Journal of Chromatography A, 1217 (2010) 45614567

Contents lists available at ScienceDirect

Journal of Chromatography A
journal homepage: www.elsevier.com/locate/chroma

Instrumental considerations for the effective operation of short, highly efcient

fused-core columns. Investigation of performance at high ow rates and elevated
temperatures
David V. McCalley
Centre for Research in Biomedicine, University of the West of England, Frenchay, Bristol BS16 1QY, UK

a r t i c l e

i n f o

Article history:
Received 29 January 2010
Received in revised form 16 April 2010
Accepted 23 April 2010
Available online 29 April 2010
Keywords:
Fused-core columns
Supercially porous columns
Efciency
Dead volume

a b s t r a c t
Fused core or supercially porous columns offer the advantages of higher efciency compared with totally
porous columns of the same particle size, but similar operating pressures. However, their performance
may be adversely affected by extra-column effects that become more signicant as the column efciency
increases, and as the diameter of the column is reduced. In this study, we show that 10 cm 0.46 cm
fused-core columns can be used on modied conventional instruments (microbore systems) without
serious loss in performance, and this approach is at present likely to yield superior results compared with
use of 0.21 cm columns (of identical efciency) on current UHPLC instruments that have minimised extracolumn volume. Furthermore, the true efciency of commercial narrow-bore fused-core columns appears
to be reduced compared with those of conventional bore, which may be due to packing difculties for
the former type. The fused-core columns in general gave excellent performance, showing no evidence
of an upturn in the Knox plots at high ow velocities and elevated temperatures. Careful control of
experimental conditions is necessary to ensure accurate data for these plots.
2010 Elsevier B.V. All rights reserved.

1. Introduction
Supercially porous or fused-core columns are a recent development of the original pellicular particles used in HPLC. In these
particles, a porous outer layer surrounds an impervious core.
The original pellicular particles typically consisted of porous layers of 12 m thickness, surrounding a solid non-porous core,
with overall particle diameter 30 m [1]. Such particles show
faster mass transfer, due to the smaller diffusion distance through
the porous layer, compared with (large) porous particles of the
same diameter, but suffered from rather poor sample capacity,
as less than 20% of the particle was porous. Later, particles of
much smaller overall diameter (5 m) with a 0.25 m thick porous
layer containing 30 nm pores were developed for the separation
of macromolecules, to take advantage of the smaller diffusion
distances for these low diffusivity solutes [2]. The efciency of
C18 columns for small molecules was reported as similar to that
of totally porous phases. Newer C18 particles of overall diameter 2.7 m, with porous layer 0.5 m thick, showed minimum
reduced plate heights (hmin ) as low as 1.5, a result replicated for
the equivalent bare silica hydrophilic interaction chromatography
(HILIC) phase [3,4]. In comparison, totally porous particles gener-

Tel.: +44 117 3282469; fax: +44 117 3282904.

E-mail address: David.Mccalley@uwe.ac.uk.
0021-9673/$ see front matter 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.chroma.2010.04.070

ally show 2.0 < hmin < 2.5. For the new particles, the porous layer
represents 75% of the volume of a totally porous particle, thus
the loading capacity is expected to be only slightly reduced. This
result was conrmed experimentally for the HILIC column [4]. It
was proposed that the high efciency of these particles might
be due to their lower internal porosity (smaller B term) in van
Deemter/Knox type equations and narrower particle size distribution and better packing (smaller eddy diffusion or A term), rather
than the smaller diffusion distance and improved mass transfer
(smaller C term) as originally suggested, at least for small molecules
[3,57].
An apparent disadvantage of these columns was the unusual
rate of increase of the plate height at high linear velocities, particularly at elevated temperature (up to 50 C) as demonstrated for the
test solute naphtho[2,3-a]pyrene [5]. Comparison of the results for
an excluded protein (bovine serum albumin) suggested that this
behaviour was not related to mass transfer kinetics in the stationary phase, but to some unexpected variation of the eddy dispersion
with the linear velocity at high temperatures, that might be connected with the roughness of the external surface of the fused-core
particles. The authors suggested therefore that if the separation
of low molecular weight compounds is performed at high velocities, the benet of using such columns might vanish. However, our
own study of similar bare silica fused-core HILIC columns did not
show any evidence of this effect at high ow rate, using similar
temperatures [4].

4562

D.V. McCalley / J. Chromatogr. A 1217 (2010) 45614567

The examination of column efciency as a function of ow rate is

clearly of much importance in the characterisation of new column
materials. However, the production of accurate van Deemter/Knox
curves is fraught with potential difculties. These include:
(a) The variable effect of instrumental dead volume on the measured efciencies. The detrimental effect will increase as the
efciency of the column increases (at a given retention factor,
k) and with decreasing k, as both decrease the volume containing the peak. The inuence of the dead volume also increases
with decreasing column internal diameter.
(b) The data collection rate may be insufcient to assess accurately
the shape of fast peaks that emerge at high ow rates.
(c) The column may be damaged by the higher pressures and ows
necessary to construct the plots.
(d) The column inuent temperature may vary at high ow rate due
to insufcient pre-heating of the mobile phase. A colder inuent
reaches the oven temperature at the walls more quickly than at
the centre, giving a radial temperature gradient that can cause
peak broadening-an effect that was rst described more than
25 years ago [811].
(e) Column frictional heating may lead to axial and radial temperature gradients [5,12]. Radial gradients (with the column core
hotter than the column walls) cause the mobile phase to travel
faster in the centre than at the walls, leading to band broadening effects. Such effects depend on many factors, including
the type of column thermostat (e.g. still air, forced air, or water
bath), but can occur even for columns packed with relatively
large (5 m) particles [13].
Dorsey and co-workers stressed that consideration of such
effects was essential to the production of meaningful plate height
vs ow curves [14]. In the light of these potential problems, and
the important consequences of the conclusions of the previous
investigation [5], we investigated further the behaviour of the same
fused-core column under similar experimental conditions, taking
note as far as possible of the potential difculties discussed above.
We were also interested in examining the effect of extra-column
bandspreading and the performance of columns of conventional
(4.6 mm) and smaller internal diameter (2.1 mm). We used both a
modied conventional HPLC instrument (microbore HPLC system) and an ultra-high pressure system (UHPLC), with the aim
of determining the inuence of extra-column effects on these
high performance columns. We employed broadly similar conditions to the previous study [5], using naphtho [2,3-a] pyrene
selected apparently for its relatively low diffusion coefcient and
high hydrophobicity, enabling the attainment of high values of the
reduced mobile phase velocity. However, it should be noted that
using high concentrations of organic modier does not inevitably
result in high reduced velocities, as the effect of decreased viscosity
of the mobile phase (allowing higher velocity) may be counteracted
by the increase in the solute diffusion coefcient.
2. Experimental
The microbore HPLC was a model 1100 binary high pressure
mixing HPLC system (Agilent, Waldbronn, Germany) with Chemstation, UV detector (1 l ow cell), and Rheodyne 7725 valve
(5 l injections). Connections were made with minimum lengths
of 0.012 cm I.D. tubing to minimise extra-column band spreading.
Temperature was regulated using the Agilent column thermostat.
The fastest data collection rate available on this instrument (14 Hz,
response time <0.12 s) was invariably used. The ultra-high-pressure
liquid chromatograph (UHPLC) was a UPLC system from Waters
(Milford, MA, USA) including Empower data handling, binary sol-

vent manager, photodiode array (UV) detector (500 nl ow cell),

and sample manager/injector valve (1 l injections from partially
lled 5 l loop with needle overll) and was used for all experiments with 0.21 cm I.D. columns. The use of 1 l injections with
the 0.21 cm column and 5 l injections from the 0.46 cm column
should result in similar detection sensitivity and column loading properties. Data collection rates were either 80 Hz (fast lter,
0.0125 s) or 20 Hz (normal lter, 0.2 s) as specied. Duplicate sample injections were made. For construction of van Deemter/Knox
plots, one data set was obtained increasing the ow rate gradually
to the maximum value, followed by a repeat data set as the ow
rate was decreased back to its minimum value. Such an approach
allows a check to be made for column damage resulting from use
at the highest ow and pressure. The columns used were Halo C18
(particle diameter 2.7 m, pore size 80 , 10 cm 0.46 cm I.D. and
10 cm 0.21 cm I.D.), from AMT (Wilmington, USA) and Discovery
C18 (particle diameter 5 m, pore diameter 180 , 15 cm 0.46 cm
I.D.) from Supelco (Bellafonte, USA).
Acetonitrile (far UV grade) was obtained from Fisher Scientic (Loughborough, UK). Naphtho [2,3-a] pyrene and uracil were
obtained from SigmaAldrich (Poole, U.K.).
System bandspreading was measured by replacing the column
with a zero dead volume connector.  was calculated from measurements of the peak width at 4.4% of peak height (5 peak width)
to take account of peak tailing typically shown for such injections.
The diffusion coefcients of naphtho[2,3-a]pyrene, as estimated from the WilkeChang equation, were taken as 1.29 105 ,
1.23 105 and 1.52 105 cm2 /s in pure ACN at 25 C, ACNwater
(90:10, v/v) at 37.5 C, and ACNwater (85:15, v/v) at 50 C, respectively [5]. Due to the low ow rates and low proportion of water
used with columns of 0.21 cm diameter, solvents were pre-mixed
when using the UHPLC system to avoid the delivery of very small
volumes by one of the pumps. For example, a ow of 0.05 ml/min of
acetonitrilewater (90:10, v/v) requires one pump to deliver accurately only 5 l/min, which we found experimentally to be beyond
the capabilities of our system.
3. Results and discussion
3.1. Determination of bandspreading of microbore and UHPLC
systems
The experimentally measured variance of a chromatographic
peak is the sum of the column and extra-column variance, the latter
in turn being the sum of the variance of the injector, detector, connecting tubing and time-based events such as sampling rate and
detector time constant [15,16].
2
2
2
exptl.
= col.
+ excol.

Extra-column dispersion can be calculated in any convenient

units of time, volume or column length. We prefer the measurement in volume units as employed by many other workers [1719].
Clearly, it is best to use the actual compound employed for the van
Deemter/Knox curve measurements to probe extra-column effects,
but we obtained additional data with uracil. There is no particular
reason to choose uracil, commonly used as an unretained marker in
RP separations, as no column is in place during the measurements.
However, it is a compound often found in analytical laboratories,
and serves therefore as a useful basis for comparative studies.
At the optimum ow of the 0.21 cm fused-core column (around
0.30.35 ml/min, see below) Fig. 1(a) shows the bandspreading of
the UHPLC system was 33.5 l2 for naphthopyrene, with slightly
higher values for uracil, based on an injection volume of 1 l. All
values were measured using acetonitrilewater (90:10, v/v) and
37.5 C, these being the median values of these parameters used in

D.V. McCalley / J. Chromatogr. A 1217 (2010) 45614567

4563

Fig. 1. System bandspreading measured for the two instruments using different probes/conditions. (a) Effect of different mobile phases and solutes for the UHPLC system.
(b) Effect of data gathering parameters on the bandspreading for naphthopyrene for the UHPLC system. (c) Comparison of bandspreading for naphthopyrene and uracil on
the microbore system.

the subsequent van Deemter/Knox plots (see below). The results are
comparable with those reported by Neue and co-workers, who have
given a detailed description of the complex factors that contribute
to bandspreading, concluding that the value of the individual contributions is difcult to predict [18]. These factors were reported
to include the dispersion in the injector and detector, the contribution of bandspreading in the tubing (which can be estimated using
the TaylorAris equation), and the time-based contribution to peak
broadening originating in the detector data collection rate and in
lters used. The tubing used between the injector and detector is
rather short in both HPLC and UHPLC applications. The TaylorAris
equation breaks down at low solute residence times. The authors
considered also an alternative treatment on the basis of work by
Golay, nally proposing a mathematical description of broadening
in the tubing using a random walk computer model. According to
Neue [18], the typical maximum as in Fig. 1(a) can be attributed
to the onset of secondary ow or turbulence, possibly caused by
the roughness of the internal surface of the tubing. Such an effect

would cause mixing of the individual ow streams, limiting the

effects of bandspreading. As the values of bandspreading depend on
the diffusion coefcients of the solutes, there should also be variation according to the mobile phase composition, particularly at
low ow rates where the TaylorAris treatment holds [18]. Indeed,
Fig. 1(a) shows larger values of the bandspreading for uracil on the
UHPLC system when measured in 10% ACN rather than 90% ACN.
These results emphasise the need for estimating the system bandspreading under the same conditions as used for the subsequent
analysis.
Fig. 1(b) shows the effect of different detector data collection
parameters 80 Hz (0.0125 s lter) and 20 Hz (0.2 s lter) on  2 for
naphthopyrene, using the UHPLC system. As the ow rate increases
above 0.5 ml/min, the values of  2 are increasingly divergent, with
 2 increasing almost linearly with ow rate for slower data collection compared with the plateau achieved for fast collection. Note
that 20 Hz itself is a faster data collection rate than that of the
microbore system, however, the lter rate was 0.2 s compared with

4564

D.V. McCalley / J. Chromatogr. A 1217 (2010) 45614567

Fig. 2. Estimation of efciency loss due to extra-column bandspreading for fusedcore columns of different dimensions, on the two HPLC systems, as a function of
retention factor. AG = microbore system; AQ = UHPLC system. Calculations based on
bandspreading values determined for naphthopyrene in ACNwater (90:10, v/v)
at 37.5 C. Upper diagram 10 cm columns with 25000 plates. Lower diagram 5cm
columns with 12500 plates.

<0.12 s used on the microbore system. Indeed, 20 Hz and 0.05 s lter

produced considerably reduced bandspreading on the UHPLC system. As the volumetric bandspreading contribution of the UHPLC
system is smaller than that of the microbore system, the detector
time-based contributions are likely to be more critical in the former
instrument.
Fig. 1(c) shows the bandspreading measured for naphthopyrene
on the microbore system at 1.5 ml/min (in the region of its optimum ow rate determined subsequently, see below), was around
25 l2 using 5 l injections. As before, a slightly higher value
(27.5 l2 ) was obtained for uracil. Bandspreading was reduced if
1 l injections were made, but the larger injections were used to
render the results compatible in terms of absolute sensitivity and
column loading with 1 l injections on the 0.21 cm I.D. column.
Fig. 1(c) shows more or less constant values of  2 as a function
of ow rate between about 0.25 and 2 ml/min. However, at ows
above 2 ml/min,  2 increased, which can be attributed to the insufcient data collection rate of the microbore system (14 Hz), that
fails to describe accurately the increasingly fast peaks emerging
from the column at high ow rate.
3.2. Predicted effect of system bandspreading on efciency of
naphthopyrene as a function of ow
Fig. 2 shows the predicted % loss in efciency of columns of
length 10 and 5 cm and true efciency (in the absence of extracolumn effects) 25,000 plates and 12,500 plates respectively, for a
0.46 cm column using the microbore HPLC, and 0.21 and 0.46 cm
columns on the UHPLC system, as a function of retention factor.
It is likely in practice that columns will be operated in the region

of their optimum ow rate, generating the maximum efciency. At

this point, the effects of extra-column band broadening are likely to
be the most serious. The spreadsheet calculations were performed
using  2 = 25 l2 for the microbore system and 3.5 l2 for the
UHPLC system, being the values corresponding to the bandspreading at the optimum ows for the 0.46 and 0.21 cm column respectively using naphthopyrene. Note that while  2 increases somewhat above 25 l2 for the microbore system above 2.5 ml/min, the
true (theoretical) column efciency declines due to mass transfer
band broadening processes, giving a concomitant reduced effect
of extra-column effects on performance. Calculations showed that
these effects of increased instrument bandspreading but reduced
(true) column efciency virtually cancelled each other out at high
ow rate on this system, verifying the approach used. Considering
Fig. 2(a) for the 10 cm column, the curve for the 0.46 cm column/microbore instrument indicates considerably less loss in efciency than for the 0.21 cm column/UHPLC instrument. For example, the loss in efciency at k = 2 is only 10% for the 0.46 cm column
but above 20% for the 0.21 cm column. At k > 2, losses in efciency
decrease, becoming very small for both systems at k = 5. However,
at k = 1, the losses are quite serious at around 20% and almost 40% for
each system, respectively. For the 5 cm columns, the losses are more
serious as the peak volumes are considerably reduced and therefore extra-column volumes become more signicant. For example
at k = 1, the losses in efciency were around 33% and 55% for the 0.46
and 0.21 cm columns, respectively. Clearly, the losses in efciency
are minimised (although still not negligible for low k) using the
0.46 cm column operated on the UHPLC instrument (calculations in
this case based again on  2 = 3.5 l2 for a 1 l injection). However,
this column/instrument combination could not be used to obtain
data at high ow rate, due to the maximum 2 ml/min available on
the UHPLC (Waters Acquity) used in these experiments.
These results have important consequences for the experimentally measured performance of columns as a function of ow rate.
Clearly, in all experimental measurements of column efciency and
plate height, the experimental value is always that of the column
plus the effect of the instrument (extra-column effects). It is possible of course to correct for the extra-column contribution, and
clearly such data treatment is necessary for a theoretical understanding of column performance, especially when the instrumental
contribution is signicant, as with the 0.21 cm fused-core columns
(see below). However, such an approach conceals the results that
are achievable practically with current instruments. A further difculty is that simple measurements of extra-column dispersion
cannot easily account for effects such as column frictional heating,
so these corrections may not be entirely accurate. To reduce the
effects of dead volume on results that are achievable practically,
longer columns are preferable, although the maximum pressure
available, and frictional heating are both factors to consider as
well as other factors such as the desired analysis time. Use of
experimental conditions to generate high solute k are also recommended. For these relatively short columns, high k at least does
not necessarily indicate prohibitively long analysis times. However,
clearly extra-column effects are detrimental to fast analysis when
a lower k might be preferred. System bandspreading values can
be greater for different solutes, and as shown in Section 3.1 are
also likely to be higher in mobile phases with a higher aqueous
content than (the unusually low) values used in these experiments, giving more serious losses in efciency than suggested by
Fig. 2.
The results have more general signicance beyond fused-core
columns e.g. for the operation of sub-2 m particle columns, which
generate similar efciency to fused-core columns. These smaller
particle columns are often used in shorter lengths due to the high
back pressures they generate, in which case instrumental bandspreading will be more signicant.

D.V. McCalley / J. Chromatogr. A 1217 (2010) 45614567

Fig. 3. Effect of ow on h for fused-core columns of different dimensions on different

instruments. The data for the 0.46 cm column was uncorrected for the (negligible) extra-column bandspreading. Mobile phase ACNwater (90:10, v/v), column
temperature 37.5 C. The reduced interstitial velocities correspond to volumetric
ow rates of 0.050.8 and 0.254.0 ml/min for the 0.21 and 0.46 cm I.D. columns,
respectively.

3.3. Measurement of Knox plots for 0.46 cm and 0.21 cm

fused-core columns on different systems
Table 1 shows the total instrument backpressure and the net column pressure for the 10 cm 0.46 cm fused-core column at a ow
rate of 1 ml/min for the various mobile phases used in the present
and previous studies [5]. The net column backpressure corrected for
the extra-column contributions is about 60 bar in each case at this
ow rate. Extrapolation of this gure would indicate a net column
backpressure of 360 bar for a 15 cm column operated at 4 ml/min,
the highest ow employed in the previous study. For our microbore system, a 15 cm column operated at 4 ml/min would generate
a total system backpressure of around 480 bar, in excess of the maximum operating pressure of our system. In any case, operating the
column at such high pressures could lead to greater frictional heating effects. With these considerations in mind, and the small losses
in efciency calculated for a 10 cm 0.46 cm column at the high k
given by naphthopyrene, we decided to use 10 cm columns for all
experiments.
Fig. 3 shows Knox plots for naphthopyrene using 0.46 and
0.21 cm fused-core columns on the microbore and UHPLC systems.
The plots are of the reduced plate height h
h=

H
dp

(1)

against the reduced interstitial velocity red,int

red,int =

Fdp
r 2 ext Dm

(2)

where F is the volumetric ow rate, dp the particle diameter, r

the column radius, ext the external porosity (determined previously as 0.43 for this column [5]) and Dm the diffusion coefcient
of the solute in the mobile phase. Highly symmetrical peaks were
obtained for naphthopyrene, with asymmetry factors very close to
1.0. Fig. 3 shows the curve for the 0.46 cm column on the microbore system was the most favourable, with values of the reduced
plate height as low as 1.5, similar to those obtained in HILIC [4] for
the bare silica version of this phase. (h = 1.5 corresponds to 25,000
plates on the 10 cm column). The retention factor of naphthopyrene was 5.4 in the mobile phase used (90% ACN at 37.5 C, see
Table 2) and Fig. 2 shows that the loss of efciency for a column

4565

with 25,000 plates is only around 2.5% at such high retention, a negligible gure. Fig. 3 shows also a partial Knox curve for the 0.46 cm
column on the UHPLC, corresponding to volumetric ow rates of
0.52.0 cm3 /min, the maximum ow rate that can be used with
our instrument. Surprisingly, the curve lies mostly slightly above
that for the same column on the microbore instrument, despite
the smaller bandspreading of the UHPLC, especially at the higher
ow rates. This result might be attributable to some small incompatibility in the temperature of the inuent to that of the column
temperature at high ows on this instrument, due to less effective
solvent pre-heating. In contrast to the UHPLC system, the microbore
system uses a serpentine capillary embedded in a massive heating block; this pre-heating system is reasonably effective, although
some cooling can take place at high set temperatures in the connecting capillary between the heating block and the column, due
to cold spots in the oven [13]. Apparently, the very small advantage
of the lower instrumental bandspreading of the UHPLC system has
no signicance at this high value of k (see Fig. 2). However, for
peaks of lower k, use of the 0.46 cm column on the UHPLC system is likely to generate the highest efciency, subject to the pump
being able to generate sufcient ow to achieve the minimum of
the Knox curve. The curve for the 0.21 cm fused-core column on the
UHPLC instrument lies clearly above that for the 0.46 cm column
on the microbore system. The predicted loss of efciency for such
a column with 25,000 plates on this instrument is still only around
6%, due to the high k of the solute. Correcting the reduced plate
height for the effects of the system bandspreading still shows less
favourable results for this 0.21 cm column (a maximum of around
20,900 plates for the true efciency) compared with the 0.46 cm
column. This result could be attributed to the greater difculties in
packing of the smaller diameter fused-core column [20]. A similar
reduction in efciency of narrower bore compared with 0.46 cm
fused-core columns (after correction for instrumental effects) was
also shown for those from a different manufacturer (Kinetex, Phenomenex, Torrance, USA, results not shown [21]). Table 2 shows
ts of these plots to the Knox equation:

h = A0.333 +

B
+ C


(3)

The tting coefcients reported for the 0.21 cm fused-core column were calculated from the data corrected for the extra-column
(instrumental) contribution to plate height, whereas for the 0.46 cm
columns, the data were uncorrected as the extra-column contributions are extremely small. The plot for the 0.21 cm column shows
A = 0.50, considerably higher than for the 0.46 cm column (A = 0.34),
indicating that differences in the packing could be responsible. Nevertheless, there is clearly some convergence of the curves for the
0.46 and 0.21 cm columns in the C term region of the plots at high
reduced velocity (see Fig. 3). Table 2 also indicates C = 0.034 for
the 0.21 cm column compared with 0.046 for the 0.46 cm fusedcore column. It is possible that this result could be explained
by greater frictional heating effects in the larger bore column,
as there is a greater difference in temperature between the core
and the walls in such columns, as heat is less easily conducted
away.
In conclusion, the combination of the 10 cm 0.46 cm column and the microbore system gave the optimum results.
In combination with the high k for naphthopyrene, this
combination also yields values of column efciency that do
not require correction for extra-column dispersion. Clearly
however, the contribution of extra-column effects needs
careful consideration for each individual experimental situation.

4566

D.V. McCalley / J. Chromatogr. A 1217 (2010) 45614567

Table 1
Backpressure measurements using different mobile phases at a ow of 1 cm3 /min for a 10 cm 0.46 cm fused core and 15 cm 0.46 cm totally porous columns, microbore
system.
Column

Mobile phase

Total system pressure (bar)

Extra-column pressure (bar)

Column pressure (bar)

10 cm 0.46 cm
fused
core

100% ACN, 25 C
90% ACN, 37.5 C
85% ACN, 50 C

78
82
81

19
22
23

59
60
58

15 cm 0.46 cm totally porous

90% ACN, 37.5 C

40

22

18

Table 2
Fits of plots of reduced plate height for naphtho[2,3-a]pyrene against reduced interstitial velocity to the Knox equation.
Column/dimensions

Instrument

Mobile phase

10 cm 0.46 cm
fused
core
2.7 m
10 cm 0.21 cm fused core 2.7 m
15 cm 0.46 cm totally porous 5 m

Microbore

100% ACN, 25 C
90% ACN, 37.5 C
85% ACN, 50 C

3.2
5.4
5.7

0.33
0.34
0.38

2.9
3.8
3.7

0.067
0.046
0.041

UHPLC
Microbore

90% ACN, 37.5 C

90% ACN, 37.5 C

4.5
3.3

0.50
0.41

4.2
7.2

0.034
0.038

The data for the 0.21 cm I.D. column was corrected for extra-column effects, whereas the data for the 0.46 cm columns is uncorrected, due to the very small contribution of
extra-column effects under the experimental conditions.

Fig. 4. Effect of ow on h for 10 cm 0.46 cm fused-core column with the microbore

system and different mobile phases/temperatures.

3.4. Comparison of Knox plots for the 0.46 cm fused-core column

at different temperatures
Fig. 4 shows the Knox plots for the 0.46 cm column at 25, 37.5
and 50 C, as used previously [5] with 100, 90 and 85% ACN, respectively. The ACN concentration was reduced as the temperature was
raised presumably to keep k approximately constant. Use of the
reduced (interstitial) velocity means that curves for different temperatures should overlay exactly, in the absence of unusual effects,
as the variation of the diffusion coefcient with temperature is
taken into account. Within experimental error, this indeed appears
to be the case. Small errors could be caused by inaccuracies in the
estimation of Dm using the WilkeChang equation, especially in the
light of reports that this is indeed likely in mobile phases of high
ACN content [22]. Variations in the temperature of the oven from
the set temperature, and variable frictional heating effects should
also be considered [13]. The results however, contrast with those
previously reported [5]. No strong upturn is shown in the curve at
high mobile phase velocity, and good performance at high ow rate
is indicated. For example, values of h exceeding 5 were previously
reported for a reduced interstitial velocity of 20 at 37.5 C in 90%
ACN, whereas a value barely above 2 is shown in the present work.
Fig. 3 also shows no upturn in the curve for the 0.21 cm column, and

Fig. 5. Effect of ow on h for 15 cm 0.46 cm totally porous (5 m) particle column.

Mobile phase ACNwater (90:10, v/v), column temperature 37.5 C. The reduced
interstitial velocities correspond to volumetric ow rates of 0.254.0 ml/min.

neither was the upturn noted for a bare silica fused-core column in
HILIC [4]. A recent report of efciency measurements with fusedcore columns in which the instrumental variance was subtracted
from the observed peak variance also showed quite at plots of H vs
ow [23]. Nevertheless, we did not study higher molecular weight
solutes, whose performance might be affected by partial exclusion
from the relatively small pores of the fused-core column used.
3.5. Comparison of Knox plots for a totally porous particle column
Fig. 5 shows a similar Knox plot for a totally porous 5 m packing column run on the microbore instrument. The column shows
excellent efciency at the optimum ow velocity for a packing
of this type, yielding a minimum value of h = 1.95, corresponding to >100,000 plates/m. Table 2 indicates a slightly higher value
of the C coefcient on the 0.46 cm fused-core column run under
the same conditions (37.5 C, 90% ACN) compared with the porous
5 m phase (C = 0.046 and 0.038, respectively). Nevertheless, this
result could be attributed entirely to frictional heating effects in the
0.46 cm fused-core phase, as the value of C for the 0.21 mm fusedcore column is only 0.034. Clearly the backpressure on the 5 m
column was much lower (see Table 1) despite its longer length,

D.V. McCalley / J. Chromatogr. A 1217 (2010) 45614567

indicating much lower frictional heating. Considering the 0.46 cm

fused-core column, values of the A and B coefcients were lower
than the totally porous column. These results agree with previous
conclusions that the good packing properties and narrow particle
size distribution of the fused-core material lead to low eddy diffusion, while the low internal porosity of the packing leads to smaller
effective diffusivity of the sample [5,6]. For small molecules at least,
these considerations appear to be more important than improved
mass transfer in the stationary phase. However, over-interpretation
of these tting coefcients for different stationary phase materials
is inadvisable using our approach, as the coefcients clearly depend
on consistent measurement of particle size. Nevertheless, measurements of the backpressure of the totally porous column indicated
a particle size of 5.2 m, according to the equation
P =

1000FL
r 2 dp2

(4)

where F is the volumetric ow rate,  is the viscosity, L the column length, r the internal radius and dp the particle size of the
packing [19]. However, this expression is only approximate, and
could lead to errors in the estimated particle size. More importantly, the equations used to describe the dependence of h on ow
(the Knox treatment) may be too simplistic; for example, there are
better ways to calculate the B contribution [5,24].
4. Conclusions
A microbore LC and a UHPLC system were carefully characterised as to their effect on the determination of van Deemter/Knox
curves for highly efcient fused-core columns. Instrumental bandspreading effects can have serious effects on the measurement of
HETP. However, for the high k solute used in these studies, the
effects were minimised and negligible when using a large diameter 0.46 cm column together with a microbore LC and smaller
than for a 0.21 cm column used with a UHPLC system. Furthermore, it appears that at present it may not be possible to pack
narrower bore fused-core columns of as high (true) efciency as

4567

conventional bore columns. These results are unfortunate, in that

the use of narrower bore columns limits the detrimental effects of
frictional heat generation, and reduces solvent consumption. We
examined the performance of the fused-core columns at high ow
rate and elevated temperature, considering carefully the problems
that may be encountered in generating accurate van Deemter/Knox
curves. No unusual features of the van Deemter/Knox plots were
reported, indicating that these columns are eminently suitable for
fast analysis.
References
[1] L.R. Snyder, J.J. Kirkland, Introduction to Modern Liquid Chromatography, John
Wiley, New York, 1979.
[2] J.J. Kirkland, F.A. Truszkowski, C.H. Dilks, G.S. Engel, J. Chromatogr. A 890 (2000)
3.
[3] J.J. Kirkland, T.J. Langlois, J.J. DeStefano, Am. Lab. 39 (2007) 18.
[4] D.V. McCalley, J. Chromatogr. A 1193 (2008) 85.
[5] F. Gritti, G. Guiochon, J. Chromatogr. A 1169 (2007) 125.
[6] F. Gritti, A. Cavazzini, N. Marchetti, G. Guiochon, J. Chromatogr. A 1157 (2007)
289.
[7] J.J. Kirkland, J. Chromatogr. Sci. 46 (2008) 244.
[8] H. Poppe, J.C. Kraak, J. Chromatogr. 282 (1983) 399.
[9] J.D. Thompson, J.S. Brown, P.W. Carr, Anal. Chem. 73 (2001) 3340.
[10] G. Vanhoenacker, P. Sandra, J. Sep. Sci. 29 (2006) 1822.
[11] R.G. Wolcott, J.W. Dolan, L.R. Snyder, S.R. Bakalyar, M.A. Arnold, J.A. Nichols, J.
Chromatogr. A 869 (2000) 211.
[12] G. Desmet, J. Chromatogr. A 116 (2006) 89.
[13] M.M. Fallas, M.R. Hadley, D.V. McCalley, J. Chromatogr. A 1216 (2009) 3961.
[14] K.M. Usher, C.R. Simmons, J.G. Dorsey, J. Chromatogr. A 1200 (2008) 122.
[15] J.G. Attwood, M.J.E. Golay, J. Chromatogr. 218 (1981) 97.
[16] M.J.E. Golay, J.G. Attwood, J. Chromatogr. 186 (1979) 353.
[17] L.R. Snyder, J.J. Kirkland, J.W. Dolan, Introduction to Modern Liquid Chromatography, 3rd ed., Wiley, Hoboken, USA, 2010.
[18] K.J. Fountain, U.D. Neue, E.S. Grumbach, D.M. Diehl, J. Chromatogr. A 1216
(2009) 5979.
[19] U.D. Neue, HPLC Columns, Wiley-VCH, New York, 1997.
[20] J.J. Kirkland, AMT Technologies, personal communication, 2009.
[21] D.V. McCalley, unpublished observations.
[22] J. Li, P.W. Carr, Anal. Chem. 69 (1997) 2530.
[23] T.L. Chester, Am. Lab. 41 (2009) 11.
[24] K. Broeckhaven, D. Cabooter, F. Lynen, P. Sandra, G. Desmet, J. Chromatogr. A
1188 (2008) 189.