volatile and semi-volatile organic compounds in complex mixtures.4_{Especially the mass spectral data generated by the standard electron ionization (EI) and} the automated identification by extended libraries make GC-MS a very versatile technique. However, GC-MS is only applicable to thermally stable and volatile compounds. Often the EI mass spectra have a reduced or missing molecular ion intensity, which makes identification hard.5_{In the identification} process, the mass spectrum is compared to a library of US National Institute of Standards and Technology (NIST) mass spectra. These libraries can be extended to more proprietary libraries. The result is a list of compounds that ranks library spectra, according to their similarities. This hit list is often used as a suggestion to the analyst and is taken into account in the identification process. To score high on the library list, a compound must be in the library and when it is in the library, it needs to have a high score in similarity between the standard spectrum and the searched spectrum. This similarity is called the match factor. To obtain a high identification probability, the analysis must produce one library spectrum with a much higher matching factor than all of the others.6_For example, when analyzing octane, fragments are found that can also be found in other normal alkanes like heptane or nonane. Since high mass ions and the molecular ion characterize a compound more than low mass ions, they carry more weight in the library search identification algorithms.7_{Because the fragmentation rate in electron ionization is often too high to obtain} molecular ion or high mass fragments, there is room for improvement.

As an addition to GC-EI-MS, chemical ionization is used as ionization techniques for GC-MS. Instead of bombarding the sample with electrons, a reagent gas like methane and/or ammonia is used, which produces reagent ions that can transfer a proton to the analyte. The stability of the resulting protonated quasi-molecular ion (MH+) is higher than the product of electron ionization. The extend of fragmentation depends on the reagent gas and the fragmentation is often different from the fragmentation in EI.8_{Thus EI and CI are complementary ionization techniques to find the} molecule of interest.

It seems like a winning team, which should not be tampered with, but the opposite is true. The NIST library only consists of EI spectra, so the recorded CI spectrum cannot directly be transferred to the library. Also, CI is less sensitive than EI and the ion source must be altered by a technician to switch between ionization techniques. On top of that, chemical ionization is ineffective for compounds with low proton affinity like aliphatic components.5

Supersonic molecular beams (SMBs) provide a solution for the current GC-MS restrictions. An SMB is formed by expansion of a gas through a small pinhole or shaped nozzle into a vacuum chamber.5_{In this expansion the carrier gas and the sample molecules obtain the same final velocity.} This homogeneous velocity establishes slow intrabeam collisions, resulting in an internal vibrational-rotational cooling of the analyte.9_{Because the analytes decrease their vibrational-rotational energy,} they have less internal energy when they are bombarded with electrons in the ionization. This results in a decrease of fragmentation.10_{Apart from the vibrational and translational cooling of} analytes, the SMB is an important system for GC-MS because it creates a unidirectional motion in space with the heavy analytes concentrating along the beam axis, because of faster radial diffusion of the helium atoms. This phenomenon is called jet separation, where the density of helium decreases much faster than the density of the analyte.5_{Through this jet separation it is possible to} discriminate between the makeup gas flow and the analytes.

Jet separation is not a new term. When gas chromatography was first coupled to mass spectrometry, packed columns were used in gas chromatography and there was a large pressure

(7)

difference that had to be overcome. Jet separators, effusion separators, and semi-permeable membranes were used to pump away the excessive carrier gas eluting from the GC.11_{Since the GC} columns developed from packed to capillary, the volumetric gas flow was reduced in such a way that pumping off the carrier gas was not necessary anymore. In addition, the mass spectrometers

changed from sector to quadrupole instruments, which have a smaller pathway and thus a smaller chance for collisions with background gas. Also, there are no high voltages in the quadrupole high source. This allows quadrupoles to have a lower vacuum than sector instruments.4_{Because the GC} capillaries are now directly inserted in the mass spectrometer, there is no option to get rid of excess carrier gas when high flows are used in the gas chromatogram. This can be a disadvantage when a high pressure in short columns is used. In the interface of the GC-SMB-MS, a high He flow, from the GC column or the makeup gas, is pumped away before the analytes enter the ionization chamber.12 Because large amounts of carrier gas can be pumped off in the SMB apparatus, it is also possible to use short, wide columns with high flow rates in the gas chromatograph. In this way, the SMB can broaden the range of compounds amenable for analysis.

In this research, a supersonic molecular beam interface between gas chromatography and mass spectrometry of Aviv Analytics has been used as an alternative to standard GC-MS. First the performance of the SMB as interface will be tested on standard mixtures. It will be investigated whether the identification procedure will be similar to the standard GC-MS procedure. The second part of the research will be to create an understanding what effect the makeup gas flow and the electron energy have on the spectral data. The supersonic molecular beam interface will be applied to the analysis of gas-to-liquid (GTL) products with carbon number distributions extending beyond the scope amenable to standard GC-MS. The capability of fast GC-MS in the SMB has been explored in the analysis of hydroperoxides, which would normally degrade in the hot column of a standard GC-MS system. Finally, pyrolysis GC-SMB-MS analysis of some natural and synthetic polymers have been performed to see how fast GC-SMB-MS could complement the current pyrolysis GC-EI-MS method. The synthetic polymers are analyzed, because they are of interest for the crude material processed by the IH2 technology. The IH2_{technology stands for the integrated hydropyrolysis and} hydroconversion (IH2) for the direct production of gasoline and diesel from biomass.53_{The feedstock} that is used in the current pilot plant contains mostly lignin and cellulose, but could also contain plastics. If Py-GC-SMB-MS can be used to differentiate between lignin, cellulose, and various types of plastics, it would be a valuable analytical technique for the pilot plant. For this reason, some

(8)

2. Background Theory

2.1. Gas chromatography

Gas chromatography was invented by Martin and James in 1952. In these days, GC was performed with packed columns.13_{In 1979, HP introduced the first capillary GC column.}14_{The capillary column} is a fused silica column with a liquid stationary phase coated on the inner wall. Depending on the analysis, column coatings differ in polarity to create the best separation between peaks. The mobile phase, or carrier gas, is mostly He and H2, while it used to be nitrogen. The gas flow is regulated and adjusted to obtain fast analysis and good separation. In most cases, GC is performed with a

temperature program.

2.1.1. Injection techniques

For gas chromatography, different injection techniques are used. The choice of a type of sampling technique depends on the sample and must be adapted to the analytical problems posed.15_{In the} next part, two inlet systems will be introduced that have been used in this research.

2.1.1.1. Split/Splitless

The first inlet systems for gas chromatography were split injectors. Since the smallest sample volumes would overload the column, especially with undiluted samples, a part of the sample volume had to be flushed away. When the sample is injected through the septum in the heated liner, the liquid sample immediately evaporates and is pushed by the carrier gas either to the GC column or the split line. This can be seen in Figure 1. The split ratio can be varied. An advantage of the split injection is that it protects the column for involatile components in the sample, but the direct downside of this advantage is that the liner discriminates between molar masses. Also thermal instable components could already degrade in the hot liner.15_{Splitless injection was invented by K. Grob when he} accidently forgot to open the split valve.16_{In this mode the sample is} injected in the hot liner and the whole evaporated volume is pushed to the GC column. Because the GC column has a lower temperature than the hot liner, the sample is thermally focused on the beginning of the column. This method has a lower chance of discrimination, but the injection is still at a hot temperature so there is still a chance of thermal degradation.17 Splitless injection can be used for very dilute solutions.

2.1.1.2. Cool On-Column (COC) injection

Grob studied the injection techniques for GC more thoroughly and found drawbacks in the available options. First, thermally labile components may be altered by the high temperature in the split inlet. Second, discrimination for components that tend to be absorbed takes place at the inlet and it can be difficult to reproduce an exact amount of sample going into the column.15, 18_{In on-column} injection the syringe goes directly into the column. Mostly a column is used with an inner diameter of 0.53 mm, which is wide enough to introduce the needle. This part of the column is deactivated, so that the analytes are not retained when injected.19_{This part of column is called the retention gap.}20 The most elementary principle of cool on column is that the sample is not injected in a hot inlet and then evaporated, but in COC the evaporation point is the same as the injection point.18_{The inlet} temperature around the column can be controlled and is usually in ‘track oven’ mode, where it is typically about 3 degrees higher than the oven. The starting temperature should be below the boiling point of the solvent so that the deactivated on-column part can be wetted, but the boiling

(9)

point should be crossed fast to evaporate the solvent directly. A disadvantage of the COC injection is that the sample is completely injected onto the retention gap, which requires the user to replace the retention gap over time. An advantage of COC injection is that the injection volume can be increased by increasing the length of the retention gap. This can be of interest in trace analysis.20

2.1.2. Retention in GC

Depending on the coating and the temperature, an analyte has a specific retention on the column which results in a specific elution time. Retention on the column is the result of the distribution of the analyte molecules in the mobile and stationary phase, which is called partitioning. Only the molecules in the mobile phase can be transported.

(1) 𝜈𝑖 = _𝑞 𝑞𝑖,𝑚

𝑖,𝑚+ 𝑞𝑖,𝑠 𝑢

In equation 1, νi is the migration speed, qi,m is the quantity of analyte in the mobile phase, qi,s the quantity in the stationary phase and u is the velocity of the mobile phase. When the analytes elute from the GC column and are detected, a retention time (tR,i) is registered, which can be expressed as a function of the column length L and the migration speed (Equation 2).

(2) 𝑡𝑅,𝑖 = 𝐿 / 𝜈𝑖

(3) 𝑡0= 𝐿 / 𝑢

In equation 3, the time t0 that the mobile phase spends in the column is related to L and the linear gas velocity u0. Equation 4 is the result of combining the three equations above.

(4) 𝑡𝑅,𝑖= �1 +_𝑞𝑞𝑖,𝑠 𝑖,𝑚� 𝑡0 (5) 𝑘𝑖 =_𝑞𝑞𝑖,𝑠 𝑖,𝑚 = 𝑐̅𝑖,𝑠𝑉𝑠 𝑐̅𝑖,𝑚𝑉𝑚 (6) 𝑡𝑅,𝑖 = (1 + 𝑘𝑖 )𝑡0 (7) 𝑘𝑖 =𝑡𝑅,𝑖_𝑡− 𝑡0 0

Factor ki, called the retention factor, differs between components. When two analytes have a similar retention factor, it will be more difficult to separate them.21_{Analytes can only be separated in a} sample if their migration speeds are sufficiently different and their widths are sufficiently minimized. The factor that shows the ability to separate two specific analytes is selectivity α.

(8) 𝛼𝑖𝑖=𝑘_𝑘𝑖 𝑖

All the chromatographic information mentioned above is general and applicable to all types of chromatography. When zooming in on gas-liquid chromatography, these equations can be changed so that they include the parameters that can be varied in the optimization process. In equation 5, the quantity in the mobile and stationary phase is already translated to average concentration times the total volume of that phase. The concentration in the stationary phase and the mobile phase can also be described as:

(9) 𝑐𝑖,𝑠 =𝑥𝑖,𝑠_𝑀𝑝𝑠 𝑠

(10)

(10) 𝑐𝑖,𝑚 =_𝑣1 𝑖.𝑚 =

𝑝𝑖

𝑍𝑖𝑅𝑅

Where xi,s is the mole fraction of the analyte in the stationary phase, ρs the density and Ms the molecular weight of the stationary phase. In equation 10, vi,m is the partial molar volume of the analyte in the mobile phase, ρi the partial pressure of the analyte and Z the compressibility

coefficient. Henry’s law states that the amount of dissolved gas in the gas phase is proportional to its partial pressure.22

(11) 𝑝𝑖= 𝑥𝑖,𝑠𝛾𝑖,𝑥∞𝑝𝑖0

In Equation 11, γ is the activity coefficient of the analyte in the liquid phase and p0_{the vapor}

pressure of the pure analyte. When equations 9, 10, and 11 are combined and the ratio between the two concentrations is compared, the following equation is obtained:

(12) _𝑐𝑐𝑖,𝑠 𝑖,𝑚 = 𝑝𝑠𝑅𝑅 𝑀𝑠𝛾𝑖,𝑠∞𝑝𝑖0 (13) 𝑘 =_𝑐𝑐𝑖,𝑠𝑉𝑠 𝑖,𝑚𝑉𝑚 = 𝑝𝑠𝑅𝑅𝑉𝑠 𝑀𝑠𝛾𝑖,𝑠∞𝑝𝑖0𝑉𝑚= 𝑅𝑅𝑛𝑠 𝛾_𝑖,𝑠∞_𝑝 𝑖0𝑉𝑚

From equation 13, the partitioning between the phases and thus the retention depends on three major factors. First, the vapor pressure of the pure solute is a strong function of temperature and so, temperature can be used as parameter to predict and affect retention. Secondly, the activity

coefficient of the analyte in the stationary phase also has a key role in the retention. Changing the type of column leads to changing the values for γ completely. The last parameter that is a major factor in the retention factor is temperature. This factor is of such importance because it can be changed during the gas chromatographic run.

Because the retention factors of components in the GC can differ incredibly, it is of interest to make sure that it changes during the run. By increasing the temperature the retention decreases and the compounds will elute faster.

2.1.3. The Golay Equation

21

The plate height of an open column is given by the following Golay equation.21, 23_{In this equation, D} m is the diffusion coefficient of the analyte in the mobile phase, Df the diffusion coefficient in the stationary phase, u the linear velocity of the mobile, dc the diameter of the column and df the stationary film thickness.

(14) 𝐻 =2𝐷𝑚_{𝑢 + 𝑓}(𝑘)𝑑_𝐷𝑐2𝑢

𝑚 + 𝑔(𝑘)

𝑑𝑓2𝑢

𝐷𝑓

Where f(k) and g(k) are given by the formulas in equation 15, where k is the retention factor. Both equations are plotted in Figure 2.

(11)

To simplify the equation two dimensionless parameters are introduced in equation 16: reduced plate height and reduced flow velocity. This gives a simplified Golay equation in equation 17.

(16) ℎ =_𝑑𝐻 𝑐 𝑎𝑛𝑑 𝑣 = 𝑢𝑑𝑐 𝐷𝑚 (17) ℎ =2_{𝑣 + 𝑓}(𝑘)𝑣 + 𝑔(𝑘) �𝑑𝑓2𝐷𝑚 𝑑𝑐2𝐷𝑓� 𝑣

When the dimensionless film thickness is introduced in equation 18, the final simplified Golay equation is written down in equation 19.24

(18) 𝛿𝑓 =𝑑_𝑑𝑓 𝑐�

𝐷𝑚

𝐷𝑓

(19) ℎ =2_{𝑣 + 𝑓}(𝑘)𝑣 + 𝑔(𝑘)𝛿𝑓2𝑣

When k is 1, which means that g(k) and f(k) are constant, the effect of multiple film thicknesses are shown by the Van Deemter plots in Figure 3 with reduced flow velocity on the x-axis and reduced plate height on the y-axis. What can be seen is that the optimal reduced plate height is the smallest when the film thickness is smaller and that when the reduced film thickness is increased, the graph tends to be sharper, i.e. there is less variation of reduced flow velocity possible without increasing the reduced plate height majorly. In Figure 4, four different retention factors are plotted for a static reduced film thickness. What can be seen is that the lower the retention factor is, the smaller the optimal reduced plate height is and that the difference between the optimal flow is not much different than other flows. With a high retention factor, the reduced plate height increased fast when a deviation from the optimal flow is made. Reducing the retention factor by increasing the temperature creates the possibility to go to higher flows, because there is a minimal difference in reduced plate height when the flow deviates from the optimal flow.

0 0.05 0.1 0.15 0.2 0 1 2 3 4 5 6 7 8 9 10 K

Factors f(k) and g(k)

f(k) g(k)

(12)

Figure 3. The effect of the reduced film thickness on the Golay equation with retention factor = 1.

Figure 4. The effect of the retention factor on the Golay equation with reduced film thickness = 0.3.

0 2 4 6 8 10 12 14 0.5 5.5 10.5 15.5 20.5 25.5 30.5 35.5 40.5 45.5 50.5 55.5 Red uc ed p la te h ei gh t ( h)

Reduced flow velocity (v)

Effect of the reduced film thickness on the Golay

equation (k=1)

δf = 0.1 δf = 0.3 δf = 0.6 δf = 1 0 1 2 3 4 5 6 0.5 5.5 10.5 15.5 20.5 25.5 30.5 35.5 40.5 45.5 50.5 55.5 Red uc ed p la te hei gh t

Reduced flow velocity

Effect of retention factor on the Golay equation

(δf = 0.3)

k = 0 k = 0.5 k = 1 k = 4

(13)

2.2. Supersonic Molecular Beam

Supersonic molecular beam GC-MS has a different and unique interface between gas

chromatography and mass spectrometry. Instead of a direct insertion of the GC column in the ion source, the sample flow goes through a different process of three general steps. These steps are indicated in Figure 5 and will be described in the next three sections.

2.2.1. Transfer Line and Makeup Gas

After the gas chromatographic separation, the column is connected to a heated transfer line. This is step 1 of the supersonic process. With the electronic flow control, an extra flow of helium is added to the transfer line. This makeup gas flow and the column flow are mixed in the black box, indicated in the schematic diagram of the supersonic GC-MS with a 1.25_{The temperature of this transfer line is} controlled and is set high to prevent a cold spot in the retention of the analytes.26_{The makeup gas} flow is also controlled and can be in constant pressure and constant flow mode.

2.2.2. Expansion from the Supersonic Nozzle

After mixing the GC flow with the makeup gas flow, the flow is forced through a supersonic 100µm nozzle. The nozzle has a conical inner shape, which creates a supersonic molecular beam after the nozzle.27_{This is step 2 in the schematic diagram. When approach this situation, it is assumed that the} both chambers before and after the nozzle are of infinite volume. Also, the nozzle vacuum chamber should be in perfect vacuum. The jet direction will be called the z-axis, which is normal to the nozzle plane. This plane is assumed to be the origin of distances, but the expansion starts before this point already at z0, with p0 and T0. The nozzle separates two different thermodynamic situations, the subsonic with a Mach number <1 and z < 0. The Mach number is the velocity divided by the velocity of sound. The other regime is supersonic with a Mach number greater than 1 and a z value higher than 0. When the gas expands from the high-pressure zone to the low-pressure zone, the atoms and molecules undergo many collisions as they move away. They gain a Mach number above 1 rapidly, which creates the zone of silence: a zone where the atoms move faster than sound, which results in silence. The internal and translational degrees of freedom are cooled. Any molecule in the expansion will eventually have about the same speed and will be effectively cooled to a temperature of a few K.28_{The temperature and mean velocity in the beam can be described by the following two laws of} Figure 5. Schematic diagram of a supersonic molecular beam interface between GC and MS.

(14)

isentropic expansion. This is a process that is both reversible and adiabatic, i.e. without heat transfer between thermodynamic systems.

(20) 𝑅_{𝑅 = 1 +}0 (𝛾 − 1)ℳ₂ 2 (21) 𝑣̅ = �2𝐶𝑝(𝑅_𝑀0− 𝑅)� 1 2 = ℳ �𝛾𝑅𝑅_{𝑀 �} 1 2

Where T and v are the temperature and mean velocity of the molecular beam, T0 the stagnation temperature, γ the specific heat ratio Cp/Cs, M the molecular weight of the gas, M the Mach number, Cp the molar heat capacity, and R the gas constant per mole.29 The resulting beam has

monochromatized properties, where the beam velocity comes from mass motion. Because of these thermodynamic laws, the makeup gas cools. In every collision, there will be a heat transfer from the analyte to the gas. In this way, the analyte will cool by losing its translational and vibrational

energy.30_{There will be more collisions in the first part of the z-axis, so the most significant cooling} will take place in the first centimeter.28

2.2.3. Free-Jet Separation

The analyte in the molecular beam will collide with the cooling helium. The formula for momentum is given by the next formula:

(22) 𝑝 = 𝑚𝑣

Where p is momentum, m is the mass and v the velocity. The momentum of the analyte is much higher than the momentum of the gas, since they have approximately the same velocity but a significant difference in mass. In this way, the analyte comes in a free-jet separation along the z-axis, while most of the helium will be pumped off. A decreasing intensity would be expected along the z-axis, but the opposite is true. With increasing z, the intensity increases and decreases after a maximum. 30_{The reason for this is that there is almost no change of direction because of the}

difference in momentum between the analyte and the helium. It is expected that the more collisions the analyte undergoes, the more energy it is lost to the helium and thus is cooled more. Increasing the length between the nozzle and the skimmer thus creates an increase in cooled analyte. There is a maximum since the helium expands and the number of effective collisions decreases with the distance. The biggest part of the cooling happens in the first part of the z-axis and from a specific maximum the cooling will not increase and the intensity will decrease. The temperature of the expanded analytes in heavy rare gases is about 5-7 K. Heavy noble gasses provide more effective vibrational cooling, which can be rationalized by the velocity slip effect. The analytes in the molecular beam are often of higher mass than the makeup gas and thus travel at lower velocities. This means that most initial collisions after the nozzle serve to accelerate the analyte to the velocity of the helium. In an argon expansion, the velocity is much lower and this results in a higher number of collisions available for cooling. The rotational temperatures achieved are about 0.1-10 K and the vibrational temperatures are about 20-150 K, varying by the use of makeup gas.31-32_{The distance of} the effective cooling depends on the ratio between diameter of the nozzle and the skimmer distance, on the gas that is used and the P0, the pressure before the nozzle.33,34 The jet will

(15)

2.3. Electron Ionization

Electron ionization or electron impact is the most used ionization technique in GC-MS. When the gas phase molecules enter the ionization chamber, they are hit by an energetic electron coming from a filament. When the collision is sufficiently effective and the energy transferred exceeds the

ionization energy, the molecule is ionized by the ejection of an electron. This results in the formation of an odd electron molecular ion with a positive charge, which can be seen in equation X.

(23) 𝑀 + 𝑒−_{→ 𝑀}+∙_{+ 2𝑒}−

What happens after the creation of the molecular ion depends on multiple parameters. The first parameter is the ionization energy, which is the energy that is needed to be absorbed by a molecule in its electronic and vibrational ground states to form an ion that is in its ground state by ejection of an electron.35_{Depending on the difference between this minimal energy needed for ejection of an} electron and the electron energy of the colliding electron, there is energy left for other

fragmentation reactions. Another parameter for what happens with the molecule after ionization is the vibrational and rotational energy levels of the molecule, which mostly depend on the

temperature the molecule is in. Since there is a gas chromatograph before the ionization and GC is often run with a temperature program, the ionized molecules contain a lot of internal energy. If molecules have high internal energy, they will fragment to a large extent; when it is low, there will be almost no fragmentation.36_{In what structure the electron ionization takes place is an important} parameter in the ionization efficiency. For this reason electron ionization sources will be discussed as well.

2.4. Electron Ionization Sources

To perform electron ionization, multiple types of ionization sources can be used. To understand how the ionization takes place in the supersonic molecular beam, the original types of ionization sources have to be examined.

2.4.1. Nier-type Ion Source

The most commonly used electron ionization source for GC-MS is a Nier type ion source, named after Alfred Nier, who

presented it in the 1940s.37_{In Figure 6 a diagram is shown of an} electron ionization or electron impact source. The GC column enters the ionization chamber and is perpendicularly hit by the electron beam coming from the filament.1_{This type of ion} source is called a Nier type, where the heated filament is positioned outside of the ionization chamber and for this reason, it is a closed ion chamber.38,37_{Because the electrons} come from outside the box and the hot filament is located outside the cage, the degree of sample degradation is limited.39

2.4.2. Brink-type Ion Source

In GC-MS with supersonic molecular beams, Brink ion sources are used.39_{Gilbert Brink designed this} type of ion source for molecular beams.40_{The ion source has a filament inside the ion source, which} is much longer than the filament in the Nier type. This means that the ionization probability is much higher in a Brink type than a Nier type. An increase in ionization probability is necessary, because a lot of analyte gets lost in the molecular beam. Another effect of the hot filament inside the cage is a higher degree of sample degradation. The axial jet of analytes flying through the ion source prevents

(16)

contact between the filament and the analyte and thus decreases degradation. In the science of molecular beams, the Brink type fly-through ion source is used throughout.41,42

2.4.3. Dual Cage Electron Ionization

A new ion cage was developed by the group of Aviv Amirav in Tel Aviv. The disadvantages of a fly-through source, i.e. sample degradation, are even of less importance in a supersonic molecular beam. Since the molecules lose their internal energy and are accelerated in the molecular beam, even when they scatter on the filament heated ion source surface, they lose their high kinetic energy. As a result, even if ionized, these charged heated analytes can be filtered out by the ion source lens, together with other thermal vacuum background.3_{The vacuum background filtration is established because of the} difference between the kinetic energy of background ions and that of analytes. The new dual cage ion source was made in a response to the incomplete background filtration in the normal Brink type ion source. This was due to the false assumption that the central ion cage was unaffected by external field: the field of the filament created an asymmetrical path for the ions. For this reason, the group created a dual cage principle where the inner cage did have an electric field close to zero. In Figure 7, a schematic diagram is shown of the dual cage EI source. The analytes first enter the source through the skimmer at 8, after which they fly through the ion formation cage (4) and the outer shielding cage (3). Outside of the outer cage, the electron emitting filament is positioned (1). Around the entire ion source, an electron repeller is located (2). The optic lenses are located behind the cage; one lens focuses the ions (6) and the other lens filters background ions (7).3

2.5. Quadrupole Mass Spectrometry

1

2.5.1. The Electric Field

The exited ions are reflected via an ion mirror into the quadrupole mass spectrometer (QMS). The QMS consists of 4 parallel rods, which are all electrically connected to the opposing rod, shown in Figure 8. The mass to charge filter is obtained by the application of direct and alternating currents to the rods, which creates an electric field. When the trajectory of the ion is not stable, it won’t reach the detector. The first set of coupled rods is set to an electrode potential, given by the following formula.

(24) 𝑉𝑡𝑡𝑡= 𝑈 − 𝑉𝑐𝑉𝑉(𝜔𝑡)

The other couple of rods has the same electrode potential, but in reverse polarity. This formula is build up with a direct current (DC) voltage U and an alternating current (AC) with V as peak amplitude and angular frequency ω. The angular frequency is described as

(25) 𝜔 = 2𝜋𝑣

With ν being the frequency in Hz. The filtering property of the mass analyzer is based on the time dependence of the potential in the x and y directions. The expression of the potential in a

quadrupole field is

Figure 7. Dual Cage Electron Ionization Source.3

Figure 8. Schematic diagram of a quadrupole mass spectrometer.2

(17)

(26) 𝜙 =𝜙_𝑟0

02(𝜆𝑥

2_{+ 𝜎𝑦}2_{+ 𝛾𝑧}2₎

Where φ0 is the applied electric potential, λ is a weighing constant for the x-coordinate, σ is a weighing factor for the y coordinate and γ the weighing factor for the z-coordinate. R0 is a constant depending on the geometry of the mass analyzer. There is no electric field along the z-axis, since it is perpendicular to the x and y fields and thus the z factor will become zero.

(27) 𝜙 =𝜙0 𝑟02(𝜆𝑥

2_{+ 𝜎𝑦}2₎

This has to satisfy the Laplace equation:

(28) 𝛿_𝛿𝑥2𝜙₂ +_𝛿𝑦𝛿2𝜙₂ =2𝜙_𝑟0𝜆 02 + 2𝜙0𝜎 𝑟₀2 =2𝜙_𝑟 0 02 (𝜆 + 𝜎) = 0 (29) 𝜆 + 𝜎 = 0 → 𝜆 = −𝜎 If we set lambda is 1, the expression is reduced to

(30) 𝜙 =𝜙0 𝑟02(𝑥

2_{− 𝑦}2₎

2.5.2. Forces on the Ion

The field equations given in paragraph 2.4.1. describe separate uncoupled coordinates and the forces working on the ion can be determined separately. The force that the ions undergo in the x and y direction is given by the following equations

(31) 𝐹𝑥 = 𝑚𝑎 = 𝑚𝑑 2_𝑥 𝑑𝑡2 = −𝑧𝑒 𝛿𝜙 𝛿𝑥 (32) 𝐹𝑦= 𝑚𝑎 = 𝑚𝑑 2_𝑦 𝑑𝑡2 = −𝑧𝑒 𝛿𝜙 𝛿𝑦

If equation 30 is rewritten with 31 and 32 for the x and y direction we get the following formula (33) 𝛿𝜙_{𝛿𝑥 =}2𝑥_𝑟 02(𝑈 − 𝑉𝑐𝑉𝑉(𝜔𝑡)) (34) 𝛿𝜙_{𝛿𝑦 =}−2𝑦_𝑟 02 (𝑈 − 𝑉𝑐𝑉𝑉(𝜔𝑡)) Thus (35) 𝐹𝑥 = 𝑚𝑑 2_𝑥 𝑑𝑡2 = − 2𝑧𝑒𝑥 𝑟₀2 �𝑈 − 𝑉𝑐𝑉𝑉(𝜔𝑡)� (36) 𝐹𝑦= 𝑚𝑑 2_𝑦 𝑑𝑡2 = 2𝑧𝑒𝑦 𝑟02 �𝑈 − 𝑉𝑐𝑉𝑉(𝜔𝑡)�

When equations 35 and 36 are rearranged, the motion of a singly charged ion in a quadrupole field is found.

(37) 𝑑_𝑑𝑡2𝑥₂+ 2𝑧𝑒

(18)

(38) 𝑑_𝑑𝑡2𝑦₂−_𝑚𝑟2𝑧𝑒

02�𝑈 − 𝑉𝑐𝑉𝑉(𝜔𝑡)�𝑦 = 0

When a new dimensionless parameter is used, phase of the alternating field ξ. Its description and its derivatives are given:

(39) 𝜉 =𝜔𝑡₂ (40) _{𝑑𝑡 =}𝑑 𝑑𝜉_𝑑𝑡_{𝑑𝜉 =}𝑑 𝜔₂_𝑑𝜉𝑑 (41) _𝑑𝑡𝑑2₂=𝑑𝜉 𝑑𝑡 𝑑 𝑑𝜉 � 𝑑 𝑑𝑡� = 𝜔2 4 𝑑2 𝑑𝜉2

The combination of 39, 40 and 41 with equation 37 and 38 now gives the Mathieu differential equations: (42) 𝑚𝜔₄4𝑑_𝑑𝜉2𝑥₂+2𝑧𝑒𝑥𝑈_𝑟 02 − 2𝑧𝑒𝑥𝑉𝑐𝑉𝑉(2𝜉) 𝑟₀2 = 0 (43) 𝑚𝜔₄4𝑑_𝑑𝜉2𝑦₂−2𝑧𝑒𝑦𝑈_𝑟 02 + 2𝑧𝑒𝑦𝑉𝑐𝑉𝑉(2𝜉) 𝑟₀2 = 0

Multiplication with 4/mω2_gives

(44) 𝑑_𝑑𝜉2𝑥₂+_𝑚𝑟8𝑧𝑒𝑥𝑈 02𝜔2− 8𝑧𝑒𝑥𝑉𝑐𝑉𝑉(2𝜉) 𝑚𝑟₀2_𝜔2 = 0 (45) 𝑑_𝑑𝜉2𝑦₂+ 8𝑧𝑒𝑦𝑈 𝑚𝑟₀2_𝜔2− 8𝑧𝑒𝑦𝑉𝑐𝑉𝑉(2𝜉) 𝑚𝑟₀2_𝜔2 = 0

To simplify this formula, two parameters a and q are defined. (46) 𝑎𝑥 = −𝑎𝑦=_𝑚𝑟8𝑧𝑒𝑈

02𝜔2

(47) 𝑞𝑥= −𝑞𝑦=_𝑚𝑟4𝑧𝑒𝑉 02𝜔2

This creates the simplified versions of equation 44 and 45

(48) 𝑑_𝑑𝜉2𝑥₂+ (𝑎𝑥− 2𝑞𝑥cos(2𝜉))𝑥 = 0

(49) 𝑑_𝑑𝜉2𝑦₂+ �𝑎𝑦− 2𝑞𝑦cos(2𝜉)�𝑦 = 0

2.5.3. Mass Analysis with the Quadrupole Mass Spectrometer

The mass analysis is achieved by scanning U and V at a fixed ratio. Since they are maintained constant ratio, there is a constant slope of a/q. This line is given as follows

(50) 𝑎_𝑞𝑥 𝑥 = 8𝑧𝑒𝑈 𝑚𝑟₀2_𝜔2 𝑚𝑟02𝜔2 4𝑧𝑒𝑉 = 2𝑈 𝑉

(19)

(51) 𝑈 =𝑟02𝜔_8𝑒2𝑎𝑥𝑚_{𝑧 = −}𝑟02𝜔_8𝑒2𝑎𝑦𝑚_𝑧 (52) 𝑉 =𝑟02𝜔_4𝑒2𝑞𝑥𝑚_{𝑧 = −}𝑟02𝜔_4𝑒2𝑞𝑦𝑚_𝑧 Since 𝑟02𝜔2

8𝑒 and 𝑟02𝜔2

4𝑒 are constant, the trajectories for a given

ion with a specific m/z value in the U-V diagram is identical to the q-a diagram. Every mass over charge ratio has its own stability region in the U-V diagram, which is shown in Figure 9. In the mass analysis, U and V are both increased over the run to filter out each specific mass. In advance, the mass range that has to be analyzed can be chosen in the method.

2.6. Pyrolysis

Pyrolysis is the thermal decomposition of a material at high temperatures under an inert atmosphere.43_{It can be used in} front of a mass spectrometer, a gas chromatograph or a gas

chromatography – mass spectrometry setup. The thermal decomposition is necessary in the case of nonvolatile materials to create volatile compounds that can be analyzed in the previously mentioned setup. Pyrolysis is a common technique and is also used in combination with molecular beam mass spectrometry.44,45

2.6.1. Lignin

To study the pyrolysis of natural materials, some standards of lignin will be used. Lignin is the most abundant organic substance on earth after cellulose. It is a structural polymer, present in all woody plants. It is unique and

essential for vascular plants to adapt to life on land. It protects the plant against gravity, wind and microorganisms, but it also seals the water-conducting system of the plant.46_{Its three monomeric precursors are} p-coumaryl alcohol, coniferyl alcohol and sinapyl alcohol, shown in Figure 10.47

2.6.2. Cellulose

Cellulose is the most abundant natural polymer and consists of monomers of D-glucose. The polysaccharide consists of crystalline and amorphous domains, depending on its source and history.48_{In this research two types of cellulose are used, one of which is a derivative of cellulose} with acetic acid.

2.6.3. Synthetic polymers

Further research will be done on manmade polymers. Those will be polyethylene, polyisobutene (PIB), polybutadiene, Kraton and Nylon. Polyethylene is a polymer of ethene, which is showed in Figure 12. Polyisobutene is a polymer of isobutene. This

polymer has a lot of methyl branches and is shown in Figure 11. Polybutadiene is a polymer of 1,3-butadiene. It polymerizes via carbon 1 and 2, which leaves a double bond for crosslinking, or via carbon 1 and 4, which leaves a double bound in the middle of the molecule (between carbon 2 and 3). Because of these two additions and the possibility to crosslink, the polymer is a thermoset. Kraton is a synthetic replacement for rubber, made by Shell from the 1950s

Figure 10. p-Coumaryl alcohol (150), Coniferyl alcohol (180) & Sinapyl alcohol (210).

Figure 12. Polyethylene Figure 11. Polyisobutene (PIB). Figure 9. Stability Diagram for a single ion mass. (Picture from Journal of Chemical Education, Vol. 75, No. 8, 1998, p. 1051)

(20)

until 2001, when the Kraton polymer business was sold. It is a block polymer, consisting of styrene (ethenylbenzene) blocks and rubber blocks, which consist of polybutadiene and polyisoprene.

2.7. NIST Library search

The use of the NIST library was already mentioned in the introduction as one of the advantages of GC-MS with electron ionization. When a spectrum is uploaded to the NIST library search, it is compared to the spectra in the library. The program provides a comparison function, which gives a match factor and a probability per library spectrum.6_{The match factor consists of the combination} of three general classes: peak series, characteristic ions, and neutral losses. The match factor is created from a linear combination of individual match factors.49,50 _{While the match factor only} includes the two spectra that are compared, the assigned probability is analysis-specific as it covers the match factors of all the other spectra that are generated.7_{Consequently, a library spectrum that} has a high match factor, does not automatically have a high result probability and, for that reason, does not necessarily end up high on the hit list. To be on top of the hit list, the compound must have a match factor that is much higher than all of the other match factors.

Because the molecular ion is increased in the supersonic molecular beam GC-MS, the generated spectrum is less similar to the library spectrum and the match factor will also decrease. On the other side, the match factors for other spectra that are not of the analyzed compound, the match factor will be even lower. For example, the spectrum of octane has a molecular ion of 114. The library spectra that will come up are probably heptane, octane, and nonane. None of the spectra in the library has an increased peak at 114, but finding the peak of 114 in nonane and heptane is not expected at all. The match factor for the spectrum of octane will be higher than that for the spectrum of nonane and heptane. Since there will be such a difference between the match factors, the library probability for octane will be highest.7,51

(21)

3. GC-SMB-MS Troubleshooting

When this research started, the supersonic molecular beam was not working and there was no signal obtained yet. An overview will now be given what steps have been taken to overcome the startup problems the SMB had. The paragraphs are indicated in Figure 13, which is a schematic diagram of the supersonic molecular beam interface.

Figure 13. Schematic diagram of the interface between the GC and the MS with the SMB: paragraph numbers indicated.

3.1. Negative flow in the EFC

When starting up the machine, the electronic flow control continuously gave a negative flow. There is a pressure control and a flow control mode in the EFC. The maximum pressure that could be used in the pressure control was 1300 mbar, and even this pressure gave a negative flow of about -6 ml/min. In the control panel of the supersonic molecular beam, the power of the turbopump was also low, as if there was no gas in the nozzle vacuum chamber. The pipes were checked whether helium was going in the flow control, which was the case. When the flow was turned off and on again three times in a row there was some change and a positive flow was obtained. A flow of 5 ml/min corresponded with 900 mbar, 6 ml/min with 1000 mbar and 7 ml/min 1100 mbar. There was a possibility that the nozzle was clogged so the cleaning procedure of the nozzle was started. The MS and the SMB were vented and the nozzle was disassembled. The parts were cleaned by rinsing them in methanol in a sonic bath. After the cleaning they were blow-dried and assembled again. After cleaning, the flow and pressure were working as they should.

3.2. Vacuum of the MS is too low

Since a high flow is now established from the nozzle, it would be expected that the vacuum of the MS should not be about the same as no flow at all. When there is no flow, the vacuum of the MS should be around 10-7_{and with flow it should be about 10}-5_{. In this situation, a flow of 10}-6_{is seen.} The first possible reason for this low vacuum could be a wrong positioning of the transfer line. The transfer line can be positioned with the screws of the XY table. When the screws were turned, they bended and had to be renewed. With the new screws, it was found that the vacuum did not increase.

(22)

lips of the skimmer were not damaged at all. With a flow meter it was checked whether the flow in the control panel was the same as the actual flow coming from the SMB. This was working

accordingly so the problem had to be found somewhere else.

The next step that was taken was to raster the XY position of the transfer line and map the pressure differences when the transfer line is moved 1 mm up/down and left/right. After the experiment, it was found that the transfer line was already in an optimal position. To overcome the difference in pressure, the transfer line was moved in the z-direction towards the skimmer (4 mm). Since most of the cooling happens directly after the nozzle, this is a good balance.

Probably the nozzle needs to be replaced to obtain the highest signal, but for now this is a defect that cannot be changed.

3.3. Electron Multiplier Voltage

When the first isothermal run was performed with a solvent, there were no peaks in the

chromatogram and the noise was also almost zero. The electron multiplier voltage was set in an absolute mode of about 1650 volts. When it was changed to an EM voltage of 1800, a small solvent peak appeared and also the PFTBA memory was showing in the noise. The peaks still missed a factor 100 in intensity and the cooling seemed off. The wires of the electronics box of the supersonic molecular beam were checked, but they seemed all right.

3.4. Ion Source Testing

There are a lot of parameters that can be altered when the ion source is optimized: electron

repeller, electron energy, outer cage voltage, ion energy, lens 1, lens 2, mirror mesh, ion mirror, lens 3 and mirror heating. When they were changed, it was found that they were not responding as they should. That is why it was decided to disassemble the ion source. In this process, it was discovered that the wire of the electron repeller and the outer cage were reversed. This was an extreme fault since the voltage on the electron repeller was at 130 V and the outer cage was only at 6 V. This is why the wire of the electron repeller could not be taken off the pin; the pin eventually broke. When this happened, the complete ion source had to be disassembled to laser weld the pin back on the electron repeller. When the ion source was assembled again and the wires were connected accordingly, the signal increased incredibly.

3.5. High Power Filament & PFTBA memory

The filament was designed for powers below 32 W, but when the filament is exposed to air, through a leak for example, it crosses the 32 watts boundary. After repairing the ion source, this was a reoccurring problem when the high temperature runs were executed. The ferrules used in the inlet and interface to the transfer line were made of graphitized vespel, which degrades under high temperatures (400 °C). As a solution, graphite vespels were used, but since they have a high likelihood to clog the nozzle again, it was decided to go with metal ferrules. Metal ferrules can handle the big temperature fluctuations. There was also a great PFTBA memory in the pipes coming from the electronic flow control. The reason for this was the extensive use of PFTBA in the

optimization process and it declined when the oven was heated to high temperatures. This should be kept in mind that there is a chance to create PFTBA memory when used often.

3.6. Band broadening through transfer line

Another problem that was manifested during the setup period was that the transfer line had a maximum temperature of 350 °C, while the Polywax® experiments were performed in the GC to a temperature of 400 °C. This means that because of the temperature difference the peaks will broaden in the transfer line, because they will have a higher retention on the column in the transfer line than in the oven. To solve this problem, a piece of deactivated fused silica capillary was put

(23)

through the transfer line and the GC was connected to the deactivated capillary in the oven of the GC. This prevented the band broadening of the later eluting compounds.

3.7. Discrimination

When the column was first installed, a split/splitless injection was used. The first experiments performed consisted of lower mass analytes. Later, when the Polywax® experiments started, this became a problem since the split inlet discriminates between low and high masses. For this reason, a retention gap was installed in the back inlet to create an on-column inlet, which follows the

temperature of the oven. This retention gap capillary is deactivated, which means that the setup consists of two retention gaps in front and behind the column, connected by press-fits.

(24)

4. Experimental Methods

4.1. Grob mixture Experiments

To examine the effect of the supersonic molecular beams on different types of compounds, the setup is tested with a modified Grob mixture, which was created within Shell to test the quality of one-dimensional and two-dimensional GC set-ups.52_{The analytes are shown in Table 1 with their} boiling points. The mix was analyzed in 4 different setups. All analyses were made on similar boiling point columns, however four different types of ionization method have been used. To compare the data of the SMB to the standard way of working, the analyses were performed in chemical ionization and electron ionization mode. With the supersonic molecular beam GC-MS, the mass spectra were recorded at 70 eV electron energy and 20 eV electron energy. The spectra are analyzed and compared to spectra of the ’NIST’ mass spectral database (https://chemdata.nist.gov) and their probabilities, matching factor and relative matching factor have been registered.

Table 1. List of compounds in modified Grob mix

No. Compound Molecular Weight

1 Toluene 92

2 2,3-Butanediol 90

3 Nonane 128

4 Naphtalene, decahydro- trans 138

5 1-Octanol 130

6 Hexanoic acid, 2-ethyl 144

7 Phenol, 2,6-dimethyl 122

8 Benzenamine, 2,6-dimethyl 121

9 Naphtalene 128

10 Dodecane 170

11 Decanoic Acid, methyl ester 186

12 Tetradecane 198 13 Cyclohexanamine, N-cyclohexyl 181 14 Hexadecane 226 15 Octadecane 254 16 Nonadecane 268 17 Eicosane 282

18 9-Octadecenoic acid, methylester 296

4.2. Mapping the Supersonic Molecular Beam

The supersonic molecular beam can be controlled through the electronic flow control. Depending on the flow through the GC, the total flow is established. Five sets of experiments have been performed where the GC was running at 3 different constant pressures and the supersonic molecular beam was changed in a constant flow and a constant pressure mode. The constant experimental settings for al experiments performed in this research are mentioned in Table 2. In Table 3 the constant settings for experiment 4.2 are described. The gas chromatograph has been operated in constant pressure mode at three different pressures: 100 kPa, 150 kPa and 220 kPa. With a back pressure of 70 kPa (700 mbar) and a temperature of 40 °C, the linear velocity of the helium is respectively 380 cm/sec, 511 cm/sec and 689 cm/sec. At 400 °C, the linear velocities are respectively 225 cm/sec, 303 cm/sec and 408 cm/sec. The makeup gas is controlled in either constant flow or constant pressure mode. The constant pressure is varied from 600 mbar to 1200 mbar with 50 mbar intervals. In the constant

(25)

flow mode, the constant flow is varied from 20 ml/min to 120 ml/min with intervals of 10 ml/min. This is shown in Table 4.

Table 2. Constant experimental settings for all the experiments performed with the GC-SMB-MS setup.

GC Oven Agilent 7890B Injection Volume 1 µL MS Agilent 5975 EM Voltage 1800 V Threshold 0 Emission 3 mA Electron Energy 70 eV

SMB instrument Aviv Analytical 5975-SMB 101-09

Table 3. Constant settings of experiment 4.2.

Sample SIMDIS standard (0.0357 g in 25 mL cyclohexane)

Column 1HT-INFERNO Zebron (5mx0.32mmx0.1µm)

GC program 60 °C (1 min hold)  400 °C (5 min hold) with 15 °C/min Injection method On-column (track oven)

Scan 50-1050 m/z

Transfer line 350 °C

Table 4. Supersonic molecular beam experiments performed.

GC pressure Constant Pressure of Makeup Gas Constant Flow of Makeup Gas 100 kPa 600, 650, 700, 750, 800, 850, 900, 950,

1000, 1050, 1100, 1150, 1200 mbar 20, 30, 40, 50, 60, 70, 80, 100, 110, 120 ml/min 150 kPa 600, 650, 700, 750, 800, 850, 900, 950,

1000, 1050, 1100, 1150, 1200 mbar 20, 30, 40, 50, 60, 70, 80, 100, 110, 120 ml/min 220 kPa 600, 650, 700, 750, 800, 850, 900, 950,

1000, 1050, 1100, 1150, 1200 mbar (not performed)

In all experiments, the peaks are integrated and exported to a text format. This format is imported into excel, where the calculations are made. The intensity of the molecular ion is compared to that of the fragments. Also the doubly charged molecular ions have been observed, so the change of the doubly charged ion is of interest in this experiment as well.5_{Since the doubly charged molecular ion} peak lies between the fragment peaks, it is important that it is visible in the spectrum. That is why the mean intensity of the fragments is calculated and compared to that of the doubly charged molecular ion. To calculate the intensity of the molecular ion, a mass range around it of 5.5 (-1.5 and +4) is summed up. For the doubly charged molecular ion, this mass range is 2.5 (-0.5 and +2). The fragment intensity is the sum of all the fragments from m/z value 50 to 115.

4.3. Effect of electron energy on high alkanes

As an addition on the previous experiment, a small experiment has been performed to find the effect of electron energies on the intensity of the molecular ion and the doubly charged molecular ion. The experimental method is given in Table 5. In total 4 experiments have been performed by combining 20 eV and 70 eV electron energy with makeup gas flows 70 ml/min and 80 ml/min. The only difference between the electron energies is that the mass spectrometer was tuned in an

(26)

optimal state for both electron energies with perfluorotributylamine (PFTBA). Again, the molecular ion intensity is compared with fragment intensity and the doubly charged molecular ion intensity. Table 5. Experimental settings of experiment 4.3.

Sample Polywax 1000 standard (0.0493 g in 15 mL cyclohexane

GC program 60 °C (1 min hold)  400 °C (5 min hold) with 15 °C/min GC constant Pressure 100 kPa

Makeup gas Flow 70 ml/min & 80 ml/min Electron Energy 70 eV & 20 eV

Injection method On-column (track oven)

Scan 50-1050 m/z

4.4. GC-SMB-MS as quantitative technique

4.4.1. Replicating GCxGC-FID and SimDis GC-FID data

To investigate whether GC-SMB-MS is a quantitative technique, two GTL samples have been analyzed. Both samples (A and B) are Fischer-Tropsch products, which consisted mostly of normal alkanes and some branched alkanes. The samples were also analyzed with GCxGC-FID and SimDis GC-FID. The quantitative data of the two analytical methods is similar and gives the mass percentage per carbon number in the sample. The GCxGC data gives the carbon distribution up to C40, while the SimDis goes up to C80. Both methods make use of FID detectors, which measure the amount of carbon in the molecules. The goal of the experiment is to determine the relative carbon number distribution in the sample, not to determine absolute quantities. The experimental settings are described in Table 6.

Table 6. Experimental settings of experiment 4.4.1, 4.4.2. and 4.4.3.

Sample GTL sample A and B (dissolved in isooctane and heated to 70 °C)

GC program 90 °C (1 min hold)  400 °C (5 min hold) with 15 °C/min GC constant Pressure 100 kPa

Makeup gas Flow 90 ml/min

Injection method On-column 70 °C (track oven)

Scan 50-1050 m/z

4.4.2. Linearity and repeatability testing in the GC-SMB-MS

Following the results of the experiments in the previous paragraphs, a linearity study was carried out on a Polywax500® standard. The experimental method was the same as used in 4.4.1 and is

described in Table 6. Three solutions were made in isooctane: Solution A 0.043 g Polywax500® in 10 mL isooctane Solution B 5 mL solution A with 5 mL isooctane Solution C 5 mL solution B with 5 mL isooctane

When the dilutions were made of solution A and B, they were both heated to 70 °C to make sure that all of the Polywax500® was dissolved. The ratio between the area of the alkanes in solution A

(27)

and B and B and C are analyzed to see whether there is a linearity between the samples. Also solutions B and C were analyzed in triplicate to see if the result is repeatable.

4.4.3. Replicating GCxGC-FID and SimDis GC-FID data with

an internal standard

Following the results of the experiments described in the previous paragraph, a new experiment was designed with an internal standard. This could correct for the inaccuracies in the process. The analytical

method was the same, but two solutions of GTL sample A were made. One of which had half the concentration of the other. The same amount of squalane was added to both the samples. The molecular structure of squalane is shown in Figure 14. This was chosen as internal standard, since it elutes between the normal alkanes.

4.5. Analysis of Hydroperoxides with SMB

The supersonic molecular beam can be used as an informative technique, because of the increased molecular ion. However, there is another advantage that increases the applicability of the SMB. There is a possibility to get rid of the helium flow in between the GC and the MS. This means that high flows in the order of 13 ml/min can be used in the GC analysis, without

sacrificing the vacuum of the MS. In normal GC-MS, where the GC column is inserted directly into the mass spectrometer, only small flows can be used. In this experiment, a proof of principle will be performed to see whether the supersonic molecular beam can be used to analyze larger hydroperoxides, such like 1-hydroxyperoxyethylbenzene (EBHP). Multiple constant isothermal runs will be carried out, from 30 °C to 70°C. The experiment started at 30 °C, because there was no possibility to actively cool the transfer line to lower temperatures. Hydroperoxide analyses should be performed at low temperatures, because they are not thermally stable and can degrade to ketones. This degradation happens in normal GC-MS, when longer columns and higher temperatures are used. The progress of thermal degradation of the hydroperoxide has been

followed in the temperature gradient and based on these results the applicability of GC-SMB-MS on larger peroxides has been assessed. The experimental settings are described in Table 7.

Table 7. Experimental settings of experiment 4.5.

Sample 36% EBHP in ethylbenzene dissolved in toluene (1000x)

GC program 30 °C / 40 °C / 50 °C / 60 °C / 70 °C GC constant Pressure 150 kPa

Makeup gas Flow 20 ml/min

Injection method Split injection with split 10 (room temperature)

Scan 40 – 400 m/z

Transfer line 30 °C / 40 °C / 50 °C / 60 °C / 70 °C

4.6. Comparing Pyrolysis GC-EI-MS with Pyrolysis GC-SMB-MS

Different solid organic materials have been pyrolyzed and analyzed both by standard GC-EI-MS and by supersonic molecular beam GC-MS. The comparison was made on two cellulose samples (Carboxymethylcellulose and Avicel), three lignin samples (Lignin Organosolv, Lignin Hydrolytic and Lignosulfaat) and two wood samples (Wheat straw and Beech wood). The settings of the standard

Figure 14. Structure of Squalane.

(28)

pyrolysis GC-EI-MS are described in Table 8. The settings of the Py-GC-SMB-MS method are described in Table 9.

Table 8. Experimental Settings of the standard pyrolysis GC-EI-MS.

GC Oven Trace GC (Interscience)

GC Column CPSil5 CB (50mx0.32mmx1.2µm)

GC Program 40 °C (10 min hold)  250 °C (10°C/min)

GC Flow 2 ml/min

Injection Split injection with split 50 (250 °C)

MS Trace MS (Interscience)

Scan 20-400 m/z

Pyrolysis instrument PyroProbe Model 5200 (CDS)

Valve Oven 250 °C

Pyrolysis Program 50 °C  750 °C (10 s hold) (20 °C/ms)

Table 9. Experimental settings of the Pyrolysis GC-SMB-MS.

GC Column 5HT-INFERNO Zebron (15mx0.32mmx0.1µm) GC Program 40 °C (1 min hold)  370 °C (15 °C/min)

GC Pressure 100 kPa

Injection Split injection with split (325 °C)

Scan 20-400 m/z

Makeup gas flow 60 ml/min

Pyrolysis Instrument PyroProbe Model 5250 (CDS)

Transfer Line 340 °C

Valve Oven 340 °C

Pyrolysis Program 50 °C  1000 °C (10 s hold) (20 °C/ms)

The standard pyrolysis GC-EI-MS has a long column installed to be able to separate the small

components resulting from the pyrolysis. The standard system has temperatures for the transfer line and valve oven set at 250 °C. The maximum temperature in the PyroProbe was changed to 1000 °C to see what larger components could come out of the pyrolysis.

4.7. Analyzing synthetic polymers with Py-GC-SMB-MS

The Py-GC-SMB-MS setup of the experimental 4.6. was tested for its feasibility for the analysis of samples containing plastics. These polymers were polyethylene, Kraton, Nylon, PIB and

polybutadiene. The same settings were used as in the Py-GC-SMB-MS analysis for lignin and cellulose (paragraph 4.6.)

(29)

5. Results & Discussion

5.1. SMB Performance testing: Grob mixture

5.1.1. NIST probability comparison

The spectra of the components in the Grob mixture have been analyzed in standard EI-MS, in GC-MS with chemical ionization, and GC-GC-MS with supersonic molecular beams with both 70 eV and 20 eV electron energy. The mass spectrum of every compound in every ionization technique, except for chemical ionization, is subsequently searched in the NIST library. The highest probability in matching compounds is given in Figure 16. The compounds analyzed are ordered on their elution time in a boiling point column. The average increase in probability between the standard GC-MS and SMB-70eV is 13.2% and 18.9% between standard GC-MS and SMB-20eV.

Figure 16. Probabilities found in NIST library search for EI, 70 eV SMB, and 20 eV SMB.

What becomes clear in the graph above is that the probability of small molecules in supersonic molecular beams does not increase. The cooling is less effective in smaller molecules and this is why there is no increase in molecular ion intensity. In the higher mass analytes, there is a higher

probability in both electron energies of the supersonic molecular beam GC-MS. Since the cooling depends on mass, this is also expected. The greatest improvements by supersonic molecular beam are seen in the acids and the esters. In for example 1-octanol, which loses water in all three spectra and thus has an absent molecular ion in all three methods, there is a major decrease in probability since the fragments are presented in different ratios. When the NIST library spectrum is searched in the NIST library itself, the probabilities are not 100%. This is shown in the following Figure 17. Except for nonadecane, the NIST spectrum always has the highest probability, but the differences are not great. What should be mentioned is that the spectrum of the NIST library often differs from the supersonic molecular beam spectra, since the molecular ion is mostly increased in SMB. This means that the SMB spectra gain probability by an increased molecular ion intensity, which gets a high weight in the search, and that the NIST spectra only gain probability in similarity.

0 10 20 30 40 50 60 70 80 90 100 Pr ob ab ili ty

Highest probabilities found in the NIST Library for three different

ionization methods

(30)

Figure 17. Probabilities found in the NIST search for the recorded spectra and the NIST spectrum.

5.1.2. Spectral comparison

5.1.2.1. Increased molecular ion

Although the comparison of matching probabilities is important, every analyst looks at the spectrum before it is send to the NIST library and the improvements made in this spectrum by SMB must be clear. From the Grob mix, some components are an example to indicate how the SMB can improve reading a spectrum.

The first thing that was promised was that, because of the cooling, the energy of the

molecule was decrease and there would be less fragmentation. In Figure 18, four spectra of eicosane are shown in EI, SMB with 70 eV, SMB with 20 eV and the spectrum of the NIST library. As can be seen, the molecular ion in both SMB spectra has improved, but the rest of the spectrum is not affected. Especially in the 20 eV spectrum, because the ratios between the fragments are also identical to the EI and NIST spectrum. When comparing this improvement to the spectra of nonane in Figure 19, the improvement of the molecular ion is not as great as it was in the high mass spectrum. This indicates that the cooling that happens in the supersonic molecular beam is not as effective on lower mass compounds as it is on the high mass compounds. For this reason, the spectrum of nonane in EI has the highest probability of the three measured compounds and probabilities in SMB are only around 10%.

On the other side, for eicosane the probability of the SMB with 70 eV is 60%, while that of the SMB with 20 eV is more than 80%. The difference in probability between these two is due to the difference in fragment ratios. The supersonic molecular beam is getting more efficient when high carbon numbers are analyzed.

0 10 20 30 40 50 60 70 80 90 100 Pr ob ab ili ty

Probabilities of spectra in NIST library search

(31)

Figure 18. Mass Spectra of Eicosane

20 eV SMB

70 eV SMB

Standard EI

NIST Reference

Figure 19. Mass spectra of nonane

Extending the scope of gas chromatography - mass spectrometry by means of supersonic molecular beams

MSc Chemistry

Analytical Sciences

Master Thesis

Extending the scope of gas chromatography – mass spectrometry by

means of supersonic molecular beams

by

Mimi den Uijl

10410465

May 2018

48 EC

September 2017 – May 2018

Supervisor/Examiner:

Examiner:

W.J.L. Genuit, Dr.

P.J. Schoenmakers, Prof. Dr. Ir.

Abstract

Table of Contents

1. Introduction

2. Background Theory

2.1. Gas chromatography

2.1.1. Injection techniques

2.1.2. Retention in GC

2.1.3. The Golay Equation

Factors f(k) and g(k)

Effect of the reduced film thickness on the Golay

equation (k=1)

Effect of retention factor on the Golay equation

(δf = 0.3)

2.2. Supersonic Molecular Beam

2.2.1. Transfer Line and Makeup Gas

2.2.2. Expansion from the Supersonic Nozzle

2.2.3. Free-Jet Separation

2.3. Electron Ionization

2.4. Electron Ionization Sources

2.4.1. Nier-type Ion Source

2.4.2. Brink-type Ion Source

2.4.3. Dual Cage Electron Ionization

2.5. Quadrupole Mass Spectrometry

2.5.1. The Electric Field

2.5.2. Forces on the Ion

2.5.3. Mass Analysis with the Quadrupole Mass Spectrometer

2.6. Pyrolysis

2.6.1. Lignin

2.6.2. Cellulose

2.6.3. Synthetic polymers

2.7. NIST Library search

3. GC-SMB-MS Troubleshooting

3.1. Negative flow in the EFC

3.2. Vacuum of the MS is too low

3.3. Electron Multiplier Voltage

3.4. Ion Source Testing

3.5. High Power Filament & PFTBA memory

3.6. Band broadening through transfer line

3.7. Discrimination

4. Experimental Methods

4.1. Grob mixture Experiments

4.2. Mapping the Supersonic Molecular Beam

4.3. Effect of electron energy on high alkanes

4.4. GC-SMB-MS as quantitative technique

4.4.1. Replicating GCxGC-FID and SimDis GC-FID data

4.4.2. Linearity and repeatability testing in the GC-SMB-MS

4.4.3. Replicating GCxGC-FID and SimDis GC-FID data with

an internal standard

4.5. Analysis of Hydroperoxides with SMB

4.6. Comparing Pyrolysis GC-EI-MS with Pyrolysis GC-SMB-MS

4.7. Analyzing synthetic polymers with Py-GC-SMB-MS

5. Results & Discussion

5.1. SMB Performance testing: Grob mixture

5.1.1. NIST probability comparison

Highest probabilities found in the NIST Library for three different

ionization methods

5.1.2. Spectral comparison

Probabilities of spectra in NIST library search

20 eV SMB

70 eV SMB

Standard EI

NIST Reference

20 eV SMB

70 eV SMB