UPTEC K 19033
Examensarbete 30 hp September 2019
Variation of chromatographic parameters in supercritical fluid chromatography of polar solutes
Victor Spelling
Teknisk- naturvetenskaplig fakultet UTH-enheten
Besöksadress:
Ångströmlaboratoriet Lägerhyddsvägen 1 Hus 4, Plan 0 Postadress:
Box 536 751 21 Uppsala Telefon:
018 – 471 30 03 Telefax:
018 – 471 30 00 Hemsida:
http://www.teknat.uu.se/student
Abstract
Variation of chromatographic parameters in
supercritical fluid chromatography of polar solutes
Victor Spelling
Supercritical fluid chromatography has seen a recent upswing and has the
potential to become a complimentary method to UPLC for routine analysis in the pharmaceutical industry. With the goal of mapping the effects of parameter variations in supercritical fluid chromatography, an investigational study was carried out. In this thesis the effects on retention, selectivity and efficiency for polar samples composed of readily available amino acids are examined, to see if and how analysis of polar solutes can be carried out. This is done by varying pressure and temperature as well as the choice of mobile phase modifier and additive on different columns. It was evident that supercritical fluid
chromatography can be applied to analysis of polar solutes. The effects on selectivity were the greatest through choice of mobile phase additive and overall retention time was mostly affected through choice of column and modifier.
ISSN: 1650-8297, UPTEC K 19033 Examinator: Christian Sköld Ämnesgranskare: Ulrika Yngve Handledare: Morgan Stefansson
1
LIST OF ABBREVIATIONS 2
POPULÄRVETENSKAPLIG SAMMANFATTNING 3
1. INTRODUCTION 4
2. THEORY 6
2.1 The SFC mobile phase 6
2.1.1 CO2 & organic modifiers 6
2.1.2 Mobile phase additives 6
2.1.3 Pressure, temperature & density 7
2.3 The SFC stationary phase 8
3. EXPERIMENTAL 9
3.1 Chemicals 9
3.2 Instrumentation 9
3.3 General procedures 9
3.3.1 Conditioning, start-up & stability 9
3.3.3 SFC chromatographic conditions 10
3.3.4 Sample preparation and storage 10
3.4 Method 14
3.4.1 Modifier solutions 14
3.4.2 Evaluation of peaks 14
4. RESULTS & DISCUSSION 15
4.1 Mobile phase additives 15
4.2 Columns 18
4.3 Modifier 19
4.4 Pressure and temperature 20
4.5 Additive concentration 21
4.6 Application on polar samples 23
5. CONCLUSION 26
BIBLIOGRAPHY 27
APPENDIX 30
2 LIST OF ABBREVIATIONS
2-EP 2-Ethylpyridine 2-PIC 2-Picolylamine 1-AA 1-Aminoanthracene
ABPR Automated back pressure regulator BEH Bridged ethylene hybrid
CSH Charged surface hybrid DEA Diethylamine
HILIC Hydrophilic interaction liquid chromatography HPLC High performance liquid chromatography Hyp L-4-Hydroxyproline
LAF Laminar air flow MeOH Methanol
MS Mass spectrometry NP Normal/Straight phase
NPLC Normal phase liquid chromatography PDA Photodiode array
RP Reversed phase
RPLC Reversed phase liquid chromatography RT Retention time
SFC Supercritical fluid chromatography SST System suitability testing
TEA Triethylamine TFA Trifluoroacetic acid
UPC2 Ultra high performance convergence chromatography UPLC Ultra high performance liquid chromatography USP U.S. Pharmacopeia
UV-Vis Ultra violet-visible
3
POPULÄRVETENSKAPLIG SAMMANFATTNING
Inom kemin används något som kallas för kromatografi för att separera, rena och detektera substanser (analyter) i olika typer av prover. För att åstadkomma detta utnyttjar man en kombination av analyternas inneboende kemiska egenskaper ihop med en väl utvald omgivning för dem att interagera med. Detta resulterar i att de rör sig olika långt och/eller snabbt och därmed uppstår en separation sinsemellan de olika analyterna i ett prov. Sättet detta utförs på varierar i instrumentell utrustning, val av det som kallas mobilfasen (gas, vätska etc.) och i vilken skala separationen utförs.
I det här arbetet har tekniken superkritisk vätskekromatografi (SFC) undersökts. Denna skiljer sig från de vanligt förekommande teknikerna gaskromatografi och vätskekromatografi.
Genom att utsätta koldioxid för högt tryck och hög temperatur försätter man den i ett så kallat superkritiskt tillstånd, i vilket det får en särskild kombination av egenskaper vad gäller exempelvis löslighet och transporthastighet. Gas- och vätskekromatografi är väletablerade tekniker medan SFC är en mindre vanligt förekommande metod för vilken nytt intresse uppstått – även fast den existerat under en lång tid. De senaste årens intresse för tekniken har att göra med dess effektiviseringspotential. Att använda koldioxid är bättre för miljön och för kemisten än traditionella lösningsmedel som används i vätskekromatografi. Dessutom är metoden snabb och billig samt mer skonsam mot molekyler som är för värmekänsliga för att analyseras i gaskromatografi, där höga temperaturer oftast förekommer. Därutöver så kan metoden användas väldigt effektivt i så kallad preparativ skala där man renar fram en substans efter syntes eller produktion. I och med att mobilfasen är koldioxid, så kan man enkelt avlägsna den direkt efter separationen genom att bara låta den dunsta bort. Detta sparar resurser i kontrast till tiden och energin som behöver läggas för att avdunsta ett, potentiellt, giftigt lösningsmedel i en annan kromatografimetod.
Fördelarna med koldioxid som mobilfas är många, men kommer också med en del nackdelar. Främsta nackdelen är att den tenderar att bara lösa upp fettlösliga substanser, medan polära molekyler inte vill lösa upp sig. Lyckligtvis är koldioxid i superkritiskt tillstånd blandbart med polära (vattenlösliga) lösningsmedel såsom metanol, och detta har utnyttjats i stor utsträckning för att kunna utföra analys och separation av polära analyter. Dessutom kan man tillsätta en sorts hjälpsubstanser till mobilfasen, utöver koldioxid och metanol. Dessa kallas för additiv och består ofta av vatten, syra/bas eller ett salt – ensamma eller i kombination. Med hjälp av additiv kan man utveckla bättre metoder och skräddarsy dem att passa just de analyter man för tillfället arbetar med. Hur allt detta fungerar i kombination är långt ifrån klarlagt och försvårar därför rutinmässig analys och separation med metoden att utföras i större utsträckning.
För att bidra till kartläggningen av metoden och hur den kan tillämpas har man i detta examensarbete reglerat inställningar, additivkoncentrationer, mängd metanol i mobilfasen etc.
Dessutom har olika höga tryck och temperaturer använts och även inkorporerandet av andra lösningsmedel än metanol har testats. Genom att undersöka inverkan av dessa variationer av instrumentella och kemiska parametrar kan man komma närmre en allmän förståelse för hur separationsmekanismerna i SFC fungerar. Dessutom hjälper arbetet till med att besvara frågan huruvida SFC är tillämpbart för analys på polära substanser, genom att hitta rätt förutsättningar för detta.
4 1. Introduction
In 1962, Klesper et al. [1] were the first to document their use of supercritical fluids as a mobile phase in chromatography. Their goal was a chromatographic mode capable of eluting heavy, heat sensitive molecules without risk of degradation, and this was accomplished by employing high pressure and temperature, beyond the critical point of the mobile phase. The technique became a complimentary method to gas chromatography and came to be known as supercritical fluid chromatography (SFC). A common statement regarding supercritical fluids is that they possess density and solvating properties resembling liquids, whilst having diffusivity and viscosity similar to that of gases [2,3]. The viscosity and diffusivity provides great separation efficiency while the density gives fast transportation times and thus yielding quicker over all analysis times when employed in chromatography [4].
Today the principal mobile phase used is carbon dioxide, albeit other gases e.g. NH3, N2O and lighter hydrocarbons can be used too, and were initially utilized [2]. Carbon dioxide has the advantage of being safe to use and better for the environment than most organic solvents commonly used in liquid chromatography. It has a solvation power similar to that of heavier organic solvents such as pentane [5], hexane or heptane [4], and it has a low critical point (Tc = 304.13 K , Pc = 73.77 bar) [2]. Similarities to the organic solvents mentioned derive from its non-existing dipole moment [6]. These overall features make the method a promising addition to analytical chemistry in pharmaceutical development. With the mobile phase consisting of carbon dioxide, SFC functions similarly to normal phase liquid chromatography and therefore it requires tweaking to be able to separate many pharmaceuticals, as they tend to be polar and contain basic functional groups [4].
Thus far, SFC has been routinely used for chiral separations because of its fast column equilibration and method development, smaller pressure drops across the column together with the benefits mentioned earlier [7]. It is also widely employed in preparative scale separation as the mobile phase is easily evaporated, which facilitates product recovery, increases overall yield and reduces the risk of hazardous residue. Recent years have seen an increased demand for analytical scale achiral separations carried out in SFC [4]. Working with the elution strength of pure CO2, SFC was for a long time only used for analysis of non-polar solutes. However, as CO2 is miscible with a range of organic solvents of different polarities, it is also possible to elute polar solutes [2,6]. The incorporation of organic modifiers, mainly methanol, into the mobile phase has been common for some time now, and has proven to greatly improve peak shape and column efficiency when separating more polar solutes [2–4,8,9]. With more recent methods incorporating small to moderate amounts of water into the mobile phase as well, the polarity has been further increased and progress is being made towards routine use of SFC for achiral pharmaceutical analysis [2–4,6,8–14]. In 2013, Liu et al. [11] demonstrated that a water content of up to 10 % in methanol modifier gradually improved peak shapes, until ultimately causing too much noise in UV detection at higher proportions. Their findings and hopeful attitudes are interesting when compared to those of Camel et al. [15] in 1991, who despite having successfully resolved samples of multiple amino acids, concluded that SFC was incapable of competing with LC for separating polar solutes at the time. Today one SFC instrument alone can cover the entire separation spectrum that LC comprises with NPLC, RPLC, HILIC, ion exchange and ion pair chromatography [2].
An important note on SFC is its name and how it often differs from what physical state the mobile phase is actually in. Even when employing only low proportions of organic modifier in the mobile phase the critical temperature is greatly increased, and the mobile phase is rather in a so-called subcritical state [2,6]. Luckily, this does not present an issue as subcritical conditions can be maintained if operations are carried out above the critical pressure. It is, however, important to keep in mind when working with SFC as there is a variation in terminology between scientists and journals, as well as over time. The manufacturers of the instrumentation
5
used in this thesis, Waters corporation, use the name convergence chromatography, based on statement from Giddings [16], therefore they call it Ultra-high Performance Convergence Chromatography (UPC2).
It is apparent that the instrumentation has come a long way and that the previously recognized drawbacks concerning stability and robustness have since been dealt with. And the columns available now are making SFC a more viable technique with good selectivity combined with the sensitivity and robustness now living up to GMP standards [17]. Thus, SFC’s potential as a complementary or alternative method to HPLC/UHPLC in pharmaceutical development and analysis is being acknowledged [2–4,13,18]. More robust and repeatable experiments can be conducted today, thanks to progress in pumping and pressure regulation as well as detection hardware.
Discussions regarding what parameters affect the retention, efficiency and selectivity the most in SFC are ongoing; also, whether the technique is applicable to analysis of polar solutes.
Therefore, this thesis aims to investigate a variation of instrumental and physicochemical parameters, to further answer these questions. This will be done by screening a selection of mobile and stationary phases with a variation of additives and comparing the results.
Experiments across a variation of temperatures, pressures, flow rates and gradients will also be carried out. By using readily available samples consisting of amino acids, that are known to be very polar and often carry charged groups, these parameters will be readily screened. The results will be evaluated in terms of efficiency and separation of samples. The findings from the parameters screened will then be utilized to try and achieve separation of molecules known to be even more polar (e.g. carbohydrates and peptides), in an optimized system.
6 2. Theory
2.1 The SFC mobile phase 2.1.1 CO2 & organic modifiers
CO2 with its inertness to most solutes, its overall abundance and its critical point within reach is today the principle mobile phase in all SFC applications. Its miscibility with most common organic solvents provides great variation and optimization opportunities for different types of solutes. It is a Lewis base as its oxygen atoms are electron pair acceptors, and it strongly interacts with other molecules in the mobile phase and with samples because of this. It also hydrogen bonds to alcohols and in small quantities form quaternary aggregates with e.g.
methanol [6]. Upon introduction of protic modifiers, such as methanol, there is a formation of an alkylcarbonic acid, just as carbonic acid forms when mixing with water [5].
Besides affecting the density and polarity of the mobile phase, and thereby the solvating power, the addition of a modifier can also affect the stationary phase through adsorption. This can result in blocking of active sites on the stationary phase, altering the characteristics of the stationary phase and/or increasing the net volume of the stationary phase which results in an increased phase ratio [2,6]. The modifier molecules can also compete with solute molecules for the hydrogen bonding sites of the stationary phase [4]. The most notable effect of the modifier is the apparent pH of a CO2:MeOH mobile phase being around the corresponding aqueous pH 5. This is a result of methylcarbonic acid formation. Upon increasing the amount of modifier, the apparent pH will decrease further, according to a recent publication by C. West et al. [5], where modifier proportions of 0–50 % methanol were tested. Being aware of the apparent pH of the mobile phase is important when discussing retention mechanisms for samples with different pKa values and the effects an acidic additive has on the stationary phase.
Even though methanol is by far the most common modifier used in SFC, others such as acetonitrile, ethanol and 2-propanol are usually implemented too. Methanol is preferred over the other alcohols as it is more polar, less dense and more compatible with mass detection [3,4,9]. However, ethanol is often mentioned as a safer, greener option and is used in a moderate amount by itself or in combination with methanol. Acetonitrile as a non-protic solvent is less compatible with SFC as it is less prone to adsorption onto the stationary phase which results in poorer peak shapes for hydrogen-bonding solutes [6]. Lone silanol groups causing “parasitic”
interactions with the solutes stems from their scarce occurrence when the silica is derivatized, resulting in decreased peak capacity. With this considered, acetonitrile can be used in combination with methanol to achieve selectivity tweaks without sacrificing the adsorption and protonation qualities sought from the modifier [3,4,6].
2.1.2 Mobile phase additives
The addition of additives to the mobile phase provides even more versatility to the technique and what solutes it can be applied to. By adding small amounts of acids, bases or salts, even the most polar of solutes are able to elute with satisfactory peak efficiency [2,3,6]; which has played a big role in the upswing in application on, usually polar, pharmaceutical compounds. The role of the additive is partly an elongation of the modifier effects of adsorption onto the stationary phase [2–4,6,19]; and furthermore, for neutral and acidic additives, a decrease in apparent pH [5]. West et al. stated that by employing modifiers containing water and/or acidic additives, higher proportions would decrease the apparent pH to below aqueous pH 2. This results from the inclusion of water, which will cause the formation of carbonic acid that is even more acidic than methylcarbonic acid. Water as an additive results in an increased polarity and solvation
7
power of the mobile phase and an overall decrease in retention times for polar solutes [9], as a result of its hydrogen bonding abilities. It has been utilized in proportions up to 10 % in methanol before causing too much noise, probably due to phase separation from immiscibility [11]. West et al. also stated that while the apparent pH will decrease with an increased portion of water and/or acidic additives, it will not increase with an increased concentration of basic additives. They also stated that the effect on mobile phase polarity from additives is insignificant “at usual concentration ranges (0.1 % or 20 mM)”. Besides affecting pH, additives can also function as ion pairing agents, ion-suppressors and as displacers in an ion-exchange mechanism [6]. What mechanisms act on retention in the presence of additives, and to what extent, are still uncertain and further studies are currently in the works.
2.1.3 Pressure, temperature & density
In the early days of SFC, the retention mechanism was tied mainly to pressure and temperature adjustments, as they resulted in density changes that affected solute solubility in the mobile phase [7]. However, it has since been shown that density changes are not as significant when working with polar solutes and subsequently employing higher modifier proportions and gradient elution. An article by T. A. Berger [20], presented the difference in density change at different modifier proportions. At 15 % methanol, the density increased from 0.75 g/cm3 to 0.94 g/cm3 upon increasing system pressure from 100 to 400 bar. While at 5 % methanol, the density changed from 0.68 g/cm3 to 0.95 g/cm3 with the same increase in pressure. At ≥20 % methanol, the density decreased with increasing pressure, while the viscosity of the mobile phase kept increasing with modifier concentration. At higher modifier proportions, the conditions are further away from supercritical ones and subsequently lead to the decreased density effect upon increasing the pressure as the mobile phase will behave more like a liquid.
When working in actual supercritical conditions, there are multiple gradients taking place during an isocratic SFC run [6]. The pressure is usually lower at the column outlet, leading to fluid density, and subsequently eluotropic strength, decreasing over the column. This pressure gradient also causes a longitudinal and radial temperature gradient which all together creates a very complicated discussion. However, when employing higher proportions of modifier and thereby working in subcritical conditions, these factors present less of an issue due to the more predictable effects of density and temperature on retention.
Upon changing the ABPR pressure for comparison purposes, but still maintaining maximum system pressure at the end of a gradient, a new flow is employed. To accommodate for change in flow rate a new gradient time can be calculated (Eq. 1). The retention times of the same samples can then be compared between gradient systems by calculating an apparent retention factor for the gradient, normalized to the flow of the first method (Eq. 2). Column dead time is determined from the front of each individual chromatogram, and system dwell volume for waters UPC2 is given as 440 µL.
!",$ = ''(
) × !",+ (1),
,-.. =(01×'))3(4
567)8309×'))
09×') (2).
8 2.3 The SFC stationary phase
The availability of stationary phase columns specifically for pharmaceutical analysis of polar solutes has seen an increase over recent years and they are being made in dimensions for high- throughput analyses [2,6,9,17]. Available columns for SFC can be classified according to the empirical linear solvation energy relationship model (LSER). Recently it has been done by performing hierarchical cluster analysis on retention data from 109 solutes on 31 different columns [17] on a LSER model with 7 descriptor terms, with two new descriptors taking solute ionization into account. This resulted in 6 clusters of SFC applicable columns, where many of those used in this project, see Table 1, landed in the same group. The 2-EP, 1-AA, 2-PIC, DEA and DIOL columns were part of the study and they exhibited retention mechanisms dominated by hydrogen bonding, especially with acidic solutes, and to an extent also displayed polar interactions. Stationary phase ligands comprised of penta-hydroxyl and a zwitterion were categorized into this group as well, albeit the columns in the study were not the same ones as the corresponding columns used in this work. Bare silica phases exhibited similar patterns in retention mechanisms but not as strongly with acidic species. Instead, stronger interactions with cationic solutes could be observed. The pentafluorophenyl column appeared to interact well with aromatic solutes and, despite carrying a positively charged surface, also cationic solutes.
Octadecyl ligands had almost no interaction with polar solutes at all. A very important note about these findings is that they were produced with an absence of additives and a low proportion of methanol in the mobile phase. This information is to be used as guidelines for understanding, but variations from it is to be expected.
The effects of the mobile phase, modifier and additive, have been discussed earlier to an extent; but besides adsorption to the stationary phase, the influence of apparent pH on the stationary phase functional groups is important. If higher proportions of methanol comes with a decrease in apparent pH, the surface silanol groups may exhibit less deprotonation while amino groups on the ligands show the contrary [6]. The change in stationary phase chemistry caused by adsorption and apparent pH of the mobile phase needs to be considered and be discussed on top of the findings in the LSER model.
9 3. Experimental
3.1 Chemicals
From Sigma-Aldrich (Saint Louis, MO, USA) 21 L-amino acids + glycine, including L-4- hydroxyproline, (≥ 99 %) along with peptides tetra-L-lysine (≥95%), tetra-L-aspartic acid (≥97%), di-L-aspartic acid, hexaglycine, penta-L-phenylalanine acetate salt (≥97%), poly-D- lysine hydrobromide and poly-L-proline. Glycylglycine (99%) was purchased from Fluorochem Ltd (Hadfield, UK). Glycylglycine amide hydrochloride was purchased from BACHEM (Bubendorf, Switzerland).
Reagent grade trifluoroacetic acid (≥99%), methanesulfonic acid (≥99%), ethanesulfonic acid (70%) in water, 3-amino-1-propanesulfonic acid (97%), (1R)-(-)-10-Campohorsulfonic acid ammonium salt, diethylamine (≥99.5%), triethylamine puriss p.a. grade (≥99.5%), hexylamine (99 ), 2-Hydroxypropyl-β-cyclodextrin (average Mw ~1380 g/mol), crown ether 18- crown-6 of purum grade (≥99%) and crown ether dicyclohexano-18-crown-6 (98%) were purchased from Sigma-Aldrich as well. The sugars 1-Amino-1-deoxy-β-D-glucose, 6-amino- 6-deoxy-d-glucose (95%), L-fucosylamine (≥98% TLC grade), Phenyl-β-D- thioglucopyranoside (HPLC grade ≥98%), Phenyl-β-D-glucopyranoside (97%), D- Mannosamine hydrochloride and D-(+)-Galactosamine hydrochloride (≥99%) were also from Sigma-Aldrich.
Gradient grade methanol, acetonitrile and 2-propanol, all ≥ 99.9 %, were purchased from Merck (Darmstadt, Germany), alongside ISO analytical grade ammonia solution 25 % in water.
95 % ethanol was purchased from CCS Healthcare AB (Borlänge, Sweden).
3.2 Instrumentation
Ultrasonic baths used were Transsonic T890 and Elmasonic S 30 H from Elma Schmidbauer GmbH (Singen, Germany). Weighing of substances was performed on a XP6 Excellence Plus XP Micro Balance from Mettler Toledo AB (Stockholm, Sweden), inside a Holten Safe 2010 class II model LAF-cabinet from Thermo Scientific (Waltham, MA, USA).
ACQUITY UPC2 system from Waters Corporation (Milford, MA, USA), consisting of Acuity UPC2 binary solvent manager, column manager, convergence manager, photodiode array detector, QDa detector and sample manager. The system has a dwell time of 440 µL.
Chromatograms were recorded with Empower® 3 software. Packed columns used are presented in Table 1.
Table 1 Packed columns used in this study.
Column name Manufacturer Particles Bonded ligand Dimensions (mm) Particle size (µm)
ACQUITY UPC2 BEH Waters FPP Bare hybrid silica 100 x 3.0 1.7
ACQUITY UPC2 BEH 2-EP Waters FPP 2-Ethylpyridine 100 x 3.0 1.7
ACQUITY UPC2 CSH Fluorophenyl Waters FPP Pentafluorophenyl 100 x 3.0 1.7
ACQUITY UPC2 Torus 1-AA Waters FPP 1-aminoanthracene 100 x 3.0 1.7
ACQUITY UPC2 Torus 2-PIC Waters FPP 2-picolylamine 100 x 3.0 1.7
ACQUITY UPC2 Torus DEA Waters FPP Diethylamine 100 x 3.0 1.7
ACQUITY UPC2 Torus DIOL Waters FPP Propanediol 100 x 3.0 1.7
ACQUITY UPLC BEH C18 Waters FPP Octadecyl 100 x 2.1 1.8
HALO-2 Penta-HILIC Advanced materials tech. SPP Penta-hydroxyl 100 x 3.0 2.0
HALO-2 HILIC Advanced materials tech. SPP Bare silica 100 x 3.0 2.0
NUCLEODUR HILIC Macherey-Nagel SPP Ammonium – sulfonic acid 100 x 3.0 1.8
NUCLEOSHELL Biphenyl Macherey-Nagel SPP Biphenylpropyl 100 x 2.0 2.7
FPP stands for fully porous particles; SPP stands for superficially porous particles.
3.3 General procedures
3.3.1 Conditioning, start-up & stability
Upon implementing columns that had not recently been in use, or which were employed for the first time, were washed for 30 minutes with methanol (50/50 CO2/Methanol v/v, 1.0 mL/min, 40 °C column, 110 bar ABPR pressure). Thereafter, columns were equilibrated with the new mobile phase for another 30 minutes with temperature and pressure conditions in accordance to the forthcoming method. Then, 0.5 µL blank samples, consisting of 50 % MeOH in water
10
(same as the sample diluent) were injected five times in succession onto the columns. A readily available SST solution consisting of 1.0 mg/mL of metoprolol, caffeine, acetophenone, 3- hydroxydiphenylamine and 4’-(1-imidazol-1-yl)acetophenone respectively, in methanol was used. The SST solution was injected twice in succession before the implementation of brand new methods and reinjected regularly to ensure system stability. SST operating conditions are presented in Table 2. The modifier was pure methanol. For the duration of the project the wash solutions consisted of 100 % methanol as seal wash, 80/20 n-heptane/2-propanol v/v as weak needle wash and 2.5 % formic acid in water/acetonitrile/2-propanol/methanol (25/25/25/25 v/v/v/v) as strong needle wash.
When maintaining the same column and mobile phase but implementing changes in the method parameters, the washing steps were not employed, and equilibration was only run for 15 minutes. This step was still followed by five blank sample injections for conditioning of the column.
Table 2 SST operating conditions performed with a linear gradient slope.
Parameter Setting
Supercritical fluid chromatography:
Gradient Time (min) % CO2 % Organic modifier
0 95 5
1.00 95 5
6.00 60 40
6.50 60 40
6.80 95 5
8.00 95 5
Flow rate (mL/min) 1.500
ABPR pressure (bar) 103
Injection volume (µL) 1.0
Column temperature (°C) 40
Sample temperature (°C) 10
PDA detection wavelength (nm) 215
Resolution (nm) 4.8
Sampling rate (points/s) 40
Weak wash volume (µL) 600
Strong wash volume (µL) 200
Seal wash time (min) 1.0
3.3.3 SFC chromatographic conditions
Unless stated otherwise, all methods used for acquiring data followed the standard operating procedure presented in Table 3, with the samples always being injected twice and the gradient slope always being linear. Different compositions of organic modifiers were employed throughout, most having methanol as an organic modifier containing 5 vol% water in combination with one or two other additives. Other modifiers and additives used are presented continuously.
3.3.4 Sample preparation and storage
The initial stock solutions for all amino acid samples (including glycine and 4-hydroxyproline) were prepared at a concentration of 10 mg/mL in 50 % methanol in water containing 100 mM TFA. Both water and TFA were utilized to facilitate dissolution. These samples were then diluted in vials with 50 % methanol to concentrations of 1 mg/mL and injected onto a diol column under standard operating conditions. From the acquired absorption spectra, new stock
11
solutions were prepared with their concentrations standardized to give approximately the same absorption peak area as each other. This was done to aid analysis of spectra by minimizing peak overlap. The amino acid sample concentrations prepared are summarized in table 4, alongside their dissociation data. Structures are presented as Appendix 1. The second batch of stock solutions contained no TFA buffer at all as it seemed to accelerate degradation and was unnecessary for dissolution. All stock solutions were kept in freezer at -18 °C to prevent degradation over time and prepared sample vials were kept at 10 °C in the sample manager for the same reason. Samples of carbohydrates and oligo-/poly peptides were prepared in 50 % methanol in water, in concentrations presented in Table 5 and Table 6 respectively.
Sometimes the amino acids were combined into 7 groups, based on a multivariate study [21]. This was done for time saving reasons and when a more generic trend in retention behavior was sought, deeming a screen of all 21 samples unnecessary. The groups were the hydrophobic (Ala, Ile, Leu, Met, Pro, Val), the alcohols (Hyp, Ser, Thr), the basic (Arg, His, Lys), the acidic and amides (Asn, Asp, Gln, Glu) and finally two groups of their own for cysteine and glycine.
Table 3 Standard operating conditions for SFC runs throughout this study with a linear gradient slope.
Parameter Setting
Supercritical fluid chromatography:
Gradient Time (min) % CO2 % Organic modifier
0 85 15
4.00 40 60
4.01 85 15
6.00 85 15
Flow rate (mL/min) 1.200
ABPR pressure (bar) 125
Injection volume (µL) 0.5
Column temperature (°C) 40
Sample temperature (°C) 10
PDA detection wavelength (nm) 215
Resolution (nm) 4.8
3D spectrum range (nm) 192 – 400
Resolution (nm) 1.2
Sampling rate (points/s) 40
Weak wash volume (µL) 600
Strong wash volume (µL) 200
Seal wash time (min) 1.0
12
Table 4 Amino acid sample concentrations at the initial screena (in 50 % methanol and 100 mM TFA) and after standardizing absorption peak areab (in 50 % methanol without additive/buffer). pKa values for backbone carboxylic group, amine group and side chain for the respective amino acids are presented as well as the pH values at their isoelectric points [22].
Amino acid Initial conc.a (mg/mL) 2nd batch conc.b (mM) pKa pKb pKx pI
Ala 10 45.0 2.34 9.69 – 6.00
Arg 10 18.2 2.17 9.04 12.48 10.76
Asn 10 15.2 2.02 8.80 – 5.41
Asp 10 39.3 1.88 9.60 3.65 2.77
Cys 10 23.4 1.96 10.28 8.18 5.07
Glu 10 61.7 2.19 9.67 4.25 3.22
Gln 10 26.5 2.17 9.13 – 5.65
Gly 10 60.5 2.34 9.60 – 5.97
His 10 0.50 1.82 9.17 6.00 7.59
Hyp 10 52.6 1.82 9.65 – –
Ile 10 31.4 2.36 9.60 – 6.02
Leu 10 36.2 2.36 9.60 – 5.98
Lys 10 30.7 2.18 8.95 10.53 9.74
Met 10 2.0 2.28 9.21 – 5.74
Phe 10 0.5 1.83 9.13 – 5.48
Pro 10 56.3 1.99 10.60 – 6.30
Ser 10 39.9 2.21 9.15 – 5.68
Thr 10 40.2 2.09 9.10 – 5.60
Trp 10 0.1 2.83 9.39 – 5.89
Tyr 10 0.4 2.20 9.11 10.07 5.66
Val 10 0.5 2.32 9.62 – 5.96
13
Table 5 Structures of the carbohydrates studied with their respective sample concentrations.
Sample Structure Conc. (mM)
D-(+)-glucosamine 1.3
6-amino-6-deoxy-D-glucose 1.3
L-fucosylamine 1.3
D-(+)-galactosamine 1.5
D-mannosamine 1.3
Phenyl β-D-glucopyranose 1.5
Phenyl β-D-thioglucopyranose 1.5
Table 6 Oligo- and poly-peptides studied and their respective sample concentrations.
Sample Conc (mM)
(Asp)2 9.7
(Asp)4 5.3
(Gly)2 19.4
(Gly)6 10.2
Glycylglycine amide 19.8
(Lys)4 11.8
Poly-Lys 1.0
(Phe)5 1.0
14 3.4 Method
3.4.1 Modifier solutions
All organic mobile phase modifiers were prepared to contain 5 vol% water as an additive in combination with another additive, unless stated otherwise. After mixing they were degassed in a sonic bath for a minimum of 20 minutes. Upon swapping or employing entirely new modifiers the solvent lines were primed for 5 minutes at a flow rate of 4.0 mL/min to ensure air and previously used modifiers were washed out. Modifiers not in use were sealed and kept for future application. All modifiers used are presented in Table 7. With these modifiers and the prepared samples readily available, system parameters and conditions were varied and screened to examine how the adjustments affected retention time and mechanisms.
Table 7 All modifiers implemented throughout the study, listed with their respective additive content and concentration. All modifiers contained 5 vol% water.
Modifier Additive Conc. (mM) Additive pKa
Methanol Trifluoroacetic acid 10 0.3
Methanol Methanesulfonic acid 10 -1.9
Methanol 3-amino-propanesulfonic acid 10 -
Methanol Ammonium camphorsulfonate 10 -
Methanol Diethylamine 10 11.1
Methanol Triethylamine 10 10.8
Methanol Hexylamine 10 10.6
Methanol Ethanesulfonic acid 2; 5; 10; 20; 50 -1.7
Methanol Ammonia 5; 10; 20; 50; 75; 100 9.3
Methanol Trifluoroacetic acid
18-crown-6 5
1.0 0.3
-
Methanol Trifluoroacetic acid
Ammonia 5
5 0.3
9.3
Methanol Ethanesulfonic acid,
18-crown-6
10 10
-1.7 -
Methanol Ethanesulfonic acid,
Dihexano-18-crown-6 10
10 -1.7
-
Methanol Ethanesulfonic acid
2-Hydroxypropyl-β-cyclodextrin 10
1.0 -1.7
-
Methanol/Acetonitrile 50/50 Ethanesulfonic acid 10 -1.7
3.4.2 Evaluation of peaks
In isocratic elution mode, peak symmetry can be evaluated (USP tailing factor, Eq 3), where W is the peak width at 5 % of the peak height and F is the time from width start point at 5 % peak height to the retention time. A perfectly symmetrical peak has a T = 1.00, while T < 1.00 represents a fronting peak and T > 1.00 a tailing peak. The efficiency can be also be determined (USP plate count, Eq. 4), where RT is the peak retention time and W is the peak width at baseline determined by tangents drawn to 61 % of the peak height. In gradient mode, calculations made with these equations are not true. However, they can be used as guidelines when comparing columns that displayed similar elution times when employing the same gradient and chromatographic conditions.
: = < ×'; (3),
= = 16(@;A)< (4).
15 4. Results & discussion
4.1 Mobile phase additives
Resulting retention times when employing different additives and injecting amino acid samples on a diol column are summarized in Figure 1. 3-Amino-propanesulfonic acid and ammonium camphorsulfonate, listed in Table 7, were also tested but they produced mostly noise and the data was disregarded.
For the basic additive ammonia, all solutes increased their retention, acidic amino acids showed a pronounced increase in retention and basic amino acids were not eluted. This is probably due to a dissociation of the carboxylic acid group in the common “amino acid part”
of the solutes, hence increasing the polarity of the molecules. The effect is even more pronounced for the acidic glutamic and aspartic acid solutes with two carboxylic groups. The lower effect on retention, for the non-acidic analytes, from the dissociation of the amino acid carboxylic group might be due to shielding or ion-pair competition from the proximate amino group resulting in a dipole or inner salt formation. The polar ammonium ion is furthermore a weak counter ion for neutralization (ion-pairing) of a negatively charged carboxylate group and thus its employment results in a high retention. Also, ammonium ions could be obstructed from ion-pairing with the basic amino acids caused by the repulsion from charged amino groups.
Consequently, a marked decrease in retention is observed for acidic amino acids (glu and asp) when utilizing the more hydrophobic alkylamine bases enhancing ion-pair formation and distribution to the less polar mobile phase. This also explains why arginine, histidine and lysine are not eluted. The overall trend of higher retention when employing basic additives would point to the samples being charged and thereby more retained by the polar stationary phase.
Likewise, an ion-pairing mechanism for acids (see TFA, methane and ethanesulfonic) with positively charged solutes would be responsible for elution of the samples. Interestingly, some selectivity changes are observed in these acidic mobile phases (samples assumed positively charged) between the carboxylic (TFA) and sulfonic acids (methyl, ethyl), respectively. This is possibly be due to differences in additional hydrogen bonding between the two species (charged solute and counter ion) as constants for ion-pair formation differ for different complexes.
Figure 1 Retention times of 21 amino acid samples injected onto a diol column plotted against the additives used in the modifier. Runs were performed under standard operating conditions and all modifiers contained 10 mM of the respective additive. Dashed lines represent no elution was observed at their end-point. Maximum amount of modifier after 4 minutes.
2 2,5 3 3,5 4 4,5
TFA
Methanesulfonic acid
Ethanesulfonic acid
Ammonia DEA TEA
Hexylamine
RT(min)
AlaArg AsnAsp CysGln GluGly HisHyp IleLeu LysMet PhePro SerThr TrpTyr Val
16
Figure 2 UV chromatograms from runs with the different additives a) ethanesulfonic acid and b) methanesulfonic acid, injected onto a diol column. The seven peaks represent 1. leucine, 2. Phenylalanine, 3. Cysteine, 4. hydroxyproline, 5. glycine, 6.
glutamic acid and 7. histidine. Lower peaks in cyan, maroon, navy and green are impurities and/or results of degradation or dimerization. The two chromatograms are in the same scale along both axes.
The apparent peak efficiency was better for sulfonate additives compared to basic additives and TFA, see Figure 2 and Figure 3 (all of these chromatograms are presented in the same scale). The efficiency was similar for the two sulfonic acids. TFA is a commonly used additive in SFC and despite this its implementation comes with highly asymmetrical peaks for amino acids of all side chain variations. The so-called parasitic interactions could be a reason for the poor peak shapes if the TFA is too weak an acid to ensure protonation of the silanol groups.
Figure 3 UV chromatograms from runs with the different additives a) TFA, b) ammonia, c) DEA, d) TEA and e) hexylamine, injected onto a diol column. The seven peaks represent 1. leucine, 2. phenylalanine, 3. hydroxyproline, 4. cysteine, 5. glycine, 6. glutamic acid and 7. histidine (only visible in figure 3a). Lower peaks in maroon are impurities and/or results of degradation or dimerization. All chromatograms are in the same scale along both axes.
a) b)
a) b) c) d)
e)
1 2 3 4 5 6 7
1 2 3 4 5 6 7
1 2 3 4 5 6
1 2 3 4 5 6 1 2 3 4 5 6
1 2 3 5 6
4
1 2 4 3 5 6 7
17
Figure 4 The difference in retention time when implementing 10 mM of two different crown ethers in the modifier are visualized and compared for 21 amino acid samples injected onto a diol column. All 3 modifiers contained 10 mM of ethanesulfonic acid.
Crown ethers have an affinity for forming complexes with protonated primary amines as their insides contain several ether oxygens that act as hydrogen bond acceptors. The result of their implementation as an additive when injecting amino acid samples, is presented in Figure 4. Ethanesulfonic acid was chosen as the additive to continue implementing in the study and was thereby employed alongside the crown ethers. Dihexano-18-crown-6 is a bulkier derivative of the other crown ether and it is more hydrophobic as well.
Lysine, having two accessible primary amines, was the most affected by crown ether addition. The same was true for glycine, which has no side chain at all. All sample molecules have at least one protonated primary amine under these conditions and subsequently exhibit a decreased retention. While proline and hydroxyproline, containing a secondary amine, were not affected by this addition to the modifier. Dihexano-18-crown-6 appeared less prone to incorporate bulkier/branched sample molecules and subsequently they are more retained when implementing this crown ether than with regular 18-crown-6. The effects observed are not exclusive to SFC but more so to the chemistry of crown ethers in general. It is possible to utilize crown ethers for separation of samples that are similarly retained without them, e.g glycine and threonine that go from co-elution to complete separation, or just to decrease retention of primary amines in general.
The implementation of 1.0 mM 2-hydroxypropyl- β-cyclodextrin as an alternative complexing agent was tested as well. This additive appeared to precipitate, even at high methanol proportions, and was disregarded as to not put the instrumentation at risk.
2 2,2 2,4 2,6 2,8 3 3,2 3,4 3,6
No crown ether 18-crown-6 dihexano-18-crown-6
RT(min)
AlaArg AsnAsp CysGln GluGly HisHyp IleLeu LysMet PhePro SerThr TrpTyr Val
18 4.2 Columns
Figure 5 Retention times of 21 amino acid samples injected onto 10 different columns and eluted under standard operating conditions with 10 mM ethanesulfonic acid in the modifier. Dashed lines represent no elution was observed at their end- point. Larger version of this diagram available in Appendix 1.
The retention times for all 21 amino acids on 10 different columns are presented in Figure 1, and column information is presented in table 1. All runs were performed under standard operating conditions with 10 mM ethanesulfonic in the modifier. No peaks were observed for the C18 and biphenyl columns as the samples appeared not to be retained and were eluted in the front.
The pentafluorophenyl column, having a low concentration of positive charges on the particle surface, exhibited lower retention than all other columns. The repelling effect of the stationary phase on positively charged solutes causes a higher distribution to the mobile phase and facilitates elution. The hydrophilic Waters columns 2-EP, DEA, AA and 2-PIC were very similar regarding retention and selectivity. The peak separation was similar as well, with multiple co-eluting peaks. To evaluate them, their respective apparent asymmetry factor and efficiency was calculated for peaks of each sample group and the average results are presented in Table 8. DEA presented poorer symmetry scores but had a higher efficiency than the 2-EP column, which was the second best. These calculated asymmetry values are only used for comparison between columns that have similar retention, as elution was done in gradient mode.
Therefore, the diol column was not included in this comparison as the samples all eluted much later. DEA was chosen to later use for trying to separate all 21 amino acids as it had the best efficiency.
Table 8 The average asymmetry and efficiency calculated from UV chromatogram peaks of seven different amino acid samples (arg, asp, cys, gly, leu, phe, ser) representing their respective groups. The asymmetry row represents average percental deviation from 1.00 for the seven peaks.
Column1 2-EP DEA 2-PIC AA
Mean asymmetry (%) 8 11 15 12
Mean theoretical plates 80 900 95 400 85 000 79 000
The bare silica columns BEH and HALO HILIC had higher retention for all samples but especially for the basic amino acids and for the secondary amines hydroxyproline and proline, which is useful from selectivity perspective. The bare silica phases should according to the
1 1,5 2 2,5 3 3,5 4 4,5
Fluoro-phenyl
BEH 2-EP DEA AA 2-PIC BEH
HALO HILIC Diol
Penta HILIC
Nucleodur HILIC
RT(min)
AlaArg AsnAsp CysGln GluGly HisHyp IleLeu LysMet PhePro SerThr TrpTyr Val
19
LSER model interact strongly with cations which could be a reason for this phenomenon. The increased retention provided some selectivity, but the peak shapes were still poor for both bare silica phases. The strong hydrophilicity of the diol and penta-HILIC columns resulted in even more retention, with histidine being completely retained by the penta-HILIC stationary phase.
All other solutes had better peak shapes on penta-HILIC than on the diol column, which presented poor peaks in general. The zwitterionic stationary phase of the Nucleodur HILIC column increased retention even further via the negatively charged sulfonate group on the stationary phase, completely retaining the basic amino acids. Furthermore, its selectivity was very poor, especially considering the overall late sample elution.
The selectivity on the pentafluorophenyl column was different from the others. Particularly, for the neutral and acidic amino acid samples which had different orders of elution. The selectivity differences on the other columns, that exhibited good efficiency and peak shape, consisted of elution order differences for basic, amidic and aromatic solutes instead. E.g.
tyrosine and tryptophan eluting before histidine and lysine on the AA column, instead of before, as they did on the 2-EP column.
4.3 Modifier
Figure 6 Retention times of 21 amino acid samples injected onto the columns a) 2-EP, b) 1-AA and c) diol for gradient elution with methanol and 50/50 methanol/acetonitrile (v/v), both containing 10 mM ethanesulfonic acid. Dashed lines represent no elution was observed at their end-point.
1,5 2 2,5 3 3,5 4 4,5
MeOH MeOH:AcN
RT(min)
1,5 2 2,5 3 3,5 4 4,5
MeOH MeOH:AcN
RT(min)
AlaArg AsnAsp CysGln GluGly HisHyp IleLeu LysMet PhePro SerThr TrpTyr Val 1,5
2 2,5 3 3,5 4 4,5
MeOH MeOH:AcN
RT(min)
a) b)
c)
20
Upon changing modifiers from one consisting of methanol with 5 vol% water and 10 mM ethanesulfonic acid to one comprised of equal volumes of methanol and acetonitrile with the same additives, see Figure 6, the most notable effect was increased retention. The lower solvent strength of acetonitrile causes a higher distribution to the stationary phase. With the methanol content being halved, another result could be an increase in apparent pH subsequently causing deprotonation of the carboxyl groups of the samples. This would make the amino acids more polar and cause further distribution toward the stationary phase. However, further studies are necessary to prove this is in fact the reason for this. Tryptophan and tyrosine appear less affected by the modifier change than other solutes in the same retention time range, albeit they still see an increase in retention. It is uncertain why they behave this way, but it could be useful for tweaking selectivity.
4.4 Pressure and temperature
Figure 7 Apparent retention factors for seven amino acids, representing their respective groups, acquired for three different ABPR pressures. 125 bar (standard operating conditions), 150 bar (flow rate 1.000 mL/min, 15-60 % modifier in 4.81 min gradient time) and 175 bar (0.900 mL/min, 15-60 % modifier in 5.33 min gradient time). Otherwise settings followed standard operating conditions. All samples were injected onto a diol column and the modifier contained 10 mM ethanesulfonic acid.
When employing higher pressures, a decreased retention was observed which is in line with earlier studies. This is due to the increasing density with the increasing pressure, which subsequently leads to a higher solvation power of the mobile phase. A visualization of this is presented in Figure 7, for seven amino acid samples representing their respective group. Here the values plotted are apparent retention factors, calculated from respective retention time, to compensate for the differences in operating conditions. All sample groups appear to react with a similar decreased retention time and no changes in selectivity was observed.
The effects of changing temperature are presented in Figure 8, as apparent retention factor values. The increase of temperature resulted in increased retention for all samples which is due to the decreased density and subsequent decrease in solvation power that accompanies it. The temperature variations were performed on a pentafluorophenyl column, since 20 °C would result in too high pressures at the high modifier proportions needed to elute all samples on a diol column. The temperature study was carried out by gradient elution of 21 amino acid samples at 1.000 mL/min flow rate and a gradient of 5-45 % methanol containing 10 mM of ethanesulfonic acid in 4.27 minutes.
3 4 5 6 7 8 9 10
125 150 175
kapp
Presure (bar)
Cys Glu Gly His Hyp Leu Phe
21
Figure 8 Apparent retention factors for 21 amino acid samples, acquired at 5 different temperatures after being injected onto a pentafluorophenyl column. Flow rate 1.000 mL/min 5-45 % modifier in 4.27 min gradient time, otherwise settings followed standard operating conditions. The modifier contained 10 mM ethanesulfonic acid.
4.5 Additive concentration
The effects of increasing the additive concentration was examined with ethanesulfonic acid on a diol column, Figure 9, and ammonia on a pentafluorophenyl column, Figure 10. The PFP column was chosen so that the amino acids with basic side chains could be studied, as they did not elute from the diol column upon implementation of basic additives.
The most notable effect is the decreased retention of the basic samples with an increase in ammonia concentration. This could mean that higher proportions of ammonia have an ion suppressive effect on solutes making them less polar, resulting in lower retention. Implementing ammonia as an additive on a pentafluorophenyl column increased the retention for all samples compared to when ethanesulfonic acid was used for screening of the columns due to the increase in apparent pH. Even at much higher concentrations, the retention mechanism was dominated by the charge carried by the solutes and the slight decrease in retention might indicate competitive effects on the stationary phase. The increased concentration of ethanesulfonic acid resulted in similar decrease in retention, probably due to increased ion-pairing and seems to level off as ion-pairing gets saturated.
Depending on the solutes in question, these results can be of use to improve selectivity.
Increasing basic additive concentration appears to decrease retention more for solutes containing basic groups. Some changes in retention order are observed when increasing the ammonia concentration, but they seem arbitrary and the chromatograms were hard to interpret as they presented very low and broad peaks. The general trend is not to be disregarded, however.
4,0 4,5 5,0 5,5 6,0
20 30 40 50 60
kapp
Temperature (oC)
AlaArg AsnAsp Cys GlnGlu GlyHis Hyp IleLeu LysMet Phe ProSer ThrTrp Tyr Val
22
Figure 9 Retention times for 21 amino acid samples injected onto a diol column, acquired at five different ethanesulfonic acid concentrations in the organic modifier.
Figure 10 Retention times for 21 amino acid samples injected onto a pentafluorophenyl column, acquired at six different ammonia concentrations in the organic modifier. The dashed, blue graph is the leucine from the ethanesulfonic acid concentration study on diol column, included to visualize scaling between the two diagrams 9 and 10.
1,9 2,1 2,3 2,5 2,7 2,9 3,1 3,3 3,5
0 10 20 30 40 50
RT(min)
[Ethanesulfonic acid] (mM)
AlaArg AsnAsp CysGln GluGly HisHyp IleLeu LysMet PhePro SerThr TrpTyr Val
1,2 1,4 1,6 1,8 2 2,2
0 20 40 60 80 100
RT(min)
[NH3] (mM)
AlaArg AsnAsp CysGln GluGly HisHyp IleLeu LysMet PhePro SerThr TrpTyr Val
23 4.6 Application on polar samples
Figure 11 ESI+ MS detection chromatograms of the amino sugars 1. L-fucosylamine, 2. D-mannosamine, 3. D-galactosamine, 4. 6-amino-6-deoxy-D-glucose and 5. D-(+)-Glucosamine (magenta) on a DEA column. Additives employed were a) 5 mM TFA + 1.0 mM 18-crown-6 and b) 5 mM TFA + 5 mM NH3.
The DEA column was chosen for separation of the amino sugars presented in Table 5. It was decided to employ TFA instead of ethanesulfonic acid as an additive to be more compatible with MS detection. It was used in combination with 18-crown-6, Figure 11a, as it had been successful in improving peak shapes for primary amines. However, the peak shapes were poor, and the best results in terms of both peak shape and separation were produced when a combination of TFA and ammonia was used without any crown ether, Figure 11b. The operating conditions for injection of the amino sugar samples are presented in Appendix 3. The poor peak shapes/overlap in Figure 11 is due to presence of anomers of the sugars and partial separation as a result of this.
Figure 12 UV chromatograms from isocratic elution of two phenyl glucosides on a DEA column at 20, 18, 16 and 14 % modifier respectively. All containing 10 mM of ethanesulfonic acid. Phenyl β-D-thioglucopyranose (the taller peak in all of the four peak couples) eluted before Phenyl β-D-glucopyranose at all modifier proportions.
Figure 13 UV chromatogram of 21 amino acids injected onto a DEA column with a low gradient slope. Settings followed the standard operating conditions except 15-35 % modifier in 12 min gradient time.
a) b)
2
3
1 4
5
2 3 5
4
1
