Unifying Chromatography to Meet Business Needs

Size: px

Start display at page:

Download "Unifying Chromatography to Meet Business Needs"

Peter Rose
5 years ago
Views:

1 Unifying Chromatography to Meet Business Needs Thomas L. Chester 7122 Larchwood Drive Cincinnati, OH Acknowledgments: Procter & Gamble Claudia Smith David Pinkston Doug Raynie Tom Delaney Grover Owens Jianjun Li Chris Ott Dimitra Simmons Jim Ziegler Lynn Cole David Innis Rosemary Hentschel Brian Haynes Don Bowling Lisa Burkes Rebecca Cunningham Steve Teremi Chris Gamsky Jason Coym Steve Page Page 1 1

2 History and overall perspective Chromatography is 1+ years old Partition chromatography, 1941 (Martin & Synge) GC, 1952 (Martin & James) HPLC, ~197 New possibilities require us to think differently Unification: Don t be ed in practice by boundaries that don t really exist. 3 unification topics: 1. Chromatography from the mobile phase perspective 2. Performance expectations 3. Parameter interactions, optimization, and opportunities. Page 2 2

3 1. The Mobile Phase Perspective Supercritical Fluid according to IUPAC and ASTM a) b) Pressure Pc Solid Liquid Critical Point Pc Solid Liquid Supercritical Fluid Region Triple Point Vapor Triple Point Vapor Tc Temperature Tc Temperature Fluids Liquids Gases Supercritical, etc. Solvating Yes No Yes, variable, and it depends on compression Compressible No Yes Yes Page 3 3

4 1. The Mobile Phase Perspective Supercritical Fluid according to IUPAC and ASTM a) b) Pressure Pc Solid Liquid Critical Point Pc Solid Liquid Supercritical Fluid Region Triple Point Vapor Triple Point Vapor Tc Temperature Tc Temperature C) pressure one-phase region available for chromatography solid liquid vapor Temperature temperature Page 4 4

5 Adding a second component requires another dimension in the phase diagram: P 1% b 1% a T a is the more volatile component The critical mixture curve is the locus of mixture critical points: Type I Binary Mixture P 1% b 1% a T Page 5 5

6 pressure b a Temperature temperature pressure Supercritical Fluid Chromatography range pressure Unnamed range a Temperature temperature OT-SFC range (with onecomponent mobile phase) a Temperature temperature Page 6 6

7 pressure Subcritical Fluid Chromatography range pressure (Extended) Enhanced- Fluidity Liquid Chromatography range a a Temperature temperature Temperature temperature Unified Chromatography from the mobile phase perspective pressure One-phase region available for chromatography a Temperature temperature Page 7 7

8 C) pressure one-phase region available for chromatography solid liquid vapor Temperature temperature Conventional LC and GC are ing cases of Unified Chromatography pressure b a GC range Temperature temperature Page 8 8

9 Unified Chromatography pressure One-phase region available for chromatography a Temperature temperature Why? Save money Reduce expense Reduce time to market 1, 2-min analyses require 139 days Reducing to 3-min analyses requires 21 days and saves 118 days For a $1M/year product: $32M in sales $3M in earnings Page 9 9

10 How can we save time? Plate generation takes time, so Improve selectivity to reduce the required plate number Generate plates faster Increase the diffusion rate Decrease critical dimensions Evidence from elsewhere See Gerd Vanhoenacker, Pat Sandra, J. Sep. Science 26, 29, for an example of the influences of temperature change on selectivity and diffusion. They took a 12-min separation that was inadequate, and resolved everything in about 2 min by increasing the temperature and the flow rate. These changes improved the selectivity (to reduce the plates required) and increased the rate at which plates are produced. Page 1 1

11 Unified techniques offer Different, potentially better selectivity Faster plates per unit time 2. Performance expectations: Unified techniques should provide the same figures of merit as HPLC, only faster. Injection and detection interfacing must be changed. Page 11 11

12 Let s use SFC as an example to look at a few problems Complaint: Peak area precision is not as good in SFC as in HPLC even when both use the same injector. Experiment: Gilson SF-3 with model 234 autosampler Specification: RSD=.5% HPLC (n=1): RSD=.38% SFC (n=1): RSD=7% (same autosampler protocol) The loop is pressurized while in the Inject position. CO 2 is liquid at ambient temperature. A. Inject position to the column inlet from the pump or mixer Page 12 12

13 When switched to Load, the loop is vented. CO 2 is gaseous at ambient temperature. B. Load position to the column inlet from the pump or mixer Let s use SFC as an example to look at a few problems Complaint: Peak area precision is not as good in SFC as in HPLC even when both use the same injector. Experiment: Gilson 234 autosampler Specification: RSD=.5% HPLC (n=1): RSD=.38% With our changes: SFC (n=15): RSD=.25% (worst case of 3 trials) RSD can be just as good in SFC J. W. Coym and T. L. Chester, J. Sep. Sci. 23, 26, Page 13 13

14 Peak area SFC calibration chart for 4-hydroxybenzyl isothiocyanate assay in mustard extract R 2 = mg/ml Let s use SFC as an example to look at a few problems Complaint: Signal/Noise ratios are worse in SFC than in HPLC Page 14 14

15 Possibilities Signal two common oversights: Analytical SFCs are sometimes furnished with prep UV detector cells (.5-mm path). Peaks with the same temporal width will produce smaller signals in concentration detectors (like UV) in a with faster flow (like SFC). Noise HPLC Injector Detector Pump A, Weak solvent (Flowcontrolled) Pump B, Modifier (Flowcontrolled) Column Oven 1 2 Mobile phase strength is constant from the injector to the outlet. Pressure is constant and fairly low in the detector. Page 15 15

16 Unified SFC Injector High-Pressure Detector Back- Pressure Regulator Pump A, Main Fluid (Flowcontrolled) Pump B, Modifier/ Additives (Flowcontrolled) Column Oven A back pressure regulator is required to prevent the MP from expanding, weakening and separating. Flow-through detectors must be operated at pressure, not atmospheric pressure. Possibilities Analytical SFCs are sometimes furnished with prep detector cells (.5-mm path). Peaks with the same temporal width will produce smaller signals in concentration detectors (like UV) in a with faster flow (like SFC). Signal: Noise: Pressure fluctuations in the BPR and detector Switching BPR a possible noise source PID BPR highly unlikely Oscillations in the detector cell or in coupled components Refractive index of compressible fluids is ~1x more sensitive to pressure and temperature changes than noncompressible fluids. Page 16 16

17 RI of CO 2 at 4 C See Sun, Y.; Shekunov, B. Y.; York, P. Chem. Eng. Commun. 19, p (23). CO 2 changes it RI value from about 1.4 as a low-density supercritical fluid to about 1.9 as a high-density supercritical fluid. Water changes only 1% from 1 to 9 C and is virtually flat with pressure. Unified SFC Injector High-Pressure Detector Back- Pressure Regulator Pump A, Main Fluid (Flowcontrolled) Pump B, Modifier/ Additives (Flowcontrolled) Column Oven temperature is changing If RI is not uniform, then eddies and convection currents will randomly refract light in the detector and create noise. Page 17 17

18 Solution? Stabilize the temperature (and the pressure) in the detector make sure the RI is uniform. Don t settle for any S/N compromise in SFC or any unified technique Find the problem and fix it. Page 18 18

19 3. Parameter Interactions, Optimization, and Opportunities In the workplace, we really don t understand the parameter interactions. We ll use HPLC as an example The Need for Modeling and a Multivariate Approach: parameter interactions Performance measures: Resolution Analysis time Pressure and Flow s (Accuracy, precision, etc.) Resolution-controlling parameters (Purnell): Efficiency (5) Selectivity (5) Retention factors (5) But these are not directly adjustable N ( ) R = α 1 S 4 α k ( k + 1) 2 2 Page 19 19

20 The Need for Modeling and a Multivariate Approach: parameter interactions Performance measures: Resolution Analysis time Pressure and Flow s (Accuracy, precision, etc.) Resolution-controlling parameters (Purnell): Efficiency (5) Selectivity (5) Retention factors (5) But these are not directly adjustable Independently adjustable parameters: Column length, L Column diameter, d c Particle size, d p Flow rate, F Stationary phase Mobile phase modifier ph Temperature Modifier concentration, %B Adjustable HPLC parameters are highly interrelated Univariate optimization will not work Leveraging the complexity Modeling and virtual experimentation Multivariate approach to optimization Combine particle size, column dimensions, and pressure and flow constraints with all other adjustable efficiency parameters to maximize the overall performance. Page 2 2

21 Examine the complexity of a simple reversed-phase separation.4.3 butylparaben propranolol naphthalene acenaphthene.2.1 uracil amitriptyline min Page 21 21

22 Vary only F and %B, L = 5 and 1 cm, Rs2., P1-MPa 3 Analysis time (min) best solution, meets Rs, not pressure-ed best solution, excess Rs, pressure-ed Column length (cm) Solutions for 5- and 1-cm, Rs = 2., 1 MPa Analysis time (min) Column length (cm) Page 22 22

23 Solutions for 5- and 1-cm Analysis time (min) not pressure ed pressure ed at: 1 MPa 2 MPa 3 MPa Column length (cm) a But at 7.5 cm Analysis time (min) not pressure ed pressure ed at: 1 MPa 2 MPa 3 MPa Column length (cm) a Page 23 23

24 At 3 MPa, 65% savings Analysis time (min) not pressure ed pressure ed at: 1 MPa 2 MPa 3 MPa Column length (cm) a The best column length depends on the pressure and on many other parameters. Loci of optimal analysis times as a function of the pressure 9 8 L =25 cm 7 pressure-ed Analysis time, min L =2 cm L =15 cm pressure-ed not pressure-ed not pressure-ed pressure-ed psi Pressure, MPa Page 24 24

25 The point of this is to show how complex the interdependencies are, even when only three variables are considered And that counter-intuitive changes in parameter values are often required to go faster 4 cold remedy actives Best analysis time vs. column length, 22 psi Minimum time for Rs= Time, min Standard HPLC, 5um Low-vol HPLC, 5um Standard HPLC, 3um Low-vol HPLC, 3um Column length, cm Page 25 25

26 Same problem, higher resolution Minimum time for Rs= time, min Standard HPLC, 5um Low-vol HPLC, 5um Standard HPLC, 3um Low-vol HPLC, 3um Column length, cm Another approach: maximum resolution with a 1-min time and 22 psi max Best Rs in 1 minutes 6 5 Rs Standard HPLC, 5um Low-vol HPLC, 5um Standard HPLC, 3um Low-vol HPLC, 3um Column length, cm Page 26 26

27 Optimization Considerations Around Particle Size Consider 3.5 µm vs. 1.7 µm particles Efficiency: H min becomes half N/L doubles, or L is half for the same N u opt doubles t R is quarter 75% time savings pressure goes up by 4x extracolumn volumes become deadly data acquisition rates must be faster Example reversed-phase separation.4.3 butylparaben propranolol naphthalene acenaphthene.2.1 uracil amitriptyline min Page 27 27

28 Same problem considering d p, %B, F and L;.1-mL extra-column volume; Rs2. Analysis time (min) constraints: 1 ml/min, 13 MPa (15, psi) 5 ml/min, 41 MPa (6, psi) 2 ml/min, 62 MPa (9, psi) Particle size ( µ m) Peaks don t behave like we thought! But, these are sparse, isocratic chromatograms. What about gradients with maximum peak capacity? Consider maximizing the peak capacity within a set time, a set retention range, and a maximum pressure set the time, set the range, set the pressure, and minimize the peak width. Page 28 28

29 Consider the peak width vs. particle size. Analysis time and retention range are fixed. No pressure. peak width (relative scale) particle size (µm) L Consider the peak width vs. particle size Analysis time and retention range are fixed No pressure peak width (relative scale) L 2L particle size (µm) Page 29 29

30 Consider the peak width vs. particle size Analysis time and retention range are fixed No pressure peak width (relative scale) L 2L 3L particle size (µm) If we start with large particles and decrease, we will eventually hit the pressure peak width (relative scale) particle size (µm) Page 3 3

31 If we start with large particles and decrease, we will eventually hit the pressure peak width (relative scale) particle size (µm) If we start with large particles and decrease, we will eventually hit the pressure peak width (relative scale) Locus of pressure-ed solutions at pressure = P particle size (µm) L and d p both increasing to reduce w at pressure = P Page 31 31

32 If we start with large particles and decrease, we will eventually hit the pressure peak width (relative scale) These are analytical solutions (that is, algebraic) with lots of simplifying assumptions particle size (µm) 3P P 2P Assumptions will eventually break down Likely outcome: an optimal d p and L combo for a given problem and pressure peak width (relative scale) Locus of pressure-ed solutions at pressure = P particle size (µm) Page 32 32

33 Preliminary numeric solutions: Peak width vs. dp fixed 6-min gradient time and fixed solute range.5 peak width, min L 2L 3L 5L P :225 bar P : 62 bar P : 1 bar dp, µm Optimization Conclusions No parameter, not even particle size, if considered alone will optimize the separation. All the adjustable parameters must be considered in concert. Savings can be big. Results can be surprising. Ultra-small particles will be most valuable for problems requiring lots of plates or lots of peak capacity. Assay methods involving a few peaks of interest will usually run faster with larger particles. Page 33 33

34 Our Current Picture of Unification: 1. Expand our choices of fluids, and add both temperature and pressure to our control parameters to achieve higher selectivity and faster plate generation for specific problems. 2. Use all adjustable parameters (including temperature and pressure) in a multivariate optimization to deliver business needs in minimum time. Additional references for the MASSEP presentation, October 18, 27: Unified Chromatography 49. Chromatography from the Mobile Phase Perspective, T. L. Chester, Anal. Chem. 69, 165A- 169A, (1997). 55. Unified Chromatography, J. F. Parcher and T. L. Chester, Eds., ACS Symposium Series, American Chemical Society, Washington, D.C., U. S. A. (2). SFC injection and detector interfacing 53. Pressure-Regulating-Fluid Interface and Phase Behavior Considerations in the Coupling of Packed-Column Supercritical Fluid Chromatographs with Low-Pressure Detectors, T. L. Chester and J. D. Pinkston, J. Chromatogr. A, 87, (1998). 64. Improving Injection Precision in Packed-Column Supercritical Fluid Chromatography, J. W. Coym and T. L. Chester, Journal of Separation Science 26, (23). 5. Injection Techniques in SFC, T. Greibrokk and T. L. Chester, Analusis 27, (1999). 71. Supercritical Fluid Chromatography Instrumentation, T. L. Chester and J. D. Pinkston, appearing in "Ewing s Analytical Instrumentation Handbook, 3rd Edition" J. Cazes, Editor, Marcel Dekker, New York (25). Modeling and multivariate optimization 66. Business-Objective-Directed, Constraint-Based Multivariate Optimization of HPLC Operational Parameters, T. L. Chester, Journal of Chromatography A, 116 (23) A Virtual-Modeling and Multivariate-Optimization Examination of HPLC Parameter Interactions and Opportunities for Saving Analysis Time, T. L. Chester and S. O. Teremi, Journal of Chromatography A, 196 (25) Business-Needs-Driven, Constraint-Based HPLC Optimization, T. L. Chester, American Laboratory, 37:24 (25), Page 34 34

Chromatography- Separation of mixtures CHEM 212. What is solvent extraction and what is it commonly used for?

Chromatography- Separation of mixtures CHEM 212 What is solvent extraction and what is it commonly used for? How does solvent extraction work? Write the partitioning coefficient for the following reaction: