MSc Chemistry
Analytical Sciences
Literature Thesis
Mass Spectrometry of Cellulose Ethers
Sophisticated approaches to decipher complex structures
by
Martijn Knoope, BSc
VU: 2667453, UvA: 12332282
July 2020
ECTS-credits: 12
Period: March 30
th2020 to July 27
th2020
Supervisor:
Examiners:
Tijmen Bos, MSc
dr. Rob Haselberg
Abstract
Cellulose ethers are chemically modified forms of naturally occurring cellulose, which is a polymer that consists of a chain of glucose units linked together by a distinct spatial structure. The cellulose ethers can be used as a thickening agent, stabilizer, viscosity modifier and serve a purpose in a wide variety of industries. For example the food, paint and pharmaceutical industry experience many benefits from these products. By means of chemical modifications, many different structures can be formed with different functional properties. For some applications as in biological processes, the exact location of the ether modification on the glucose unit is essential. However in other industrial applications, the distribution of these modifications over the polymer chain can be more relevant since it might influence characteristics such as solubility and viscosity. Also, the choice of cellulose ether will influence the properties of the material. In this work, non-ionic cellulose ethers are described including methyl-, ethyl-, hydroxyethyl- and hydroxypropyl cellulose as well as mixed cellulose ethers such as hydroxypropyl methylcellulose.
By studying the relationship between the chemical structure and the functional properties, new cellulose ethers can be developed more efficiently. In this literature review, mass spectrometry (MS) is evaluated as a useful tool to describe certain features of these cellulose ethers. The analysis of these materials can be highly complex; a simple methylcellulose with either O-CH3 or O-H substitution on the three reactive hydroxyl groups can already consist of 𝑎𝑛= 23= 8 different glucose-based monomers. For mixed cellulose ethers,
this goes up to 64 different glucose residues without considering secondary reactions. By combining the power of GC-FID and GC-MS, quantitative data on the monomeric level is obtained using the effective carbon response for quantitation and MS for structure identification. This setup avoids the requirement of reference standards. Also, MSn has shown to be successful in describing the degree of substitution (DS) and the substituent distribution of methylcellulose by monomer analysis. These distributions are compared in this work to random models, based on the reactivity of the main reactive sides, to define the degree of heterogeneity within the anhydroglucose unit (AGU). With this data, specific properties of the materials can be explained.
Even more complicated is the analysis of the distribution along the polymer chain since each AGU can link together randomly. Analysis on the oligomeric level, after partial hydrolysis of the bulk material, has led to insights of higher and lower substituted regions along the polymer chain in comparison to a random model, based on binomial statistics. Analysis has shifted from FAB-MS to state of the art MALDI-TOF-MS instruments, where quantitative data on the oligomeric level is obtained after isotopic labelling of methylcellulose, and after additional labelling with quaternary ammonium for hydroxyalkyl ethers. Hyphenation of ESI-IT-MS with LC allowed the analysis of cellulose ethers with a relatively higher degree of polymerisation in a quantitative manner. It also gave rise to the ability to perform ESI-MSn on methylcellulose to gain further insight into the substitution distribution within the oligomer. This altogether leads to a better understanding of the material which can be used to develop cellulose ethers more efficiently.
Acknowledgements
I would like to express my special thanks to Tijmen Bos MSc, who gave me the opportunity to work on this challenging project and for the guidance and flexibility during the process of writing this thesis. I am especially grateful for the time and online communication efforts via e-mail and Skype during this exceptional time where physical meetings have not been possible due to the coronavirus (COVID-19) measures. I would also like to thank dr. Rob Haselberg for reviewing this work and being the first examiner. Moreover, my thanks go to dr. Andrea Gargano for being the second examiner of this project.
Martijn Knoope, Amsterdam, July 2020
Abbreviations
AGU Anhydroglucose unit ACN Acetonitrile
AcOH Acetic acid
CE Capillary electrophoresis CI Chemical ionisation
CID Collision-induced dissociation CIP Cahn-Ingold-Prelog
CL Chain length
CRM Charge residue model DHB Dihydroxy benzoic acid DI Direct infusion
DP Degree of polymerisation DS 1 Degree of substitution EC Ethylcellulose
ECN Effective carbon number ECR Effective carbon response EHEC Ethyl hydroxyethylcellulose EI Electron impact
ESI Electrospray ionisation FAB Fast-atom bombardment FID Flame ionisation detection GC Gas chromatography HBMC Hydroxybutyl methylcellulose HEC Hydroxyethylcellulose HEHPC Hydroxyethyl hydroxypropylcellulose HPC Hydroxypropylcellulose (HP)LC (High-performance) Liquid chromatography HPMC Hydroxypropyl methylcellulose IEM Ion evaporation model
IR Infrared
IT Ion trap
LIF Laser-induced fluorescence
1 Degree of- and Molar subsitution (DS/MS) are
mABA/oABA Meta-/Ortho- aminobenzoic acid MALDI Matrix-assisted laser
desorption/ionisation MC Methylcellulose MEC Methyl ethylcellulose MHEC Methyl hydroxyethylcellulose MS Mass spectrometry
MS 1 Molar substitution MW Molecular weight
NMR Nuclear magnetic resonance SIMS Secondary-ion mass spectrometry TFA Trifluoracetic acid
TOF Time-of-flight UV Ultraviolet
aronym of mass spectrometry (MS) and for
Table of contents
1 Introduction ... 1 2 Background Theory ... 3 2.1 Glucose ... 3 2.2 Glucopyranose ... 3 3 Native Cellulose ... 6 3.1 Structure of Cellulose ... 6 3.2 Supramolecular structure ... 74 Non-ionic cellulose ethers ... 9
4.1 Substituent distribution within the glucose unit ... 10
4.2 Substitution distribution along the polymer chain ... 16
5 Mass Spectrometry ... 21
5.1 Instrumentation ... 22
5.1.1 Ionisation ... 22
5.1.2 Mass analysers ... 23
5.2 Analytical strategies ... 24
5.2.1 Degree of substitution and substituent distribution of the glucose unit ... 25
5.2.2 Substitution distribution along the polymer chain ... 28
5.3 Evaluation and discussion ... 35
6 Conclusion ... 37
1 Introduction
Polymers consist of a long chain of repeating units: the monomer. This chain is formed by polymerisation reactions where the monomers react consecutively with each other to form the polymer. Research has been done on these types of compounds, which has led to the development of new materials. Nowadays, it is hard to imagine life without these products due to their many applications, in for example medication, nutrition, clothing, buildings, paper, wood, et cetera. [1]
Efforts in different fields of science have led to the development of highly complex and diverse polymer structures. Considerable efforts are being made in the research and development of modified natural polymers. [2] Especially today, where there is an increasing demand for the industry to invest in renewable and green resources. One natural polymer that is of high interest is cellulose, which is biosynthesised in plants. It is the structural basis of the cell wall of plants and is also a part of human diets as a source of fibre. Cellulose is widely used as its natural form in material like paper and wood. Since cellulose in its natural form has several unwanted features, such as its hydrophobicity, various chemical modifications can be made to mitigate these properties. [3] This enlarges the applicability for, e.g. the food and pharmaceutical industries with a high potential for more applications. This is due to the modifiable, non-toxic, renewable and biodegradable properties. [4]
These modifications are usually made by etherification of the free hydroxyl groups of the glycosidic monomer. The characteristics of the cellulose are tuned by chemical processes to fit the right application. However, this results in highly complex structures with potential heterogeneity within the monomeric unit, along the polymer chain and in the bulk material. [5] This heterogeneity can result in either wanted or unwanted properties of the material.
Various techniques can be used to analyse these different functionalised glucose units. The polymer can either be hydrolysed by acidic treatment or depolymerized utilising other techniques such as enzymatic degradation, reductive cleavage and ultrasonic treatment. [5,6] After sample pre-treatment the sample can be analysed with high-performance liquid chromatography (HPLC) [7] or capillary electrophoresis (CE) [8] followed by ultraviolet (UV) or laser-induced fluorescence (LIF) detection. [9] However, the method of choice is still gas chromatography (GC) – flame ionisation detection (FID) together with GC – mass spectrometry (MS) which can be used for identification. [5] Moreover, it is also possible to determine the substitution pattern by either solid-state- or solution 13C-NMR with [10] or without [11] the need for complete hydrolysation. [12]
More complex is the analysis of the distribution of the substituents along the polymer chain. Reaching a sound of understanding of these structures and correlate this to the functional properties is a crucial task. Within the UNMATCHED program, universities and companies are working together to find a way to close this gap in knowledge. The structure must be deciphered by using better separation and detection methods.
the focus will be on structure identification using MS. Also, theoretical models will be discussed, since these are often used in combination with MS to interpret the data obtained from these complex systems. MS is a very interesting analytical tool to obtain structural data but also faces some challenges due to the high complexity of cellulose ethers. [5] Within this literature thesis, the focus will be on different approaches to obtain structural information of cellulose derivatives of non-ionic ethers. After a general introduction of the cellulose polymer and its monomeric composition is given in chapter two and three, a substantial part will focus on modelling of the ether substituents within the AGU and along the polymeric chain in chapter four, followed by strategies for the actual analyses of different cellulose derivates such as methyl-, ethyl- and hydroxypropyl cellulose with MS in chapter five. This thesis aims to describe the different types of structural information that can be obtained from cellulose ethers by MS. Furthermore it aims to critically evaluate the different mass spectrometric instrumentations and approaches based on the opportunities and limitations for various ether analogues and the type of structural information. In the end, in chapter six, the findings of this work will be concluded.
2 Background Theory
2.1 Glucose
Cellulose is a naturally occurring polysaccharide that consist of several β(1⟶4) linked D-glucose units. Glucose (C6H12O6) is a saccharide that exists of an aldehyde group, six carbon atoms and five hydroxy groups. It can exist both in an open chain and in a cyclic form. Glucose is often drawn in an open chain in the Fischer projection as shown in Fig. 1. Characteristic for glucose is the hydroxy group on carbon-3 (C-3) which is heading in the opposite direction in comparison to the hydroxyl group on C-2,4,5. A C-2 epimer of glucose is mannose, and a C-4 epimer is galactose, here the hydroxy on the specified carbon atom is heading in the opposite direction relative to glucose. When all the hydroxy groups are shifted to the other side, a non-superimposable mirror image is created. The nomenclature of the two enantiomeric forms of glucose is dependent on the substitute arrangement of the chiral centre on the fifth carbon. The most common form of glucose that naturally occurs is D-glucose, where the hydroxy group is facing to the right. The mirror image is L-glucose, where the hydroxy group is facing to the left, which can be chemically synthesized. The D/L system is a relative measurement as the C-5 stereocenter is related to the stereochemistry of glyceraldehyde where D-glucose has the same configuration as (R)-glyceraldehyde and L-glucose has the same configuration as (S)-glyceraldehyde. The R/S nomenclature is an absolute method where the substitutes are assigned by priority according to the Cahn-Ingold-Prelog (CIP) priority rules. [1,2]
2.2 Glucopyranose
The formation of a ring structure of glucose is the result of an intramolecular reaction between the aldehyde at C-1 and the hydroxy group on C-5. This results in a pyranose ring: a cyclic hemiacetal (R2C(OH)OR’ [R’ ≠ H]) forms of monosaccharides in which the ring is six-membered (a tetrahydropyran), hence the systematic name glucopyranose. When the aldehyde reacts on the hydroxy group attached to C-4, a furanose is formed. The most simple way to display a cyclic monosaccharide is by using the Haworth representation where the ring is a planar structure (Fig. 2). This representation is ideal for defining which Fig. 1. Fischer projection of the glucose enantiomers: D-glucose (a) and L-Glucose (b). The D and L forms are assigned based on the structure in the box, note that for the R/S isomeric assignments the absolute configuration is different for the chiral centres.
a) b)
Fig. 1. Fischer projection of the glucose enantiomers: D-glucose (a) and L-Glucose (b). The D and L forms are assigned based on the structure in the box, note that for the R/S isomeric assignments the absolute configuration is different for the chiral centres.
substituents are oriented axial or equatorial with respect to the ring. When D-glucose has reacted to a pyranose ring structure, the hydroxymethyl group on C-5 is facing up relative to the ring This structure is described as D-glucopyranose. The opposite is true for L-glucopyranose, where the hydroxymethyl group faces down. [15]
Fig. 2. Haworth projection of α-D-glucose (a) where the hydroxy group on the first carbon atom (C-1) is faced down and in the opposite direction of the hydroxymethyl group on C-5 and β-D-glucose (b) where the hydroxyl group on the first carbon is faced up in the same direction as the hydroxymethyl group on C-5. The direction of this hydroxymethyl group is characteristic of D-glucopyranose and β-D-glucopyranose is the characteristic building block of native cellulose. The intramolecular reaction in order to create the ring-structure introduces an extra chiral centre on C-1. The position of the hydroxy group is described as alpha (α) and beta (β) and is relative to the position of the C-6. In the α-form, the hydroxyl group on C-1 is on the opposite side of the ring relative to C-6 and in the β-form, the hydroxyl group on C-1 is on the same side of the ring as from C-6. These two forms of isomers are called anomers and are able of interconverting in solution trough the linear from.
The Haworth projection does not give a correct representation of the shape of the ring structure. All the carbon atoms are sp3 hybridized, so the ideal angle for this tetrahedral structure is 109.5 degrees. In a ring structure, as suggested by the Haworth representation, the angles will be forced to 120 degrees which makes the structure unstable. In order to gain stability, the glucopyranose will adapt to a chair configuration (Fig. 3). Looking at this conformation, the angles of the bond are 110.9 degrees which make the structure more stable. [16]
All the substitutes on the ring are either axially (parallel to the ring axis) or equatorially (perpendicular to the ring axis) located. Any axial group other than hydrogen imparts instability to the pyranose ring. This is the result of 1,3-diazal interaction, which are repulsive forces between two axial substitutes. In equatorial position, the substituents are further away from each other leading to lower repulsing forces and therefore more stability. In α-D-glucopyranose, the hydroxy group on C-1 is in the axial position and therefore less
a) b)
a) b)
Fig. 3. Schematic representation of a 4C
stable than β-D-glucopyranose where this hydroxy group is in the equatorial position. This phenomenon has been experimentally confirmed in the past by Rao et al. [16] There are two types of chair conformations known for the enantiomers, the most stable form is again where the (most) substitutions are equatorially orientated, which is a 4C1-chair conformation. This results in no instability factors for β-D-glucopyranose. For a 1C4-chair of the β-conformation there will be instability factors on C-1,2,3,4. Among repulsing forces also other mechanisms play a role in the stability, but they all favour the 4C1-chair conformation for β-D-glucopyranose. [16]
3 Native Cellulose
Anselme Payen (1795 – 1871), a French agriculturalist, was the first one who separated wood into its components. He obtained a white fibrous material, which he called ‘cellulose’. The natural form of cellulose that he described back in the days, as just a white powder, will be described to more detail in this paragraph regarding its chemical structures. [17]
3.1 Structure of Cellulose
Cellulose is present in all major plants in different amounts and it serves as the structural basis of the cell wall. It is also responsible for the strength of the plant cells since it forms straight supramolecular fibres of high tensile strength. The cellulose is naturally produced by photosynthesis under the influence of sunlight where β-D-glucopyranose units in the 4C1-chair configuration are bound to each other by polycondensation. During the reaction, water is removed. The resulting glucose monomeric unit is referred to as the anhydroglucose unit (AGU), where anhydro- denotes the removal of water from the glucopyranose. The AGUs are joined together by a 1,4-β-D-glycosidic bond. The β refers to the special orientation of the hydroxy group on C-1. If the hydroxy group is pointing upwards, a β-glycosidic bond is formed. At the opposite, if the hydroxy group is pointing down, an α-glycosidic bond is formed. The numbers 1 and 4 refer to the carbon number of the hydroxyl group that takes part in the covalent bonding. For example, starch consists of α-D-glucopyranose where the hydroxy group on C-1 is facing down, leading to 1,4-α-D-glycosidic linkages. The 1,4-β-glycosidic bond results in higher stability and is responsible for the straight structure of cellulose. The two glucose groups at the end of the cellulose chain are slightly different form each other. It consists of a non-reducing end group and a reducing end group (Fig. 4). At the non-reducing end, the pyranose ring stays intact with the hydroxyl group on C-4. Where on the other side, the intact pyranose ring is in equilibrium with an aldehyde structure, responsible for the reducing properties. [18]
Fig. 4. Schematic representation of the cellulose polymer, where: the monomeric unit (AGU) is shown in square brackets, the dimer cellobiose is also indicated including the characteristic 1,4-β-glycosidic bond as well as the non-reducing and non-reducing end group which forms an equilibrium with the pyranose structure and the aldehyde. Reproduced from Credou and Berthelot. [18]
The length of a cellulose chain is described by its degree of polymerisation (DP), which describes the number of monomeric units. [15] The degree of polymerisation in native cellulose is given as an average value since it consists of multiple chains with different DP. Apart from this polydispersity, also the average DP of cellulose varies from 1,000 to 30.000 monomeric units which is strongly dependent on its origin. In
wood (40-50 wt% cellulose) a DP of 10,000 monomer units is common and in natural cotton (90 wt% cellulose) a DP of 15,000 is found, whereas isolated cellulose has most commonly a DP varying from 800 to 3,000. The average chain length (CL) of a cellulose polymer is about 500 – 15,000 nm. Within the scope of the UNMATCHED programme, the focus is on wood pulp and some other small species of cellulose. [3,19]
Within and between cellulose strains, different interactions influence the properties of the cellulose. The most important one is the hydrogen bonding between adjacent AGUs or between different cellulose chains. Another form of hydrogen bonding which plays a part is the C–H···O interaction where CHn is the proton donor and also hydrophobic interaction such as van der Waals forces can influence the properties of the polymer. [19]
The three unbounded hydroxyl groups on every AGU can interact with different hydroxyl groups of adjacent AGUs or with another cellulose chain through hydrogen bonding. This can happen in various ways leading to different polymorphs of cellulose with extensive hydrogen bond networks. Infrared (IR) spectroscopy and solid-state carbon-13 (13C) nuclear magnetic resonance (NMR) spectroscopy analysis revealed how these cellulose chains and AGUs are interconnected. This network is composed of intramolecular as well as intermolecular bonds. The main hydrogen bond is formed between the hydroxyl group on C-3 and the oxide on the ring of the adjacent AGUs (O3H-O5’). Together with the 1,4-β-D-glycosidic bond, these are the key factors in the stiffness of the cellulose polymer. While the different polymorphs show various hydrogen bonding networks, this O3H-O5’ bond is a common characteristic. However, in different polymorphs of cellulose other hydrogen bond networks exist. In native cellulose, for example, the AGUs are also intermolecularly connected by an O2H-O6’ hydrogen bond as shown in Fig. 5. [2,3]
Fig. 5. Intramolecular hydrogen bonding of cellulose I indicated by the black dotted lines. Reproduced from Credou and Berthelot. [18]
3.2 Supramolecular structure
Native cellulose I consists of both intramolecular and intermolecular hydrogen bonds. Intermolecular bonds are formed between the hydroxyl group on C-6 and the oxygen on C-3 (O6H-O3’’), resulting in a parallel structure, as shown in Fig. 6. In cellulose II the supramolecular structure differs as a result of a different hydrogen bond network. Here, the polymer chains are interconnected by an O6H-O2’’ bond instead of the O6H-O3’’ bond. Cellulose I can be converted into Cellulose II by regeneration (dissolution followed by crystallization) or mercerization (treatment with aqueous sodium hydroxide). Since the antiparallel structure of cellulose II is thermodynamically more stable, this conversion cannot be reversed. From these two polyforms, respectively cellulose III1 and III2, can be created by treatment with ammonia. Moreover, heat
treatment of cellulose III leads to cellulose IV1 and IV2. These reactions are reversible where the starting materials, cellulose I and cellulose II, can be recovered.
Fig. 6. Crystal structure the polyforms cellulose I (left) and cellulose II (right). Intramolecular hydrogen bonds are indicated by the dark dotted lines and intermolecular hydrogen bonds by the light dotted lines. Both the polyforms share the same O3H-O’ intermolecular hydrogen bonds, but cellulose I also consist of an O2H-O6’ hydrogen bond. This affects the intermolecular hydrogen bonding network which is characterised by the O6H-O3’’ and 06H-O2’’ hydrogen bond for respectively for the two crystal structures. Reproduced from Credou and Berthelot. [18]
Native cellulose I crystalizes simultaneously in two ultrastructures, Iα and Iβ. The co-existence of two distinct crystalline forms of native cellulose has been known for years. Where one form is present in higher ratios in bacteria and algal whereas the other form is dominantly present in especially vascular plants. [20] The dimorphs can coexist in a cellulose sample but also in each microfibril and are found to vary between samples of different origin. Cellulose Iα consists of a triclinic unit cell including one chain and Iβ has a monoclinic unit cell inducing two parallel chains. The triclinic shape makes it difficult to analyse the exact structure because of cylindrical averaging in fibre analysis. Progress in further structure elucidation has been made by the discovery of tunicin, a nearly pure Iβ phase (~90%), obtained from Halocynthia roretzi, and nearly pure Iα phase (~90%) from Glaucocystis. It has also been shown that Iα is metastable and can be converted into the Iβ phase in the solid phase. The difference in structure is mainly due to a difference in hydrogen bonding. In Iβ the hydrogen bonds are distributed over a region of better geometry. [18,21,22] When several cellulose chains are structurally ordered on top of each other a crystalline region is formed as a result of strong intermolecular hydrogen bonding. In this geometry the cellulose is strong and insoluble in most solvents. When the different polymers have a less well-defined shape an amorphous region is created. This results in a lower number of hydrogen bonds leading to free reactive groups which can interact with other components such as water. These hydrophobic parts lead to the swelling of the cellulose but due to the hydrophilic crystalline regions it will not dissolve. The different regions are held together by polymers with CL longer than the ordered regions. [18] When the features of the native cellulose are modified by means of etherification of the hydroxyl groups, the sample is first being homogenised. With this process, the crystalline structure is broken to create a sample where the reactive groups are accessible for the same amount of reactant. [4] This subject will be covered in the next chapter, where etherification of the hydroxyl groups is described.
4 Non-ionic cellulose ethers
Ever since the beginning of recorded history, cellulose played an important role in the daily life of humans. The ancient Egyptians used cotton (>98% cellulose fibres) for clothing and fibres from the papyrus plant for paper. Nowadays, cellulose is also used for its appealing features such as its rigid and high crystalline structure which makes it an ideal structural engineering material together with its insolubility characteristics in water. However, the insoluble nature of cellulose, as a result of the extensive hydrogen bond network, limits also its industrial applications. Therefore, native cellulose is often modified either by chemical or other processes. The substitution of the hydroxy groups by various alkyl groups disturbs the inter- and intramolecular hydrogen bond systems which leads to better solubility in various solvents. The first cellulose ether was prepared after methylation, therefore called methylcellulose (MC), and was reported in 1905. [23] Nowadays, cellulose ethers are dominantly used in the industry. Table 1 lists some common cellulose ethers that are commercially available with single as well as mixed substituents. Minor industrial produces are indicated in parentheses. Note that only non-ionic cellulose ethers are included in this work, but are often referred to as just cellulose ethers.
Table 1. Cellulose ether derivatives with the associated substitutions and their abbreviations. [23,24] Abbreviation Substituent group Structure Abbreviation of commercial products a
Me Methyl -CH3 MC, (MEC), MHEC, HPMC, (HBMC)
Eth Ethyl -CH2CH3 EC, (MEC), EHEC,
HE 2-hydroxyethyl -(CH2CH2O)n-H b HEC, MHEC, EHEC, (HEHPC)
HP 2-hydroxypropyl -(CH2CH(CH3)O)n-H b HPC, HPMC, (HEHPC)
HB 2-hydroxybutyl (CH2CH(C2H5)O)n-H b (HBMC)
a In parentheses: minor products, ending “C” means “Cellulose” b Generally n=1,2,3
These cellulose ethers have different properties, dependent on the nature of the substituent. Therefore, they can be used in a wider field of applications in industry. Apart from the nature of the substituent, also the degree of substitution (DS) as well as the distribution of the substituents along the chain influences their properties. The DS describes the average number of hydroxyl group per AGU that are terminated by a substituent. This value ranges from zero for no substitution to three for a fully substituted AGU. When a substitute is non-terminating, it introduces a new free hydroxyl group for further (secondary and higher-order) substitution. This is defined as the molar substitution (MS), the average number of substitutes per AGU. For example, HEC and HPC can have higher numbers of MS with no theoretical upper limit. When all the possible substituents are terminative, the DS is per definition equal to the MS. On the other side, when substitutes contain hydroxy groups, the difference between MS and DS shows the proportion of side-chain substitution. The DS, MS and the proportion of side-chain substitution, as well as the substitutes itself, all together determine the final properties of the derivatized cellulose. However, as a rule of thumb above a DS of 1.0, most of the cellulose derivatives are soluble in neutral water. [2] Non-modified cellulose is not soluble in water and most other solvents due to the high degree of hydrogen bonding.
The cellulose ethers can be described on different hierarchical levels. Within this work, the cellulose ethers are being described and analyses based on the monomeric and oligomeric level, respectively the first and second hierarchical levels. By performing homogeneous analysis, the resulting data can be used to describe the third hierarchal level which describes the complete sample of modified cellulose chains. [5]
4.1 Substituent distribution within the glucose unit
The first hierarchical level describes the smallest entity, the monomeric unit called AGU, which consist of three reactive hydroxy groups on C-2, C-3 and C-4. When the cellulose is treated with methyl chloride under the right reaction conditions, the hydroxyl groups can be replaced by methoxide leading to MC. As a result of this reaction, multiple different monomers are formed. The number of possible monomers is defined as 𝑎𝑛 where a = number of different functional groups (in this case either OH or OMe) and 𝑛 =
number of hydroxyl groups available per AGU. For MC, with only one type of substituent, this results in a 23= 8 possible glucose residues. The analyses become more complex for mixes of different substituents.
For the production of HPMC, the methylation and hydro alkylation are performed simultaneously leading to four different substituents (-[CH2CH(CH3)O]nH and -[CH2CH(CH3)O]nCH3 with 𝑛 = 0 𝑜𝑟 1) and thus 43= 64 different monomeric units if no secondary reactions take place. More diverse structures are generated when 𝑛 > 1, where secondary and tertiary reactions occur together with the formation of diastereomers from the HP-substitute. [5]
The substitution pattern can be calculated by using various models. These calculations are based on the reaction rates (𝑘) of the hydroxyl groups on C-2, C-3 and C-6 (respectively: 𝑘2, 𝑘3 and 𝑘6) and the
reaction time, which results in the substitution pattern depended on the DS. Spurlin [25] was the first one who attempted to model the substitution pattern within the AGU in 1939, he stated the following assumption to simplify the model. (i) The CL is significantly longer so that the end groups can be neglected, (ii) substitution within a given monomeric unit, does not affect the reactivity of the remaining hydroxy groups, (iii) the ratio of reactivity between the three hydroxyl groups are independent of each other and remain constant during the reaction (pseudo-first-order kinetics), and (iv) all the AGUs are equally accessible to the same concentration of substituents. [25-31]
The following symbols were introduced and are still generally used: the mole fraction of a specific single regioisomer is indicated by 𝑠𝑖, were 𝑖 describes all the substituted positions (or zero for none). For example,
𝑠23 describes an AGU where the hydroxy groups have been substituted exclusively on the second and third
position. On the other hand, 𝑥𝑖 is used to describe the total mole fraction of AGU that haven been substituted
on the specified position. For example, 𝑥2 is the mole fraction of all the regioisomers which have a substitute
on the second position (𝑠2 , 𝑠23, 𝑠26, 𝑠236). The relationship between 𝑥𝑖 and 𝑠𝑖 is described in Eq. (1-3).
x2= s2 + s23+ s26+ s236 (1)
x3= s3 + s23+ s36+ s236 (2)
To indicate the molar fraction of AGUs that are of un- (c0), mono- (c1), di- (c2), and trisubstituted (c3), 𝑐𝑖
is employed. The interrelation between 𝑐𝑖 and 𝑠𝑖 is described in Eq. (4-7).
c0= s0 (4)
c1= s2+ s3 + s6 (5)
c2= s23+ s26 + s36 (6)
c3= s236 (7)
According to the model of Spurlin, the 𝑥𝑖 values can be calculated, based on the relative reactivities of the
hydroxyl groups expressed by the first-order rate constants (𝑘𝑖), see Eq. (8-11). For evidence of the
first-order kinetics, the reader is referred to the original literature from Spurlin. [25]
xi= 1 − e−Bki (8)
x2= 1 − e−Bk2 (9)
x3= 1 − e−Bk3 (10)
x6= 1 − e−Bk6 (11)
Where factor B is a rate constant with the dimension of time (𝐵 ∝ 𝑡). Longer reaction times lead to higher DS, for comparable relative reactivities. The DS, originally referred to as S, per AGU is equal to the sum of 𝑥2, 𝑥3 and 𝑥6 and the weighted sum of 𝑐𝑖, as described in Eq. (12-13). [26]
DS = ∑ xi i=2,3,6
(12)
DS = 0c0+ 1c1+ 2c2+ 3c3 (13)
The 𝑥𝑖 value is commonly used for calculation. However, industry as well as academia are usually more
interested in the specific stereoisomers (𝑠𝑖) and the degree of substitution (𝑐𝑖). The mole fractions of the
specific isomers can be calculated according to the model of Spurlin, were also the reaction rates for unsubstituted hydroxyl groups are considered, since the regioisomers describe substitution on exclusively the given positions. The calculations of the mole fractions of the specific stereoisomers are given in Eq. (14-21). [25] s0= e−B(k2+k3+k6) (14) s2= (1 − eBk2)e−B(k3+k6) (15) s3= (1 − eBk3)e−B(k2+k6) (16) s6= (1 − eBk6)e−B(k2+k3) (17) s23= (1 − eBk2)(1 − eBk3)e−Bk6 (18) s26= (1 − eBk2)(1 − eBk6)e−Bk3 (19) s36= (1 − eBk3)(1 − eBk6)e−Bk2 (20) s236= (1 − eBk2)(1 − eBk3)(1 − eBk6) (21)
As mention before, the 𝑠𝑖 and 𝑐𝑖 values are interrelated, see also Eq. (4-7), and therefore the molecular
fraction of un- to trisubstitution can directly be calculated. Eq. (22-25) can be used and are given in a simplified manner. [25] c0= s0= e−B(k2+k3+k6) (22) c1= e−B(k2+k3)+ e−B(k2+k6)+ e−B(k3+k6)−3e−B(k2+k3+k6) (23) c2= e−Bk2+ e−Bk3+ e−Bk6− 2e−B(k2+k3)− e−B(k2+k6) −2e−B(k3+k6)+ 3e−B(k2+k3+k6) (24) c2= 1 − e−Bk2− e−Bk3− e−Bk6+ e−B(k2+k3)+ e−B(k3+k6) −e−B(k2+k3+k6) (25) The molar fractions of 𝑐𝑖 are dependent on factor B, which results in higher DS and MS for longer reaction
times. The effect of the DP on the relative mole fractions is given in Fig. 7. In Spurlin’s model, the reaction rates (ki) of the three hydroxyl groups are equal: k2:k3:k6=1:1:1, so no hydroxyl group is favoured. [25] Therefore, the relative molar fraction of s2, s3 and s6 for monosubstitution and s23, s26 and s36 for disubstitution will be equal independent on the total DS. However, the molar fraction of un- to trisubstituted AGU will variate dependent on the DS. The total monomer composition is described as the sum of the fraction of all the regioisomers as well as the unsubstituted (𝑠0) and trisubstituted (𝑠236) forms:
∑ 𝑠0, 𝑠2, 𝑠3, 𝑠6, 𝑠23, 𝑠26, 𝑠36, 𝑠236= ∑ 𝑠𝑖= 1 𝑜𝑟 100% or by the sum of the mol fraction of un-, mono-,
di- and trisubstituted monomers: ∑ 𝑐0, 𝑐1, 𝑐2, 𝑐3= ∑ 𝑐𝑖= 1 𝑜𝑟 100%. [5,25]
Fig. 7. Diagram of the theoretical mole fraction, 𝑐𝑖, of un-, mono-, di-, and trisubstituted AGU as a function of the DS.
The calculations are based under the assumptions of Spurlin’s model, e.g. equal reactivities of the three hydroxyl groups: k2:k3:k6=1:1:1. Reproduced from Mischnick and Momcilovic. [5]
Reuben and Conner determined the relative rate constants for carboxymethylation: k2:k3:k6=2.14:1.00:1.58. As a result of the regiospecific reactivities, the summarized profiles of c1 and c2 will no longer be perfectly symmetrical images. With a larger difference in reactivities, the more deviation from symmetry is being observed due to a steeper decrease of c0 and at the same time a delayed increase in the mole fraction of c3. The difference in ki also leads to a favoured regioisomer of the mono- and disubstituted AGU. The relation between the mole fractions of s2, s3 and s6 for monosubstitution and s23, s26, s36 for disubstitution of carboxymethylcellulose as a function of the DS is described in Fig. 8, based on the reactivities found by Reuben and Conner [27].
Fig. 8. Diagram of the theoretical mole fraction of the specific regioisomers 𝑠𝑖 for mono substitution (left) and
disubstitution (right). The calculations are based on the relative rate constants of CMC of k2:k3:k6=2.14:1.00:1.58.
Reproduced from Mischnick and Momcilovic. [5]
Later, Spurlin rephrased the second (ii) assumption to: “The only interference of substitution with a glucose unit occurs between position 2 and 3”. This was done after experimental proof, based on monomer data, that the reactivity of the second and third hydroxy groups are correlated for some derivatives. [28] The substitutes which show this type of intermolecular interactions include methyl-, ethyl- and hydroxyethyl- celluloses. Here the reactivity on the third position was enhanced by a factor of 3-5, as a result of primary substitution on the second position. [28-31] The relative rate constants that were found for MC by Reuben and Conner were: k2:k3:k3’:k6=1.24:0.28:0.82:1.00, indicating a three-fold increase in reactivity. [26] These interactions leading to enhanced reactivities are often referred to as intramonomeric effects for neighbouring hydroxyl groups and intermonomeric effects if these interaction happens between the hydroxyl group of adjacent AGUs.
Based on these observations of intramonomeric interactions, Reuben further developed the model for Spurlin by introducing an addition rate constant: 𝑘3′ (Model II). [26] They also investigated the reversed effect,
where substitution on the third position would influence the substitution on the second position (Model III, 𝑘2→ 𝑘3′) but this effect was not found. He also simplified the formulas by defining the operative quantity
𝑝𝑖 as the probability for having an unsubstituted hydroxyl group as described in Eq. (26). [26,30]
𝑝2,6= 1 − xi= eBki (26)
Therefore, the probability of having an unsubstituted AGU on all positions (𝑠0) is described by the
product of the three probabilities, shown in Eq. (27).
s0= p2p3p6 (27)
Inherently of the probability of no substitution, the probability of having substitution is defined by the complementary event shown in Eq. (28).
xi= 1 − pi (28)
Therefore, the mole fraction of a trisubstituted AGU is described by the probability of substitution at all the positions, shown in Eq. (29).
s236= x2x3x6 (29)
The probability for substitution on the second and sixth position can be calculated according to Spurlin, but for the substitution on the hydroxyl group attached to C-3 the substitution on C-2 has to be taken into account, according to the simplified Eq. (30) as described by Reuben. [27]
p3(Reuben)= 1 − x3=
1 − x2+ s0+ s6− s3− s36
2(1 − x2)
(30) The corresponding 𝑠𝑖 values for Reuben’s model (Model II) can be calculated according to Eq. (31-38),
using the model from Spurlin for substituents on the second and sixth position. [25,26]
s0= p2p3(Reuben)p6 (31) s2= p6(s2+ s26) (32) s3= p2p6(1 − p3(Reuben)) (33) s6= p2p3(Reuben)x6 (34) s23= p6(1 − p2p3(Reuben)− s2− s26− p2(1 − p3(Reuben))) (35) s26= x6(s2+ s26) (36) s36= p2(1 − p3(Reuben))x6 (37) s236= x6(1 − p2p3(Reuben)− s2− s26− p2(1 − p3(Reuben))) (38)
Apart from single substitution reactions, Reuben also developed a model for secondary reactions where an extra rate constant, k’, has been introduced. [26] With this modification also the different reactivities of secondary substitution were considered. Although many attempts have been made to further improve the models from Spurlin and Reuben, it can still only serve as a reference model where the experimental data fits or not. [5,25,26]
There are two common ways described in literature to determine the relative rate constants of cellulose ethers. First, according to Spurlin, the relative reaction rates can be experimentally determined using the simplified Eq. (39).
ki− ln(1 − xi) (39)
Another way to determine the reactive rate constants was described by Reuben, who plotted − 𝑙𝑛(𝑐0)
against − 𝑙𝑛(1 − 𝑥𝑖) where the slope is equal to 𝑘𝑖, seen in Fig. 9. [27]
The calculations of the monomeric composition as a function of the DS and reactivities can be compared to experimental data. The difference between the regioisomers can be described as ∑ ∆𝑠𝑖 or as an average
heterogeneity parameter, as shown in Eq. (40). [30] A low value for the heterogeneity parameter indicates a more heterogeneous distribution and a higher value indicates a more homogeneous distribution.
H = (∑ ∆si2)
Fig. 9. Experimental determination of the 𝑘𝑖 accoring to Reuben where the slope divided by the sum of the slopes
equals the relative rate constant. The slopes for the 𝑘2,3,6 are respectively 0.470, 0.220 and 0.248. Deviding 𝑘𝑖 by
𝑘2 leads to the relative rate constants of 2.14:1.00:1.58. Reproduced from Reuben. [27]
There are several reasons why there could be a variation between the calculated and the experimental data. (i) The second assumption of Spurlin’s model is invalid, and Reuben’s model might be a better fit, (ii) the is variation in reactivities as a result if interchain connections such as hydrogen bonding (macrostructure), and (iii) besides the variation of reactivity as a function of the orientation, the reactivities can also change as a function of time. The main cause of a time-dependent change in reactivity is the transition from a homogeneous reaction to a reaction in multiple phases (heterogeneous). [5] Since many parameters can influence the kinetics of the reaction, it is hard to state on favoured position. This was shown by Mischnick and Momcilovic who compared with experimental 𝑘𝑖-values for CMC, which were
described in the literature and gathered under various reaction conditions by different analytical methods, see Table 2. The exact reaction condition leading to the different reactivities can be found in the original literature.
Table 2. Reported relative rate constants (𝑘𝑖) for the carboxymethylation of cellulose
Study Ref. 𝑘2 𝑘3 𝑘6
Abdel-Malik and Yalpani [11] 2.1 1.0 1.5
Reuben and Connor [27] 2.1 1.0 1.6
Baar et al. [32] 3.0 1.0 2.1
Kragten et al. [33] 1.8 1.0 1.3
Zeller et al. [34] 2.4 1.0 1.8
Niemelä and Sjöström [35] 3.1 1.0 2.6
Ho and Klosiewicz [36] 2.0 1.0 1.5
Timell and Spurlin [37] 1.0 1.0 2.0
Croon and Purves [38] 2.0 1.0 2.5
Also, for hydroxy alkylated methyl celluloses, such as HPMC and HEMC the reactivities of the different hydroxyl groups can strongly vary dependent on the reaction conditions. Typically, the higher degree of substitution is found on the O-6 position with values between 35-56%, followed by O-3 subsection with 29-45% and the O-3 showed the lowest reactivity with typically a degree of substitution between 10 and 20%. In terms of relative reactivities, the average will be k2:k3:k6=2.47:1.00:3.13 just to give an indication. [40]
4.2 Substitution distribution along the polymer chain
The second hierarchical level describes the degree of substitution as well as the orientation of the different monomeric units along the polymer chain. The AGUs can arrange in a nearly unlimited number of sequences. For single substituted cellulose polymers with a DP of and without considering secondary reactions, there are already 820≈ 1.15 × 1018 different sequences theoretically possible. Therefore, sequence analyses will
not be useful, since the odds are low that one chain coexists in the bulk material. Models are more often used to describe a certain cellulose ether.
When all the AGUs are accessible with the same probability during the whole course of the reaction, according to Spurlin’s assumptions, the substitutes would spread in a random matter along the polymer chain. This has been visualized in Fig. 10 for the degree of substitution along the polymer chain (𝑐𝑖). This
random pattern is also described as homogenous, uniform or regular in literature. When not all the hydroxyl groups on every AGU react with the same probability, a more heterogeneous pattern will be created along the polymer chain or between polymer chains. This more ordered structure, in comparison to the random model, can be the result of e.g. residual crystalline region where the local reactivities are lower in comparison to the amorphous region. Heterogeneity can exist between chains (first-order), not shown here, or within a single polymeric chain (second-order). Other distributions that can be described are a more regular or a block-like pattern. [5]
Fig. 10. Schematic representation of the distribution of un-, mono-, di- and trisubstituted AGUs along the polymer chain. Reproduced from Mischnick and Momcilovic. [5]
In principle, there are two different approaches available for the evaluation of the substituent distribution along the backbone. One is based on quantitative oligomer analysis after random degradation using e.g. hydrolysis, followed by statistical evaluation. The second method uses enzymes for selective glycosidic bond cleavage with glucose determination, followed by SEC and MS analysis of the obtained oligomers. However, this method is less common, since not all the degradation mechanism are well understood. [41] Within this work only random hydrolysis by acid treatment is described.
As previously described, the models form Spurlin and Reuben can be used to calculate the monomeric composition, as well as the distribution of the DS (c0,1,2,3) of AGUs for a given average DS, see Fig. 7. [25,26] For example, at an average DS of 1.5, a mole fraction of 0.1 will be un- and trisubstituted and 0.4 will be mono- and disubstituted. In order to check if a specific synthesis has followed one of the models, the experimental data can be compared to the calculated data. The same can be done along the polymer chain where the distribution of the number of substitutes along an oligomer with a specific DP (≥2) is compared against experimental data. Usually, a low DP is used under the assumption that the oligomeric fractions reflect the situation in the polymer chain. This can be justified by analysing a small part of the polymer chain (DP ≥2) instead of breaking it down to its monomeric forms, some structural information is being preserved. The advantages of oligomers with a lower DP are that the concentration is higher after partial hydrolysis which increases detection sensitivity. [42]
Table 3. Possible substitution patterns for a trimer. The relative probabilities P and the mol % are calculated based on monomer data from a HPMC sample, see Cuers et al. [43]
𝑛(𝑅) 𝑛(𝑐3) 𝑛(𝑐2) 𝑛(𝑐1) 𝑛(𝑐0) Relative probabilities P Mole % 9 3 0 0 0 1 0.000 8 2 1 3 0.001 7 2 1 3 0.004 7 1 2 3 0.011 6 3 1 0.039 6 2 1 3 0.008 6 1 1 1 6 0.086 5 1 2 3 0.175 5 2 1 3 0.471 5 1 1 1 6 0.183 4 2 1 3 0.996 4 1 2 3 1.912 4 1 1 1 6 0.740 3 3 1 2.585 3 1 2 3 0.782 3 1 1 1 6 8.079 2 2 1 3 16.386 2 1 2 3 8.535 1 1 2 3 34.622 0 3 1 24.384
Based on the known molecular ratios of 𝑠𝑖 and 𝑐𝑖 the distribution of the number of substituents on a specific
dimer, trimer or oligomer for a random distribution can be calculated using Bernoulli binomial distribution statistics. [30] For an oligomer with a DP of 3 there are several combinations possible of non-, mono-, di-, and trisubstituted AGUs resulting in zero to nine substitutes along the chain, since the maximum DS per AGU is three. These combinations are shown in Table 3 where 𝑛(𝑅) are the number of substitutes on a trimer and 𝑛(𝑐𝑖) are the number of un- to trisubstituted AGUs within the trimer.
The probability of the different substitution patterns can be calculated based on the 𝑐i values, using
multinomial distribution calculations, as shown in Eq. (41). For the calculated values shown in Table 3, the following data was used from real monomer data of a HPMC sample: 𝑐3= 0.006681, 𝑐2= 0.072894, 𝑐1=
0.295681 and 𝑐0= 0.624744. According to Eq. (13), the DS/MS of this substituent is 0.46.
P = DP! n(c⁄ 3)! ∙ n(c2)! ∙ n(c1)! ∙ n(c0) (41)
The mole percentage for every individual substitution pattern is calculated according to Eq. (42). mole % = 100p ∙ c3n(c3)∙ c2n(c2)∙ c1n(c1)∙ c0n(c0) (42)
Finally, the substitution pattern can be calculated by summing all mole fractions of the same number of substitutes, 𝑛(𝑅). This leads to the mole fractions, as shown in Table 4, for the given example, describing a random distribution.
Table 4. Calculated substitution distribution pattern for a cellulose derivative with a DP of 3 and a DS of 0.46 n(R) Calculated distribution (%) 0 24.384 1 34.622 2 24.921 3 11.447 4 3.648 5 0.829 6 0.133 7 0.015 8 0.001 9 0.000
This random distribution will only be found for experimental data when all Spurlin’s assumptions are fulfilled. Deviation from the random pattern can indicate various distributions along the polymer chain as shown in Fig. 11. This data is based on an oligomer with a DP of 4 and a DS of 1.5. A more heterogeneous pattern, in comparison to the random model, can be identified by a flatter curve, which can be the result of crystalline regions as discussed before. The more regular pattern, on the other hand, shows a narrower distribution in comparison to the random model. This could be observed when the reactivities change during the cause of
the reaction by a change in polarity or by changing local accessibility because of e.g. steric hindrance or electrostatic repulsion. A distorted (broadened) pattern is the result of superimposing different random patterns which result in an average pattern that deviates from the random model. The bimodal pattern is observed when two competing mechanisms take place leading to two different reaction rates for the same sample.
Fig. 11. Diagrams of a calculated random distribution (shown in grey) and deviating distributions (experimental) which can indicate a more heterogeneous, more regular, distorted or bimodal distribution (shown in black) for an oligomer with a DP4 and an average DS of 1.5. The degree of substitution per oligomer is described as the number of reacted groups, 𝑛 (𝑅) and given as a mol fraction in the percentage of the total amount. Reproduced from Mischnick and Momcilovic. [5]
Still, for these models, the relative reactivities of the three hydroxyl groups are of great interest. When one hydroxyl group has a relative higher reactivity, the curve will shift as c0 decreases faster and c3 is formed at higher average DS. This is schematically shown in Fig. 12, for a DP of 3 and an average DS of 1 (average of 3 substitutes (n) per trimer). If there is no regioselectivity, the distribution would follow the black line. Its shows that there is a relatively large deviation (polydispersity) in the number of substitutes. If one hydroxyl group has a higher reactivity in comparison to the other two, all the AGUs will mainly be monosubstituted, leading to a high mol percentage of trimer with a total of 3 substitutes. The difference between the random calculated distribution and the experimentally determined distribution can be expressed
by the heterogeneity parameter, which slightly deviates from the calculation for the monomer composition. Here the difference in DS is taken, instead of the difference between the specific regioisomers, as given in Eq. (43) for a trimer (DP=3). [30]
H3= √∑[∆DS(n)]2 9
n=0
(43)
Fig. 12. Diagram of the theoretical distribution for 𝑛𝑖 substituents, where 𝑖 ranges from 0 to 6 in a dimer (left) and
from 0 to 9 for the trimer (right), both with an average DS of 1. The black line indicates no regioselectivity (k2≈k3≈k6)
and the grey line indicates one strongly favoured reaction side, for example, k2: k2>>k3≈k6. The mole fraction is given
5 Mass Spectrometry
Mass spectrometry has developed to be the method of choice for the analysis of polysaccharides for its separation efficiency, high sensitivity and rapid analysis. Much work on mass spectrometry has focused on structure elucidation and sequencing. However, one major limitation of MS is that the analysis is not inherently quantitative. The signals in MS are described as relative intensities which are proportional to the concentration but also affected by a lot of other parameters such as ionisation efficiency, matrix effects and ion transmission. In Fig. 13, an MS spectrum is shown of HPMC with an average DS for methyl (DSme) of 2.0 and an average molecular substitution of HP (MSHP) of 0.2. Although the relative substitution of HP is low, the relative intensities are high. [5] This is explained by the fact that the measured intensities are a product of various characteristics of the analyte, such as size, basicity (positive mode), structure and polarity and of the system itself like the dimensions, type of analyser etc. [41,44]
Fig. 13. ESI-MS spectrum of HPMC, hydrolysed to its monomeric forms. The DSme of the polymer is 2.0 and the MSHP
is 0.2. Even though the methyl substituents were present in a higher concentration of one order of magnitude, the spectrum is dominated by intense peaks which belong to AGUs with HP substitution. Reproduced from Momcilovic and Adden. [45]
For most MS applications, a rough estimation of molar ratios is enough, but for substitution-pattern analysis, a more quantitative way of performing MS is required. The most critical factor is the ionisation efficiency/yield, which is strongly affected by the substitution of the AGU. Generally, the ion formation of those neutral carbohydrates is dependent on the formation of cationic adducts, such as sodium. With a higher DS and MS, more reaction sides are introduced, causing an increase in intensity. Especially for oligomers with a low DP this can be of huge influence since the reaction sides are fewer. Another important parameter is the mass transfer in the analyser, which will be discussed later. However, in order to overcome all the discrimination effects in various stages, a chemical uniform structure is required (polarity, mass etc.) for quantitative MS. For cellulose ethers, this is not always that straight forward, so a deeper understanding of the various parts of the mass spectrometer and insight in the critical points for quantitative determination
5.1 Instrumentation
In this section, an overview will be given of the various critical components within a mass spectrometer which will be evaluated in terms of efficiency for monomer as well as oligomer analysis.
5.1.1 Ionisation
The most widely applied hard ionisation technique is EI. The initial electrons are formed in a thin filament which emits electrons. This filament is heated electrically to incandescence or to a temperature where it emits free electrons at a low pressure of 10-4 Pa. These free electrons are then focused using electric as well as magnetic fields. This creates an electron beam, usually with an energy of 70 eV (1 eV = 23 kcal mole-1) which is forced through a slit. During the ionisation process, the electron cloud of the neutral molecule in the gas phase interacts with the ion beam and absorbs a part of the energy. This leads to the ejection of one of its own electrons, resulting in a positive charge. This sample ion is defined as the molecular ion. Usually, the electron that is the least strongly bound will be ejected. However, as a result of the high amount of energy, almost any electron could theoretically be ejected to produce the ion. The more energy that is absorbed by the component, the higher order of fragmentation will occur. From this fragmentation pattern, the initial structure can be elucidated. This, together with reference libraries, makes EI the way to go for structure identification. A downside of this extensive fragmentation is that the molecular ion peak will be smaller in comparison to softer ionisation techniques, where less energy is absorbed. This makes identification and quantitation a tougher process. Due to these high energies, EI has a limited mass range of up to 103 Da. This makes it not useful for oligomeric or polymeric analyses, but a good tool for monomeric identification. [47]
For the analysis of cellulose oligomers, a soft ionisation technique is required to preserve the molecular ion peak, without losing quantitative information by fragmentation. In the 80s until the early 90s, the first analyses on cellulose derivatives were performed by fast-atom bombardment (FAB)-MS. Briefly, in FAB the analytes are dissolved in a matrix of a high boiling point component such as glycerol or m-nitrobenzyl alcohol. The high boiling point of the matrix is required since the matrix and sample are introduced in a high vacuum. Usually, Xe ions are first ionized using an EI source and accelerated using a specific potential. These fast-moving ions are then neutralized by collision with a cloud of neutral Xe atoms. Then, hence the name, the sample matrix is bombed with these fast-moving Xe atoms. As a result of the impact and the momentum of the Xe atoms, the sample gets ionized. [48] FAB is often seen as an improved version of secondary-ion mass spectrometry (SIMS) where a beam of ions, such as Ar+, is directly focused on the sample, which sputters out secondary ions. Molecular ions in FAB are often of the kind [M+H]+ or metal ion clustered. However, it shows also some fragmentation, intense background from the matrix, and it has a limited mass range with a maximum of 5,000 Da. [49] Therefore, the field has mainly shifted to other softer-ionisation techniques such as matrix-assisted laser desorption/ionisation (MALDI) and electrospray ionisation (ESI) which were developed more recently.
The development of MALDI in the first decade of the twenty-first century let to some large developments in the analysis of cellulose and its derivatives. Especially due to the higher mass range, generally up to
200 kDa, together with enhanced sensitivity of several orders of magnitude. [49] In contrast to FAB where the sample is in a liquid form, in MALDI the sample is commonly crystallized in a specific matrix. This matrix is chosen to absorb the light of a short laser pulse. During the irradiation, the absorbed light is transferred to heat which desorbs the analytes as well as the matrix itself and forms ions. This results in a ‘plume’ of super-compressed gas where the charge is passed to analyte molecules. This results in single negative as well as positive charged ions. [49] For a quantitative determination based on absolute relative intensities, the ion yield of the ionisation must be equal for all components. For this chemical uniformity is important since the molecular size and polarity influences how the compounds crystalize in the matrix and therefore also influence the ion yield. Another important reason why a chemical uniformity is required, was outlined before where the number of reaction sides are discussed which also influences the ion transfer, or complex formation with sodium for example, in the ‘plume’ of matrix ions and sample molecules. Important to note is that as a result of the high vacuum, more volatile (less polar) and low molecular weight (MW) compounds can be partially lost. [41]
Where FAB and MALDI are pulsed techniques due to the laser initiation, ESI is a constant process where solvated molecules are brought in the gas phase and ionized under an intense electric field. The sample is brought into a vacuum as a charged aerosol through a capillary under the influence of a strong electric field between the capillary end and a counter electrode. Two main mechanisms have been going around explaining the final transition from these charged aerosols to charged analyte ions. According to the ion evaporation model (IEM), a small ion can be ejected from the surface of a nanodroplet, resulting in a single ion. The driving force is the higher electrical charge on a smaller area, as a result of evaporation. Another way of explaining the ion formation is by the charge residue model (CRM) that suggest that the solvent evaporates until it reaches the size of the analyte. As a result of the Rayleigh limit, the charge is transferred to the analyte. This model has shown to be more applicable to larger molecules. [49] According to the IEM model, discrimination in ion yield can be explained for components with a difference in polarity, since this influences the distribution in the droplets from the core to surface. In cellulose ethers, especially unsubstituted hydroxy groups and substitutes which introduces a new hydroxyl group could affect the overall polarity. However, this effect tends to levels off at higher DP and DS. [41,50]
5.1.2 Mass analysers
After ionisation, the ions are transported to the mass analyser as a beam of ions. The first mass analysers that were introduced were the (double-focus) sector instruments. After acceleration, the ions are focused based on their kinetic energy and the momentum. The ions are selected and detected based on a distinct trajectory that is followed for ions with discrete m/z values. The trajectory is usually defined based on the magnetic field strength. Besides the fact that the principles of a sector instrument are still used in many mass spectrometers, the instruments itself are not generally used anymore. The main reason for this is that they have been outperformed in terms of sensitivity and resolving power while being more costly and harder to operate. [47]
Currently, the most commonly used mass analysers in cellulose derivative analyses are the TOF-MS and the IT-MS. TOF-MS is mainly used in combination with MALDI since they are both pulsed, non-continuous, techniques which allow high mass ranges. After the ions reacted to the inlet of the TOF, they are accelerated in a high vacuum under the influence of a pulsed electric potential in the acceleration grid. In this acceleration grid, the ions get approximately the same kinetic energy, which is defined as the mass times the velocity squared. Therefore, heavier ions will move slower through the drift tube and will arrive at a later point in time at the detector. Classical TOF instrumentation consists of a linear drift tube, but modern instrumentation consists often of a reflectron where the ions are being reflected into the direction of the ion source. This allows for correction of small differences in initial kinetic energy and increases the resolution. Since no other fields are applied, TOF-MS has a high ion transmission efficiency for a wide range of masses. This makes it possible to analyse components with a mass of 1 MDa for the linear TOF mass spectrometer. For most applications, this mass range would be sufficient. However, cellulose ethers tend to have masses up to 1.6 MDa. Moreover, a lot of different substituted polymeric chains will be present at only very low concentrations; therefore, fragments of these components are usually analysed [5]. The mass range of TOF instrument with a reflectron is one order of magnitude lower but has an increased resolving power.
The IT is often used in combination with ESI and is quite like a quadrupole with a relatively low mass range up to 1 kDa. The ions are trapped using an oscillating RF field and a superimposed DC electric field. The analytes are separated based on m/z and detected after ejection out of the trap by varying the RF potential. Mass spectrometers with an IT show generally higher sensitivity because the ions can be collected over a certain period. However, the resolving power is often limited to a single mass unit. [51] The IT has the possibility of performing MSn in a single analyser since the separation is performed in time instead of in space. In MSn collision-induced dissociation (CID) can be applied, where the analyte ions are fragmented by using a collision gas. This process is comparable to EI, but here the kinetic energy of the collision gas is transformed into internal energy of the analytes instead of by the energy by the 70 eV electron beam. The resulting fragmentation pattern will be different since the energy provided by the collision gas is often well below 70 eV. Using MSn in combination with fragmentation, structure elucidation can be performed, and more quantitative data can be obtained based on the specific fragmentation reactions. In comparison to EI, the analytes can be introduced in a solvent instead of as gas molecules, which is problematic for larger, high boiling point, structures. A big difference of the IT with the TOF is the limited mass range up to 2 kDa
5.2 Analytical strategies
To unravel the complex structure of the cellulose ethers, the polymer can be analysed on different hierarchical levels. The following section gives will a literature overview of different strategies using MS, to gain information on the DS, heterogeneity within the glycosidic unit and over the polymer chain using the principles that are described before in this work. An overview of the different methods, used to decipher the structure along the polymer chain, is given in Table 5.
5.2.1 Degree of substitution and substituent distribution of the glucose unit
An important starting point for research is to unravel the linkages between the structure of a cellulose derivative and the functional properties of the monomer analysis. The monomer data can give information about the overall average DS as well as the distribution of substituents over the different hydroxyl groups from which the reactivities can be calculated as well as the deviation from random models from Spurlin and Reuben. [25,26]
The analysis is generally performed by a very powerful combination of FID for quantification and GC-MS for identification. A high-resolution GC separation is required to separate the different monomers. For a simple HEMC ether, already 43= 64 different AGU patterns are possible including H and Me endcapping
without considering HE-tandem reactions (O-H, O-Me, O-HE-H, O-HE-Me). Identification is usually achieved by electron ionisation (EI). Since EI is a hard ionisation technique, it creates fragment ions, which are used for the identification which helps to determine the exact monomer composition. [52] The cellulose derivative is first per(deuterio)methylated, and then hydrolysed, reduced and acetylated to form an acetylated glucitol, as shown in Fig. 14. Methylation of the unsubstituted hydroxyl groups is required to prevent the intramolecular acetal formation. [55] The resulting structure is very similar to the Fischer projection of the glucose as shown earlier in Fig. 1. However, the aldehyde group is being reduced, resulting in another hydroxyl group on C-1. The product that is formed from reducing a monosaccharide is called an alditol. The reactive hydroxyl groups are either methylated or substituted by another derivative and the other non-reactive hydroxyl groups on the alditol are acetylated (C-1,3,4 and C-5). [41,53]
Fig. 14: Schematic representation of partially etherified and acetylated glucitol of HEC. The reactive hydroxyl groups have been indicated by 2, 3 and 6. The letters a, b, c and d indicate different fragments that are observed after EI. These fragments are used for identification of the modified monomeric unit. Reproduced from Lindberg et al. [52] The fragmentation pattern for methylated alditols follows a standard principle as described by Björndal et al. [54] This makes identification for other derivatives relatively straight forward. This substitution on the second position can directly be determined from the major fragment a, as depicted in Fig. 14. This fragment has different masses, dependent on the MS and the mass of the substituent itself. For HEC the mass of fragment a will be 144 + 𝑛2∙ 44 where 𝑛2 is the number of HE substituents on the second position and
44 the mass of the substituent. The next fragment gives the sum of the substituents on the third and the sixth position (𝑚/𝑧 = 233 + 𝑛3∙ 44 + 𝑛6∙ 44). It also gives a strong secondary fragment (not shown here)
with a mass difference of 60 (𝑚/𝑧 = 𝑏 − 60). Moreover, the sum of the degree of substitution on the second and third position is given a weak signal of the fragment c (𝑚/𝑧 = 161 + 𝑛 ∙ 44 + 𝑛 ∙ 44) and
