TOPICS IN LIQUID CHROMATOGRAPHY Part 2. Optimum Bed Density [OBD™] Columns: Enabling Technology for Laboratory-Scale Isolation and Purification Dramatic improvement in stability, lifetime, efficiency, peak shape, and reproducibility results from continuing investigation of, and innovation in, the design of particles, columns, and packing processes.

DRUG DISCOV ERY CHALL ENGES P REPA RAT IV E LC For modern pharmaceutical research laboratories to fill new drug pipelines with viable candidates, thousands of synthetic and natural compounds must be isolated, identified, synthesized, purified, characterized, screened, and tested. Tools and techniques that optimize and streamline these processes shorten development cycles, increase competitive advantage, and, ultimately, save lives. Laboratory-scale chromatographic isolation and purification plays a central, and rapidly expanding, role in these efforts. Purifying the contents of huge combinatorial libraries, compound collections, or batch products from parallel synthesis requires prep LC to perform efficiently and reliably in a high-throughput environment. An obvious solution is to maintain the speed and resolution of analytical LC by using the same small [5-10 µm] particles in a larger diameter column of identical length at a flow rate scaled up by the ratio of column cross-sectional areas. This demands a system with high-pressure and high-flow-rate capabilities, as well as detectors able to handle higher sample concentrations, all under automated control. However, it is not so simple! Though manufacturers have scaled up their tube diameters, packing stations, and protocols, typical prep HPLC columns withstand only a few hundred injections. Column lifetime is variable and, all too often, unacceptably and unpredictably short—leading to system downtime, lost productivity, and even loss of valuable samples. Already operating at higher back pressures, additional stress is placed on small-particle column beds by injecting slugs of sample dissolved in a powerful, but highly viscous, solvent like dimethyl sulfoxide [DMSO, 2.47 cP], by running fast gradients [especially ones with significant viscosity changes], and by failing to remove debris or to anticipate sample solubility problems, particularly in the initial gradient mobile phase. All these factors can expose the weaknesses in bed structures formed by imperfect packing processes and lead to collapse, voiding, plugging, and premature column failure. Scientists at Waters pioneered packing methods for high performance of small- and large-diameter particles and columns more than three decades ago. They have continued to research fundamental principles and to refine packing processes in parallel

2

with our materials science programs to develop new, higher performance LC particles. In response to input from separations scientists worldwide, they sought to produce small-particle prep LC columns with the predictable performance [efficiency, peak shape, lifetime, reproducibility] typically associated with analytical-scale columns. Their success will be detailed here.

A Brief History of Packing Preparative LC Columns T he primary goal in preparing any LC column is to create a homogeneous packed bed, that, in turn, permits flow uniformity at all points throughout its porous structure. Once formed within a suitable container, it is equally important to introduce the sample into this bed in as narrow and as uniform a band as possible across the entire cross-section of the bed. Finally, the bands of sample components separated in their passage through the bed must be removed uniformly and efficiently from the outlet without further spreading or dilution. Martin and Synge espoused these principles in their Nobel-prize-winning work on partition LC published in 1941.1 Chromatographers today are still struggling to achieve in practice the efficiency expected from such packing and performance perfection. Research results in the area of column packing must be interpreted cautiously. Beware of too much clean thinking about dirty experiments, done with poorly characterized materials, illconceived devices, and improper controls. T his may lead to columns not only filled with, but designed and built upon an unstable theoretical foundation of, sand. Fundamental, pioneering studies on prep LC column design and packed bed formation were conducted at Waters in the mid1970s [Figure 1], beginning with the development of the first commercial 10-µm HPLC columns [µBondapak ® C18 ]. 2 While details remain proprietary, a review of large-diameter bed structure characteristics, packing methods, and unpublished experiments can be found in reference 3. An excellent discussion of the design and packing of analytical HPLC columns can be found in reference 4.

a movable piston.8 A slurry of 10-µm particles was poured into the vertically-mounted column. Once the inlet cap was secured, the piston was moved upward from the bottom to effect the packing. In contrast to radial compression, this scheme was later described in its commercial embodiment as axial compression.

Figure 1. These hitherto unpublished 1975 Waters R&D prototype photos show how thin-plastic-walled, 5.7 x 30 cm PrepPak® cartridges, the initial embodiment of the invention of radial compression,5 could easily be dissected, making directly visible what never had been seen before: the actual shape of sample bands at various stages in their path through the bed. Such experiments led to theory-changing observations and innovations in packing and column inlet distributor design.3

A. From Dry Packing to Dynamic Compression Dry packing is principally used for particles > 25–30 µm in diameter. Early work was done using large [> 50 µm], irregular silica particles. We confirmed that larger diameter prep LC columns have two advantages: ■■

The volume of packing near the wall is a much smaller percentage of the total bed volume; this reduces somewhat the relative magnitude of the bandspreading component due to the higher permeability of void spaces near the wall.

■■

Particle bridges are less likely to form at larger diameters. Bridging occurs when axial pressure becomes redirected from particle to particle outward in a radial direction more or less perpendicular to the column wall, preventing particles below the bridge from being subjected to the exerted pressure.3

A concern in filling larger diameter columns is particle-size segregation in both axial and radial directions. Our investigation of dry-filling processes found the RTP [rotate–tap–pour] procedure to be best at preventing this phenomenon.3,6 Then, we made the discovery that combining an effective dry filling process with the application of pressure on a flexible column wall in a direction perpendicular to the axis of flow created a very uniform bed crosssection, eliminating wall chanelling, bridging, and internal voids within the packed bed structure.3,5 About the same time as our discovery of the advantages of radial compression for dry-packed beds, a group in France, independently, was constructing a prep LC column that contained

To compare the effectiveness of both techniques, carefully controlled experiments were performed in 1977, using the same 55-105 µm silica to prepare two prep LC columns of approximately the same length, one radially, and one axially, compressed.9 While preparing the latter, it was observed that after pouring the particle slurry into the open column, the particles settled by gravity into a loose bed before the end cap could be affixed, fluid connections made, and the piston activated. The piston only served to remove the void above the bed and maintain its structure. It did very little to pack the bed. Increasing the piston pressure caused bridging to occur and made it very difficult to extrude the bed when emptying the column. By comparison, it was found, under normalized conditions of linear flow velocity, sample concentration and load, that radial compression formed a much more efficient [329 vs. 714 µm plate height], higher density [0.42 vs. 0.34 g/mL] bed. Radial and axial compression are often referred to as dynamic techniques.10 Each is used to remove or avoid voids. Radial compression is much more effective at eliminating wall channels and maintaining bed homogeneity. It is important to note that either technique can fall short of theoretical expectations for efficiency and performance if the instrumentation for its practice is improperly designed or used. Also, in both cases, it is critical that a uniform bed of particles be created first, before applying compression.

B. Slurry Packing and Dynamic Compression Slurry packing is preferred for particles ≤ 20–25 µm since, in the dry state, they behave more like a fluid, rather than ‘rocks’, due to electrostatic or electrokinetic forces between them. Early methods allowed gravity to settle the particles suspended in a fluid, but it was soon found that a high-pressure filtration process formed more uniform beds. For particle diameters ≤ 10 µm, schemes involving the use of high-density, balanced-density, or high-viscosity fluids were tested. Variables that may be tailored for a specific type or shape of particle to induce uniform bed formation include flow rate, pressure, slurry concentration, and slurry solvent composition. For a good review of the science and art of slurry packing, see reference 7.

3

As manufacturers gained confidence in creating good analytical LC columns, they generally applied the same 5-10 µm-particle packing methods to prep LC columns up to 50 mm I.D. T hese were intended for high-throughput pharmaceutical applications. However, as stated earlier, frustration with short unpredictable column lifetimes has led to a search for a way to produce consistently superior prep LC columns.

0.700 Optimum Packed-Bed Density

[g/cm3]

Dynamic compression techniques are not without drawbacks. Too much pressure may fracture particles, creating less homogeneous, denser, lower-permeability beds. The requisite hardware is often cumbersome and costly. Meanwhile, in the last two decades, due to advances in its capability to yield efficient, reproducible, stable beds, slurry packing has obviated compression schemes in laboratory-scale LC columns.

care must be taken not to over-compress, or disrupt in any nonuniform way, this portion of the bed. For prep LC columns with a smaller aspect ratio [length/diameter], high-pressure slurry packing of small particles cannot achieve the optimum bed density characteristic of well-made analytical LC columns [Figure 2]. Too much axial compression applied at the inlet can break particles, build bridges, and lower local bed permeability [Figure 3].

Packed-Bed Density

Radial compression technology was extended to analytical-scale LC columns in 1977. T hicker-walled-plastic, 8 mm I.D. RadialPak™ cartridges were first filled by slurry packing 10-µm spherical particles and then operated in special chambers that applied force radially along the entire length of the tube to eliminate wall voids.11 Controlled studies confirmed that this combination of slurry packing and radial compression gave the most uniform and the highest density beds.12

0.675

4.6 x 100 mm

Analytical LC Columns

0.650

19 x 50 mm

0.625

Standard Packing Procedures Insufficient to Achieve Optimum Packed-Bed Densities in Larger-Diameter Columns

19 x 100 mm 0.600

50 x 100 mm

Prep LC Columns

0.575 0.550 0

5

10

15

20

25

Column Aspect Ratio — Length/Diameter

Figure 2. Average bed density for conventional analytical-LC-column-slurrypacking procedures vs. column aspect ratio.

Development of Waters Preparative OBD Columns In 2001, Waters scientists and engineers began a research program, seeking to understand why prep LC columns fail prematurely. They learned that beds of silica particles, generally considered to be relatively rigid and incompressible, in fact, exhibited elastic behavior under the fluid pressures used in packing and operating columns. With smaller particles in larger-diameter, shorter-length columns, these forces lead to bed-structure changes and cut column life short. Sophisticated equipment was built to measure the effects on bed structure and density of a variety of applied mechanical [compression] and hydraulic [packing fluid or mobile-phase drag] forces. Variables studied included column dimensions [I.D. and length], particle morphology and type, and solvent composition. Bed density was characterized in both axial and radial directions, in both uncompressed and compressed states. We re-confirmed that, with every type of column and particle size and shape, whether dry- or slurry-packed, particles near the inlet, for a variety of reasons, ultimately form the least homogeneous portion of the bed. Also, during the final capping process, great 4

4

Optimum Bed Density

Packed-Bed Density

Axial-Ram-Compressed Slurry Packed

High-Pressure Slurry Packed Column Inlet

Column Outlet

Figure 3. Effect of packing method on average bed density across column length.

We then sought to marry high-pressure slurry packing with a carefully calculated modicum of axial compression localized at the less-dense inlet end of the bed [Figure 4]. With careful tuning of the packing process for each particle type and column geometry, predictable, uniform density profiles were obtained throughout the column.13 A new, elegant column was designed to incorporate a pair of specially designed distributors and dual, chemically inert seals made to prevent leaks at variable and high operating pressures [Figure 5].

High-Pressure Slurry Packing Localized Bed Expansion Near Inlet

Axial-Ram Slurry Packing Localized Compression Near Inlet

XTerra Prep MS C18 19 x 50 mm, 5 µm 5000

OBD Process Optimum Bed Density Throughout

Prep OBD LC Column*

Efficiency [plates]

4000

Figure 4. Optimized bed density in an OBD column is achieved by judicious application of a tuned amount of axial force only at the column inlet.

3000

2000

Traditional Prep LC Column*

1000

0

Initial

Cycle 1

Cycle 2

Cycle 3

Backpressure: 490

510

510

320

370

370

510 psi* 370 psi*

USP Tailing: 1.03

1.04

1.04

1.04*

1.01

1.15

1.45

1.47*

Waters is a primary producer and supplier of over 30 different silicaand hybrid-particle packings. Over three decades of experience with the interplay of particle properties, column design, and packing technique enable the achievement of optimal performance for all the Optimum Bed Density [OBD] products.14

Performance of Preparative OBD LC Columns Extensive evaluation in laboratories both at Waters and in the pharmaceutical industry, conducted during the three-year study on OBD preparative columns, verified improvements in lifetime, reproducibility, and performance. Column efficiency was shown to approach that of corresponding analytical LC columns. For example, SunFire™ analytical and prep LC columns exhibit an average reduced plate height of < 2.3.

XTerra Prep MS C18 OBD 30 x 100 mm, 5 µm

3.0

Maximum Spec = 2.8

2.5 2.0 0 1

2

3

4

5

6

7

8

9 10 11 12 13 14 15 16 17 18 19 20

Test Number Column 1

USP Tailing Factor

Figure 5. An exploded view of the elements of an empty OBD column. Inlet is at top right.

Reduced Plate Height

Figure 6. Accelerated stability test comparison. OBD columns are dramatically more stable than traditional prep LC columns.15 Higher back pressure indicates a higher bed density. Consistency of plate count and U.S.P. Tailing Factor shows that the OBD packed bed remains homogeneous and stable.

Column 2 XTerra Prep MS C18 OBD 30 x 100 mm, 5 µm

1.4

Maximum Spec = 1.2

1.2

1.0 0 1

2

3

4

5

6

7

8

9 10 11 12 13 14 15 16 17 18 19 20

Test Number

Figure 7. Long-term stability of OBD columns is excellent during occasional use over a 7–8 week span. Each line break between data points represents a solvent-switch stress cycle.16

5

A. Superior Stability

C. Effective Maintenance of Efficiency and Resolution

An accelerated lifetime test was devised to punish small particle prep LC columns with cycles of solvent viscosity changes. As can be seen in Figure 6, the bed stability and performance of the new OBD columns is outstanding. A similar longer-term test [Figure 7] demonstrates little variation in reduced plate height and USP peak tailing factor over a 7–8 week period. Note that a reduced plate height value near 2 indicates a high-quality well-packed column.

Once a baseline separation is achieved, the real challenge in a high-throughput environment is to see how long adequate resolution is maintained. Results from an experiment designed to do this are shown in Figure 9. Five consecutive cycles were run, each consisting of ten serial injections followed by flushing with 900 column volumes of 70:30 [v:v] water:acetonitrile with 1% TFA mobile phase. Maintenance of an average resolution of 1.77, with a standard deviation of only 0.07, demonstrates the superior stability of the OBD column.

3

Day 1: Column 1 Total Mass Load: 30 mg in DMSO

2

3 Column Resolution

Injection 14

1

3

Day 10: Column 1 Total Mass Load: 30 mg in DMSO

2

Injection 955

1

Average of 50 injections = 1.77 ± 0.07

2

1 10

2

1

3

4

5

30

6 min

50

Number of Injections

XTerra Prep MS C18 OBD, 19 x 50 mm, 5 µm

Column

New Column

After 1000 Injections

Plates

USP Tailing Factor

Plates

USP Tailing Factor

1

4,278

1.14

3,927

1.17

2

4,430

1.05

4,129

1.07

3

3,943

1.02

4,054

1.08

4

4,093

1.03

3,945

1.13

4,186

1.06

4,014

1.11

Average

% Change in Efficiency -8.2 -6.8 2.8 -3.6 -4.0

Figure 8. Four OBD columns after more than 10 days of continuous operation and 1000 injections demonstrate bed stability, long lifetime and consistent column-to-column performance.17

B. Improved Reproducibility and Column Lifetime Reproducible columns with predictable lifetimes are key to maintaining a high-throughput-purification laboratory. As a test, four XTerra Prep MS C18 OBD, 19 x 50 mm, 5 µm columns were subjected to the continuous trauma of 1000 DMSO sample injections. Data from these experiments and representative chromatograms are shown in Figure 8. No significant change in backpressure was observed. Efficiency decreased on average only 4%, and USP tailing factors changed very little. T hese results demonstrate the excellent lifetime, ruggedness, and reproducibility of preparative OBD columns.

Figure 9. Resolution was maintained for a separation of a pair of fungal metabolites.18

An additional benefit of improved efficiency is the ability to run faster prep separations, thereby improving productivity. In Figure 10, a 50% increase in flow rate is shown to reduce the cycle time for a separation of 5 bases from 15 to 10 minutes. Productivity improves by 33%, without compromising the separation.

XBridge™ Prep MS C18 OBD, 19 x 50 mm, 5 µm Column

3

Flow Rate: 23.9 mL/min Backpressure: 700 psi

5

Cycle Time: 15 min

2

1 0

2

4

6

3 1

0

2

2

4

4

5

8

10 min

Flow Rate: 36.2 mL/min Backpressure: 1000 psi

4

Cycle Time: 10 min

6 min

Figure 10. With good efficiency and more effective use of the entire bed, OBD columns can be run at higher flow rate to increase producivity.19

6

D. Scale-Up with Confidence

Summary:

As described above, preparative OBD columns ensure performance equivalent to that of analytical columns. This simplifies scalingup a separation from analytical to prep LC. An example is shown in Figure 11. Nimodipine was separated from its impurities on a SunFire C18 analytical column. A total load of 6 mg on the analytical column permitted baseline resolution between peaks for impurity 2 and its parent compound. The scale-up factor is simply the ratio of the cross-sectional area of the larger- to that of the smaller-I.D. column; 17 x 6 mg = 102 mg. Clearly, these results indicate the ease of direct scale-up from analytical to preparative OBD columns.

Our research has shown that optimum bed density and uniformity can be achieved reproducibly in high-throughput prep LC columns. Doing this requires a thorough understanding of the interplay between particle characteristics, packing process parameters, and column hardware design. Properly executed, this expertise, developed over decades, delivers dramatic improvement in column lifetime as well as the efficiency, peak shape, and back pressure typical of an analytical column.

SunFire C18 4.6 x 100 mm, 5 µm Total Mass Load: 6 mg

1

Nimodipine 3

making many thousands of injections on many thousands of preparative OBD columns without voids and bed collapse! Truly this is the dawn of a bright future for high-throughput isolation and purification.

4

2

SunFire Prep OBD C18 19 x 100 mm, 5 µm Total Mass Load: 102 mg

1

Nimodipine 3

A survey of 200 chromatographers in the pharmaceutical industry indicated that 24% of them expected a prep LC column lifetime of no more than 1–200 injections.21 Only 6% of them experienced 1000 or more injections. Imagine their joy now at successfully

— Patrick D. McDonald, Raymond P. Fisk, Steven M. Collier, Fang Xia, Diane Diehl, Jie Cavanaugh, Gary S. Izzo, Thomas Grady, Jonathan L. Belanger, Darcy Shave

4

2 2

4

6

8

10

12

14 min

Figure 11. Scale-up from analytical to prep LC is easy using OBD columns. 20

7

References: NOTE: Where indicated below, use bold search term following * to access PDF document in Waters Library at . 1. A.J.P. Martin and R.L.M. Synge, Biochem. J. 35: 1358–1368 [1941] 2. P.D. McDonald, James Waters and His Liquid Chromatography People: A Personal Perspective, Waters Whitepaper *WA62008: 20 pp [2006] 3. P.D. McDonald and B.A. Bidlingmeyer, Strategies for Successful Preparative Liquid Chromatography, Chap. 1 in: J. Chromatogr. Lib. 38: 1–103 [1987], see esp. pp. 82–88 and references therein 4. U.D. Neue, HPLC Columns: Theory, Technology, and Practice, New York: Wiley-VCH [1997] 5. P.D. McDonald and C.W. Rausch, Radial Compression of Packed Beds, U.S. Patent 4,250,035 [Filed: 1975; Issued: 1981] 6. S .T. Sie and N. van den Hoed, Preparation and Performance of High-Efficiency Columns for Liquid Chromatography, J. Chromatogr. Sci. 7: 257–266 [1969] 7. U.D. Neue, op. cit., Column Packing and Testing, Chap. 5, pp. 93–105; see esp. pp. 93–101 and references therein 8. E. Godbille and L. Tondu, Chromatographic Device, U.S. Patent 3,996,609 [1976]; E. Godbille and P. Devaux, Use of an 18-mm I.D. Column for Analytical- and Semi-Preparative-Scale High-Pressure Liquid Chromatography, J. Chromatogr. 122:317–329 [1976] 9. Radially compressed column: a PrepPak ® Cartridge [I.D. 5.7 cm] used in a PrepLC ™/System 500 instrument; axially compressed column [I.D. 8 cm] prepared and used in a Jobin-Yvon™ Optical ChromatoSpac Prep 100 System; experimental details in: P.D. McDonald, C.W. Rausch, H.M. Quinn, J-D. Dauchy, An experimental comparison of radial and axial compression in Prep LC, Poster C16, 14th International Symposium on Chromatography, London, September 15, 1982 10. G . Guiochon, T. Farkas, H. Guan-Sajonz, J-H. Koh, M. Sarker, B.J. Stanley, T. Yun, Consolidation of particle beds and packing of chromatographic columns, J. Chromatogr. A 762:83–88 [1997] 11. C .W. Rausch, Y. Tuvim, U.D. Neue, Chromatographic Cartridge and Holder, U.S. Patent 4,228,007 [Filed: 1979; Issued: 1980] 12. N eue, op. cit., pp. 66–70 13. G .S. Izzo, R.P. Fisk, J. Belanger, D.E. Ziniti, Y. Tuvim, Chromatographic Column and Methods for Controlling Sorbent Density, U.S. Patent Application, US 2005/0224414 A1 [Filed: 2005] 14. F. Xia, J.Y. Cavanaugh, D.M. Diehl, G. Izzo, D. Shave, M. Savaria, T. Grady, M. Wanninger, D. Ziniti, B. Francis, R. Fisk, J. Belanger, D. Morrison, Stability and Reliability: New Approaches in Preparative HPLC Column Design, Pharm. Canada 5(1): 10–12 [2004]; Optimum Bed Density [OBD] Preparative Columns, Waters Brochure *720001456EN [2006] 15. Figure 6 Conditions: Columns: Both XTerra® Prep MS C18 19 x 50 mm 5 µm; one packed in traditional manner/hardware; one using OBD technology. Efficiency, back pressure, and U.S.P. tailing factor were measured: sample:

acetone [unretained] + acenaphthene [retained, test analyte]; mobile phase: acetonitrile:H2O 50:50 [v:v]; flow rate: 18 mL/min. Each of three 50-min accelerated stability test cycles consisted of five rounds of: 5 min at 100% H2O, followed by switch to 100% MeOH for 5 min, all at 18 mL/min. After each cycle, efficiency, backpressure and U.S.P. tailing factor were measured as before. 16. F igure 7 Conditions: Columns: XTerra Prep MS C18 OBD 30 x 100 mm 5 µm; efficiency and U.S.P. tailing factor measured as in Figure 6, except flow rate: 45 mL/min [linear velocity: 6.35 cm/min]. During 7–8 weeks of occasional testing between idle periods, each column was subjected to three accelerated stability test cycles, as described above, and its efficiency and tailing factor were checked just before and just after each cycle; line breaks on graphs indicate the points at which a cycle was run. 17. F igure 8 Conditions: Columns: XTerra ® Prep MS C18 OBD 19 x 50 mm 5 µm. Sample: Tylosin, sulfathiazole and ketoprofen dissolved in DMSO; total concentration: 35 mg/mL. Mobile phase A: H2O with 1% formic acid; Mobile phase B: acetonitrile with 1% formic acid. Gradient: 5% to 75% B; duration: 4.5 min. Flow rate: 18 mL/min. 18. F igure 9 Conditions: Column: XTerra Prep MS C18 OBD 19 x 100 mm 5 µm. Sample: Econazole and miconazole [3.2 mg/mL each] dissolved in DMSO; total concentration: 6.4 mg/mL. Mobile phase A: H2O with 0.1% trifluoroacetic acid. Mobile phase B: acetonitrile with 0.1% trifluoroacetic acid. Gradient: 5% to 90% B; duration: 10 min. Flow rate: 18 mL/min. Injection volume: 1000 µL; total sample load: 6.4 mg. After every tenth injection, column was flushed with 900 column volumes of H2O:acetonitrile 70:30 [v:v] with 1% trifluoroacetic acid; 50 injections total. 19. F igure 10 Conditions: Column: XTerra Prep MS C18 OBD 19 x 50 mm 5 µm. Analytes: 1–Labetolol [50 mg/mL], 2–Quinine [50 mg/mL], 3–Diltiazem [50 mg/mL], 4–Verapamil [100 mg/mL], 5–Amitriptyline [50 mg/mL]; all in DMSO. Injection volume: 660 µL; total load: 198 mg on column. Detection: UV at 260 nm. Mobile phase A: 0.1% DEA in H2O. Mobile phase B: 0.1% DEA in acetonitrile. First Run: Flow rate: 23.9 mL/min. Gradient: 1.91 min hold at 5% B, linear gradient to 95% B over 8 min, 1 min hold at 95% B, return to initial conditions, re-equilibration: 4 min. Second Run: Flow rate: 36.2 mL/ min. Gradient: 1.27 min hold at 5% B, linear gradient to 95% B over 5.33 min, return to initial conditions, re-equilibration: 2 min. 20. F igure 11 Conditions: Analytical column: SunFire C18 4.6 x 100 mm 5 µm. Prep LC column: SunFire™ Prep C18 OBD 19 x 100 mm 5 µm. Mobile Phase A: H2O with 0.1% formic acid. Mobile Phase B: acetonitrile with 0.1% formic acid. Sample: Nimodipine [30 mg/mL] in DMSO. Detection: UV at 290 nm. Instrument: Waters AutoPurification™ System. Peaks 1 through 4 are impurities in the standard. Analytical column: Flow rate: 1.4 mL/ min. Injection volume: 200 µL [6 mg]. Gradient: 1 min hold at 30% B, linear to 90% B over 10 min, 1 min hold at 90% B, return to initial conditions, re-equilibration: 4 min. Prep LC column: Flow rate: 23.9 mL/min. Injection volume: 3400 µL [102 mg]. Gradient: 1.97 min hold at 30% B, linear to 90% B over 10 min, 1 min hold at 90% B, return to initial conditions, re-equilibration: 4 min. 21. M . Andrews, J. O’Connor-Fix, HPLC Column Market, Waters Corporation Survey [June 2003]

Waters, XTerra, µBondapak, and PrepPak are registered trademarks of Waters Corporation. OBD, XBridge, SunFire, Radial-Pak, AutoPurification, PrepLC, and T he Science of W hat’s Possible are trademarks of Waters Corporation. Jobin Yvon is a trademark of Horiba Jobin Yvon Inc. All other trademarks are the property of their respective owners. ©2012 Waters Corporation. Produced in the U.S.A. June 2012 720001939EN IH-PDF

Waters Corporation 34 Maple Street Milford, MA 01757 U.S.A. T: 1 508 478 2000 F: 1 508 872 1990 www.waters.com