Bachelor Thesis Chemistry
Optimization of a comprehensive two-dimensional liquid
chromatography method for the separation of natural and
synthetic dyes
by
Lotje Bijkerk
27-06-2017
Studentnumber
10718826
Research institute
Van ’t Hoff Instituut of Molecular
Sciences
Research group
Analytical Chemistry
Supervisor
Dhr. prof. Dr. ir. P.J. (Peter)
Schoenmakers
Daily supervisor
Samenvatting
Het is plezierig om naar kunst te kijken, onder andere door de mooie kleuren die aanwezig kunnen zijn in een kunstwerk. Al eeuwen lang maken mensen voorwerpen mooier door ze te kleuren. In de loop der tijd is de manier van kleuren veranderd, evenals de kleurstoffen die gebruikt worden. Daar waar er eerst alleen natuurlijke kleurstoffen werden gebruikt (die hun oorsprong vinden in plant- en insectextracten), worden er sinds de toevallige ontdekking in 1865 van mauveïne, de eerste synthetische kleurstof, voornamelijk synthetische kleurstoffen gebruikt. Synthetische kleurstoffen worden door mensen gemaakt, en hun structuur wordt zodanig aangepast dat er een mooie, heldere kleur ontstaat. Het maakproces van synthetische kleurstoffen is daarnaast minder arbeidsintensief dan het extraheren van kleurstoffen uit natuurlijke bronnen. Maar kleuren vervagen in de loop der tijd. De kleurstoffen worden blootgesteld aan bijvoorbeeld warmte en UV-licht (zonlicht). Deze omstandigheden kunnen ervoor zorgen dat kleurstoffen degraderen, wat betekent dat de moleculen van de kleurstoffen in kleinere stukjes breken en kapot gaan. Dit kan vervaging als gevolg hebben, maar het is ook mogelijk dat de nieuwe kleinere moleculen een ander
soort kleur hebben. Dit roept de vraag op of de kleuren die wij in oude schilderijen zien, wel de oorspronkelijke kleuren zijn. In Afbeelding 1 is hier een voorbeeld van weergegeven. Aan de linkerkant van de afbeelding is te zien wat de kleuren waren toen Van Gogh het schilderij
schilderde in 1888, en aan de rechterkant van de afbeelding zijn
de kleuren te zien die wij nu waarnemen wanneer we dit schilderij zien hangen in het museum. Dat ziet er totaal anders uit!
Om deze reden is het belangrijk dat er onderzoek wordt gedaan naar kleurstoffen. Wanneer je bijvoorbeeld een schilderij wil onderzoeken, wil je weten welke kleurstoffen er zijn gebruikt en of deze kleurstoffen zijn gedegradeerd. Om dit te kunnen bepalen, moeten de moleculen van de verschillende kleurstoffen gescheiden worden. Dit kan gedaan worden met vloeistofchromatografie. Vloeistofchromatografie is een scheidingsmethode waarbij gebruik wordt gemaakt van een kolom met een vaste fase waarmee verschillende moleculen verschillende interacties hebben. Zo kan bijvoorbeeld gescheiden worden op lading (een molecuul van -1 zal langer blijven plakken aan de vaste fase dan een molecuul van -2) of op hydrofobiciteit (een molecuul dat minder van water houdt zal langer op de hydrofobe vaste fase blijven plakken dan een molecuul wat zich liever in de waterige loopvloeistof bevindt). Omdat er eindeloos veel verschillende soorten kleurstoffen zijn, en er in een schilderij ook een mix van natuurlijke en synthetische kleurstoffen kan voorkomen, is het gebruik maken van één soort scheiding niet genoeg. Daarom wordt er in dit project gebruik gemaakt van een scheiding in twee dimensies. Dit betekent dat er twee soorten scheidingen tegelijk (!) plaatsvinden. Na elke kleine scheiding in de eerste dimensie wordt dit groepje gescheiden moleculen doorgegeven naar de tweede dimensie waar deze moleculen dan op een zo anders mogelijke manier van elkaar worden gescheiden. Dit zorgt ervoor dat er zoveel mogelijk verschillende soorten moleculen gescheiden kunnen worden. Een twee-dimensionale methode opzetten is ingewikkeld. De methode moet aan allerlei eisen voldoen, zoals dat er zoveel mogelijk scheiding moet zijn, het niet te lang mag duren, maar nog belangrijker: dat de resultaten herhaalbaar zijn. In dit project is er daarom gefocust op de optimalisering van een twee dimensionale methode om zowel natuurlijke en synthetische kleurstoffen in één keer zo goed en zo herhaalbaar mogelijk te scheiden.
Afbeelding 1: 'De Slaapkamer' door Vincent van Gogh in 1888 (links) en 2015 (rechts).
Optimization of a comprehensive
two-dimensional liquid chromatography method
for the separation of natural and synthetic dyes
Abstract
The analysis of natural and synthetic dyes is increasingly a topic for studies because of the importance of research on cultural heritage objects. Samples from cultural heritage objects (e.g. historic paintings and textiles) can contain many different sorts of dyestuffs and their degradation products. When using liquid chromatography as a separation method, the use of a two-dimensional setup can increase the peak capacity and the number of sorts of dyestuffs that can be separated. In this project, a two-dimensional method is optimized for the separation of a mix of 74 different natural and synthetic dyestuffs. In the first dimension, strong anion exchange is used and in the second dimension reversed phase is used, coupled together in a comprehensive 2D-LC method (LC x LC). In this project, it is kept in mind while optimizing the LC x LC method that the method has to be compatible for a mass-spectrometer. The mass-spectrometer can identify which dyestuff is present in which peak of the chromatogram, when coupled to de LC x LC.
Table of contents
1. Introduction ... 5
1.1. Dyestuffs ... 5
1.2. Aim of project ... 6
2. Theory ... 7
2.1. Extraction and analysis ... 7
2.2. Strong anion exchange ... 7
2.3. (Ion-pair) Reversed phase ... 8
2.4. Comprehensive two-dimensional liquid chromatography... 9
2.5. Mass Spectrometry ... 11 3. Experimental ... 12 3.1. Instrumental ... 12 3.2. Chemicals ... 12 3.2. Analytical conditions ... 13 3.2.1. Sample preparation ... 13 3.2.2. Methods ... 13
4. Results and discussion ... 16
4.1. Strong Anion Exchange ... 16
4.2. Reversed Phase ... 20
4.3. LC x LC ... 21
4.3.1. Long vs. short modulation time for the second dimension ... 21
4.3.2. 50 vs. 75 mM ammonium sulphate in mobile phase B of the second dimension ... 22
4.3.3. Step gradient in anion-exchange ... 23
4.3.4. Normal gradient in anion-exchange and 40% B in reversed phase ... 24
4.3.5. Step gradient in anion-exchange and 40% B in reversed phase ... 25
4.3.6. Optimized LC x LC separation ... 26 4.4. Future perspectives ... 29 5. Conclusion ... 30 6. Acknowledgments ... 32 7. References ... 33 8. Appendix ... 34
1. Introduction
1.1. Dyestuffs
Before the nineteenth century, objects were brightened up with
dyestuffs originating from natural sources (plants and insects).1
After the discovery of the first commercial synthetic dye mauveine, in 1856 by William Perkin, the production of synthetic
dyes increased enormously.2 Mauveine itself had a short lasting
commercial lifetime, but its discovery let to the invention of a lot
of new and better dyestuffs.3 Nowadays there are over 400
synthetic dyes. The colouring matters are used in paintings, textiles and food but also in quantitative and qualitative medical analysis methods. The analysis of dyestuffs itself is gaining more importance in the field of cultural heritage studies because it can
chart the originally used dyes and their degradation products.4
With a successful qualitative and quantitative method for the analysis of the dyestuffs in cultural-heritage objects, it is possible to gain knowledge about the original appearance and the
creation-process.5 A recent example in which dyestuff analysis can be of
great importance is the discovery of a 17th century dress in a wreck in the Waddenzee (Figure 1). It is exceptional that such an old piece of textile is found in a good state so it is the perfect opportunity to gain more knowledge about the dyestuffs used in that time. Besides the characterization of the used dyestuffs, the degradation products can be studied to give an idea of what the original state of the object looked like. In Figure 2 the painting ‘De Slaapkamer’ by Vincent van Gogh is shown with its original colours (left) and with the colours that are now visible (right). In
almost 130 years of existence, the used dyestuffs in the painting are degraded with visible different colours as a result. In this case the analysis and the identification of the degradation products were very useful to regain the original state of the painting.
Figure 1 : 17th century dress found in the Waddenzee.
Figure 2: 'De Slaapkamer' by Vincent van Gogh in 1888 (left) and 2015 (right).
Figure 3: influence of the use of different mordants to dyed textile.
As mentioned before, there are many different sorts of dyes. The dyestuffs differ in their
structure and can be acidic, basic, direct or mordant (reactive).5 The first three groups contain
water-soluble dyes, with the anionic and cationic dyes in the first two groups. Direct dyes can be used to colour fibres without extra functional group in contrast to mordant dyes. A mordant is a metal that is
added to the dyestuff when the dyestuff itself has little or no affinity for the fibres.5 The metal forms
a coordination complex with the dyestuff and can be of influence for both the colour and the intensity of the colour on a piece of textile. In Figure 3, the effect of mordants on different (natural) dyestuffs on wool can be seen.
1.2. Aim of project
The analysis of natural and synthetic dyes is increasingly a topic for studies because of the importance of research on cultural heritage objects. In most of these studies the dyestuffs are split up in sorts (e.g.
exclusively acidic or only sulfonated dyes) or only a few separate dyestuffs are studied.1,6,7,8,9 But as
said before, samples from cultural heritage objects can contain many different dyestuffs and possibly also their degradation products. It is therefore necessary that an analysis method is found that is usable for all dyestuffs and that needs as less sample as possible to keep the cultural heritage object intact. Because of the fact that the samples can contain many different dyestuffs, there is chosen to work with a two-dimensional setup. In this project both dimensions (strong anion exchange and ion-pair reversed phase) were first optimized in a one-dimensional setup. Once both dimensions are optimized apart from each other in separating megamix (a mix of 74 different natural and synthetic dyestuffs), they were coupled together in a two-dimensional setup. From there on, the separation of megamix is further optimized on the two-dimensional setup. The ultimate goal of this project is to set up an optimized two-dimensional method which is compatible for the mass-spectrometer so that all the small separated fractions can be transferred from the second dimension to the mass-spectrometer to identify which dyestuff is present in which peak in the chromatogram.
2. Theory
2.1. Extraction and analysis
The analysis of dyes in historical objects is of great importance, but is also very difficult. First, the number of possible dyes in an object is very high and together with a variety of degradation products of all these dyes, many different products can be found in only one sample. Second, the dyes can be
impure (they were made for their bright colour instead of pure compounds) and chemically instable.5
Moreover, when researching a cultural heritage object, it is necessary to use as little as possible from
the object to prepare a sample so the integrity of the object is not disturbed.3 Therefore an effective
extraction method is needed to get the dyestuffs out of the object. For the extraction of dyestuffs from wool there is already an effective method, in which only 1 mg wool is needed. The dyestuffs are
extracted with the method shown in Scheme 1 and dissolved in acetonitrile/DMSO (1:1).10 This
extraction method is set up to make analysis with high performance liquid chromatography and mass spectrometry possible. These two analysis methods are an obvious choice because high performance
liquid chromatography combines reproducibility, speed and high resolution.1 Besides that, it can be
combined with analysis-systems for peak identifications such as mass spectrometry. In this project both strong anion-exchange and ion-pair reversed-phase liquid chromatography are used. For structure determination, it is possible to use mass spectrometry. In the next three sections, these methods are further explained.
Scheme 1: extraction method for extraction of dyestuffs from wool.10
2.2. Strong anion exchange
In ion-exchange chromatography, molecules are separated on the basis of the affinity of the molecules
for the charged, stationary phase of the column.11 A separation is achieved between differently
charged stationary phase. Molecules containing more charged groups have more affinity to the oppositely charged stationary phase and will thus have more retention. Ion-exchange chromatography is a suitable method for the analysis of many synthetic dyes because the dyes vary
largely in charged states.5 In this study, anion exchange is used because most charged dyes contain a
negative group. The stationary phase for an anion-exchange column is positively charged, and so the more negatively charged the analyte, the more retention. There are two different mechanisms for ion exchange: weak and strong. In weak ion exchange the charge of the analytes, and with that their affinity for the stationary phase, are pH dependent. The separation can only be performed in a narrow pH range, while strong ion exchange can be performed within a wide pH range because the number of charges on a strong ion exchanger is independent of the pH. With strong ion exchange, the affinity of the analytes is determined by the concentration of salts from the mobile phase in the column. In the gradients used while working with strong ion exchange, the concentration of salts is gradually built up. This results in a competition mechanism in which the salt will replace the analytes in order of the charge of the ions. Eventually all the analytes will lose their retention because an excess of salts is being added. To give reproducible results, it is necessary to remove the (strong bonding) salts from the stationary phase before analyzing a new sample. When injecting a new sample while the salts of the mobile phase are still bonded to the stationary phase, the analytes will show less or no retention because the stationary phase is still saturated with the salts of the mobile phase. Because of this reason, a third mobile phase is added with an excess of weak interacting salts. While ‘flushing’ with this mobile phase, the excess of salts from the third mobile phase will push the salts of the second mobile phase away. Given the fact that it is a weak interacting salt, the column is easily re-equilibrated with the first mobile phase (containing no salts). In section 3.2.2 and 4.1 the conditions needed for good and reproducible anion-exchange results are further explained.
2.3. (Ion-pair) Reversed phase
Reversed phase HPLC is a form of high performance liquid chromatography in which molecules are separated on the basis of hydrophobicity. A reversed-phase column is built up from a hydrophobic, nonpolar (silica modified with octadecyl alkyl (C18) chains) stationary phase through which a polar mobile phase flows. Hydrophobic molecules have higher affinity for the stationary phase than
hydrophilic molecules, resulting in a higher retention time.12 In this study, reversed phase is used
because dyestuffs contain conjugated ring structures, which are hydrophobic and thus give retention. When focused on the degradation products of dyestuffs it is important to take in mind that the degraded products will be smaller than the original dyestuffs and will thus be less hydrophobic. With
reversed-phase liquid chromatography, it is possible to demonstrate the degradation products because of the difference in hydrophobicity. For these two reasons, reversed-phase liquid chromatography is an obvious choice for the analysis of dyestuffs.
As mentioned before, dyestuffs can be separated by reversed phase, but because many dyestuffs have a (negative) charge, it is important to use an ion-pair when practising reversed phase. Because of the charge, the molecules can have hydrophilic interactions with the mobile phase, which results in less retention with the stationary phase. When an ion-pair is used, the charged dyestuffs are neutralized by a counter-ion (the ion-pair), which helps to create more retention. The fact that besides reversed phase also ion exchange and (possibly) mass spectrometry are used as analysis methods in the two-dimensional analysis, makes it more difficult to choose a suitable ion-pair. Firstly,
the ion-pair has to be low in UV absorption so it is not detected in the chromatogram.9 Secondly, the
ion-pair has to be volatile enough to be compatible with the mass spectrometer.9,13 The analysis of
aromatic sulfonated compounds (including azo-dyes) is widely studied.9,13,14 These studies show that
when using LC-MS methods, trialkylamines are possible ion-pairs. Both Fuh13 and Storm14 show that
triethylamine is a suitable ion-pair for the liquid chromatography analysis of highly sulfonated compounds (azo dyes) and is volatile enough for the mass spectrometer when coupling the LC to the MS.
2.4. Comprehensive two-dimensional liquid
chromatography
As pointed out in the first section, the dyestuff samples are complex because one sample can contain any number of different dyestuffs. Also, on the one hand, there is a lot of similarity between the molecules within the groups of dyestuffs but on the other hand, there are great differences between the different groups. This makes analyzing dyestuff samples difficult. Besides that, the concentration of dyestuffs in the samples can be low because it is important to keep the objects of which the samples are taken as intact as possible. These reasons indicate that analyzing dyestuff samples with a one-dimensional method is not sufficient. Combining two liquid chromatography methods can increase the possible number of analytes that are possible to separate in one time and can improve the peak
resolution and capacity.15 These advantages lead to the decision to work with a (comprehensive)
two-dimensional method when separating dyestuffs. In comprehensive liquid chromatography (LC x LC), all effluent is transferred from the first-dimension column to the second-dimension column in small
volume fractions.15 After the analytes have passed through both columns, they reach the detector.
that were passed on from the first dimension to the second dimension, are combined.15 The overall
chromatogram will thus span over the time of the first-dimension gradient. Because of this reason, the first-dimension gradient is very slow, so that there is enough time for the peaks of the first dimension to pass the effluent to the second dimension. This is done with the help of a switching valve, built up from two loops: one that collects the effluent from the first dimension and one that transfers its content to the second dimension. A schematic
overview of the switching valve is shown in Figure 4. The valve switches in the modulation time: the time of the second-dimension gradient. The time of the second-second-dimension gradient is for this reason very short; to be sure all the small fractions from the first dimension can be separated in the second dimension.
When working with a two-dimensional setup, it is important to keep the concept of orthogonality in mind.
Orthogonality is used, in separation science, as a measure of independency between separation
techniques.15 The peak capacity of a LC x LC chromatogram is higher when more orthogonal
mechanisms are used. In this project, a two-dimensional setup is made from strong anion exchange (first dimension) and ion-pair reversed phase (second dimension). These separation mechanisms are orthogonal to each other and are obvious choices to separate dyestuffs, as explained in the sections above. This means that first the molecules will be separated on basis of their charge, from there the differently charged groups will be further separated with ion-pair reversed phase. This will result in a chromatogram in which hopefully all dyestuffs can be separated. As said in section 1.2, the ultimate goal of this project is to set up an optimized two-dimensional separation method which is compatible with the mass-spectrometer. When the molecules from all the different peaks from the second dimension are send to the mass spectrometer, it is possible to identify the dyestuffs which are present in the sample. Because of the fact that a two-dimensional setup gives better peak capacity and therefore more separation is possible, mass spectrometry is easier to combine than when using a one-dimensional setup. With a two-one-dimensional separation less dyestuffs will be present in one peak, which results in a more pure fraction that will be passed on to the mass spectrometer. The mass-spectrometer is then better capable to ionizing and detecting the molecules in the fraction. The combination of a two-dimensional setup and mass-spectrometry will give a very strong indication of the used dyestuffs in the samples.
Figure 4: Coupling of columns to the switching valve. (Source: Agilent OpenLAB CDS program).
2.5. Mass Spectrometry
Mass spectrometry is one of the most powerful analysis methods because it can give both quantitative
and qualitative information with high sensitivity.12 The mass spectrometer produces and weighs ions
from injected samples from which the molecular weight can be obtained.16 The molecules from the
sample are brought into the gas phase from where ions are produced. The (ion-)molecules are then separated in their mass-to-charge ratio (m/z). The mass spectrometer contains an electrostatic field which will accelerate the charged molecules. Thus the mass spectrometer is only capable of detecting
charged components.16 There are a few methods in which molecules are taken into the gas phase and
are ionized and there are also more possibilities in how the ions are then detected. The ions in this study will be produced by electrospray ionization (ESI). With the ESI method, a small flow of liquid is made into a fine mist, from which dissolved sample molecules are released from droplets by
evaporation of the solvent.16 As said in the last section, it is possible to couple the liquid
chromatographer to the mass spectrometer. In liquid chromatography, the sample is separated with the help of a mobile phase, and the sample is therefore eluted in this mobile phase. Mass spectrometry however, needs a sample as pure as possible. The solvent is evaporated, as indicated above, and it is therefore necessary that the solvent (the mobile phase of liquid chromatography in this case) is volatile enough for quick evaporation. It is also for this reason that the possible ion-pairs for the reversed phase separation are limited. As said in the previous section, the ion-pair must be very volatile when used in combination with mass spectrometry. After ionization, the molecules will be detected with, in this study, a time-of-flight analyser. Ideally all ions have the same kinetic energy.
This will cause a separation between the mass of the ions because of the kinetic energy equation Ekin
= ½ mv2. This equation shows that lighter molecules will have a higher velocity, so they will reach the
detector earlier than the heavier molecules.12
Mass spectrometry is a perfect (extra) analysis method after using two-dimensional liquid chromatography because it can give information about which molecules are present in which peaks of the chromatograms. This information is very useful when identification of used dyes and their degradation products is needed.
3. Experimental
3.1. Instrumental
For the optimization of the gradients for the first and the second dimension, a one-dimensional setup was used on a Shimadzu HPLC (HPLC 3). The machine was built up from a liquid chromatograph (LC-10AD), a communication bus module (CMB-20A), an UV-VIS detector (SPD-20A), a mixer (FCV-10AL)
and a degasser (DGU-20A5). The system was controlled by the Shimadzu Real Time Analysis program
(LCsolution Version 1.21 SP1). The 1D-LC data was processed with Microsoft Office Excel 2013. The two-dimensional setup was done on an Agilent 1290 Infinity 2D-LC system (Agilent, Waldbronn, Germany). The machine was built up from an Agilent 1100 series autosampler, two Infinity 1290 binary pumps (G4220A), two Infinity 1290 thermostated column compartments (G1316C) and an
Infinity 1290 diode-array detector (G4212A) operated by an Max-Light Cartridge Cell (G412-6008, V0
= 1.0 µL). To switch from the first to the second dimension, a 2-position 8-port valve (G4236A) with two 60-µL loops was used. The two- dimensional setup was controlled by the Agilent OpenLAB CDS Chemstation Edition (Rev. C.01.04[35]) software. The 2D-LC data was processed with Transform.
The used strong anion exchange column was an Agilent PL-SAX column (PL1951-3802, 150 x
2.1 mm i.d., 8-µm particles, 1000-Å pore size). The used reversed phase column was an Agilent
ZORBAX Eclipse Plus C18 Rapid Resolution HT (959941-902, 50 x 4.6 mm, 1.8 µm particles, 1000-Å
pore size). In the one-dimensional setup, there was coupled an in-line filter (Agilent 1290 in-line filter) to both the anion exchange column and the reversed phase column to protect the columns from insoluble dyes. In the two-dimensional setup, the in-line filter was coupled to the first-dimension column, the strong anion-exchange column.
3.2. Chemicals
The 74 authentic dyestuffs (natural and synthetic from the period 1850 – 1920) were obtained from the reference collection of the Cultural Heritage Agency of the Netherlands (RCE, Amsterdam, The Netherlands). It is possible that the samples contain impurities and degradation products. The dyestuffs were dissolved in Acetonitrile (ACN, LC-MS grade, obtained from Avantor Performance Chemicals, Deventer, The Netherlands) and DMSO (obtained from SAFC supply Solutions, Steinheim, Germany). The mobile phases were prepared with in house purified milli-Q water (Millipore Q-POD, Millipak Express, 0.22 µm), Acetonitrile (ACN, AR grade, obtained from Avantor Performance Chemicals, Deventer, The Netherlands). The salts and buffers used in the mobile phases for strong
anion exchange were ammonium sulphate (BioXtra ≥99.0%), sodium chloride (BioXtra ≥99.5%), Trizma hydrochloride (BioPerformance Certified, cell culture tested, ≥99.0%) and Trizma Base (Reagent grade; minimum ≥99.9%), all obtained from Sigma-Aldrich (Darmstadt, Germany). The buffer and ion-pair for the mobile phases for reversed phase were prepared with formic acid (reagent grade, ≥95%) and triethylamine (TEA, purity ≥99.5%), both obtained from Sigma-Aldrich (Darmstadt, Germany).
3.2. Analytical conditions
3.2.1. Sample preparation
All optimization measurements are done with separation of megamix 60 ppm. This mix contains all natural and synthetic dyestuffs from the period of (1850 – 1920), 74 dyestuffs in total, provided by the Cultural Heritage Agency of the Netherlands. From each dyestuff a 5000-ppm solution of ACN/DMSO 1:1 (v/v) was present, from which 60 ppm solution was made by putting together 100 µL of each dyestuff with a supplement of ACN/DMSO 1:1 to 10 mL.
3.2.2. Methods
As written in section 2.1, the used column for anion-exchange liquid chromatography was an Agilent PL-SAX. Two different gradient setups were used, both having the same mobile phases and ‘flush gradient’ after the gradient for the peak area. For the one-dimensional setup, the flowrate was set to 1.0 mL/min. The injection volume was approximately 20 µL. The mobile phases consisted of (A) 60% water, 40% ACN and 5 mM of Tris buffer (pH=7); (B) 60% water, 40% ACN, 5 mM of Tris buffer (pH=7) and 50 mM ammonium sulphate and (C) 1M NaCl in water. The ‘flush gradient’ consisted of 15 column-volumes of C (5.25 min) followed up by 35 column-volumes of A (12.25 min). The first gradient setup (for 1D measurements) had a time program of 31 min: 0 - 0.5 min isocratic at 100% A; 0.5 – 10.5 min a linear gradient to 100% B; maintained at 100% B for 2 min; 12.5 – 13.0 min a linear gradient to 100% C; maintained at 100% B for 5.25 min; 18.25 – 18.75 min a linear gradient to 100% A; maintained at 100% A for 12.25 min. The second gradient setup was a so-called step-gradient. The gradient setup (for 1D measurements) had a time program of 22 min with a flow rate of 1.0 mL/min: 0 – 1.5 min gradually to 20% B; 1.5 – 2.2 min gradually from 20% to 60% B; 2.2 – 3.7 min gradually from 60% to 100% B, maintained at 100% B for 0.1 min; 3.8 – 9.75 min isocratic at 100% C; 9.75 - 22 min isocratic at 100% A.
The used reversed-phase column was an Agilent ZORBAX Eclipse Plus C18 RRHT column (section 2.1). For the one-dimensional setup, the flowrate was set to 1.0 mL/min. The injection volume was approximately 20 µL. Two mobile phases were used: (A) 95% buffer, 5% ACN (v/v) and (B) 95% ACN, 5% buffer. The buffer consisted of 5 mM triethylamine (TEA, ion-pair) in water, brought to pH=3 with formic acid. As with ion-exchange, two different gradient setups were used. The first gradient setup (for 1D measurements) had a time program of 16 min: 0 – 0.5 min, isocratic at 100% A; 0.5 – 12.5 min a linear gradient to 100% B, maintained at 100% B for 0.5 min; 13.0 – 15.0 min a linear gradient to 100% A, maintained at 100% A for 1 min. The second gradient setup (for 1D measurements) also had a time program of 16 minutes, but started at 40% of mobile phase B: 0 – 0.5 min, 40% B; 0.5 – 12.5 min a linear gradient from 40% to 100% B, maintained at 100% B for 0.5 min; 13.0 – 15.0 min a linear gradient from 100% to 40% B, maintained at 40% B for 1 min.
For comprehensive two-dimensional liquid chromatography, the same anion-exchange and reversed-phase columns were used with the same mobile phases. The injection volume was 10 µL. The modulation time was 1.5 min. For the LC x LC measurements, the 1D gradients were modified and translated to a 2D setup. Both the anion-exchange and the reversed phase gradients had to be modified for two reasons. Translating the anion-exchange gradient from 1D to 2D (from a flowrate from 1.0 mL/min to 0.02 mL/min) will result in an anion-exchange gradient that will take approximately 11 hours, which is too long. The gradient is modified to a shorter one, which will take approximately 3.5 hours. The reversed phase gradients had to be modified because of the modulation-time in the 2D measurements. The modulation-modulation-time in the 2D setup is 1.5 min, a program modulation-time of 16 min is therefore not applicable. The reversed phase gradient is modified to a shorter one, so that it fits in the modulation time of 1.5 min. The first dimension (anion-exchange) had a flowrate of 0.02 mL/min in the peak area and a flowrate of 1.0 mL/min in the flush gradient. The first anion exchange (first dimension) gradient setup had a time program of 208 min: 0 – 5.0 min isocratic at 100% A (0.02 mL/min); 5.0 – 185 min gradually to 100% B (0.02 mL/min); 185 – 190 min gradually to 100% C (0.02 mL/min); 190 – 195.75 min isocratic at 100% C (1.0 mL/min); 195.75 – 208 min isocratic at 100% A (1.0 mL/min). The step-gradient for anion exchange (first dimension) also had a time program of 208 min: 0 – 5.0 min isocratic at 100% A (0.02 mL/min); 5.0 – 75.0 min gradually to 20% B (0.02 mL/min); 75 – 110 min gradually to 60% B (0.02 mL/min); 110 – 185 min gradually to 100% B (0.02 mL/min); 185 – 190 min gradually to 100% C (1.0 mL/min); 190 – 195.75 min isocratic at 100% C (1.0 mL/min); 195.75 – 208 min isocratic at 100% A (1.0 mL/min). The modulation time in the 2D setup was 1.5 min, and the time program of the second dimension was therefore also 1.5 min. The flow rate was set to 2.4 mL/min. One second dimension gradient started at 0% B: 0 – 1.20 min
gradually from 0% to 100% B; 1.20 – 1.30 to 100% A, maintained at 100% A till 1.50 min. The other second dimension gradient started at 40% B: 0 – 1.20 min gradually from 40% to 100% B; 1.20 – 1.30 to 40% B, maintained at 40% B till 1.50 min.
4. Results and discussion
4.1. Strong Anion Exchange
The used gradient is built up from three different mobile phases as is explained in section 3.2.2. Mobile
phase C (1M NaCl in H2O) is used to remove the sulphate anions from mobile phase B (50 mM
(NH4)2SO4) from the positive stationary phase of the column. The NaCl is added in excess to be sure
that it is strong enough to remove all the sulphate anions even though the bond between the Cl anions and the stationary phase of the column is not strong. To remove the excess of NaCl and to re-equilibrate the column, mobile phase A (contains no salts) is used. To be sure the results are reproducible, it is necessary to know how much column-volumes C and A are needed to bring back the column to equilibrium. Firstly, the number of column-volumes C is optimized while the number of column volumes A is held at a constant excess to be sure that only the influence of the quantity of C is being shown. After the optimized number of column-volumes C, the number of column volumes A is optimized while keeping the number of column-volumes of C constant at the optimized number. This is schematically presented in Table 1- 3.
In Figure 5 the chromatograms of megamix with different gradients for the optimization of C (Table 1 – 2) are shown. During the optimization of C, the number of column-volumes of A (for flushing after C) is held constant at 40 column-volumes to be sure there is used enough of mobile phase A. The measurements were performed in duplo to demonstrate if the gradients gave reproducible results. The figure shows that gradient 5 had the best reproducible results. This shows that 15 column-volumes of C works best for this column. During the optimization of the number of column-column-volumes of A, therefore flushing with mobile phase C will be held constant at 15 column-volumes.
Table 1: strong anion exchange gradient setup with variables X, Y and Z (see Table 2 and Table 3).
Time A (%) B (%) C (%) 0 100 0 0 0.5 100 0 0 10.5 0 100 0 12.5 0 100 0 13.0 0 0 100 X 0 0 100 Y 100 0 0 Z 100 0 0
Table 2: schematic overview of optimizing number of column volumes of A with the variables X, Y and Z (Table 1). Gradient Column-volumes of C X Y Z 1 35 25.25 25.75 39.75 2 30 23.50 24.00 38.00 3 25 21.75 22.25 36.25 4 20 20.00 20.50 34.50 5 15 18.25 18.75 32.75
Figure 5: Megamix on strong anion exchange with different gradients for the optimization of column-volumes of mobile phase C. Used column is Agilent B.041. Specifications of the used gradients are given in Table 1 - 2.
The gradients that were less reproducible (gradients 1 - 4) than gradient 5, show differences in peaks in the area of the mono-anions. This can be explained by the fact that the salt used in mobile phase B and C, ammonium sulphate and sodium chloride, have a -1 charge. The mono-anions of the sample have to compete with these chloride anions directly from the moment of injecting when too much column volumes of C is used. The mono-anions of the sample will have less retention because the chloride anions are taking their place on the column. There is less difference in the peak area of the di-anions of the sample because they will bind stronger to the stationary phase of the column than the chloride anions. This results in less competition and therefore a smaller difference in the retention time is seen when more or less sodium chloride is used.
In Figure 6 the chromatograms of megamix with different gradients for the optimization of A (Table 1 and Table 3) are shown. Gradient 6 shows the best reproducibility and the best divided peaks. In gradient 6 there is flushed with 15 column-volumes of C and 35 column-volumes of A (Table 3). These numbers of column volumes will be used for flushing after each ion-exchange measurement to be sure that the ion-exchange column is fully re-equilibrated and that the results are reproducible.
Table 3: schematic overview of optimizing number of column volumes of A with the variables X, Y and Z (Table 1).
Gradient Column-volumes of A X Y Z
6 35 18.25 18.75 31.00
7 30 18.25 18.75 29.25
8 25 18.25 18.75 27.50
Figure 6: Megamix on strong anion exchange with different gradients for the optimization of column-volumes of mobile phase A. Used column is Agilent B.041. Specifications of the used gradients are given in Table 1 and Table 3.
In the 2D setup, the flowrate of the first dimension (ion-exchange) is very slow (0.02 mL/min). It therefore takes a long time to use the gradient of B which is used in the optimization gradients. The time of a 2D setup would take approximately 11 hours when the optimized gradient is used. It is better and more sustainable to have a shorter gradient. For this reason the very first 2D measurements were
done with the gradient for B which was used before in this research.5 Mobile phase B gradually goes
up to 100% in 180 minutes, which corresponds with 5.14 column-volumes. The results of this 2D setup were not reproducible. When the used gradient was translated back to a 1D setup with a flow of 1.0 mL/min, it showed that the analytes with a charge of -2 did not come off the column with this little amount of column-volumes of B. This is shown in Figure 7 (left). The molecules with a di-anionic charge have a retention time of approximately 6 minutes, while C is added after 4.5 minutes. This explains why the results in the 2D setup were not reproducible. Because the time is wanted to be as short as possible, another solution had to be found other than a longer gradient for B. For this reason, it is tested if increasing the concentration of ammonium sulphate in mobile phase B helps to get all the analytes off the column in a limited time. Three different concentrations were tested (50 mM, 75 mM and 100 mM) in the translated gradient from the 2D setup. In Figure 7 the results of 50 mM (left) and 75 mM (right) ammonium sulphate are shown. With a concentration of 75 mM ammonium sulphate, all the analytes come off the column before flushing with C. The 100 mM solution gives better separated peaks (not shown here), but because of the danger of clogging and the possible use of a mass spectrometer, there is chosen to work with 75 mM ammonium sulphate in mobile phase B. When a new Agilent column (B.048) was used, the 50 mM ammonium sulphate in mobile phase B gave better results. In Figure 8 the translated gradient on the new Agilent column is shown with 50 mM (left) and 75 mM (right) ammonium sulphate. There is thus worked with 50 mM mobile phase because it gave better and more reproducible results (Figure 8) and because an as low as possible concentration of ammonium sulphate is preferred to prevent clogging and to take into account the possible use of a mass-spectrometer. From these results, it can be stated that a higher concentration of ammonium sulphate in mobile phase B is needed when the column is older and in a less good state.
Figure 7: Megamix with translated 2D ion-exchange gradient with 50 mM (left) and 75 mM (right) ammonium sulphate in mobile phase B on Agilent Anion Exchange column B.041.
4.2. Reversed Phase
The used reversed phase gradient on 1D is based on the article of Pirok et al.5 The mobile phases used,
contained triethylamine (TEA) as an ion-pair instead of TMA or TBA because TEA is more compatible
with the mass-spectrometer.13,14 Two different gradients were tested for the reversed phase
separation of megamix. The first one tested, starts at 100% mobile phase A (95% buffer and 5% ACN)
and 0% of mobile phase B (95% ACN and 5% buffer).5 The second gradient tested starts at 40%
mobile phase B. It is chosen to test the second gradient for two reasons. The first reason is that the first gradient shows no peaks before the gradient is at 40% B, so starting at 40% B will ensure that the peaks will come earlier in the gradient. The second reason is that the ‘breakthrough effect’ will be reduced in the 2D setup when the second dimension (reversed-phase) will start at 40% of mobile phase B because the mobile phase used in the first dimension (anion-exchange) also consists of 40% acetonitrile. The fractions of the first dimension are injected into the second dimension and the concentration of ACN in the mobile phase used in the first dimension is much higher than in the mobile phase of the second dimension. The analytes will than rather retain in the solution of the first dimension than in that of the second dimension, which results in less retention in the second dimension. This breakthrough effect is reduced when an injection solvent is used that is as weak an
eluent as possible.17 The use of a gradient that starts at 40% of mobile phase B in the second
dimension is therefore a logical choice. The chromatograms of both gradients in a one-dimensional setup are shown in Figure 9 and Figure 10. The peaks in the gradient that starts at 40% B (Figure 10) come earlier than the peaks in Figure 9. The reason for this effect is explained above. An important thing to take into account when the peaks shift to a shorter retention time is that degraded dyestuffs are less hydrophobic and will therefore have a shorter retention time. For this reason, the shift cannot
Figure 8: Megamix with translated 2D ion-exchange gradient with 50 mM (left) and 75 mM (right) ammonium sulphate in mobile phase B on Agilent Anion Exchange column B.048.
be too great because otherwise the degraded products will disappear in the t0 peak. The effect of the
(possibly) reduced breakthrough effect can only be seen on 2D measurements.
Figure 9: Megamix on reversed phase with the gradient starting at 0% of mobile phase B. (see section 3 for specifications of gradient and mobile phase).
4.3. LC x LC
The aim of this project is to optimize the 2D method used in the publication of Pirok et al.5 Besides
the optimization of the method, it is the aim to be able to separate more dyestuffs (74 instead of 54), a mix of natural and synthetic dyestuffs instead of only synthetic dyes, and to make the method reproducible. The 2D chromatogram of their research is shown later in this section. After the separate optimization of the first and second dimension, different 2D setups were tested and they are compared to see which gradient works best for the separation of megamix (74 different synthetic and natural dyestuffs). In the first dimension (strong anion exchange) the re-equilibration of the column, the mobile phases and the gradient were optimized in a one-dimensional setup. Furthermore, the second-dimension gradients (ion-pair reversed phase) were tested on good separation and reproducibility in a one-dimensional setup. With these optimized dimensions, different combinations were tested in a two-dimensional setup. In the next sections, the chromatograms of the different gradients are shown and compared to each other. Finally, the best tested gradient is compared to the
chromatogram in the article of Pirok et al. to show the improvements.5
4.3.1. Long vs. short modulation time for the second dimension
Firstly, it is tested if it is better to use a longer (3 minutes) or shorter (1.5 minutes) modulation time for the second dimension. A longer modulation time could improve the separation in the second dimension because the reversed phase gradient is more spread out. A disadvantage of a longer modulation time is the fact that bigger fractions of the first dimension will be send to the second dimension. This means that obtained separation in the first dimension will be partly lost when
Figure 10: Megamix on reversed phase with the gradient starting at 40% of mobile phase B. (see section 3 for specifications of gradient and mobile phase).
sending it to the second dimension. In Figure 11 the chromatogram of the separation of megamix is shown with a modulation time of 3 minutes and in Figure 12 the chromatogram with a modulation time of 1.5 minutes is shown. The modulation time of 1.5 minutes shows more separation between the di- and tri- anionic molecules but the separation between the mono-anionic molecules is not very different apart from the fact that all peaks in in the chromatogram in Figure 11 are more intense and broader. This effect is seen in more chromatograms, but it is still unclear what the cause for this problem is. Because there is more separation when a shorter modulation time (1.5 minutes) is used, there is chosen to work with a modulation time of 1.5 minutes (90 seconds) in the following chromatograms.
4.3.2. 50 vs. 75 mM ammonium sulphate in mobile phase B of the second
dimension
As described in Section 3.1, the concentration of ammonium sulphate in mobile phase B of anion-exchange (60% water, 40% ACN, 5 mM tris buffer pH=7 and a to be determined mM of ammonium sulphate) has influence on the separation of natural and synthetic dyes. On the first used Agilent anion exchange column (B.041), 50 mM of ammonium sulphate was not enough for all analytes to lose their retention (Figure 7) in the relatively short time when the 2D gradient is translated to a 1D gradient. When 75 mM ammonium sulphate was used, all the analytes lost their retention and the result was reproducible (Figure 7). When switched to a 2D setup, a new Agilent anion-exchange column was used (B.048) and therefore both mobile phases were tested again (Figure 8). In Figure 13 and Figure
14, the 2D results of these measurements are shown. When 50 mM ammonium sulphate is used, more
peaks were obtained. For this reason, and also because as less ammonium sulphate as possible is
Figure 11: UV chromatogram (254 nm) of LC x LC (1st dimension
strong anion exchange and 2nd dimension reversed phase)
separation of megamix 60 ppm with a modulation time of 3 minutes in the second dimension.
Figure 12: UV chromatogram (254 nm) of LC x LC (1st dimension
strong anion exchange and 2nd dimension reversed phase)
separation of megamix 60 ppm with a modulation time of 1.5 minutes in the second dimension.
0 25 50 75 100 125 150 175 200 175 150 125 100 75 50 25 0
First Dimension retention (min) - Ion exchange
S e c o n d D im e n s io n r e te t io n ( s ) - R e v e rs e d P h a s e C 1 8 0 437500 875000 1312500
Signal Intensity in mAu
0 25 50 75 100 125 150 175 200 88 75 63 50 38 25 13 0
First dimension Retention (min)_Ion exchange
S e c o n d d im e n s i o n R e t e n t i o n ( s ) _ r e v e r s e d p h a s e ( C 1 8 ) 0 1375000 2750000 4125000 5500000
wanted when the LC x LC is coupled to the mass-spectrometer, there is chosen to work with 50 mM ammonium sulphate in mobile phase B of anion exchange.
4.3.3. Step gradient in anion-exchange
As described in section 3.2.2 a step-gradient for the second dimension is tested. In this step- gradient, the time to go from 0% to 20% of mobile phase B is spread out, the time from 20% to 60% is speeded up and the time to go from 60% to 100% is again spread out. This setup is chosen because most analytes lose retention between 0% and 20% and between 60% and 100% of mobile phase B. In the time between 20% and 60% there are two long peaks, which could be shortened with a steeper gradient. In Figure 15 the 2D result of a step gradient in the first dimension is shown. This chromatogram indeed shows shorter peaks in the time between 20% and 60% of mobile phase B. The peaks in the other areas are not better separated, but the first peak in the first dimension is shifted from 25 to 30 minutes. This is a result of the less steep gradient between 0% and 20% of mobile phase B. This shift is useful because it shows that it is possible to have more separation between the analytes with a neutral or positive charge and (almost) no retention. In the next sections both the normal and step gradient are therefore used to compare if a step gradient in combination with different reversed- phase gradients works better than a normal gradient combined with the same tested reversed-phase gradients. 0 25 50 75 100 125 150 175 200 88 75 63 50 38 25 13 0
First dimension retention (min)-Ion exchange
S e c o n d d im e n s i o n r e t e n t io n ( s ) R e v e r s e d P h a s e ( C 1 8 ) 0 1375000 2750000 4125000 5500000
Signal intensity in mAU
Figure 13: UV chromatogram (254 nm) of LC x LC (1st dimension
strong anion exchange and 2nd dimension reversed phase)
separation of megamix with mobile phase B in the 1st dimension
containing 50 mM ammonium sulphate). See section 3.3.2 for specification of the gradient.
Figure 14: UV chromatogram (254 nm) of LC x LC (1st dimension
strong anion exchange and 2nd dimension reversed phase)
separation of megamix with mobile phase B in the 1st dimension
containing 75 mM ammonium sulphate). See section 3.2.2 for specification of the gradient.
0 25 50 75 100 125 150 175 200 88 75 63 50 38 25 13 0
First dimension Retention (min)_Ion exchange
S e c o n d d im e n s i o n R e t e n t i o n ( s ) _ r e v e r s e d p h a s e ( C 1 8 ) 0 1375000 2750000 4125000 5500000
4.3.4. Normal gradient in anion-exchange and 40% B in reversed phase
As described in section 3.2.2 and section 4.2 there are two different reversed phase gradients used. In the first gradient, there is started with 0% of mobile phase B (95% ACN and 5% buffer) whether in the second gradient there is started with 40% of mobile phase B. The 2D result of this reversed phase gradient in combination with the ‘normal’ anion exchange gradient is shown in Figure 16. In this chromatogram the reversed-phase peaks (second dimension) come earlier, which fill up the loose space (0 – 40 seconds) which is present in the 2D chromatograms in which there is started with 0% of mobile phase B (Figure 13 and Figure 14). In this chromatogram, all peaks are more intense and broader for a still unknown reason. Due to this, it is not possible to accurately compare the degree of breakthrough effect in the gradient which starts at 0% and the gradient which starts at 40%.0 25 50 75 100 125 150 175 200 88 75 63 50 38 25 13 0
First Dimension retention (min) - Ion exchange
S e c o n d D im e n s io n r e t e t io n ( s ) R e v e r s e d P h a s e C 1 8 0 218750 437500 656250 875000 Signal Intensity in mAU
Figure 15: UV chromatogram (254 nm) of LC x LC (1st dimension strong anion exchange and 2nd dimension reversed
phase) separation of megamix with a step gradient in the 2nd dimension). See section 3.2.2 for specifications of the
gradient. 0 25 50 75 100 125 150 175 200 88 75 63 50 38 25 13 0
First Dimension retention (min) - Ion exchange
S e c o n d D im e n s io n r e t e t io n ( s ) R e v e r s e d P h a s e C 1 8 0 218750 437500 656250 Signal Intensity in mAU
Figure 16: UV chromatogram (254 nm) of LC x LC (1st dimension strong anion exchange and 2nd dimension reversed phase)
separation of megamix with the 2nd dimension gradient starting at 40% mobile phase B (95% ACN and 5% buffer)). For
4.3.5. Step gradient in anion-exchange and 40% B in reversed phase
The last combination tested, is a combination of the step-gradient in the first dimension and starting at 40% of mobile phase B in the second dimension. The result of this combination is shown in Figure
17. The first peak in the first dimension is again shifted to 30 minutes, and more separation between
the neutral and positive charged molecules (in t0) is obtained in reversed phase. The reversed phase
peaks are starting at 5 seconds instead of 40 seconds, so the separation in reversed phase is more spread out. Also the peaks between 20% and 60% of mobile phase B in the first dimension are shortened. This combination of gradients can be seen as an improvement in many prospects for the above named reasons. But also in this chromatogram the peaks are more intense and broader for an unknown reason, which results in the fact that this chromatogram is not good enough to accurately
compare to the chromatogram shown in Figure 12 (normal 1st dimension gradient and starting at 0%
B in the 2nd dimension). The peaks were too intense, which results in a ‘cut off’ of the peaks in
translating the signals into a 2D chromatogram. The peaks are therefore blunt instead of sharp, and in the 2D chromatogram this results in thick, broad and intense signals. A few options that can cause the bad condition of the peaks are excluded by testing in a one-dimensional setup. The reversed-phase column (Agilent ZORBAX C-18) was still in good shape and gave the same separation of megamix as earlier measurements. Moreover, the newly made megamix was tested and showed no differences with the former megamix. There were made new mobile phases in cleaned bottles. None of these variables had influence on the intensity and the broad of the peaks in the two dimensional chromatograms. In future research, it could be interesting to test this combination of gradients again when the shape of the peaks is recovered, because the combination looks promising.
0 25 50 75 100 125 150 175 200 88 75 63 50 38 25 13 0
First Dimension retention (min) - Ion exchange
S e c o n d D im e n s io n r e t e t io n ( s ) R e v e r s e d P h a s e C 1 8 0 218750 437500 656250 Signal Intensity in mAU
Figure 17: UV chromatogram (254 nm) of LC x LC (1st dimension strong anion exchange and 2nd dimension reversed phase)
separation of megamix with a step gradient in the 1st dimension and starting at 40% mobile phase B in the 2nd dimension). For
4.3.6. Optimized LC x LC separation
After testing different gradients in both the first and second dimension and making different combinations between them, the obtained chromatograms can be compared to the chromatogram
published by Bob Pirok et al.5 This chromatogram is shown in Figure 18. In Figure 19 the
chromatogram with the newly optimized gradient is shown. In this setup, the first dimension (anion exchange) is built up with the first, normal, gradient described in section 3.2.2. Mobile phase B of the first dimension contains 50 mM ammonium sulphate and the flushing gradient for re-equilibration is used (section 4.1). In the second dimension, the gradient starts with 0% of mobile phase B from the second dimension. The results that show the gradient that starts at 40% of mobile phase B were not trustworthy because of too intense and broad peaks, so there is chosen to show the results that have the gradient starting at 0% mobile phase B. Two different 2D chromatograms are shown with different matrices. The first chromatogram (Figure 19) shows the results with matrix 1. This figure clearly shows that it is possible to separate both natural and synthetic dyes in one sample. The chromatogram shows a good separation between neutral, mono-, di- and tri- negatively charged molecules. Besides that, there is more separation in the second dimension, reversed phase, within these differently charged groups of molecules. Because of an optimized flushing gradient, the column is re-equilibrated after each measurement, which results in more reproducible measurements. The reproducibility of the optimized gradient is shown in Figure 20. In this figure, two measurements using the same gradient and same sample are overlaid. The yellow/red signals show the first measurement and the blue surrounded signals show the second measurement. The signals of the neutral and positively charged molecules and the mono- and di- anionic molecules are perfectly reproducible. The signals of the tri-anionic molecules show small differences. The good reproducibility of the 2D measurements is a big improvement and is due to the optimized flushing gradient for the first dimension (strong anion exchange).
In Figure 19, the UV chromatogram of the LC x LC separation of megamix is shown with the optimized gradient. In the chromatogram, there are vertical blue (means low intensity) stripes visible. These stripes appear when the matrix is moderated in a way that as much peaks as possible are visible. When looked at the direct output of 2D signals in the OpenLAB CDS program, no differences are spotted at the time where the blue lines appear in the transformed chromatograms. In Figure 21, the UV chromatogram of the LC x LC separation of megamix is shown unedited. In this figure, the stripes are less present and when observed good, no peaks disappear, but only the intensity of the peaks is lower. The origin of the stripes is still unknown. The first possibility is that a few dyestuffs are present in the megamix with an excessively high concentration, but when a new megamix with
extra checked proportions was prepared, the stripes were still there. Another possibility is that the used mobile phases are contaminated, but after making new, clean mobile phases, the stripes were still there. The stripes are however no interruption for the separation and the peaks in the chromatogram are still visible because they have a much higher intensity than the stripes.
0 25 50 75 100 125 150 175 200 88 75 63 50 38 25 13 0
First dimension Retention (min)_Ion exchange
S e c o n d d im e n s i o n R e t e n t i o n ( s ) _ r e v e r s e d p h a s e ( C 1 8 ) 0 1375000 2750000 4125000 5500000 Signal Intensity in mAU
Figure 18: UV chromatogram (254 nm) of LC x LC (1st dimension anion-exchange without optimized flushing gradient and 2nd dimension reversed phase with TMA as ion-pair) of 54 synthetic dyes. Modulation time of 2 minutes and injection volume of 20 µL. (Source: Pirok , B.W.J.; Knip, J.; Bommel, van, M.R. & Schoenmakers, P.J. Characterization of synthetic dyes by
comprehensive two-dimensional liquid chromatograpy combining ion-exchange chromatography and fast ion-pair reversed-phase chromatography. J. Chromatogr. A, 1436, 144 (2016))
Figure 19: UV chromatogram (254 nm) of LC x LC (1st dimension strong anion exchange with optimized flushing
gradient and 2nd dimension reversed phase with TEA as ion-pair) separation of megamix (74 natural and synthetic
0 25 50 75 100 125 150 175 200 88 75 63 50 38 25 13 0
First Dimension retention (min) - Strong Anion-Exchange
S e c o n d D im e n si o n r e te n ti o n ( s) I o n -P a ir R e v e rs e d -P h a se C 1 8
Figure 21: UV chromatogram (254 nm) of LC x LC (1st dimension strong anion exchange with optimized flushing gradient and 2nd dimension reversed phase with TEA as ion-pair) separation of megamix (74 natural and synthetic dyestuffs). See section 3.2.2 for specifications of the gradients. The chromatogram is unedited (raw data).
Figure 20: Overlay of two different LC x LC UV chromatograms (254 nm) of the separation of Megamix (1st dimension strong anion
4.4. Future perspectives
As in all research, there are still many improvements possible. At first, the origin of the too intense and broad peaks has to be discovered. Secondly, the origin of the vertical stripes in the chromatograms has to be unmasked. When these two points are improved, it would be interesting to test the combination of the step-gradient in the first dimension and the second dimension starting at 40% mobile phase B again. When a chromatogram is obtained with less intense and less broad peaks, the combination of these gradients can be accurately compared to the one shown in Figure 21, which was the best combination tested in this project. The combination of a step gradient in the first-dimension and the second first-dimension starting at 40% mobile phase B looks promising because of the possibility of less breakthrough effect and less loose space in the second dimension. Furthermore, the long peaks in the first dimension are shortened when this combination is used.
Another very important subject for further research is the coupling of the mass spectrometer to the LC x LC. In this project, there was not enough time to cover this coupling. Nevertheless, it is always remembered when modifying the 2D method that it had to be compatible with the mass spectrometer one day. When the mass spectrometer is coupled to the LC x LC, all peaks could be identified and this would be a big reinforcement of the obtained data.
A step further in the separation of samples that contain both natural and synthetic dyes is the separation of degraded samples. In Figure 22 a degraded megamix (4 hour radiation from a 365 nm UV lamp) is separated in the best tested LC x LC method. This chromatogram shows a lot of separated peaks, but as in more chromatograms, these peaks are too broad and too intense. In further research a more optimized LCxLC method could be developed in which a degraded megamix can be fully separated. 0 25 50 75 100 125 150 175 200 88 75 63 50 38 25 13 0
First Dimension retention (min) - Ion exchange
S e c o n d D im e n s io n r e te t io n ( s ) - R e v e rs e d P h a s e C 1 8 0 218750 437500 656250 875000 Signal Intensity in mAU
Figure 22: UV chromatogram (254 nm) of LC x LC (1st dimension strong anion exchange and 2nd dimension reversed phase) separation of degraded megamix. See section 3.2.2 for specifications of the gradients.
5. Conclusion
During this project, a two dimensional LC x LC separation of a mix of 74 natural and synthetic dyes (megamix) was optimized. Firstly, both dimensions were optimized on a one-dimensional setup.
For the first dimension, strong anion exchange, the concentration of ammonium sulphate in mobile phase B was optimized at 50 mM for the used Agilent SAX column (B.048). Furthermore the re-equilibration of the column was optimized by creating a flushing gradient which was built up from 15 column-volumes of mobile phase C (1M NaCl in water) and 35 column-volumes of mobile phase A (60% water, 40% ACN and tris-buffer to pH=7). The flushing gradient was done after each measurement to be sure that the column was again in a good state and that the results were reproducible. Two different gradients were tested: a ‘normal’ gradually gradient and a step gradient. Both gradients showed reproducibility and were tested as functional in a two-dimensional setup. With a step-gradient the loose space and long peaks in the chromatogram were decreased. Because of too broad and too intense peaks in the 2D chromatogram in which the step gradient was used, it was difficult to accurately compare the normal and step gradient. But both gradients looked promising.
For the second dimension (ion-pair reversed phase) as well, two gradients were tested. The difference between the gradients was the starting percentage of mobile phase B (95% ACN and 5% buffer). In the first gradient, it started at 0% mobile phase B, while in the second gradient it started at 40% mobile phase B. Increasing the starting percentage of B resulted in the peaks coming earlier in the second dimension (and thus the peaks were more spread out) and probably led to less breakthrough effect. This cannot be said with complete certainty because of the too broad and too intense peaks in the 2D chromatogram.
The optimized dimensions with their different gradients were tested in several combinations in a two-dimensional setup. The 2D chromatograms showed that a step-gradient in the first dimension (strong anion exchange) shortened the long peaks in the zone of 20% to 60% of mobile phase B. Moreover, the first peak shifted from 25 min to 30 min which was favourable. For the second dimension (ion-pair reversed phase) starting at 40% of mobile phase B seemed to be of good influence on the chromatograms. The combination of the step-gradient in the first dimension and starting at 40% of mobile phase B in the second dimension therefore seemed like a promising LC x LC method. But in the end, the comparison between the optimized method and the method used by Pirok et al. was made with the combination of a normal gradient (with flushing gradient) in the first dimension and starting at 0% mobile phase B in the second dimension (for specifications see section
3.2.2).5 This was done because most other chromatograms contained too broad and too intense peaks
and also showed vertical stripes. The chromatogram that was compared showed normal peaks and was therefore more suitable for comparison. When the chromatograms were compared, it could be concluded that the results were reproducible, that there was more separation in the optimized method and that it was possible to separate both natural and synthetic in one sample.
The ultimate goal in this research is to develop a reproducible LC x LC – MS method, which can separate and identify the molecules in a (degraded) dyestuff mix sample at the same time. In this project, the coupling between the LC x LC and the MS is not yet accomplished, but steps were made in the direction to a trustworthy and reproducible method.
6. Acknowledgments
First of all, I would like to thank my daily supervisor Bob Pirok and his master students Sanne Berbers and Noor Abdulhussain. I learned a lot from all their insights and advises. Together they ensured that I critically thought through every step I took in this project. Secondly, I would like to thank Peter Schoenmakers for the chance to do this bachelor project in the Analytical Chemistry group at the University of Amsterdam. And last but not least I would like to thank my lab partners Nienke Meekel, Laura de Wal and Ole Hofman for useful brainstorm sessions and for making my time on the lab even more fun.
