1. Pushing the envelope on separation speedPushing the envelope on separation speed::
“knowing* when to push it and when to back off”
An empirical study of precisely how fast HPLC can beAn empirical study of precisely how fast HPLC can be
performed while maintaining separation qualityperformed while maintaining separation quality
Mark J.Mark J. HaywardHayward
“Follow the red text!”
Mark J.Mark J. HaywardHayward
&& QingpingQingping HanHan
Lundbeck ResearchLundbeck Research USAUSA
ParamusParamus, NJ 07652, NJ 07652
Title credit to Chuck
Yeager: first person to
break sound barrier
*Investment in
knowledge is crucial!

2. It’s all about speed! (velocity)
To achieveTo achieve resolution in reverse phaseresolution in reverse phase
LC,LC, one generally must have retentionone generally must have retention
•• No retention = no separationNo retention = no separation
elutes in void volumeelutes in void volume
•• Must sweep multiple column volumes (k’)Must sweep multiple column volumes (k’)
k’ must be greater than 2k’ must be greater than 2
5 < k’ < 10 very often5 < k’ < 10 very often optimaloptimal (sometimes k’ = 20+ required)(sometimes k’ = 20+ required)5 < k’ < 10 very often5 < k’ < 10 very often optimaloptimal (sometimes k’ = 20+ required)(sometimes k’ = 20+ required)
•• Sufficient k’ produces resolutionSufficient k’ produces resolution,, much moremuch more
so than particle diameter or column lengthso than particle diameter or column length (fine tuning at most)(fine tuning at most)
ToTo get resolution fast one must sweepget resolution fast one must sweep
column volumes quicklycolumn volumes quickly
(k’/min(k’/min ≥≥≥≥≥≥≥≥ 5 whereas typically k’/min = 1)5 whereas typically k’/min = 1)
•• Shorten columnShorten column
•• IncreaseIncrease velocityvelocity (flow)(flow)
10x faster than “normal”10x faster than “normal”
10 mm/s = 0.0224 mi/hr10 mm/s = 0.0224 mi/hr
or Mach 2.9 xor Mach 2.9 x 1010--55
Not as risky as Yeager!Not as risky as Yeager!

3. Transformative philosophy
Knox and Saleem shaped the way we think of fast LC.
Predicted columns 1 cm in length with particles <1 m in
diameter yielding 5000 plate separations in 20 s.
This made perfect sense four decades ago when the separation
performance was primarily mass transfer limited because
particle diameters were greater than 50 microns.
Since the mass transfer and hence separation efficiency would
be expected to be inversely proportional to the square of
particle diameter, it was naturally assumed that as particle
diameter was decreased, column length would also would
reduced such that the flows and pressures would be fairlyreduced such that the flows and pressures would be fairly
similar to those already demonstrated or in use.
The result of this pursuit would be essentially the same
separations as before, but much faster due to the short
column, small particle combination.
However, despite more than four decades of significant effort,
Knox and Saleem’s prediction has not been achieved because
we have not been able to shorten the column to 1 cm without
paying a severe price in the separation as a result of the
separation efficiency not being mass transfer limited.
The separation speed still depends on mass transfer, but the
approach for increasing it may not be the same (trade-offs).
J.H. Knox, M. Saleem, J. Chromatogr. Sci. 1969, 7, 614.

4. Observations on gettingObservations on getting resolution fastresolution fast
(increasing k(increasing k’ / unit time)’ / unit time)
Shortening the columnShortening the column
•• Lowers pressureLowers pressure
•• Reduces resolutionReduces resolution
•• Lost resolution might be restored with smaller particlesLost resolution might be restored with smaller particles
•• Even with smaller particles, 50 mm seems to be anEven with smaller particles, 50 mm seems to be an
empirical minimumempirical minimum (subjective, but consensus acceptability)(subjective, but consensus acceptability)
Reducing particle diameterReducing particle diameter
•• Increases separation efficiency (resolution per unitIncreases separation efficiency (resolution per unit
column length), but not nearly as much as theoreticallycolumn length), but not nearly as much as theoreticallycolumn length), but not nearly as much as theoreticallycolumn length), but not nearly as much as theoretically
predictedpredicted (thus limits how much column can be shortened)(thus limits how much column can be shortened)
•• Pressure increases rapidlyPressure increases rapidly ((αα 1/d1/dpp
22)) andand even with neweven with new
high pressure pumps, 1high pressure pumps, 1--22 µµmm seems to be an empiricalseems to be an empirical
minimumminimum (nothing smaller demonstrated)(nothing smaller demonstrated)
Increasing velocityIncreasing velocity (and k’ per unit time)(and k’ per unit time)
•• Increases pressureIncreases pressure ((∆∆∆∆∆∆∆∆PP αααααααα F eg. Darcy’s LawF eg. Darcy’s Law approximatelyapproximately))
•• Practical limits on pressurePractical limits on pressure (50(50--75% rating)75% rating) & flow& flow (100%(100%
rating)rating) for the pump are primary limitersfor the pump are primary limiters ofof velocityvelocity
•• Autosampler performance limits reduction in column IDAutosampler performance limits reduction in column ID
as a means to increasing velocityas a means to increasing velocity (maintain “infinite diameter”)(maintain “infinite diameter”)
•• Must eventually increase temperature in order toMust eventually increase temperature in order to
operate under optimized conditionsoperate under optimized conditions

5. Practical limitationsPractical limitations of Knox approachof Knox approach
Injection process limits the benefitsInjection process limits the benefits
of smaller particlesof smaller particles ((σσσσ2
injection process >> σσσσ2
column C-term))
Smaller particles only slightlySmaller particles only slightly
increase optimum velocity and higherincrease optimum velocity and higherincrease optimum velocity and higherincrease optimum velocity and higher
temperature must be used if velocitytemperature must be used if velocity
is to be tripled or more whileis to be tripled or more while
maintaining optimal conditionsmaintaining optimal conditions
Injection process limits the ability toInjection process limits the ability to
reduce column diameterreduce column diameter (“infinite diameter”)(“infinite diameter”)

6. BackgroundBackground –– Variance,Variance, AKA peak widthAKA peak width
σσσσ2
observed = σσσσ2
injection process (80%) + σσσσ2
column (20%) + σσσσ2
extra-column (<1%)
σσσσ2
injection process is as much as 80% of σσσσ2
observed in an otherwise
optimized LC system (assumes “good” column).* The
minimum contribution from the column appears to be ≈≈≈≈ 20%.
Once σσσσ2
column becomes relatively small, it is difficult to
positively influence σσσσ2
observed with the column (particularly C-term)!
“The contribution of the sampling device is particularly“The contribution of the sampling device is particularly
deleterious since, for a 2 µµµµL injection, the maximum solute
concentration in the peak that enters into the column is nearly
ten-fold lower than that of the sample.” **
Sample is generally observed to be diluted 20Sample is generally observed to be diluted 20 –– 50 fold upon50 fold upon
injection!injection!
σσσσ2
extra-column can readily be made negligible.
The ultimate speed and separation efficiency in LC is not
limited by mass transfer efficiency in the column (i.e. not
limited by dp).* Injection process is the limiting factor!
*F. Gritti, A. Felinger, G. Guiochon, J. Chromatogr. A, 2006, 1136, 57.

7. Practical consequences of beingPractical consequences of being
injection process limitedinjection process limited (top down view):(top down view):
Practical consequences of beingPractical consequences of being
injection process limitedinjection process limited (top down view):(top down view):
Expected improvements with particle size reductionExpected improvements with particle size reduction
level off* (pressure can even slow separationlevel off* (pressure can even slow separation**).**).
*J. Kofman, Y. Zhao, T. Maloney, T. Baumgartner, R. Bujalski, Am. Drug Discovery 2006, 1, 12.
**T.L. Chester, S.O. Terami, J. Chromatogr. A, 2005, 1096, 16.
Peak Width as a Function of Particle
Size for Reserpine
Peak Width as a Function of Particle
Size for Met-Enkephalin
Minimum σσσσ2 exiting column slightly larger (+20%) than σσσσ2 entering column
(HPLC or UPLC, by connecting UV to inj valve)
Best half dozen columns all yield about the same performance (C18 Luna and Sunfire shown).
Velocity = 7 mm/s, T = 45°C, L = 50 mm, column diameter = 4.6 mm HPLC & 2.1 mm UPLC.
0
0.5
1
1.5
0 2 4 6 8 10
Particle Diameter (um)
PeakWidth(s)
HPLC Measured
UPLC Measured
Theory (C term)
0
0 .5
1
1.5
0 2 4 6 8 10
Particle Diameter (um)
PeakWidth(s)
HPLC Measured
UPLC Measured
Theory (C term)

8. Practical consequences of beingPractical consequences of being
injection process limitedinjection process limited
(bottom up view allows us to see what’s happening)(bottom up view allows us to see what’s happening)::
Practical consequences of beingPractical consequences of being
injection process limitedinjection process limited
(bottom up view allows us to see what’s happening)(bottom up view allows us to see what’s happening)::
In carefully executed experiments using UPLC, starting with the smallest particlesIn carefully executed experiments using UPLC, starting with the smallest particles
and working upward in diameter, a small particle diameter effect can be seen.and working upward in diameter, a small particle diameter effect can be seen.
However, the effect appears to be A term onlyHowever, the effect appears to be A term only (column dependent multi(column dependent multi--path dispersion)path dispersion)..
In cases where σσσσ2
improves with sub 3
µµµµm particles, the
benefits are small,
CANNOT offset the
Variance as a Function of Particle Diameter @ 7 mm/s
0.14
VarianceforReserpine(seconds
Not mass
transfer
15% reduction
in peak width
costs 420%
increase in CANNOT offset the
rapidly increasing
pressure, & DO NOT
result in peaks
narrower than may
be achieved with
other 3 µµµµm particles
(with smaller A term).
Result: lower velocities!
Must have C term
impact on Van Deemter
equation to improve
speed with ↓↓↓↓ dp.
(σσσσ2
EC ∝∝∝∝ F)0.02
0.04
0.06
0.08
0.1
0.12
1 3 5
Particle Diameter (micrometer)
VarianceforReserpine(seconds
squared)
Theoretical C term
Measured BEH on UPLC
Theoretical A term
Currently, readily
achievable levels
using 3 –3.5 µµµµm
with HPLC or
UPLC. Minimum
σσσσ2 exiting column
> σσσσ2 entering
column.
transfer
limited!*
No evidence
of resistance
to mass
transfer
(C term)
Multi-path dispersion
likely primary contributor
to σσσσ2 in dp trend of 2.1
mm columns (A term).
*F. Gritti, A. Felinger, G. Guiochon, J. Chromatogr. A, 2006, 1136, 57.
increase in
pressure
(3.5 to 1.7 µµµµm
transition)

9. Particles are not enough:Particles are not enough:
Must use temperature too!Must use temperature too!
Van Deemter Curves at Two
Different Particle Sizes
(3 & 5 micron diameters shown)
Variance(peakwidth/2)2
Van Deemter Curves at Two Different
Temperatures
Variance(~PeakWidth)
Room Temp
(20C)
Elevated Temp5 µm
0 2 4 6
Velocity (flow rate in mL/min)
Variance(peakwidth/2)
0 2 4 6
Velocity (~Flow mL/min)
Variance(~PeakWidth)
Elevated Temp
(30C)
Optimum velocity is proportional to 1/particle diameter
Optimum velocity is proportional to e-k/RT
5 µm
3 µm

10. Ideal column diameter – depends on performance of injector.
Well known in literature, see: L.R. Synder & J.J. Kirkland, “Introduction to Modern
Liquid Chromatography,” 1979, 2nd Ed., John Wiley & Sons: New York (left figure)
“Infinite Diameter Effect” or dispersion at column wall
Peak Width vs. Column Diameter for
Met-Enkephalin at Constant Velocity and
Retention Time
Practical consequences of beingPractical consequences of being
injection process limited:injection process limited:
Practical consequences of beingPractical consequences of being
injection process limited:injection process limited:
3 decades ago
0 1 2 3 4 5 6 7
Column Diameter (mm)
Variance(~peakwidth)
4.6 mm ID looks like way to
go (ordinary HPLC).
These curves can be flattened below 1 mm diameter by using direct on-column syringe injection.*
*Henry, R.A., in Modern Practice of Liquid Chromatography, J.J. Kirkland ed., Wiley-Interscience: New York, 1971.
Multi-path dispersion can
become a primary
contributor to σσσσ2 when
HPLC column diameter is
reduced (1 µµµµL injection).Now, using
Waters Alliance
2795
UPLC moves flat portion down to 2 mm

11. Operation under “infinite diameter” conditionsOperation under “infinite diameter” conditions
gives best separation efficiency. Reducinggives best separation efficiency. Reducing
diameter below that significantly sacrificesdiameter below that significantly sacrifices
separation efficiency.separation efficiency.
Column diameter must be scaled to deliveredColumn diameter must be scaled to delivered
injection volume to get best separation efficiencyinjection volume to get best separation efficiency
and speed.and speed.
Delivered injection volume (2Delivered injection volume (2σσσσσσσσ) can be measured) can be measured
Practical consequences of beingPractical consequences of being
injection process limited:injection process limited:
Practical consequences of beingPractical consequences of being
injection process limited:injection process limited:
Delivered injection volume (2Delivered injection volume (2σσσσσσσσ) can be measured) can be measured
by connecting UV detector directly to injectionby connecting UV detector directly to injection
valve.valve.
Instrument (Instrument (autosamplerautosampler) choice is one way to) choice is one way to
reduce column diameter for improved sensitivityreduce column diameter for improved sensitivity
without sacrificing separation efficiency.without sacrificing separation efficiency.
Key volume / column diameterKey volume / column diameter to maintainto maintain
efficiencyefficiency::
•• 22σσσσσσσσ ≈≈≈≈≈≈≈≈ 5050 µµµµµµµµLL col. dia. 4col. dia. 4 –– 6 mm6 mm (ordinary HPLC)(ordinary HPLC)
•• 22σσσσσσσσ ≈≈≈≈≈≈≈≈ 1010 µµµµµµµµLL col. dia. 1.5col. dia. 1.5 –– 2.1 mm2.1 mm (example: UPLC)(example: UPLC)
•• 22σσσσσσσσ ≈≈≈≈≈≈≈≈ 0.20.2 µµµµµµµµLL col. dia. 0.2col. dia. 0.2 –– 0.3 mm0.3 mm (example:(example: EksigentEksigent Express)Express)

12. Peptide mix separated using
different column inside diameters
4.6 mm ID - Waters Alliance
2.1 mm ID - Waters UPLC
0.3 mm ID - Eksigent Express
Instrument choice as a solution toInstrument choice as a solution to
being injection process limited:being injection process limited:
Nearly equal gradient performance possible at column IDNearly equal gradient performance possible at column ID ≥≥≥≥≥≥≥≥ 0.3 mm0.3 mm
Instrument choice as a solution toInstrument choice as a solution to
being injection process limited:being injection process limited:
Nearly equal gradient performance possible at column IDNearly equal gradient performance possible at column ID ≥≥≥≥≥≥≥≥ 0.3 mm0.3 mm
Sample: HPLC peptide
mix - Sigma H-2016
(different lots).
Stationary phase: Inertsil
ODS3, 3 µµµµm, 50 mm
length (different lots).
Sample injection
volumes: 1000, 500, &
0 0.2 0.4 0.6 0.8 1
Retention Time (min)
Intensity
volumes: 1000, 500, &
150 nL respectively.
Mobile phase: buffer (0.2
% HOAc) & ACN ramped
from 1 to 30% in 1 min.
Instruments and
scientists were all
different in different labs.
Smaller ID yieldsyields higher
sensitivity αααα 1/(column ID)2
Sensitivity makes extra
effort worthwhile.
Velocity = 7 mm/s (i.e. flows = 5000, 1000, & 20 µµµµL/min respectively).

13. GoalGoal: find the edge of the: find the edge of the envelopeenvelope
(in a systematic way)(in a systematic way)
For each of the following, find the limit producingFor each of the following, find the limit producing
thethe highesthighest velocityvelocity without sacrificing resolution:without sacrificing resolution:
•• PressurePressure ∆∆PP (Make / model LC pump)(Make / model LC pump)
•• Flow rateFlow rate (Make / model LC pump)(Make / model LC pump)
•• Particle diameter dParticle diameter dpp
•• Scale (HPLC / UPLC)Scale (HPLC / UPLC)
•• Stationary phase (silica, polymer, hybrid)Stationary phase (silica, polymer, hybrid)
•• Separation type (isocratic, gradient,Separation type (isocratic, gradient, eluenteluent choice)choice)•• Separation type (isocratic, gradient,Separation type (isocratic, gradient, eluenteluent choice)choice)
Assume 50 mm column lengthAssume 50 mm column length
Assume “good” column (i.e. silica with negligibleAssume “good” column (i.e. silica with negligible
secondary interactions driving separation)secondary interactions driving separation)
Assume injection is HPLC (Assume injection is HPLC (22σσinjectioninjection ≈≈ 5050 µµL, fixedL, fixed
full loop w/overfillfull loop w/overfill) or UPLC () or UPLC (22σσinjectioninjection ≈≈ 1010 µµL,L,
partial loop, heart cut etc.partial loop, heart cut etc.) and treat separately) and treat separately
Assume gradient reverse phase withAssume gradient reverse phase with acetonitrileacetonitrile
(fastest), then eventually compare with isocratic,(fastest), then eventually compare with isocratic,
other solvents, and polymer stationary phasesother solvents, and polymer stationary phases

14. Pressure as a function of speed for a
range of particle diameters
500
600
Pressure(bar)
2 micron
Practical
pressure
limit for
Agilent 1200
Shimadzu XR
Plotting the speed limits of HPLC
(ordinary fixed full loop injection, stars indicate additional limits
imposed by 5 ml/min pump flow limit and column ID)
= 4.6mm ID = 4.0mm ID = 3.0mm ID (constrained by 5 mL/min and 60°C)
0
100
200
300
400
0 5 10 15 20 25 30
Eluent velocity (mm/s)
Pressure(bar)
2.5 micron
3 micron
4 micron
5 micron
Practical
pressure
limit for
ordinary
HPLC
Practical
temperature
limit (60°C) for
classic silica
based columns
BEH only
3 mm ID
requires repacking
column & 10 mL/min

15. Highlights of the speed limitHighlights of the speed limit
plots for ordinary HPLCplots for ordinary HPLC
Essentially any HPLC can be operated at 7 mm/s (fast)Essentially any HPLC can be operated at 7 mm/s (fast)
with 4.6 x 50mm xwith 4.6 x 50mm x 33--3.53.5µµµµµµµµm columns (widely available)m columns (widely available)
Maximum velocity can be increased to near 10 mm/sMaximum velocity can be increased to near 10 mm/s
by using 4 mm ID columns with minimal impact onby using 4 mm ID columns with minimal impact on
separationseparation (however fewer columns are available in 4 mm ID)(however fewer columns are available in 4 mm ID)
Maximum velocity can be further increased to 15 mm/sMaximum velocity can be further increased to 15 mm/s
by using 3 mm ID columns but a noticeable impact onby using 3 mm ID columns but a noticeable impact onby using 3 mm ID columns but a noticeable impact onby using 3 mm ID columns but a noticeable impact on
resolution seems likelyresolution seems likely (however,(however, eveneven fewer columns arefewer columns are
available in 3 mm ID)available in 3 mm ID)
55 µµµµµµµµm BEH 3 mm ID column could be used to achieve 25m BEH 3 mm ID column could be used to achieve 25
mm/s but with loss of resolutionmm/s but with loss of resolution (might still be useful)(might still be useful)
Sub 3Sub 3 µµµµµµµµm particles will limit velocity with ordinarym particles will limit velocity with ordinary
HPLCs (350HPLCs (350--400 bar pressure limits)400 bar pressure limits)
Enhanced pressure LCs (Agilent 1200Enhanced pressure LCs (Agilent 1200 –– 600 bar limit)600 bar limit)
can achieve high velocities with sub 3can achieve high velocities with sub 3µµµµµµµµm particles butm particles but
velocity begins to be limited as particle diametervelocity begins to be limited as particle diameter
approaches 2approaches 2 µµµµµµµµmm

16. Plotting the speed limits of UHPLC
(partial loop injection, stars indicate additional limits imposed by 1
or 2 ml/min pump flow limit [Accela or Acquity] and column ID)
= 2.1mm ID = 1.5mm ID = 1.0mm ID (constrained by 2 mL/min and 60°C)
Pressure as a function of speed for
a range of particle diameters
800
1000
Pressure(bar)
1.7 micron
Practical
pressure
limit for
Accela or
Acquity
1 mL/min
1.0mm ID
0
200
400
600
800
0 10 20 30 40 50
Eluent velocity (mm/s)
Pressure(bar)
1.7 micron
2 micron
2.5 micron
3 micron
3.5 micron
Practical
temperature
limit (60°C) for
classic silica
based columns
BEH only
i.e. >60°C
1.0mm ID
1.3 mL/min
Acquity only

17. Highlights of the speed limitHighlights of the speed limit
plots forplots for UHPLCUHPLC
Limited instrument hardware (pump &Limited instrument hardware (pump & autosamplerautosampler), flow), flow
rate, and column choices (1 or 2 ml/min max) yields fewerrate, and column choices (1 or 2 ml/min max) yields fewer
optionsoptions
22 µµµµµµµµm and smaller particles limits maximum velocitym and smaller particles limits maximum velocity
Use of 1.7Use of 1.7 µµµµµµµµm particles limits maximum velocity to less thanm particles limits maximum velocity to less than
10 mm/s10 mm/s
Pressure only gets you 2.5x improvement in velocityPressure only gets you 2.5x improvement in velocity
Maximum velocity can be further increased by using 1.5 andMaximum velocity can be further increased by using 1.5 and
1.0 mm ID columns but a noticeable impact on resolution (1.0 mm ID columns but a noticeable impact on resolution (infinf
diadia) is observed (bigger cross sectional steps) is observed (bigger cross sectional steps –– 2x each)2x each)
Very high velocity (40 mm/s) can be achieved, but there areVery high velocity (40 mm/s) can be achieved, but there are
very, very few columns that can be optimized at thesevery, very few columns that can be optimized at these
velocities (hybrid 3.5velocities (hybrid 3.5 µµµµµµµµm particles only)m particles only)
Difference in peak width between 1.7 and 3.5Difference in peak width between 1.7 and 3.5 µµµµµµµµm BEHm BEH
particles is 15% and difference in pressure is 420%.particles is 15% and difference in pressure is 420%.
Maximum velocity for silica based columns (15 mm/s) isMaximum velocity for silica based columns (15 mm/s) is
easily achieved for most any column with 2.5easily achieved for most any column with 2.5 µµµµµµµµm or largerm or larger
particlesparticles

18. If you have an Agilent 1050/1100/1200,If you have an Agilent 1050/1100/1200, Waters HWaters H--UPLC,UPLC, MetaloxMetalox,,
oror SeleritySelerity column oven, or other setup with preheating thencolumn oven, or other setup with preheating then
you are ready! Just use the suppliedyou are ready! Just use the supplied preheaterpreheater..
Otherwise, all you need is a mobile phaseOtherwise, all you need is a mobile phase preheaterpreheater
(inexpensive).(inexpensive). Examples include:Examples include:
––SeleritySelerity CaloraThermCaloraTherm (0.2(0.2--1010 mLmL/min)/min) activeactive onon--demand heating.demand heating.
––AgileSLEEVEAgileSLEEVE ((≤≤≤≤≤≤≤≤11 mLmL/min L=40 cm tubing req.)/min L=40 cm tubing req.) a compact tube oven.a compact tube oven.
TemperatureTemperature: How to control: How to control TT??
**Ordinary air ovens are totally insufficient!Ordinary air ovens are totally insufficient!
∆∆∆∆∆∆∆∆TT ≥≥≥≥≥≥≥≥ 1212 C for the entire column throughout the gradient is common @ high pressure.**C for the entire column throughout the gradient is common @ high pressure.**
Even water baths are not enough!**Even water baths are not enough!**
––AgileSLEEVEAgileSLEEVE ((≤≤≤≤≤≤≤≤11 mLmL/min L=40 cm tubing req.)/min L=40 cm tubing req.) a compact tube oven.a compact tube oven.
––JJ--KEM Sci. PrepKEM Sci. Prep (20(20--150150 mLmL/min)/min) active but high mass dampenedactive but high mass dampened..
––Once you haveOnce you have activeactive eluenteluent preheating, simple air based columnpreheating, simple air based column
heating can keep the column steel at the desiredheating can keep the column steel at the desired TT.***.***
∆∆∆∆∆∆∆∆TT ≤≤≤≤≤≤≤≤ 44 C for the entire column throughout the fastC for the entire column throughout the fast
gradientgradient is achievable on all scales (prep too).is achievable on all scales (prep too).
SmallerSmaller ∆∆∆∆∆∆∆∆TT correlates well with narrower peaks!correlates well with narrower peaks!
[Also, lower steel mass often correlates with smaller[Also, lower steel mass often correlates with smaller ∆∆∆∆∆∆∆∆TT.].]
*R.J. Perchalski, B.J. Wilder, Anal. Chem. 1979, 51, 774.
*R.G. Wolcott et al., J. Chromatogr. A 2000, 869, 211.
**A. de Villiers, H. Lauer, R. Szucs, S. Goodall; P. Sandra, J. Chromatogr. A 2006, 1113, 84.
***B.A. Jones, J. Liq. Chromatogr. Related Technol. 2004, 27, 1331.
Selerity
CaloraTherm
Pre-Heater
makes temperature
control easy!

19. Velocity vs. temperature for:Velocity vs. temperature for:
elution mode and column choiceelution mode and column choice
Optimal Velocity vs. Temperature
135
160
SeparationTemperature(C)
Fit gradient ACN - silica and BEH
Observed gradient ACN - silica and BEH
Isocratic ACN silica - TN/Guiochon etal
Iso (& grad) MeOH silica - TN/Guiochon etal
BEH or
polymer
only
(particle
stability)
TN-MeOH
F. Gritti, G.
Guiochon,
Anal. Chem.
2006, 78,
5329.
TN-ACN
F. Gritti, A.
Felinger, G.
10
35
60
85
110
0 5 10 15 20 25 30
Optimum Eluent Velocity (mm/s)
SeparationTemperature(C)
Iso (& grad) MeOH silica - TN/Guiochon etal
Iso (& grad) ACN polymer - MN/Carr etal
Range where both
analyte and silica
stability are well
established
BEH or silica with
reduced H2O
content only
(not polymer)
stability)
Limited stability for silica
MeOH Gradient ACN
Felinger, G.
Guiochon, J.
Chromatogr.
A 2006,
1136, 57.
MN-ACN
B. Yan, J.
Zhao, J.S.
Brown, J.
Blackwell,
P.W. Carr,
Anal. Chem.
2000, 72,
1253.

20. Optimal temperature as a function of
particle diameter at 7 mm/s
52
54
Temperature(C) Measuring T effects in isolationMeasuring T effects in isolation
Optimum T as a function of dp
It’s all about mass transfer; when you need more, T helps!
•Optimum T is
independent of dp when
the separation is not
mass transfer limited
(dp = 1.7-3.5 µµµµm).
•As dp increases
(dp = 5-8 µµµµm) and
separation becomes
partially mass transfer
limited, optimum T
44
46
48
50
1 3 5 7 9
Particle diameter (um)
Temperature(C)
Inertsil ODS3
X-Bridge BEH
SunFire
Luna (2)
Fit line
limited, optimum T
increases with dp.
•2σσσσobserved still <1 s for
dp = 5 & 8 µµµµm.
•T may be alternative to
smaller dp as T is a route
to at least partially
achieve added mass
transfer (from diffusion)
ordinarily gained from
smaller dp.
Observed phenomena @ dp ≤≤≤≤3.5 µµµµm consistent with injection process limited.
Observed phenomena @ dp ≥≥≥≥5 µµµµm consistent with: υυυυoptimum = (B/C)1/2 ∝∝∝∝ D/dp.

21. Temperature offers much greater ability than otherTemperature offers much greater ability than other
techniquestechniques (particle diameter)(particle diameter) to achieve higher velocitiesto achieve higher velocities
Temperature is a way to partially achieve the benefitsTemperature is a way to partially achieve the benefits
of smaller particlesof smaller particles
•• Most mass transfer restoredMost mass transfer restored
•• Doesn’t address A term (multipath dispersion)Doesn’t address A term (multipath dispersion)
Choice of particle type matters!Choice of particle type matters!
•• Polymer phases slow velocity way down presumably due toPolymer phases slow velocity way down presumably due to
Highlights of temperature effect plotsHighlights of temperature effect plots
•• Polymer phases slow velocity way down presumably due toPolymer phases slow velocity way down presumably due to
surface/film penetration/diffusion (even in gradient mode)surface/film penetration/diffusion (even in gradient mode)
•• Ordinary silica limited by stability (solubility) toOrdinary silica limited by stability (solubility) to ≤≤60ºC or60ºC or ≤≤ 1515
mm/s (very fast with ACN gradientmm/s (very fast with ACN gradient –– 1010--15x ”normal”)15x ”normal”)
•• BEH stability allows even higher velocities (30 mm/sBEH stability allows even higher velocities (30 mm/s
demonstrated) due to increased temperature stabilitydemonstrated) due to increased temperature stability
Gradients with acetonitrile are truly fast!Gradients with acetonitrile are truly fast!
•• Isocratic operation limits improvement to about 3x ”normal”Isocratic operation limits improvement to about 3x ”normal”
•• Other solvents (alcohols) behave like isocratic operation evenOther solvents (alcohols) behave like isocratic operation even
in gradient modein gradient mode (ACN gradients appear both unique & crucial)(ACN gradients appear both unique & crucial)

22. Flow matters too!Flow matters too!
Highest speed is usually achieved atHighest speed is usually achieved at
maximum flowmaximum flow
Slower flow results in larger gradientSlower flow results in larger gradient
delaysdelays
UHPLCUHPLC with 1.7 microm particleswith 1.7 microm particlesUHPLCUHPLC with 1.7 microm particleswith 1.7 microm particles
yields a 15 s gradient delay atyields a 15 s gradient delay at
beginning of chromatogram and nobeginning of chromatogram and no
retained peaks (or usefulretained peaks (or useful
chromatogram before RT = 15 s)chromatogram before RT = 15 s)
Next slide shows minimization of theNext slide shows minimization of the
delay (3delay (3--4 s)4 s)

23. Results – Scalable speed (velocity) increases
while maintaining resolution (totally routine)
Peptide mixture at 7 mm/s, 45C and <250 bar
0
0.01
0.02
0.03
0.04
Intensity(AUat275nm)
Waters Alliance HPLCWaters Alliance HPLC
7 mm/s (t7 mm/s (too = 0.12 min)= 0.12 min)
4.6 x 50 mm, d4.6 x 50 mm, dpp = 3= 3 µµmm
Flow 5Flow 5 mLmL/min/min (200 bar peak)(200 bar peak)
Average 2Average 2σσ = 0.62 s= 0.62 s0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (min)
Peptide mixture at 14 mm/s, 60C and <500 bar
0
0.01
0.02
0.03
0.04
0 0.1 0.2 0.3 0.4
Time (min)
Intensity(AUat275nm)
ACN 1-30% gradient, Inertsil ODS3
Average 2Average 2σσ = 0.62 s= 0.62 s
Peptide mix (test mix for evaluating 2 hrPeptide mix (test mix for evaluating 2 hr
proteomics gradient separations)proteomics gradient separations)
WatersWaters AcquityAcquity UPLCUPLC
14 mm/s (t14 mm/s (too = 0.06 min)= 0.06 min)
2.1 x 50 mm, d2.1 x 50 mm, dpp = 3= 3 µµmm
Flow 2Flow 2 mLmL/min/min (400 bar peak)(400 bar peak)
Average 2Average 2σσ = 0.32 s= 0.32 s

24. Fast Open Access is totally robustFast Open Access is totally robust
HP 1100 / VGHP 1100 / VG--Platform (Platform (1515 year oldyear old system)system)
Time
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
AU
5.0
1.0e+1
1.416e+1
Range: 1.368e+1
3: UV Detector: 240_400
(1)
1.13
(3)
1.40
Time
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
AU
5.0
1.0e+1
4.0
4.109
Range: 3.463
3: UV Detector: 240_400
(3)
1.13
(5)
4.0
Same sample
A random
process
development
chemistry
sample
DAD chromatograms
k’ = 17
in
2 min
Time
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
AU
2.0
(5)
1.40
Time
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
AU
2.0
Same sample
2 weeks later
(re-diluted)
+/- ion
mass
spectra
30k samples
per year and
99+% uptime
demonstrated
for LC/MS with
UV (DAD) and
ELSDm/z
200.00 300.00 400.00 500.00 600.00 700.00 800.00 900.00
%
0
100
1:MSES+
1.1e+004
Combine (65:69-(49:52+82:84))
281.1
251.5237.7
283.5
354.8391.2 563.7552.8512.5471.9
m/z
%
0
100
2:MSES-
1.6e+003
Combine (64:69-(49:51+82:84))
279.3
178.7 233.3 339.4283.4
621.5393.6
+/- ion
mass
spectra

25. You can push your ordinary HPLCYou can push your ordinary HPLC
furtherfurther (requires cleaner samples as you approach pressure limits)(requires cleaner samples as you approach pressure limits)
Data for
Waters
Alliance
T = 60°C
Injection
and other
software
overhead
takes longer
than
separation!
(even in inject
ahead mode)

26. Speed vs. particle diameter
a comparison using UPLC with Column Manager and PDA
Selerity
CaloraTherm
Pre-Heater
makes
temperature
control easy!
50°C
A random lead optimization compound in solubility assay
6.00E-02
8.00E-02
1.00E-01
1.20E-01
1.40E-01
AU
Close up
3.00E-03
5.00E-03
7.00E-03
9.00E-03
BEH
ODS3
Higher resolution while simultaneously higher velocity
(k’/unit time) got there faster (3 µµµµm vs. 1.7 µµµµm).
Increased quality and speed at the same time with >2µµµµm
particles!
50°C
-2.00E-02
0.00E+00
2.00E-02
4.00E-02
6.00E-02
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (min)
AU
BEH C18 - 0.8 mL/min
Inertsil ODS3 - 1.5 mL/min
-1.00E-03
1.00E-03
0.1 0.2 0.3 0.4 0.5 0.6

27. ConclusionsConclusions
Ultra fast separations can be readilyUltra fast separations can be readily
achieved onachieved on anyany HPLC systemHPLC system
The keys are:The keys are:
•• Matching column diameter to autosamplerMatching column diameter to autosampler
performanceperformance
•• Choosing the right stationary phase (there areChoosing the right stationary phase (there are
at least a half dozen choices that can exceedat least a half dozen choices that can exceed
the autosampler performance)the autosampler performance)the autosampler performance)the autosampler performance)
•• Tuning temperature (use plot) to matchTuning temperature (use plot) to match
velocityvelocity
•• Using acetonitrile gradients! (CRUCIAL)Using acetonitrile gradients! (CRUCIAL)
The highest velocity while maintainingThe highest velocity while maintaining
resolution occurs with 3resolution occurs with 3--3.53.5 µµm particlesm particles
regardlessregardless of the instrumentof the instrument
The techniques described here have beenThe techniques described here have been
shown to be robustshown to be robust (7 mm/s)(7 mm/s) in >10in >1066 analysesanalyses

28. Conclusions UHPLCConclusions UHPLC
The techniques described here have been shownThe techniques described here have been shown
to be robust (14 mm/s) in >to be robust (14 mm/s) in >101055 analysesanalyses
These many analyses have been performed usingThese many analyses have been performed using
33--3.53.5 µµm silica Cm silica C88 and Cand C1818 particles from severalparticles from several
manufacturersmanufacturers
The column life times were ”normal” (2000The column life times were ”normal” (2000
injections) and in the best case 12000 injectionsinjections) and in the best case 12000 injections
(without deterioration of peak(without deterioration of peak symmetry/widthsymmetry/width))
Even higher velocities (30 mm/s) have beenEven higher velocities (30 mm/s) have beenEven higher velocities (30 mm/s) have beenEven higher velocities (30 mm/s) have been
studiedstudied using BEH columns to further understandusing BEH columns to further understand
the temperature velocity relationshipthe temperature velocity relationship
Velocities >15 mm/sVelocities >15 mm/s may have limited impactmay have limited impact
with current instruments becausewith current instruments because thethe injection +injection +
SWSW overheadoverhead can take as long as the separation,can take as long as the separation,
but that may be changingbut that may be changing (ex: Shimadzu(ex: Shimadzu XR)XR)
The autosampler seems to be the primary limitThe autosampler seems to be the primary limit
on extracting further benefit from the column,on extracting further benefit from the column,
but that may bebut that may be changingchanging (broader industry focus)(broader industry focus)

29. Outcome milestonesOutcome milestones
PProjects withrojects with 50000 samples50000 samples
completed in 2 monthscompleted in 2 months (6 instruments / scientist achieved)(6 instruments / scientist achieved)
Sustained group output >4000Sustained group output >4000
chromatograms per month for each /chromatograms per month for each /
all chromatographersall chromatographers (need, not capacity(need, not capacity limitedlimited))all chromatographersall chromatographers
•• PhysicoPhysico--chemical assayschemical assays (AI and formulations)(AI and formulations)
•• Stability samplesStability samples
•• ID and purityID and purity
•• Purification (100 ml/min fully scalable)Purification (100 ml/min fully scalable)
•• Biomarker analysisBiomarker analysis
•• ADME assaysADME assays