2918 Electrophoresis 2002, 23, 2918–2928

Anurag S. Rathore1 Joule heating in packed capillaries used in capillary

Kimberly J. Reynolds2
Luis A. Colón2
1
Technology Development,
Bioprocess and Formulation
Technology, Pharmacia Co.,
Chesterfield, MO, USA
2
Department of Chemistry,
University at Buffalo,
State University of New York,
Buffalo, NY, USA
electrochromatography
Effective heat dissipation is critical for reproducible and efficient separations in electri-
cally driven separation systems. Flow rate, retention kinetics, and analyte diffusion
rates are some of the characteristics that are affected by variation in the temperature
of the mobile phase inside the column. In this study, we examine the issue of Joule
heating in packed capillary columns used in capillary electrochromatography (CEC).
As almost all commonly used CEC packings are poor thermal conductors, it is
assumed that the packing particles do not conduct heat and heat transfer is solely
through the mobile phase flowing through the system. The electrical conductivity of
various mobile phases was measured at different temperatures by a conductivity
meter and the temperature coefficient for each mobile phase was calculated. This
was followed by measurement of the electrical current at several applied voltages to
calculate the conductivity of the solution within the column as a function of the applied
voltage. An overall increase in the conductivity is attributed to Joule heating within the
column, while a constant conductivity means good heat dissipation. A plot of conduc-
tivity versus applied voltage was used as the indicator of poor heat dissipation. Using
theories that have been proposed earlier for modeling of Joule heating effects in cap-
illary electrophoresis (CE), we estimated the temperature within CEC columns. Under
mobile and stationary phase conditions typically used in CEC, heat dissipation was
found to be not always efficient. Elevated temperatures within the columns in excess
of 237C above ambient temperature were calculated for packed columns, and about
357C for an open column, under a given set of conditions. The results agree with
recently published experimental findings with nuclear magnetic resonance (NMR)
thermometry, and Raman spectroscopic measurements.

Keywords: Capillary electrochromatography / Conductivity / Electrophoresis / Joule heating

EL 5075

1 Introduction ment of stationary-phase materials tailored for CEC per-

1.1 General aspects
Capillary electrochromatography (CEC) has gained popu-
larity among separation scientists due in part to its low
solvent and sample consumption, high mass sensitivity,
good selectivity, and high efficiencies, as well as its
amenability to use in conventional capillary electrophore-
sis (CE) equipment. Over the past decade, researchers
have focused on solving column technology difficulties
such as column preparation, frit fabrication, and develop-
Correspondence: Dr. Luis A. Colón, Department of Chemistry,
University at Buffalo, State University of New York, 578 NS Com-
plex, Buffalo, NY 14260-3000, USA
E-mail: lacolon@buffalo.edu
Fax: 1716-645-6963
Dr. Anurag S. Rathore, Technology Development, Bioprocess
and Formulation Technology, Pharmacia Corporation, GG3K,
700 Chesterfield Pkwy North, Chesterfield, MO, 63198, USA
E-mail: anurag.s.rathore@pharmacia.com
Fax: 1636-737-7281
formance, as it has been reviewed recently [1]. Attention
has also been directed toward fundamental understand-
ing of the parameters that control the electroosmotic flow
(EOF) [2–4]. Due to the dual nature of typical CEC col-
umns, that is, a segment which is filled with a packing
material, and an open-tube segment filled only with
mobile phase, two vastly different scenarios are present
within the same capillary. The material and architecture
of the packed bed mandate that the packed segment of
the column has its own characteristic zeta potential, linear
flow rate, and electrical conductivity, which are different
from that of the open segment. The differences in electri-
cal conductivity of the two sections lead to a nonlinear
distribution of the total applied field over the column. Hor-
váth and co-workers [2, 3] have presented equations suit-
able for determining the conductivity and electrical field
strength over the individual open and packed segments
of CEC columns. It has been suggested that these para-
meters can be used in conjunction with EOF data to yield
information about the packing structure and porosity of
the packed bed [5]. Conductivity measurements for open

 2002 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 0173-0835/02/1709–2918 $17.501.50/0
Electrophoresis 2002, 23, 2918–2928 Joule heating in CEC 2919

tubes have also been used to determine the surface con-
ductivity and zeta potential of the silica surface [6]. For the
sake of simplicity, “electrical conductivity” will be referred
to as just “conductivity” in this paper and “thermal con-
ductivity” will be mentioned in full.
developed theory to quantify temperature rise in CE [7].
We find that a temperature increase is higher in an open-
tube than in a packed column, under similar experimen-
tal conditions.

With regard to CEC columns, heat generation and tem- 1.2 Theory
perature profiles in the open and packed segments are
also expected to differ based on the different characteris- 1.2.1 Flow of ions through open and packed
tics of the two sections of the column [4]. These para- capillaries
meters affect EOF, retention kinetics, diffusion of analytes,
When an open tube with fixed charges at the tube wall is
and ultimately can affect the efficiency and reproducibility
filled with an electrolyte solution, an electrical double
of a separation. It is therefore important in CEC separa-
layer is formed [19–21]. The double layer contains a
tions to determine the extent of heat generation in both
higher concentration of counterions than the bulk solution
segments of the column, and the resulting variation in con-
and so to maintain electroneutrality, the bulk electrolyte
ductivities and electroosmotic mobility. Information per-
outside has the same amount of excess co-ions. If ionic
taining to heating effects within the column is most readily
conduction through the bulk electrolyte is the dominant
accessed through the study of the electrical current
mechanism of ionic migration, the conductivity of an elec-
through the capillary. Nonlinear relationships between cur-
trolyte-filled cylindrical capillary, s, is expressed as [3]
rent and potential drop across an open or packed capillary
are common in the literature of CE and CEC and several iL
sˆ (1)
attempts have been made to model this nonlinear behavior VA
[7–15]. In a recent report, flow-field dynamics studies in where i is the current flowing through a capillary of length,
open and packed segments of capillaries by direct motion

encoding of fluid molecules using pulsed magnetic field
gradient nuclear magnetic resonance (NMR) were per-
formed [16]. Others have used NMR thermometry [17]
and Raman spectroscopic measurements [18] for measur-
ing temperatures inside the capillaries in CE. Temperature
increases of up to 307C inside the capillaries have been
reported in these studies. Our objective in the work pre-
sented here is to examine the issue of Joule heating in
packed columns using solvent systems commonly used
CE and CEC
L, and cross-sectional area, A, when a potential drop,
V, is applied across the capillary. Equation (1) was used
to calculate conductivity for open tubes (e.g., Table 1).
The conductivity of an electrolyte solution is independent
of the dimensions of the capillary tube and can be
expressed in terms of the concentration charge and
mobility of the constituent ions as follows [3, 22]
X
sopen ˆ F 2 C z2j vj xj (2)
j

for CEC separations. We have experimentally measured where F is the Faraday constant, C is the molar con-
conductivities of the mobile phase and estimated the tem- centration of the buffer and zj, vj and xj are the valency,
perature within the CEC capillary column using previously mobility, and number of moles of jth ionic species per

Table 1. Calculation of the temperature coefficient of electrical conductivity, a, for different buffers
used in the experiments using the conductivity meter; buffer pH 8.0

Buffer Temperature Conductivity Temperature Average a

(7C) (mS/cm) coefficient of s (K–1) (K–1)
…s=sO † 1
aˆ
T TO

80% MeCN – 10 0.500

20% 100 mM Tris 20 0.650 0.030 0.025
(20 mM total Tris) 30 0.780 0.020
80% MeCN – 10 0.290
20% 50 mM Tris 20 0.370 0.028 0.024
(10 mM total Tris) 30 0.440 0.019
80% MeCN – 10 0.062
20% 10 mM Tris 20 0.077 0.024 0.022
(2 mM total Tris) 30 0.093 0.021
2920 A. S. Rathore et al. Electrophoresis 2002, 23, 2918–2928

mole of buffer. It follows from Eq. (2) that the conductivity
of sufficiently dilute electrolytes increases linearly with the
molar concentration of the electrolyte.
In absence of Joule heating and other nonlinear effects,
the conductivity of an ionic solution is a constant and so
current varies linearly with the applied potential drop and
the electric field strength according to the following
expression
 
sA
iˆ V ˆ …sA†E (3)
L Power dissipation in a capillary has been known to cause
where E is the electric field strength across the capillary.
In order to simplify the analysis of a column that is com-
pletely packed, it can be represented by a hypothetical
open tube [2]. This tube has an equivalent length, Le, that
is the same as the distance traveled by a neutral and inert
tracer traveling through the packed segment [2]. The
equivalent length of this hypothetical tube is an empirical
quantity which takes into account the porosity, tortuosity
and other characteristics of the packed column. An alter-
nate approach is to consider that the hypothetical tube
has the same length, L, as that of the packed column,
but has an equivalent conductivity, se, that accounts for
the geometrical characteristics of the packed column.
This case can be represented as follows
 
se Ae
iˆ V (4)
L
where, e is the column porosity and accounts for the
reduced free cross-sectional area in a packed column.
The conductivity (se) of this hypothetical tube is an
empirical quantity that needs to be calculated from
current-voltage measurements with the packed column.
Equation (4) was used to calculate the conductivity for
the case of packed columns (e.g., Tables 2 and 3).
1.2.2 Effect of Joule heating on flow of ions
heating of the electrolyte, a phenomenon often referred to
as Joule heating [7–15]. The amount of heat generated in
an electrical system can be simply calculated using the
following expression
V 2 t V 2 sAt
J…t† ˆ iVt ˆ i2 Rt ˆ ˆ (5)
R L
where, J is the heat generated (Joules) as a function of
time (seconds), t, when a potential drop is applied across
a capillary of resistance, R, area of cross-section, A, and
length, L. This heat is dissipated across the capillary wall
via natural or forced cooling. It follows from Eq. (5) that the
problem of Joule heating gets worse with increasing cap-
illary diameter, potential drop, and time. The effect of
time, however, is for a very short period, since once equi-
librium between heat generation and dissipation to the
surroundings has been established, time has no effect.

Table 2. Calculation of theoretical temperature inside a packed column at different voltages

Buffer V (kV) i (mA) s (mS/cm) DTref (K) k Bi0oA T (oC)

80% MeCN – 5 2.79 0.474 25.0

20% 100 mM 10 5.79 0.492 0.194 0.070 26.6
Tris (20 mM 15 9.16 0.519 0.437 0.105 28.9
total Tris) 20 13.48 0.572 0.777 0.139 32.5
25 19.21 0.652 1.215 0.174 38.1
30 26.01 0.736 1.749 0.209 0.061 47.1
80% MeCN – 5 1.17 0.199 25.0
20% 50 mM 10 2.40 0.204 0.082 0.044 26.1
Tris (10 mM 15 3.73 0.211 0.183 0.066 27.4
total Tris) 20 5.21 0.221 0.326 0.088 29.6
25 6.93 0.235 0.509 0.111 32.6
30 9.03 0.256 0.734 0.133 0.040 36.9
80% MeCN – 5 0.26 0.044 25.0
20% 10 mM 10 0.52 0.044 0.018 0.020 25.4
Tris (2 mM 15 0.79 0.045 0.040 0.030 26.0
total Tris) 20 1.07 0.045 0.072 0.040 26.7
25 1.37 0.047 0.112 0.050 27.8
30 1.69 0.048 0.162 0.060 0.022 29.1

Column, 15 cm675 mm; packing, 1 mm bare silica; column porosity, 0.4; buffer, pH 8.0; thermal con-
ductivity of solution of 0.61 W m–1K–1 and temperature coefficient of electrical conductivity, a, as
calculated in Table 1
Electrophoresis 2002, 23, 2918–2928 Joule heating in CEC 2921

Table 3. Calculation of theoretical temperature inside a packed column at different voltages

Buffer V (kV) i (mA) s (mS/cm) DTref (K) k Bi0oA T (oC)

80% MeCN – 5 2.22 0.378 25.0

20% 100 mM 10 4.62 0.392 0.155 0.062 26.5
Tris (20 mM 15 7.36 0.417 0.348 0.093 28.6
total Tris) 20 10.72 0.455 0.619 0.124 31.9
25 15.01 0.510 0.968 0.156 36.9
30 19.95 0.565 1.394 0.187 0.053 44.8
80% MeCN – 5 0.88 0.150 25.0
20% 50 mM 10 1.79 0.152 0.061 0.038 25.5
Tris (10 mM 15 2.74 0.155 0.138 0.058 26.2
total Tris) 20 3.72 0.158 0.245 0.077 27.1
25 4.72 0.160 0.383 0.096 28.4
30 5.93 0.168 0.552 0.115 0.061 30.1
80% MeCN – 5 0.41 0.069 25.0
20% 10 mM 10 0.83 0.070 0.028 0.025 25.2
Tris (2 mM 15 1.23 0.070 0.064 0.037 25.5
total Tris) 20 1.67 0.071 0.114 0.050 25.9
25 2.12 0.072 0.178 0.062 26.5
30 2.56 0.073 0.256 0.075 0.063 27.1

Column, 15 cm675 mm; packing, 0.5 mm bare silica; other conditions as in Table 2

In the presence of the EOF in an open capillary, ions can
be transported through convective as well as conductive
mechanisms. The convective mechanism can be ignored
if [11]
dueo rcp
<3 (6)
l ner:
where d is the capillary diameter, ueo is the EOF velocity,
and r, cp , and l are (respectively) the density, specific
heat capacity, and heat conductivity of the buffer solution.
For most CE and CEC systems, the left hand side in Eq. (6)
is on the order of 10–3 and so the condition in Eq. (6)
is always satisfied. For such systems, the temperature
profile in the capillary can be assumed axially uniform
through the capillary. These conclusions are also applic-
able to packed capillaries in CEC if the packing is non-
conductive and surface conductivity through the double
layer around the particles is negligible [2, 3]. However, as
Morris and co-workers [18] have reported, if there is a
temperature sink (such as a metal plate) in touch with the
capillary end, the resulting end effects could be appreci-
able.
1.2.3 Temperature increase due to Joule heating
Several attempts have been made to model the effects of
Joule heating and to calculate the resulting temperature
profile in a CE capillary [7–15]. After reviewing the litera-
ture, the simplest useful approach that we found to pre-
dict the temperature inside a capillary due to Joule heat-
ing effects was that proposed by Bello and Righetti et al.
[7]. This approach will be briefly described in this section
and will be used in our analysis of CEC packed capillaries.
The buffer conductivity (s) is generally understood to be
linearly dependent on temperature in the following man-
s ˆ so ‰1 ‡ a …T To †Š (7)
where T is the temperature of the buffer in the capillary, a
is the temperature coefficient of conductivity and so is the
conductivity measured at the reference temperature, To.
The Biot number is often used to characterize the heat
transfer across the capillary from the internal heated buf-
fer to the surrounding capillary wall and external coolant.
The Biot number reaches a maximum when heat is pro-
duced within the capillary at an equal or higher rate than
it can be dissipated. When the experiments are con-
ducted at the reference temperature, the Biot number
can be simply expressed as [7]:
k2
BioA ˆ (8)
2…1 so =s†
where k2 = aDTref is termed as the autothermal parameter.
Further, DTref is known as the characteristic temperature
difference and is defined as follows:
so AV 2
DT ref ˆ (9)
pwL2
2922 A. S. Rathore et al. Electrophoresis 2002, 23, 2918–2928

Table 4. Calculation of theoretical temperature inside an open capillary at different voltages

Buffer V (kV) i (mA) s (mS/cm) DTref (K) k Bi0oA T (oC)

80% MeCN – 5 4.96 0.674 25.0
20% 100 mM 10 10.59 0.719 0.173 0.064 27.2
Tris (20 mM 15 17.47 0.791 0.389 0.097 30.3
total Tris) 20 25.97 0.882 0.691 0.129 35.5
25 37.85 1.029 1.080 0.161 44.2
30 54.55 1.235 1.555 0.193 0.041 59.7
80% MeCN – 5 2.81 0.382 25.0
20% 50 mM 10 5.77 0.392 0.098 0.048 26.4
Tris (10 mM 15 9.15 0.414 0.220 0.073 28.2
total Tris)% 20 13.18 0.448 0.391 0.097 31.0
25 18.09 0.492 0.611 0.121 35.2
30 23.49 0.532 0.880 0.145 0.037 41.4
80% MeCN – 5 0.50 0.068 25.0
20% 10 mM 10 1.02 0.069 0.017 0.020 25.3
Tris (2 mM 15 1.54 0.070 0.039 0.031 25.8
total Tris)% 20 2.08 0.071 0.070 0.041 26.4
25 2.62 0.071 0.109 0.051 27.2
30 3.23 0.073 0.157 0.061 0.026 28.3

Capillary, 30 cm675 mm; buffer pH, 8.0; detection at 214 nm at 15 cm; thermal conductivity of solu-
tion, 0.61 W m–1K–1 and temperature coefficient of electrical conductivity, a, as calculated in Table 1

DT ref
where w is the thermal conductivity of the buffer solution T ˆ To ‡ (14)
2Bi0oA k 2
and can be taken as 0.61 Wm–1K–1 [7]. The temperature
inside the capillary at a particular voltage can then be pre- Equations (11)–(14) were used to estimate the tempera-
dicted using the following expression [7]: tures inside the packed columns (Tables 2 and 3).

DT ref
T ˆ To ‡ (10)
2Bi0oA k 2 2 Materials and methods
where the Biot number used, Bi’oA, is the average of
experimentally determined Biot numbers at 25 and 30 kV.
2.1 Materials and chemicals
Equations (7)–(10) were used to calculate temperature Fused-silica capillaries of 75 mm, 50 mm, and 12 mm ID
inside open tubes (e.g., Table 4). This method allows sim- were purchased from Polymicro Technologies (Phoenix,
ple calculation of the temperature inside the capillary AZ, USA). All capillaries were flushed with 0.1 M sodium
using the current-voltage measurements done at the hydroxide (Fisher, Fair Lawn, NJ, USA) and with water
reference temperature. If the measurements are done at before use. Open capillary columns 30 cm in total length
a different temperature, then the above expressions are were prepared with a detection window at 15 cm. To
different, requiring complicated relationships as described ensure that no coating material was in contact with the
in detail in the original reference [7]. buffer solutions during the experiments, 6 cm of poly-
imide was removed from both ends of each capillary.
For the case of packed columns that are used in CEC, it
Packing materials for the packed capillaries were 1 mm
was assumed that the packed segment can be replaced
bare silica material, donated by EIChrom (Darien, IL,
by a hypothetical open tube as described earlier [2, 3] and
USA), and 550 nm bare silica, fabricated in-house as
so Eqs. (7)–(10) can be rewritten as follows:
described elsewhere [23]. HPLC acetone was purchased
se ˆ se;o ‰1 ‡ a…T To †Š (11) from Mallinckrodt (Paris, KY, USA) and HPLC-grade
methanol from Spectrum Chemical (Gardena, CA, USA).
k2 HPLC-grade acetonitrile, naphthalene, tris(hydroxy-
BioA ˆ (12)
2…1 se;o =se † methyl)aminomethane (Tris), and tris(hydroxymethyl)ami-
nomethane hydrochloride (Tris-HCl) were obtained from
se;o AV 2 e
DT ref ˆ (13) Aldrich (Milwaukee, WI, USA). Water was purified through
pwL2
Electrophoresis 2002, 23, 2918–2928
a Milli-Q UV Plus water purification system fed from a
Milli-RO 10 Plus reversed osmosis system (Millipore,
Bedford, MA, USA). Aqueous portions of the mobile
phases were made in the following way: appropriate
volumes of equimolar aqueous Tris and Tris-HCl solutions
were combined to achieve Tris solutions having the
desired pH. Measured volumes of aqueous solution were
mixed with measured volumes of organic solvent; the
resulting solution was filtered through a 0.45 mm filter and
degassed by sonication for 10 min. Naphthalene solution
(1 mM) used for EOF marker in open tubes was prepared
by diluting a stock solution (500 mM in acetonitrile) of the
compound with the mobile phase. The electroseparation
system was constructed in-house; the inlet end of the
system consisted of a 30 kV power supply (Glassman
High Voltage, Whitehouse Station, NJ, USA) connected
to the inlet vial via platinum electrode. The capillary was
placed horizontally from the inlet vial to the grounded
outlet vial, and was passed through a variable wavelength
UV detector (Thermo Separation Products, San Jose,
CA, USA) for analyte detection at the optical window.
The detector was placed on its side to allow for align-
ment of the capillary in a horizontal position. The current
was measured by recording the voltage across a resistor
placed in series with the capillary column at the outlet
end. A plexiglass box enclosed the high-voltage end of
the system for safety. The signal from the UV detector
was fed into an A/D converter board (DT2804, Data
Translation, Marlboro, MA, USA) and analyzed by
GRAM386 for Chromatography (Galactic Industries
Salem, NH, USA).
2.2 Preparation of packed capillaries
Capillary columns, 40 cm in length, were flushed with
0.1 M aqueous sodium hydroxide followed by flushing
with water. Initial retaining frits were fabricated from 1 mm
bare silica by drawing a small plug of the material slurried
in water into the column and heating for 3 s. Once the
retaining frit was in place, the column was packed using
either a pressure or electrokinetic packing method. The
1 mm silica was packed using a pressure packing method
at 8000 psi with a particle concentration of 50 mg/mL in
acetone, and methanol as the packing solvent. An empty
2 mm ID stainless steel HPLC column was used as the
slurry reservoir. The 550 nm silica was packed using an
electrokinetic packing procedure [24] employing 30 kV,
with 200 mg/mL silica slurried in 4 mM phosphate in
methanol. After packing, all columns were flushed with
water on a pump at 8000 psi. During flushing, an outlet
frit was formed by threading the column through a hole
in a Nichrome wire fitted to a soldering gun. Heat was
applied for 3 s to form a frit at the end of the packed bed.
Joule heating in CEC 2923
A second (inlet) frit was formed by sliding the wire 15 cm
down the packed bed and again heating for 3 s. The
column was depressurized and the excess open tube
was cut off, leaving a 15 cm packed column.
2.3 Measurement of conductivity at different
temperatures
To obtain the temperature coefficient of conductivity, the
conductivity of each mobile phase studied was measured
as follows. Each mobile phase (100 mL) was placed in a
temperature bath at 10.07C and equilibrated for 30 min.
The conductivity was measured using a Cole-Parmer
conductivity meter equipped with a gold dip-type cell
(Vernon Hills, IL, USA). The cell was rinsed in triplicate
with the solution to be measured using a disposable
pipet, followed by immersion of the cell into the solution
for 30 s to read the conductivity of the solution. Subse-
quently, the temperature was raised to 20.07C and
30.07C and the procedure was repeated at each tempera-
ture. The measured values for the different mobile phases
used are listed in Table 1.
3 Results and discussion
3.1 Heat inside packed capillaries
CE and CEC are typically performed in capillary columns
,100 mm ID to dissipate Joule heating. It is a common
practice to graph current measurements versus applied
voltage (i.e., Ohm’s plots) in order to assess column heat-
ing. A positive deviation from linearity in such plots indi-
cates that Joule heating is significant in the system stud-
ied. We constructed Ohm’s plots for two packed col-
umns using phases typically used in CEC, mixtures of
acetonitrile and aqueous Tris buffer. For comparison pur-
poses, Ohm’s plots for three open tubes under identical
mobile phase conditions were also constructed. Figure 1
shows the Ohm’s plot for the two packed columns, using
mobile phases containing two different electrolyte con-
centrations, acetonitrile/100 mM Tris, pH 8 (80/20 v/v)
and acetonitrile/50 mM Tris, pH 8 (80/20 v/v). The positive
deviation from linearity is indicative of inadequate dissipa-
tion of Joule heating in these two systems. The deviation
is larger for the mobile phases containing the higher con-
centration of electrolyte. This behavior is similar to that
observed in open tubes as illustrated in Fig. 2. Good heat
dissipation is realized in open tubes with small inner dia-
meters (Fig. 2A). It is clear that using these typical CEC
mobile phase conditions, heat is not effectively dissi-
pated.
2924 A. S. Rathore et al. Electrophoresis 2002, 23, 2918–2928

Figure 1. Ohm’s plots for 75 mm ID packed capillaries,

capillary length in both cases was 15 cm, entirely packed,
with particle size as indicated. (A) 80% acetonitrile/20%
50 mM Tris; (B) 80% acetonitrile/20% 100 mM Tris at pH 8.

3.2 Temperature increase due to Joule heating

Table 1 presents the calculations of the temperature coef-

ficient of conductivity for an acetonitrile/water mobile
phase at four different electrolyte concentrations. It is
seen that the conductivity increases and that the average
coefficient calculated for the different electrolyte concen-
trations are very similar to each other; this is similar to
values reported in the literature for aqueous buffers [7].
Tables 2 and 3 show the temperature inside two packed
columns at different applied voltages according to Eq.
(14), using Eqs. (4) and (11)–(13). For comparison, Table 4
shows temperature calculations for an open tube under
similar experimental conditions, using Eqs. (1) and (7)–
(10). The data presented in these tables show that the cur-
rent and the conductivity increase with the potential drop
Figure 2. Ohm’s plots for (A) 12; (B) 50; and (C) 75 mm ID
capillaries. Tris buffers solution pH 8; capillary length in all
cases, 30 cm.
and so does the predicted temperature inside the capil-
lary. The predicted temperature is higher at higher elec-
trolyte concentrations, as Joule heating increases with
electrolyte concentration. One can also find a tempera-
ture increase of about 237C inside the packed capillaries
at the highest applied voltage for the mobile phase con-
taining the highest electrolyte concentration. In the case
Electrophoresis 2002, 23, 2918–2928 Joule heating in CEC 2925

of the open tube, the temperature increase inside the

column under the same mobile phase conditions is about
357C, even when the electric field strength in the open
tube is half of the one in the packed columns.

The heat generated inside the packed columns is less

than their open capillary counterpart at similar applied
electric field strengths. From Eq. (5), one can see that
the severity of Joule heating increases with increasing
ionic concentration (conductivity), potential drop, and the
cross-sectional area of the capillary. In a packed column,
the electric conductivity is through the thin channels
separated by the packing particles. The actual cross sec-
tion of the column contributing to the conduction is less
than that of an open tube, depending on the column
porosity. This is reflected by the lower temperatures
inside the packed columns as seen in Tables 2 and 3
compared to the open tubes (Table 4).

3.3 Conductivity-electric field strength

relationships

The effect of the applied voltage on conductivity can be

illustrated by means of conductivity plots using Eq. (4)
(i.e., conductivity as function of applied voltage). Since
conductivity remains constant in a solution, providing
that the temperature of the solution is fixed, conductivity
plots for systems where heating is effectively dissipated
will show a slope close to zero. An increase in tempera-
ture of the solution within the capillary is reflected as an
increase in the measured conductivity. Figure 3 shows
conductivity plots for the two packed columns for which
Ohm’s plots are presented in Fig. 1, at the two electrolyte
concentrations. Conductivity plots for the open tubes at
three different internal diameters in Fig. 4 show the 12 mm
ID column with the slope near zero as the voltage is
increased, indicating good heat dissipation. Having a dif-
ferent organic modifier in the mobile phase yields a similar
trend (see Fig. 5). The deviation from the slope of zero
observed for the 75 and 50 mm diameter columns are
indeed better appreciated in the conductivity plots than
the deviations seen in the Ohm’s plots. In Fig. 2B, for
example, the Ohm’s plot for the 50 mm ID capillary
appears to be close to a linear fit, while a deviation is
clearly noticed in the conductivity plots (see Fig. 4B). Cal-
culations based on the model developed by Righetti and
co-workers [7] indicate that the temperature within the
capillary could rise from 257C to 377C. However, a user
could easily conclude from examining the Ohm’s plot
that there is negligible heating in the system.
In Fig. 6, electroosmotic mobility for the above-mentioned
case is plotted as a function of the applied field. Using the
50 mm ID capillary and mobile phases containing 20% of
Figure 3. Conductivity plots for 75 mm ID packed capil-
laries using (A) 80% acetonitrile/20% 50 mM Tris and
(B) 80% acetonitrile/20% 100 mM Tris. Other conditions
as in Fig. 1.
50 mM Tris and 20% of 100 mM Tris, 10% and 40% overall
increases in the mobilities are observed, respectively.
Mobility dependence on applied field, if unrecognized,
can lead to irreproducible separations; it is also an indica-
tor of incomplete heat dissipation. Examining the conduc-
tivity plots (Fig. 4) corresponding to these conditions,
however, we see that the plots clearly illustrate the heat-
ing effect, showing a 15% and 30% increase in value for
the two different concentrations of Tris considered. Ana-
lyzing electrical current data by constructing conductivity
plots highlights the solution heating within the column and
can serve as an effective tool for CE and CEC users for
recognition of ineffective heat dissipation. Therefore, we
suggest the use of conductivity plots instead of Ohm’s
plots to assess column heating.
Joule heating caused by power dissipation leads to an
increase in the temperature of the electrolyte inside the
capillary and, thus, higher ionic mobilities. This results in
an increase in electrical conductivity of the solution inside
of the capillary at higher potential drops. It follows then
2926 A. S. Rathore et al. Electrophoresis 2002, 23, 2918–2928

Figure 4. Conductivity plots for 12, 50, and 75 mm ID

capillaries. The mobile phases used were (A) 80% aceto-
nitrile/20% 50 mM Tris, and (B) 80% acetonitrile/20%
100 mM Tris. Other conditions as in Fig. 2.

from Eq. (3), that the relationship between the current and
the potential drop becomes a nonlinear relationship with
the current increasing disproportionately at higher volt-
ages in comparison to that predicted by a linear relation-
ship. These effects of the Joule heating on the conductiv-
ity have been modeled and the following relationship has
been proposed by Davis and co-workers [8, 9]:

s ˆ so ‡ a0 E m (15)

where, a’ and m are constants and so is the conductivity
at 257C that can be measured at very low voltages as the
reference temperature To. Assuming that a packed col-
umn is a hypothetical open tube with an equivalent con-
ductivity (se), the data obtained for the packed capillaries
and to a counterpart open tube, under similar conditions,
was fitted to Eq. (15) and presented in Figs. 7–9. It can
be seen that conductivity is not a constant, increasing
with the electric field strength due to Joule heating for
the mobile phase containing 20% of 50 and 100 mM Tris
Figure 5. Conductivity plots for (A) 12; (B) 50; (C) 75 mm
capillaries using different organic modifiers in the mobile
phase. The mobile phases used were 80% indicated
organic/20% 100 mM Tris. Other conditions as in Fig. 2.
(see Figs. 7A, 8A and 9A). With the exception of the col-
umn packed with the smallest particle size at the lowest
electrolyte concentration, the data fitted the proposed
relationship in Eq. (15) very well (R2 values above 0.97 for
Electrophoresis 2002, 23, 2918–2928
Figure 6. Electroosmotic mobility observed in 50 mm
open tubes. Conditions as in Fig. 2.
Em vs. se). In the case of the column packed with 0.5 um
particles with a total electrolyte concentration of 2 mM,
secondary effects caused by possible double-layer inter-
Joule heating in CEC 2927
actions may have influenced the current/conductivity
measurements, which can lead to the observed deviation
from Eq. (15).
4 Concluding remarks
Using experimental measurements of electrical current at
different voltages and theories that have been proposed
to model Joule heating in CE, the resulting rise in tem-
perature inside the CEC columns have been estimated. It
is assumed that the packing particles do not conduct heat
and heat transfer is solely through the mobile phase flow-
ing through the system. Plots of conductivity vs. applied
voltage have been proposed as better indicators of the
efficiency of heat dissipation in the separation system.
Under mobile and stationary phase conditions typical for
CEC, temperature increases of up to 237C above ambient
temperature were estimated for the packed columns
under study and about 357C for an open tube. We think
that such an increase in temperature in a packed bed
Figure 7. Dependence of the
electrical conductivity on (A)
electric field strength and (B)
the electric field strength as pre-
dicted by Eq. (11) (m is the fitted
parameter of the equation and
is equal to 3 in our calculations)
for the case of a 75 mm ID capil-
lary column (15 cm) packed with
1 mm bare silica filled with differ-
ent mobile phases. Raw data is
presented in Table 2.
Figure 8. Dependence of the
electrical conductivity on (A)
electric field strength and (B)
the electric field strength as
predicted by Eq. (11) (m is the
fitted parameter of the equation
and is equal to 3 in our calcula-
tions), capillary as in Fig. 7,
except 0.5 mm bare silica. Raw
data is presented in Table 3.
2928 A. S. Rathore et al.
under CEC conditions can contribute to problems such as
bubble formation, which have long plagued CEC separa-
tions, particularly at the frits due to the restrictive nature of
the sintered material. There is a lack of reports in the liter-
ature providing experimental temperature measurements
inside packed columns under truly CEC conditions. Such
measurements will provide proof of our treatment pre-
sented here and will be an area for future work. Our results
with the open tubes, however, agree with published ex-
perimental findings using direct motion encoding of
fluid molecules using pulsed magnetic field gradient
NMR [16], NMR thermometry [17], and Raman spectro-
scopic measurements [18].
Financial support was provided by the National Science
Foundation, USA under grant CHE-9614947.
Received March 13, 2002
5 References
[1] Colón, L. A., Maloney, T. D., Fermier, A. M., J. Chromatogr. A
2000, 887, 43.
[2] Rathore, A., Horváth, C., Anal. Chem. 1998, 70, 3069.
[3] Rathore, A., Wen, E., Horváth, C., Anal. Chem. 1999, 71,
2633.
[4] Choudhary, G., Horváth, C., J. Chromatogr. A 1997, 781, 161.
[5] Wan, Q., J. Phys. Chem. B 1997, 101, 4860.
[6] Arulanandam, S., Li, D., J. Colloid Interface Sci. 2000, 225,
421.
Electrophoresis 2002, 23, 2918–2928
Figure 9. Dependence of the
electrical conductivity on (A)
electric field strength and (B)
the electric field strength as pre-
dicted by Eq. (11) (m is the fitted
parameter of the equation and
is equal to 3 in our calculations)
for the case of a 75 mm ID open-
tube capillary (30 cm) filled with
different mobile phases. Raw
data is presented in Table 4.
[7] Bello, M. S., Chiari, M., Nesi, M., Righetti, P. G., Saracchi,
M. J., Chromatogr. 1992, 625, 323–330.
[8] Yu, L., Davis, J. M., Electrophoresis 1995, 16, 2104–2120.
[9] Seals, T. H., Davis, J. M., Murphy, M. R., Smith, K. W., Ste-
vens, W. C., Anal. Chem. 1998, 70, 4549–4562.
[10] Grushka, E., McCormick, R. M., Kirkland, J., J. Anal. Chem.
1998, 16, 241–246.
[11] Poppe, H., Kraak, J. C., Huber, J. F. K., van den Berg, J. H. M.,
Chromatographia 1981, 14, 515–523.
[12] Knox, J. H., McCormack, K. A., Chromatographia 1994, 38,
207–214.
[13] Gobie, W. A., Ivory, C. F., J. Chromatogr. 1990, 516, 191–210.
[14] Bello, M. S., Righetti, P. G., J. Chromatogr. 1992, 606, 95–10.
[15] Bello, M. S., Righetti, P. G., J. Chromatogr. 1992, 606, 103–
111.
[16] Tallarek, U., Rapp, E., Scheenen, T., Bayer, E., As, H. V.,
Anal. Chem. 2000, 72, 2292–2301.
[17] Lacey, M. E., Webb, A. G., Sweedler, J. V., Anal. Chem.
2000, 72, 4991–4998.
[18] Llu, K. K., Davis, K. L., Morris, M. D., Anal. Chem. 1998, 70,
1998, 3744–3750.
[19] Hunter, R. J., Foundations of Colloid Science, Clarendon
Press, Oxford 1986, p. 395.
[20] Hunter, R. J., Zeta Potential in Colloid Science Principles
and Applications, Academic Press, London 1988, p. 11.
[21] Hiemenz, P. C., Rajagopalan, R., Principles of Colloid and
Surface Chemistry, 3rd Ed., Marcel Dekker, New York 1997,
p. 499.
[22] Masliyah, J. H., Electrokinetic Transport Phenomena, Aos-
tra, Alberta 1994, p. 99.
[23] Reynolds, K. J., Colón, L. A., J. Liq. Chromatogr. Relat. Tech.
2000, 23, 161–173.
[24] Yan, C., US Patent # 5,453,163, Sept. 26, 1995.