Diffusive Monitor 1

The Diffusive
Monitor
February 2006
Issue 15
Health and Safety Executive
Committee of Analytical
Requirements - Working Group 5
Inside this Issue
1
By way of introduction
1
Current News
4
Effect of temperature on
long-term diffusive
sampling
6
Validation of techniques
for benzene and VOC
measurements in Flanders
10
Packing integrity of
thermal desorption tubes
and traps
20
Discrimination during
calibration of thermal
desorption methods

Registering your interest
in Diffusive Monitor. p. 23 Monitor. p. 23

By way of Introduction By way of Introduction
Welcome to the fifteenth edition of The
Diffusive Monitor, which is a free
publication of the Health and Safety
Executive CAR Committee (Committee
of Analytical Requirements), Working
Group 5. This working group is
concerned with workplace, indoor and
environmental applications of diffusive
sampling for assessing air quality.
Welcome to the fifteenth edition of The
Diffusive Monitor, which is a free
publication of the Health and Safety
Executive CAR Committee (Committee
of Analytical Requirements), Working
Group 5. This working group is
concerned with workplace, indoor and
environmental applications of diffusive
sampling for assessing air quality.

The newsletter was started in May 1988
as a consequence of the Diffusive
Sampling Symposium held in
Luxembourg the previous year and is
published approximately once a year. It
contains articles on diffusive monitoring
techniques and applications, and is a
useful source of information on
European and international
standardisation in this area and of
sampling rate data. Contributions are
mostly from members of the Working
Group, which has an international
membership.
The newsletter was started in May 1988
as a consequence of the Diffusive
Sampling Symposium held in
Luxembourg the previous year and is
published approximately once a year. It
contains articles on diffusive monitoring
techniques and applications, and is a
useful source of information on
European and international
standardisation in this area and of
sampling rate data. Contributions are
mostly from members of the Working
Group, which has an international
membership.
Contributions to the newsletter are not,
however, intended to be exclusively
from CAR/WG 5, and any reader is
welcome to submit an item for
consideration. The only limitations are
that articles should have some diffusive
sampling application and should not be
too obviously commercial.
Contributions to the newsletter are not,
however, intended to be exclusively
from CAR/WG 5, and any reader is
welcome to submit an item for
consideration. The only limitations are
that articles should have some diffusive
sampling application and should not be
too obviously commercial.
The newsletter has a world circulation
of some 200 people, all of whom have
specifically requested the publication,
so if you wish to contribute articles, you
can be assured of a wide and receptive
audience. Articles are not peer-
reviewed, but are subject to the Editors
discretion. A Word template for authors
is recommended and is available from
the Editor on request.
The newsletter has a world circulation
of some 200 people, all of whom have
specifically requested the publication,
so if you wish to contribute articles, you
can be assured of a wide and receptive
audience. Articles are not peer-
reviewed, but are subject to the Editors
discretion. A Word template for authors
is recommended and is available from
the Editor on request.
Copies of this newsletter and previous
issues back to no.12 (July 2001) may be
downloaded from the Health and Safety
Laboratory website at
http://www.hsl.gov.uk/publications/diffuse-
Copies of this newsletter and previous
issues back to no.12 (July 2001) may be
downloaded from the Health and Safety
Laboratory website at
http://www.hsl.gov.uk/publications/diffuse-
monitor.htm
No registration is necessary to
download a copy. However, if you want
to be placed on a list to be notified when
a new issue is published contact the
Editor mike.wright@hsl.gov.uk (see
page 23).

Current News
Committee of Analytical
Requirements (CAR) reconstituted
HSE/CAR first met and decided its
terms of reference in 1976. CAR
replaced the even older 'Committee on
Toxic Substances in Air' responsible for
the simple field tests known as the 'Red
Book' methods. CAR continued to meet
about twice a year for 25 years which is
a long time in the life of any advisory
committee. CAR met twice in 2005 on 8
June and 27 October at HSL Buxton.
Both membership and terms of
reference were under review. Minutes
of these and future meetings will be
published on the HSL website.
Essentially the proposed new ToF were:
To advise on the need for new or
improved tools to assess exposure
to harmful materials in workplaces
covered by the UK Health and
Safety Executive and Local
Authorities;
To evaluate and peer review the
tools to ensure they are fit for
purpose;
To disseminate information about
the assessment tools to potential
users;
To encourage good quality in the
application of the tools.
1 . . . . . . . . . . . . . . . . . . . . . . .
February 2006
TC264/WG11 is tasked with developing
performance requirements for diffusive
sampling of ambient and indoor air.
Guides to selection, use and
maintenance have now been published
as EN13528 part 3 and EN 14412 [7, 8].
The 5 part standard of TC264/WG13
(EN 14662) for benzene in ambient air
was published in 2005 [9-13]. Note that
the final titles differ slightly from the
prEN versions listed in The Diffusive
Monitor issue 14. EN 14662-1 and EN
14662-3 are declared as suitable EU
reference methods for the manual and
automatic procedures respectively. Parts
2, 4 and 5 are indicative methods,
although their performance has been
shown to be more or less equivalent to
the reference methods. With publication
of EN 14662 for ambient benzene
comes the advent of a valuable resource.
It collates much information on
laboratory intercomparisons and
measurement uncertainty that was
previously scattered around the grey
literature.
CEN air quality standards
Issue 14 of The Diffusive Monitor
described the progress of various work
items in expert working groups (WGs)
reporting to Technical Committees of
the Comit Europen de Normalisation
(CEN) up to the end of 2003. The
programmes of TC137 (workplace
exposure) and TC264 (air quality)
derive mainly from the requirements of
the Chemical Agents Directive
(CAD)(98/24/EC) and the Ambient Air
Directive (96/62/EC) respectively [1,2].
It was previously reported that in 2003
the European Commission had given a
mandate to TC137 in order to
implement Article 3(10) of CAD,
namely standardised methods for the
measurement and evaluation of
workplace air concentrations in
relation to occupational exposure limits
shall be developed in accordance with
Article 12(2).. where .. the Commission
shall draw up practical guidelines of a
non-binding nature. Task groups of
TC137/WG2 during 2005 completed
their evaluation of measurement
methods for about 100 priority
substances, including all those with
European Limit Values (IOELVs). The
final project report will be placed on the
website of BGIA St Augustin Germany
(Berufsgenossenschaftliches Institut fr
Arbeitsschutz),
http://www.hvbg.de/d/bia/index.html
probably before mid-2006.
Meanwhile TC137/WG2 has been
revising the measurement performance
requirement standard EN 482:1994 [3].
The 1994 document is a rather
pragmatic standard and the best that
could be achieved at the time. Various
subsidiary performance standards such
as EN 838 (diffusive sampling) [4] and
EN 1076 (pumped sampling) [5] were
in the future and could not be referred
to. PrEN 482:2005 has been through
CEN enquiry and TC137 has taken the
decision to go for formal vote. It retains
the numerical performance requirements
for comparison with limit values and
takes the cause and effect approach to
calculating uncertainty which is more in
accordance with recent thinking and
with the GUM [6].
TC264/WG12 (inorganic gases in
ambient air) has recently published
standard methods using automated
measuring systems namely, EN 14211 -
NO
x
by chemiluminescence [14], EN
14212 -SO
2
by UV fluorescence [15]
and EN 14625 ozone by UV
photometry [16]. There is also a great
need for indicative diffusive methods
for ambient NO
x
, SO
2
and ozone. Such
methods are commonly used in Europe,
but without the benefit of European
standards. It is hoped that such a
proposal will eventually be supported
by the Commission and carried forward
by TC264/WG11.
ISO air quality standards
Workplace, ambient and indoor air
quality aspects are all dealt with in sub-
committees of TC146. In TC146/SC2
(workplace atmospheres) projects in
preparation include a revision of the 2-
MP derivative method for isocyanates
[17], dibutylamine and chloroformate
derivative methods [18, 19], a draft
standard on mercury vapour [20] and a
committee draft on inorganic acids [21].
TC146/SC4 (general aspects) is revising
ISO 20988 a guide to estimating
uncertainty specifically for air quality
measurements [22] and ISO 9169 on the
definition of general tests of automatic
measuring systems [23]. ISO 9169
should not be mistaken for a type
approval standard however. It advises
you how to conduct tests for
repeatability, lack of fit, drift etc. in
general terms, but acceptance criteria
are outside its scope and must be
defined elsewhere.
ISO TC146/SC6 (indoor air) continues
development of the ISO 16000 series.
Since 2004 standards have been
published on sampling strategy [24, 25],
active sampling of carbonyl compounds
[26], diffusive sampling of
formaldehyde [27], active sampling of
VOCs [28] and emission testing of
building products and furnishings [29-
31]. In preparation are draft standards
on VOC measurement strategy [32],
characterising ventilation [33] and
determination of nitrogen dioxide [34,
35].
UK Methods for the Determination of
Hazardous Substances (MDHS)
For a list of titles and revision history in
the UK MDHS series see
http://www.hsl.gov.uk/publications/mdhs_list.htm
All MDHS titles currently in print are
available for download from the HSE
site at
http://www.hse.gov.uk/pubns/mdhs/index.htm.
All 'in print' titles remain as priced
publications if paper copies are ordered
from HSE Books. Older titles 'out of
print' or 'withdrawn' are unavailable
from HSE Books, although they can
always be obtained from the British
Library. BL index them as a serial
publication and the HSL website
explains how to get them by this route.
Twenty-three of these older 'withdrawn'
titles have been identified as containing
information which is not factually
incorrect and may be useful, therefore
their position is under review. If it is
possible to host them on the HSL
website then HSL will endeavour to
include links to the back-up data reports
in the grey literature that are otherwise
difficult to obtain through normal
library services.
2 . . . . . . . . . . . . . . . . . . . . . . . .
The Diffusive Monitor, 15
6. Guide to the Expression of Uncertainty in
Measurement (ISO, 1995, equivalent to EN
13005:1998 equivalent to BS PD 6461 Part
3:1995).
Proficiency testing news
HSL operates several PT schemes for
laboratories measuring workplace and
ambient substances and we offer some
materials/media not found in similar
schemes elsewhere. See the following
link for details of the scope and how to
join:
7. EN 13528-3:2003 Ambient Air Quality
Diffusive samplers for the determination of
concentrations of gases and vapours
Requirements and test methods. Part 3:
Guide to selection use and maintenance.
http://www.hsl.gov.uk/proficiency-
testing/index.htm
Several participants have enquired
about the accreditation status of their
scheme. HSL has recently begun the
process of proving conformity to ISO
Guide 43/ILAC Guide 13 and aims to
complete its UKAS external assessment
by the end of 2006.
Following the commissioning of new
facilities on the Buxton site, HSL is
now offering validated PT
samples/media for sale, including BTX,
chloroalkanes, isocyanates,
formaldehyde, nitrite for ambient NO
2
,
mineral fibres, quartz, Pb/Cd/Cr and
welding fume. Further details and order
forms are available through this link:
http://www.hsl.gov.uk/proficiency-
testing/standards-materials.htm#oh

References
1. Council Directive 98/24/EC on the
protection of the health and safety of
workers from the risks related to chemical
agents at work (1998).
2. Council Directive 96/62/EC on ambient air
quality assessment and management (1996).
3. EN 482:1994 Workplace atmospheres -
General requirements for the performance of
procedures for the measurement of chemical
agents.
4. EN 838:1995Workplace atmospheres
gases and vapours Requirements and test
methods.
5. EN 1076:1997 Workplace atmospheres
Pumped sorbent tubes for the determination
of gases and vapours Requirements and
test methods.
8. EN 14412:2004 Indoor Air Quality
concentrations of gases and vapours Guide
for selection use and maintenance.
Standard method for the measurement of
benzene concentrations Part 1: Pumped
sampling followed by thermal desorption
and gas chromatography.
benzene concentrations Part 2: pumped
sampling followed by solvent desorption and
gas chromatography.
benzene concentrations Part 3: Automated
pumped sampling with in situ gas
chromatography.
benzene concentrations Part 4: Diffusive
sampling followed by thermal desorption
and gas chromatography.
13. EN 14662-5:2005 Air Quality Standard
method for the measurement of benzene
concentrations Part 5: Diffusive sampling
followed by solvent desorption and gas
chromatography.
14. EN 14211:2005 Ambient air quality
Standard method for the measurement of the
concentration of nitrogen dioxide and
nitrogen monoxide by chemiluminescence.
concentration of sulphur dioxide by
ultraviolet fluorescence.
concentration of ozone by ultraviolet
photometry.
17. ISO/DIS 16702 Workplace air quality --
Determination of total organic isocyanate
groups in air using 1-(2-
methoxyphenyl)piperazine and liquid
chromatography.
18. ISO 17734-1 Determination of
organonitrogen compounds in air using
liquid chromatography and mass
spectrometry -- Part 1: Isocyanates using
dibutylamine derivatives.
19. ISO 17734-2 Determination of
organonitrogen compounds in air using
liquid chromatography and mass
spectrometry -- Part 2: Amines and
aminoisocyanates using dibutylamine and
ethyl chloroformate derivatives.
20. ISO/DIS 20552 Workplace air --
Determination of mercury vapour -- Method
using gold-amalgam collection and analysis
by atomic absorption spectrometry or atomic
fluorescence spectrometry.
21. ISO/CD 21438-1 Workplace atmospheres -
Determination of inorganic acids by ion
chromatography -- Part 1: Non volatile acids
(sulphuric acid and phosphoric acid).
22. ISO/DIS 20988 Air quality -- Guidelines for
estimating measurement uncertainty.
23. ISO/FDIS 9169 Air quality -- Definition and
determination of performance characteristics
of an automatic measuring system.
24. ISO 16000-1:2004 Indoor air -- Part 1:
General aspects of sampling strategy
Sampling strategy for formaldehyde
Determination of formaldehyde and other
carbonyl compounds -- Active sampling
method
Determination of formaldehyde -- Diffusive
sampling method
Determination of volatile organic
compounds in indoor and test chamber air by
active sampling on Tenax TA sorbent,
thermal desorption and gas chromatography
using MS/FID.
Determination of the emission of volatile
organic compounds from building products
and furnishing -- Emission test chamber
method.
and furnishing -- Emission test cell method.
and furnishing -- Sampling, storage of
samples and preparation of test specimens.
32. ISO/DIS 16000-5 Indoor air -- Part 5:
Measurement strategy for volatile organic
compounds (VOCs)
33. ISO/DIS 16000-8 Indoor air -- Part 8:
Determination of local mean ages of air in
buildings for characterizing ventilation
conditions
34. ISO/CD 16000-14 Indoor air -- Part 14:
Sampling strategy for nitrogen dioxide
(NO2
)
35. ISO/CD 16000-15 Indoor air -- Part 15:
Measurement of nitrogen dioxide (NO
2
)

3 . . . . . . . . . . . . . . . . . . . . . . . .

February 2006
Effect of temperature on long-term diffusive sampling how to properly evaluate
effective diffusive uptake rates
Theo Hafkenscheid, NMi van Swinden Laboratory, PO Box 654, 2600 AR, Delft, Netherlands. E-mail: thafkenscheid@nmi.nl

over the sampling period is acceptable for practical
purposes.
Introduction
The effective uptake rate in diffusive sampling is given by
the equation Approach
The error in using one value for the diffusion coefficient at
the mean sampling temperature is given by
E D
L
A
U
eff
= (1)
( )
( )
dt
T D
T D
t
e
t
|
|
.
|
\
|
=
0
1
1
(3)
where
A = cross-section area of the diffusion pathway (usually in
cm
2
)
L = diffusion path length (usually in cm)
In order to calculate the magnitude of this error for various
situations, relationships have to be established between T
and T .
D = diffusion coefficient of the measured compound in air
(usually in cm
2
/s)
E = sampling efficiency of the diffusive sampler.
The dependence of U
eff
on temperature (variations) of the air
sampled is determined by the temperature dependence of the
diffusion coefficient D and the sampling efficiency E.
Daily sampling period
For a 24-hour sampling period the variations in temperature
are assumed to be based on a cosine-shaped relationship as
follows
The dependence of D on temperature is given by a Maxwell-
type equation
|
.
|
\
|
= t a T T
d
12
cos

(4)
( ) ( )
n
T
T
T D T D
|
|
.
|
\
|
=
0
0
(2)
where
where
a
d
= amplitude of the cosine function, i.e., half of the
difference between daily extremes in temperature variations
T = actual temperature
t = time in hours.
T
0
= reference temperature for D, usually 20 or 25 C
n = exponent with value between 1.5 and 2. Input of (4) in (2), assuming the coefficient of the Maxwell
equation to be 2 (worst-case scenario with respect to
variations in D) and subsequent application of (3) leads to
the following expression for e:
Equation 2 can be used to estimate values of D from a
known value at a reference temperature, and a known actual
temperature T. This is particularly relevant for large
deviations of the actual sampling temperature from that at
which U
eff
has been determined. For example, if the mean
actual sampling temperature is 0 C, while U
eff
has been
determined at 20 C, the resulting error in D is about 10-
13%. In long-term ambient air sampling such deviations of
the actual temperature from the temperature at which U
eff
has been determined are not uncommon.

2
2
2T
a
e
d
= (5)
For a mean temperature of 250 K and a maximum daily
temperature variation of 30 K, the error resulting from using
a diffusion coefficient at 250 K instead of a proper
expression of D as a function of T is found to be 0.2% only.
Moreover, the actual temperature will exhibit variations,
both in the form of diurnal variations, and in longer-term
trends. Hence, application of eq. (2) is not straightforward.
In principle, a relationship between D and T based on eq. (2)
should be used as input to eq. (1). This is rather
cumbersome; preferably, one unique temperature
representative of the sampling conditions should be found to
which eq. (2) can be applied.
Two-week sampling period
For a 2-week sampling period the variations in temperature
are assumed to be based on a combined linear trend and a
cosine-shaped relationship as follows
|
.
|
\
|
|
.
|
\
|
+ = t a
t
T T T
d
12
cos
2
1
336

(6)
The hypothesis in this paper is that the error resulting from
the use of a diffusion coefficient at the mean temperature
where

4 . . . . . . . . . . . . . . . . . . . . . . . .
= difference between mean temperature at the
beginning and end of the sampling period, i.e. the drift in
the mean temperature over the sampling period.
T
Input of (6) in (3) assuming the coefficient of the Maxwell
equation to be 2 (worst-case scenario with respect to
variations in D) and subsequent application of (3) leads to
the following expression for e:

2
2
2
2
2 12 T
a
T
T
e
d
+
= (7)
For a mean temperature of 250 K, a drift of 10 K and a
maximum daily temperature variation of 30 K, the error
resulting from using a diffusion coefficient at 250 K instead
of a proper expression of D as a function of T is still only
0.2%.
One-year sampling period
For monitoring of ambient air levels of e.g. benzene and
comparison with the limit value for a one-year reference
period, sequential diffusive sampling, for example during 27
consecutive 2-week periods, may be applied. In principle,
for each of the 2-week periods a different diffusion
coefficient should be used, based on the actual temperature
conditions.
However, it is also possible to use one value for D at the
annual mean temperature without introduction of large
errors. It is assumed that the variations in temperature are
described by two superimposed cosine-shaped functions:
|
.
|
\
|
|
.
|
\
|
= t a t a T
d y
12
cos
4380
cos

T (8)
where
a
y
= amplitude of the cosine function for the yearly
contribution, i.e., half of the difference between yearly
extremes in daily mean temperatures.
The resulting error found by combining eqs. (8), (2) and (3)
is

2
2 2
2T
a a
e
d y
+
= (9)
For an annual mean temperature of 273 K, a maximum daily
temperature variation of 30 K and a yearly variation between
daily mean temperatures of 50 K, the error resulting from
using a diffusion coefficient at 250 K instead of a proper
expression of D as a function of T is 0.7%. However,
application of the above implies that the annual mean
concentration value can only be calculated by processing all
2-week measurement results a posteriori.
Conclusions and recommendations
Based on modeling of temperature variations over various
sampling periods, it has been shown that when using a
diffusion coefficient and, consequently an effective uptake
rate - at the mean temperature over the sampling period, the
effect of temperature variations may safely be ignored.
Only a small error in the order of magnitude of <1% is
introduced in comparison with the use of a proper
expression of D as a function of T. This error is acceptable
when compared to other contributions to the uncertainty in
the effective uptake rate. Therefore, it is recommended to
use a (calculated) effective uptake rate at this mean
temperature for temperatures deviating considerably from
the temperatures at which the effective uptake rate has been
determined. Acceptable errors may depend on the
application and can be used to calculate maximum allowable
temperature deviations.
However, there are some additional considerations. Firstly,
the above considerations apply to uptake rates expressed in
units of volume per unit time. Many effective uptake rates
are expressed in units of ng/(ppm.min) or pg/(ppb.min),
relating the mass sampled to the compound concentration in
dimensionless units and sampling time. The uptake rate
expressed in this way is related to the uptake rate expressed
in units of cm3/min by a factor Vm/M

m
eff eff
V
M
cm U ppm ng = min) / ( min) / (
3
U
(10)
where
M = molar mass of the compound (kg),
V
m
= molar volume of the compound (m
3
) at the temperature
at which U
eff
has been determined.
Because of the introduction of V
m
, the temperature
dependence of Ueff on T reduces to eq. (11).
1
0
0
) min, / ( ) min, / (

|
|
.
|
\
|
=
n
eff eff
T
T
T ppm ng U T ppm ng U
For simplicity it has been assumed that the reference
temperatures for diffusion coefficient and molar volume are
the same. This assumption does not alter the basic
dependence of U
eff
on T by an exponent of (n-1). Secondly,
for non-ideal samplers, i.e. samplers for which E<1, the
sampling efficiency generally increases with decreasing
temperature due to increasing sorbent strength and vise
versa, leading to a partial compensation for the effect of the
changing diffusion coefficient [1]. In this case, application
of eqs. (2) or (11) in full will lead to an overcompensation of
the effective uptake rate for temperature dependence. A
detailed discussion of the extent of this compensation effect
is beyond the scope of this paper.
Concluding, the above may be used as a guideline, but
should be applied with proper care and consideration.
References
1. CEN 13528-3. Ambient Air Quality - Diffusive samplers for the
determination of concentrations of gases and vapours - Guide to
selection, use and maintenance. CEN, Brussels.
5 . . . . . . . . . . . . . . . . . . . . . . . .

February 2006
Validation of thermal desorption monitoring techniques for benzene and VOCs in
environmental air in Flanders
F. Maes
a
, R. Bormans
a
, D. Poelmans
a
, L. Verbeke
a
, E. Goelen
a
and E. Wauters
b

a
VITO, Environmental Measurements, Boeretang 200, 2400 Mol, Belgium.
b
Flemish Environment Agency, Vlaamse Milieu
Maatschappij, Department of Monitoring Networks and Research, CDVP Laboratorium, Krijgslaan 281/S2, 9000 Gent, Belgium.
Two different kinds of samplers were validated:
Introduction
Perkin Elmer tube filled with the sorbent
Carbopack B (300 mg/ 60-80 mesh); noted as PE; The Flemish Environment Agency VMM [1] has set up a
monitoring network in Flanders to observe the
concentrations of pollutants within the scope of the ozone
directive 2002/3/EG. The concentrations of VOCs, such as
benzene, toluene, xylenes and ethylbenzene (BTEX) were
measured by means of continuous monitors. Because of the
high investment and maintenance costs the number of
sampling points is rather limited. Later on, VITO [2] and
VMM cooperated to monitor successfully the BTEX
concentrations by means of diffusive sampling with solvent
desorption in the lab. An alternative for VOCs is the thermal
desorption of diffusive samplers; this technique is more
sensitive and is easily automated. On the other hand, the
environmental conditions can influence the sampling
because of the use of the relatively weak sorbents. The
objective of this project is to validate two existing diffusive
monitors appropriate for thermal desorption in order to use
as additional measurements in the permanent monitoring
network.
Radiello, packed with 350 mg Carbograph 4 (35-50
mesh); noted as thermal Radiello.
Both sorbents are of the same type: graphitised carbon black,
treated above 2500 C.
After this laboratory validation, the samplers were tested in a
field validation in which the results were compared with the
results of the reference method (active sampling with GC
analysis) and with the results of the Radiello sampler for
solvent desorption (solvent Radiello).
Materials and methods
Test chamber
An experimental set-up based on a glass test chamber was
developed to execute the validation experiments. It consists
of two concentric glass tubes. The inner tube is the exposure
chamber and contains a fan at the end to allow various wind
speeds. The whole system is placed in a freezer which
contains heating elements. A temperature of 0 C can be
reached when the freezer is working. To reach 30 C the
freezer serves as an isolator to retain the produced heat from
built in heating elements. All temperatures are PID
controlled.
The theoretical basis of diffusive sampling is Ficks first law
which describes the speed of mass transport resulting from a
concentration gradient, namely the difference in
concentration of a compound on the sorbent and in the air [3,
4]. This mass transport determines the uptake rate, which is
characteristic for the sorbent and for the shape of the
sampler. There are quite a few factors which can influence
the uptake rate and related uncertainty, therefore it is
important to validate any kind of diffusive sampler before
use. The protocol to be followed is described in the CEN
standard EN 13528-2 [12]. According to the standard,
following parameters have to be tested:
The required dynamic standard atmospheres are generated
by means of an adapted version of a capillary dosage unit
[11]. The calculation of the generated compound
concentrations is done starting from balance measurements
(dosed pollutants in g/min) in a known carrier gas flow
(l/min).
1. determination of the blank value;
Analytical method
2. desorption efficiency;
3. determination of the uptake rate;
After sampling, the Radiello cartridge has to be transferred
to an empty Perkin Elmer (PE) tube. The Carbopack B PE
tubes can be analysed after exposure. The samplers were
analysed with a thermal desorber (UNITY
TM
Thermal
Desorber from Markes International Ltd) interfaced with a
GC (Agilent GC Plus 6890) and a mass selective detector
(HP 5973 MSD). Further information about a similar
method has been described in Ref. 3.
4. effect of wind speed and sampler orientation;
5. storage of the samplers after exposure;
6. sampler integrity (leak test);
7. determination of back diffusion;
8. effect of concentration and exposure time on uptake
rate;
9. effect of temperature and humidity on uptake rate.

Calibration of the GC-MS unit is done by liquid spiking on
blank adsorption tubes. Known compound concentrations
are gravimetrically prepared in methanol. The required
amount of this solution is vapourised into a continuous flow
of an inert gas and transferred to the blank tube.

6 . . . . . . . . . . . . . . . . . . . . . . . .
Results and discussion
Determination of the blank value
The blank value is determined by analysing a calibration
tube several times until the peak values are constant. In
practice, this occurred after three desorptions. The result of
the first desorption corresponds to the calibration value, the
blank value can be calculated from the third desorption
results. This way of determining blank values excludes some
apparatus variables. During the whole research, the blank
values have been determined.
The highest blanks for both samplers are found for benzene,
up to 30 ng for benzene. Typical blanks ranged from 0.3-15
ng. The blank value of the PE sampler is about three times
lower than Radiello sampler, but the latter compensates it
with a much higher uptake rate (see Determination of uptake
rates). The blank value is a limiting factor for the PE
sampler considering the chosen sampling times in
environmental conditions (7 days).
In literature, comparable blank values for Carbograph 4 and
Carbopack B have been determined [3-6].
Desorption efficiency
The blank values are constant after three successive
desorptions, at which point there is 100 % desorption. The
desorption efficiency (DE) can be determined by
comparison of the first desorption value with the total
desorption. Under the selected analytical conditions, the DE
of the PE tubes amounts to 97 %, for Radiello it is almost
100 %. These values show that both sorbents are suitable for
thermal desorption applications. According to EN 13528-2,
DE should be above 95 %, but it is not used as a correction
factor in the final processing of the analytical results. The
calibration tubes and the samples have the same DE.
Determination of the uptake rates
In these tests, the diffusive samplers were exposed under
various conditions of exposure time and concentration levels
(see Table 1).
Table 1 Test conditions to determine uptake rates.
Test number 4 6a 6b 7 9
B 15.5 5.10 5.10 6.06 15.6
T 63.1 20.8 20.8 24.6 63.1
EB 12.3 4.05 4.05 4.71 12.1
mp-X 22.1 7.28 7.28 8.53 21.9

compound-
concentration
in g/m

o-X 10.6 3.48 3.48 4.06 10.4
exposure time (min) 7200 10140 20291 7080 2700
Relative humidity (%) 42 5 39 5 39 5 39 5 36 5
Temperature (C) 18 1 18 1 18 1 18 1 18 1
Wind speed (m/s) 0.14 0.14 0.14 0.14 0.14

Regression curves for each compound were made: amount in
ng compound on the tubes as a function of the loads of the
exposed samples in g/m/min. Each test represents a point
in the curve. The slope of the curve corresponds with the
uptake rate. Note that according to EN 13528-2, the uptake
rate should be determined solely with the results from test
6a, with an exposure time of 1 week. The uptake rates
resulting from the approach used in this research combine a
set of tests with different exposure times and concentration
levels thus giving reliable uptake rates. The correlation
coefficients indicate potential effects of exposure time
and/or concentration levels.
The experimentally determined uptake rates are listed in
Table 2.
Table 2 Experimentally determined uptake rates per
compound (BTEX).
Radiello (Carbograph 4) Perkin-Elmer (Carbopack B)

Uptake
rate
ml/min
RSD %
(*)
R
Uptake
rate
ml/min
RSD %
(*)
R
B 30.6 12 0.96 0.42 16 0.52
T 33.2 6.6 0.99 0.52 5.6 0.98
EB 31.5 8.7 0.99 0.47 8.9 0.96
mp-X 30.7 7.3 0.99 0.47 10 0.94
o-X 31.5 10 0.99 0.46 13 0.93
* RSD is derived from the standard deviation of the uptake
rates calculated per test.
The most volatile compound, benzene, causes the most
problems with determining the uptake rate, especially for the
PE sampler. This is indicated by the lower correlation
coefficients for benzene compared to the other compounds.
The PE sampler suffers from the low uptake rates which are
about a factor 65 lower than those for Radiello. As a result,
the total load on the PE tubes is sometimes less than three
times the blank value. The linearity increases strongly when
the blank values are negligible compared to the load, as with
Radiello. Comparable values of experimentally determined
uptake rates can be found for Radiello [4, 7], and for PE [8].
Sampler orientation and wind speed
The PE sampler is placed parallel in standard conditions, i.e.
with the diffusion opening on the outlet side of the test
chamber. A perpendicular position (to the wind direction)
has also been tested. The standard position for the Radiello
is perpendicular, i.e. the length axis of the sampler is
perpendicular to the wind direction. In the parallel test, the
Radiello axis is parallel to the wind direction.
This tests are performed relatively. It means that
calibration of the thermal desorber is not necessary as the
tubes are compared mutually. The analysis value of the
standard position is set to 100 %, and the other position is
related to this value. Benzene has for both samplers the
highest deviation, 98.3 % for Radiello and 105 % for PE.
Other compounds show no differences in orientation. The
influence of sampler orientation seems to be not significant,
yet the RSD values increase slightly with the parallel
position.
7 . . . . . . . . . . . . . . . . . . . . . . . .

February 2006
The aim of the wind speed experiments is to evaluate the
effect of wind speed on the uptake rates. According to the
manufacturers Radiello uptake rate is invariant with wind
speed between 0.1 m/s and 10 m/s. For PE wind speed
should be above 0.7 cm/s. Just like the sampler orientation
test, the results are processed relatively. Both tests are done
in the same conditions, only altering the wind speed. At
simultaneous analysis of the test tubes it is possible to
investigate the effect of wind speed by comparing the
analytical results for each type of sampler. It turns out that
the uptake rates are invariant at 0.14 m/s and 0.74 m/s. It is
possible that these two wind speeds are chosen too close to
each other to result in significant differences. Yet it is
obvious that the RSD values increase at the high wind speed,
and this especially with the PE sampler. The PE sampler is
not equipped with a sufficiently effective membrane, and as
a result the static layer is not well protected against wind
turbulences. This static layer is essential to obtain complete
molecular diffusion. These results seem to be in
contradiction with the results found in literature [4, 7].
Storage stability
According to EN 13528-2 the storage stability of the
samplers has to be evaluated for all target compounds. After
sampling the samplers are stored for two weeks in a
container under nitrogen flow. The analytical results of the
stored samplers and samplers analysed immediate after
sampling should not differ more than 10 %. The deviations
for Radiello exceed the 10 % value for more compounds
(-11 % for EB and -12 % for mp-X, both for Radiello).
These deviations are probably originating from the
analytical repeatability of the method since the deviation for
benzene, the most volatile compound, is less than 10 %.
Same comment applies to the PE tubes, but here the
collected amount against the blank value is a critical factor,
which results in a higher measurement uncertainty. All
things considered, storage stability at room temperature
causes no problems with regard to the criterion in the
standard.
Back diffusion
Using non-ideal sorbents implies the possibility of
desorption when the vapor concentration just above the
sorbent surface is higher than in the surrounding
atmosphere. This phenomenon is called back diffusion and
occurs more frequently with weak sorbents that typically
have a small breakthrough volume for the compounds during
active sampling. This possibility of back diffusion was
investigated by exposing two sets of samplers to a test
atmosphere, removing one set and then exposing the
remaining set to clean air. The test design permits a relative
comparison of results. Compounds which show smaller
peaks from tubes exposed to clean air indicate possible
losses as a result of back diffusion. Carbopack B sorbent
tends not to loose compounds. The Radiello sorbent,
Carbograph 4, does show some losses. For benzene, about 7
% of the collected amount is lost during the exposure to
clean air. Other compounds, with a lower vapour pressure,
experience this effect less (about 3 % for toluene, the other
compounds show no back diffusion). This effect has to be
kept in mind when interpreting measurement results of the
Radiello sampler. Literature [4] reports the determination of
back diffusion during 7 days alternate exposure to a high
concentration then zero air.
Effect of concentration level and exposure time on
uptake rates
Tests with different exposure times (24 hours, 7 days and 14
days) and concentration levels (6.1 g/m and 15.5 g/m for
benzene) were used to investigate the possible effects on the
uptake rates. The relative exposure time tests show that for
Radiello the expected mass uptake ratios, namely 1:7:14, are
found, except for the most volatile compound, benzene. So
the uptake rates for exposure times until 2 weeks are
invariant, except for benzene. The slight decline is relatively
small. This phenomenon is caused by displacement and
saturation effects at the vapour/sorbent interface [10].
The exposure time test 24 hours 7 days has been
performed for PE, but the results are not shown as the
collected amounts are too small to yield reliable results. In
contrast to Radiello, the uptake decreases to a larger extent
(- 23 % for benzene). As a consequence, the uptake rates can
only be applied in a limited and strictly validated area.
About the same conclusions for the concentration level
effect: Radiello uptake rates differ slightly, within the global
sampling and analytical method. The PE sampler
experiences higher uptake rates when exposed to lower
concentration levels. Comparable tests have been reported in
Ref. 4.
Effect of temperature and relative humidity on uptake
rates
The effect of temperature on the uptake rate of a diffusive
sampler can be calculated by the following formula:
5 . 1
25
298
273
(
+
=
T
Q Q
T
(1)
The low temperature test has been done with low humidity,
and the high temperature test with high humidity. These
tubes are analysed together with loaded tubes from a test in
standard conditions, so a relative comparison is possible.
Out of the peak surfaces of the reference test, it is possible to
calculate the theoretical peak surfaces of both tests, taking
the different loaded amounts and formula 1 into account.
These theoretical values are then compared with the real
values from which the deviation is determined. For the PE
sampler the deviation of the real value with respect to the
theoretical value for the low temperature test is under 6 % ,
except for benzene. For the high temperature test, the
deviations are negligible.
8 . . . . . . . . . . . . . . . . . . . . . . . .
For the Radiello sampler, low test deviations rise in a
negative manner, from -15 % from the most volatile to the
least volatile compound. High test deviations are smaller
than 5 %, again except for benzene. All in all, the practice
confirms the deviations calculated from the theoretical
formula of less than 15 %.
Table 3 Field validation: comparison of four sampling methods;
reference method (A), Perkin Elmer (B), thermal-Radiello (C),
solvent-Radiello (D), at two locations.
Field validation
Field validations were carried out in order to test the
samplers for use in real-life environmental conditions and
the presence of likely chemical interference from species
such as other VOCs and ozone. The samplers were mounted
in shelters to protect them from high air speed and
precipitation while maintaining sufficient air movement for
sampling. The samplers were placed in four typical areas
spread over Flanders: rural locations, urban backgrounds,
busy crossroads and industrial areas. Two locations were
chosen for each area type. Two sampling periods were
performed: one with winter conditions (0 C), the other with
spring conditions (10 C).
A comparison was done for four types of sampling methods
on two of the eight locations: the reference method, PE,
thermal Radiello and solvent Radiello diffusive sampling
(see Table 3). For Radiello, the uptake rates supplied by the
manufacturer were used to calculate ambient air
concentrations. For the PE sampler the uptake rates
determined in the laboratory validation were used, except for
benzene [8].
For measuring period 1, the deviations between the lowest
and highest concentrations are less than 25 % for all
compounds. Taking the measurement uncertainty into
account and the fact that it is a field validation, these
differences are small. In location 8, the deviations amount
up to 40 %. Ignoring the PE results for benzene and o-xylene
(possible blank problems) in measuring period 2, the
deviation at location 7 reaches a maximum value of 27 %
with o-xylene. The other deviations are less than 25 %. The
deviations at location 8 are higher (as in measuring period
1), up to 36 % for ethylbenzene. Another comparison is
reported in Table 4: deviations from the solvent Radiello
with respect to thermal Radiello and the PE sampler.
Table 4 Deviation of solvent-Radiello results from thermal-
Radiello and PE.

Measuring period 1
number of results
Measuring period 2
number of results
Deviation
Solv-Rad.
vs. therm-Rad.
Solv-Rad.
vs. PE
Solv-Rad.
vs. therm-Rad.
Solv-Rad.
vs. PE
15 % 18 19 25 16
15 - 30 % 14 15 14 7
30 % 8 6 1 17

About half of the 40 measurements are less than 15 %
deviation for each comparison. The PE problem appears in
measuring period 2. Seventeen measurements deviate more
than 30 % from the solvent Radiello. The similarity between
both Radiellos is remarkable in measuring period 2. An
important conclusion of these field validation results is the
fact that the results of the thermal Radiello mainly for
benzene are systematically higher than the results from the
solvent Radiello. This means that the applied uptake rate for
the thermal Radiello is possibly too low, which is also the
conclusion from the laboratory validation.
Conclusions
Both samplers, Radiello with Carbograph 4 as sorbent and
Perkin Elmer (PE) filled with Carbopack B, have been
subjected to the different tests as described in EN 13528-2
(2001). The sorbent characteristics and the effects of the
environmental conditions were tested in the laboratory
investigation. Then both samplers were exposed in the field,
for a comparison between the reference method (active
sampling followed by GC analysis) and the solvent
Radiello, which has already been used for a longer time for
ambient air monitoring. It turns out that the PE sampler
shows a limiting factor. The high blank values and the low
uptake rates cause analytical problems. The Radiello
sampler avoids this problem thanks to the higher uptake
rates, which were compared to literature values. For
Radiello, the uptake rates are slightly higher (10-20 %). For
PE, the uptake rates are situated in the range of values
found in literature, except for benzene which suffers the
most from the analytical problem. However, in spite of
these problems, the parameters to be tested were determined
successfully. The other sorbent properties (storage test, leak
test) showed no problems, except for back diffusion. For
Radiello there is a loss of 7 %. The wind speed, sampler
orientation and the relative humidity (environmental
conditions) did not influence significantly the uptake rates.
The theoretical formula which describes the temperature
dependence of the uptake rates is confirmed in the
temperature test. Longer exposure times lead to a decrease
of the uptake rates: for PE until 23 % for benzene (less for
the other compounds) and for Radiello only for benzene (-
measuring period 1 measuring period 2
loc. 7 concentrations in g/m loc. 7 concentrations in g/m
method
B T EB mp-X o-X B T EB mp-X o-X
A 3.2 5.2 1.2 3.3 0.94

1.8 4.2 0.9 2.5 0.7
B 3.6 5.6 1.2 3.1 1.1

3.2 4.1 0.8 2.7 1.2
C 3.5 4.4 1.1 3.3 0.91

1.9 3.9 0.9 2.7 0.9
D 3.0 4.9 1.4 4.0 0.98

1.6 3.9 1.0 3.0 0.6
loc. 8 concentrations in g/m

loc. 8 concentrations in g/m
A 4.3 8.1 1.6 3.7 2.5

1.7 5.2 0.9 2.1 1.2
B 3.1 10 1.6 3.6 1.6

2.5 5.9 0.9 2.8 1.1
C 3.7 8.3 1.6 4.1 1.9

2.3 6.7 1.2 2.8 1.2
D 3.4 9.3 2.0 4.8 1.5

1.9 6.3 1.4 2.9 0.9
9 . . . . . . . . . . . . . . . . . . . . . . . .

February 2006
10 . . . . . . . . . . . . . . . . . . . . . . . .
14 %). This effect is theoretically explainable and limits the
measuring range where the same uptake rate is valid. A
change in concentration level has more effect on the PE
uptake rate than on the Radiello uptake rate.
3. P. Bruno, M. Caputi, M. Caselli, G. de Gennaro, M. de Rienzo,
Reliability of a BTEX radial diffusive sampler for thermal desorption:
field measurements, Atmospheric Environment, 39, 2005, 1347-1355
4. Anne Pennequin-Cardinal, Herv Plaisance, Nadine Locoge, Olivier
Ramalho, Sverine Kirchner, Jean-Claude Galloo, Performances of
the Radiello diffusive sampler for BTEX measurements: influence of
environmental conditions and determination of modelled sampling
rates, Atmospheric Environment, 39, 2005, 2535-2544
A comparison with the reference method was done at two of
the eight measuring points of the field validation. During the
first measuring period there is a good correlation between
the different sampling methods, thermal and solvent
Radiello, PE sampler and the reference method. This
correlation was also good in the second period, ignoring the
PE sampler which possibly experienced blank problems. The
deviations of the solvent Radiello against both samplers to
be validated are smaller than 15 % for about the half of the
measurements of the field validation.
5. P. Prez Ballesta, R. Connolly and N. Cao, Blanks on Carbograph 4,
The Diffusive Monitor, 13, September 2002.
6. U. Wideqvist, V. Vesely, C. Johansson, A. Potter, E. Brorstrm-
Lundn, K. Sjberg, T. Jonsson, Comparison of measurement
methods for benzene and toluene, Atmospheric Environment, 37,
2003, 1963-1973.
7. Fondazione Salvatore Maugeri; Instructions for benzene, toluene and
xylenes sampling by Radiello and thermal desorption recovery,
Edition 01/2004.
8. R.H.Brown, Environmental use of diffusive samplers: evaluation of
reliable diffusive uptake rates for benzene, toluene and xylene; J.
Environ. Monit., 1, 1999, 115-116. About half of the 40 measurements are less than 15 %
deviation for each comparison. The PE problem appears in
measuring period 2. Seventeen measurements deviate more
than 30 % from the solvent Radiello. The similarity between
both Radiellos is remarkable in measuring period 2. An
important conclusion of these field validation results is the
fact that the results of the thermal Radiello mainly for
benzene are systematically higher than the results from the
solvent Radiello. This means that the applied uptake rate for
the thermal Radiello is possibly too low, which is also the
conclusion from the laboratory validation.
9. Bo Strandberg, Anna-Lena Sunesson, Katarina Olsson, Jan-Olof
Levin, Gran Ljunqvist, Margit Sundgren, Gerd Sllsten, Lars
Barregard, Evaluation of two types of diffusive samplers and
adsorbents for measuring 1,3-butadiene and benzene in air,
Atmospheric Environment, 39, 2005, 4101-4110.
10. Physical dynamics and analytical chemistry of diffusive sampling
devices, VITO report nr. 2000/MIM/R/128.
11. E.Goelen, M.Lambrechts, F.Geyskens and T.Rymen, Development
and performance characteristics of a capillary dosage unit with in situ
weight sensor for the preparation of known amounts of gaseous VOCs
in air; Intern. J. Anal. Chem., 1992, 217 225.
12. EN 13528-2:2002. Ambient air quality Diffusive samplers for the
determination of concentrations of gases and vapours Requirements
and test methods, CEN, Brussels.
References

1. http://www.vmm.be

2. http://www.vito.be/english/index.htm
Methodology for assessing the packing integrity of adsorbents within thermal
desorption tubes and traps
Andrew Tipler
a
, Neil T. Plant
b
and Michael D. Wright
b

a
PerkinElmer LAS, 710 Bridgeport Avenue Shelton CT, USA,
b
Health and Safety Laboratory, Harpur Hill, Buxton SK17 9JN UK

Introduction
from coconut charcoal. As
can be seen, although the
particles are slightly
eccentric in shape, they
have rounded edges and are
of uniform size. Such
material should produce a
nicely packed tube or trap.
This paper describes new technology that was developed to
help evaluate the integrity of the packing inside thermal
desorption (TD) tubes and traps. Adsorbent packing
materials are either produced from natural materials or may
be synthesized using proprietary techniques. The ideal
packing will be made up from symmetrical particles of
uniform size and will be packed into the TD tubes and traps
in a regular pattern so as to minimize the gaps between
adjacent particles. Unfortunately, this ideal is rarely met as
the particles can be irregular in shape and of varying size.
The process of packing the tubes and traps and the
subsequent handling and use may further degrade the
packing uniformity.
Figure 2 shows how the
same packing material
appears after moderate use
in a TD trap. Note how the
particles are distorted and
fragmented into fines.
Clearly, a trap packed
with this material is going
Figure 1 shows an image of a carbon black packing
(Carbotrap C, ex. Supelco) taken from a fresh bottle. This is
a very delicate and friable packing material that originates Figure 1 Example of an unused
carbon black adsorbent
11 . . . . . . . . . . . . . . . . . . . . . . . .

To obtain a predictable diffusive uptake rate, the
temperature must be monitored in order to determine the
diffusive coefficient. The geometry of the diffusion gap at
the tube inlet must be carefully controlled for predictable
results. Note that the diffusive uptake rate is inversely
proportional to the diffusion gap path-length. For the
standard PerkinElmer diffusive monitoring tubes, this path-
length is 15 mm. The bed-length of the adsorbent in these
tubes is typically of the order of 50 mm. A settlement of
only 1 mm may increase the diffusion path by this length
and will reduce the diffusive uptake rate by 6%. This
relationship is approximately proportional so that a 2 mm
settlement will reduce the diffusive uptake rate by 12%.
Such deviations represent very significant analytical errors
and so some means of detecting settlement of the packing is
desirable in order to detect these potential effects.
to have different flow and retention characteristics from that
if packed with the material shown in Figure 1.
Figure 3 Settlement of adsorbent in a glass
tube creating a void
Figure 2 Carbon black
adsorbent after moderate use
Besides the distortion of the adsorbent particles there is the
possibility that the packing
will settle and leave air
gaps or voids in the
adsorbent bed. Figure 3
shows a glass tube in which
this has happened. Where
voids are present, gas
(sample stream or
desorption gas) will take
the path of least resistance
and travel through the void
rather then the adsorbent
thus compromising tube or
trap performance. It is
important to note that the
trap from which the
packing shown in Figure 2
was taken was performing
reasonably well but apparently not as well as it might.
Theory
Work performed in the mid-nineteenth century by Darcy,
Poiseuille and Hagen modelled the flow of fluids through
porous media and tubular vessels. Based on these
relationships, Equation 2 represents the flow-pressure
relationship through a TD tube or trap packed with an
adsorbent bed. This is essentially the same equation that
describes flow through a GC packed column but, because
the pressure drop across the tube or trap is (relatively) low,
gas compressibility terms may be neglected.

.

=
L
p A B
F
o
o
( 2)

where:

F
o
is the gas flow rate at the tube outlet,
B
o
is the specific permeability coefficient of the
packing ,

A is the tube inner cross sectional area,
L is the adsorbent bed length,
is the viscosity of the gas,

p is the pressure drop across the packing.

Particle distortion, fragmentation and settling may also
affect the flow rate of gas passing through the adsorbent bed
and thus lead to errors in sampling volumes and analytical
split ratios.
Of special interest in the permeability coefficient, B
o
, this is
a geometric term that is based on the particle size and the
spacing between adjacent particles in essence, it represents
the packing integrity. For a tube of trap of given diameter,
packing bed-length and gas viscosity, the value of B
o
will be
proportional to the flow rate through the adsorbent bed
divided by the pressure drop across it. Thus by determining
both the flow rate and the pressure drop through a tube or
trap, a measure of the packing integrity may be derived.
In the case of diffusive monitoring, we can consider Ficks
first law as shown in Equation 1.
z
A D
U
m

= (1)
where:
Existing methods
U
m
is the diffusive uptake rate for a given analyte,
To date, users wanting to monitor the tube packing
impedance normally used something like the manual system
shown in Figure 4. The impedance of traps was not usually
monitored because it required periodic removal of a trap
from the TD during use.
D is the diffusion coefficient for a given analyte,
A is the tube diffusion gap cross-sectional area,
z is the tube diffusion gap path-length.

February 2006
12 . . . . . . . . . . . . . . . . . . . . . . . .
p
a
is the absolute pressure of the gas entering the tube
using the automated method,
As shown in Figure 4, a fixed flow rate of gas (normally
nitrogen or air) was supplied by a mechanical mass flow
controller to the inlet of the tube under test. The outlet of the
tube was vented to atmosphere. The pressure drop across the
tube during the test was indicated by either a water
manometer or a sensitive (gauge) pressure transducer. The
application and results of such a system are reported in a
previous issue of The Diffusive Monitor [1]. Whereas the
manual method gave good results, the process was very
labor intensive and exposed the tubes to a lot of additional
handling and potential sources of contamination and so a
more secure and automated method was sought.
T
m
is the absolute temperature of the packing with the
manual method,
Gas at 50 or
100mL/min
Water
Manometer
Tube Under
Test
Mass
Flow
Controller
Pressure
Transducer
Vent
T
a
is the absolute temperature of the packing with the
automated method,
m
is the gas viscosity of the gas flowing through the
packing using the manual method,
a
is the gas viscosity of the gas flowing through the
packing using the automated method.

Proof of concept work
To validate the applicability of Equations 2 and 3 to thermal
desorption tubes, a series of tests were conducted using the
test rig shown in Figure 5. Only tubes were examined as
only the results from these needed to be correlated against
manual measurements.
Figure 4 System used to make manual measurements of
tube impedance. In this system, carrier gas is applied at a fixed pressure to the
tube inlet. An electronic mass flow controller is attached to
the tube outlet. This mass flow controller operated using the
thermal conductivity principle. The advantage of this
technology is its ability to maintain the set flow rate as the
upstream pressure changes this is not the case with
mechanical flow controllers.
Transducers
P
Water manometer
Tube under Test
P
Electronic
Flow
Controller
Electronic
Pressure
Regulator
Constraints imposed by automating the measurements
Although automation of the method appears to be an
attractive option, there will be some significant differences
between it and the manual method. The main difference is
that within a TD instrument, the gas flowing through the
tube and trap will be the carrier gas and will not be air and
possibly not nitrogen. Another change will be that the tube
and trap will be located within existing plumbing circuits
and so may be at elevated pressures at the time the
impedance measurement is made. Finally the adsorbent bed
may be held at a temperature other than ambient during the
measurement.
Figure 5 Test rig to study relationship between flow rate
and pressure drop across thermal desorption tubes.
A high sensitivity differential pressure transducer (10 kPa
range) was connected across the tube and a second (gauge)
transducer (100 kPa range) measured the pressure applied to
the tube. A water manometer was used to calibrate the
differential pressure transducer. Table 1 gives details of the
test equipment used in this study and Figure 6 shows a
photograph of the assembled system.
Pressure drop measurements made at a different pressure,
gas type and temperature can be easily related to results
from manual measurements by applying Equation 3.
a
m
a
m
m
a
a m
T
T
p
p
p p
= (3)
where:
p
m
is the corresponding pressure drop across the packing
using the manual method,
p
a
is the corresponding pressure drop across the packing
using the automated method,
p
m
is the absolute pressure of the gas entering the tube
using the manual method,
13 . . . . . . . . . . . . . . . . . . . . . . . .

Table 1 Components used in test rig to validate pressure-
flow characteristics of TD tubes
Figure 7 shows the results obtained from two types of
thermal desorption tubes run on the test rig. These data show
a reasonable reproducibility in the measured pressure drop
across each group of tubes and clearly there is a significant
difference in the pressure drop between the two types as
expected. These data were highly encouraging.
Pressure Regulator PerkinElmer Clarus 500 PPC (600 kPa)
Back-Up Pressure Regulator Parker mechanical (650 kPa)
Differential Pressure Transducer Motorola/Freescale MPX5010 (10 kPa)
Gauge Pressure Transducer Motorola/Freescale MPX5100 (100 kPa)
Mass Flow Controller Tylan 0-100ml/min
Manometer Water in Tygon tubing
Equation 2 indicates that, with all other conditions kept
constant, the pressure drop across the packed bed should be
directly proportional to the flow rate of gas passing through
it. Figure 8 shows a plot of pressure drop versus flow rate
for a selected tube and demonstrates that this equation is
valid.
The next test was to try and relate the
pressure drop results at an elevated
pressure against results taken with the tube
at ambient pressure.
Figure 9 shows a plot of results from a
batch of tubes measured under two applied
pressures. For Equation 3 to be valid, the
slope of the line produced should be the
ratio of the two absolute pressures (with all
other conditions being kept constant). In
this case, very good agreement is
demonstrated as shown in Table 2 an
error of only 1.7% between the predicted
and actual pressure ratio result is observed.
This result confirms that pressure drop
results at elevated pressure may be used to
accurately generate results that would be
expected if the measurements were taken
at ambient pressure.

Table 2 Comparison between predicted and observed
(Fig. 9) pressure ratios from pressure drop measurements at
ambient and elevated pressures.
Figure 6 Photograph of test rig to validate pressure-flow
characteristics of TD tubes
Electronic
Flow
Controller
Electronic
Pressure
Regulator
Power
Supply
DVM
GC
Backup
Regulator
Transducer
Tube
Flow
Controller
Electronic
To
Manometer
S St ta ai in nl le es ss s S St te ee el l T Tu ub be e, , R RS SD D = = 4 4. .7 7% %
G Gl la as ss s T Tu ub be e, , R RS SD D = = 7 7. .2 2% %
P
r
e
s
s
u
r
e

D
r
o
p

Tube no.
20 15 10 5
2.0
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0
p
a
kPa p
m
kPa Ratio Slope Error, %
151 101 1.50 1.52 1.7

Table 3 Comparison between predicted and observed
(Fig.10) viscosity ratios from pressure drop measurements
with nitrogen and helium at ambient temperature.
N
2
cP He cP Ratio Slope Error, %
0.0176 0.0196 0.90 0.92 2.2

A similar test was used to study the effects of different
carrier gases and hence gas viscosity. Figure 10 shows
pressure drops measured from a batch of tubes when tested
with two different carrier gases. The slope of the line fitted
to these data should be the same as the ratio of the viscosity
of the two gases used. The data in Table 3 show good
Figure 7 Pressure drops across 20 glass and 20 stainless steel
tubes packed with Tenax TA with 50ml/min nitrogen flowing
through each at an inlet pressure of 50 kPa.
February 2006
14 . . . . . . . . . . . . . . . . . . . . . . . .
agreement (only a 2% error) between the experimental and
theoretical viscosity ratios thus confirming that results
obtained using one type of carrier gas can be related to those
obtained using another. Interestingly, the expected ratio does
not change significantly with temperature. If a nominal
temperature of 25C is assumed, then the maximum error in
the pressure drop measurement would be about 1%. Such an
error is statistically small and so temperature deviations such
as these can be ignored in the results.
y = 0.0118x
R
2
= 0.9991
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 20 40 60 80 100
Flow Rate, (mL/min)
P
r
e
s
s
u
r
e

D
r
o
p
,

(
k
P
a
)
Practical implementation
Figure 11 shows a schematic diagram that illustrates one
approach to the implementation of technology to automate
impedance measurements.
Figure 8 Pressure drop versus flow rate for helium at
50 kPa, Air Toxics tube.
In this implementation, there are two independent pneumatic
circuits that have been designed to optimize the normal, 2-
stage, thermal desorption analytical process. Such an
approach is also ideal for tube and trap impedance
measurements as it allows a standard pressure of 50 kPa to
be applied to the tube while allowing the flexibility in the set
pressure applied to the trap that is necessary to maintain
carrier gas into the GC column throughout the measurement
process.
y = 1.5177x
R
2
= 0.9965
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.0 0.5 1.0 1.5
Pressure Drop at 50kPa
P
r
e
s
s
u
r
e

D
r
o
p

a
t

0
k
P
a
y = 0.9173x
R
2
= 0.9978
0.52
0.54
0.56
0.58
0.60
0.62
0.64
0.66
0.68
0.70
0.72
0.58 0.60 0.62 0.64 0.66 0.68 0.70 0.72 0.74 0.76 0.78
Pressure Drop with Helium (kPa)
P
r
e
s
s
u
r
e

D
r
o
p

w
i
t
h

N
i
t
r
o
g
e
n

(
k
P
a
)
Figure 9 Pressure drop at ambient pressure versus 50 kPa
on steel and glass tubes packed with Tenax TA, nitrogen
carrier.
G Gl la as ss s
S St ta ai in nl le es ss s
S St te ee el l
2.0
In both instances a flow rate of 50 ml/min is allowed to flow
through the packing. Electronic mass flow controllers are
used on the outlets to regulate these flows. Separate
differential pressure sensors are used to measure the pressure
drops across the tube and trap. A further gauge pressure
sensor is used to monitor the tube inlet pressure so that the
equivalent manually measured pressure drop may be
calculated. The impedance measurement is made on each
tube and on the trap during each analysis and prior to tube
desorption. In both cases, the flow direction of gas during
the impedance measurement is the same as the sampling
direction; this minimizes the risk of analyte breakthrough.
The connections to the pressure transducers are separated
from the pipes through which gas will flow during the
measurements. This ensures that pressure drops within the
pipes do not contribute to the measurement.
One aspect of this implementation that was unexpected was
the pressure drop observed across the filters that protect the
rotary valve from possible damage from particles admitted
with the gas flows. Table 4 shows the pressure drops
measured across different types of particle filter that may be
used in this location.
Figure 10 Glass Air toxics tubes with nitrogen and helium
carrier gases at 50 kPa and 50 ml/min.

15 . . . . . . . . . . . . . . . . . . . . . . . .

Table 4 Pressure drops across 9 mm filters with helium at
50 kPa and 50 ml/min
Filter Type
Pressure
Drop,
kPa
-1

Imped. Strength Inertness
High Temp.
Tolerance
PTFE
Standard 1.041 High Medium High Poor
Zitrex 6104 0.903 High Medium High Poor
Glass Fibre
Glass Fibre A 0.752 High Low High High
Glass Fibre B 0.655 High Low High High
GF/B* 0.281 Medium Low High High
GF/D* 0.175 Medium Low High High
EPM 2000* 0.372 Medium Low High High
QMA* 0.398 Medium Low High High
Steel
1m screen** 0.035 Low High High*** High
* Ex. Whatman
** Ex. Valco
*** After Sulfinert
TM
coating, ex. Restek
As has been already demonstrated, the pressure drop across
a tube under these conditions may be 0.6kPa or lower. This
means that the pressure drop across some of these filters will
cause a very significant error in the impedance
measurements; in a few instances, including the Standard
filters used on these instruments to date, the pressure drop
across the filter may be much higher than that of a tube.
Inspection of Table 4 indicates that the three filters made
with stainless steel screens give the best physical
characteristics including a very low pressure drop across

Particulate Filters to Protect
Valve
Rotary Gas
Valve
50mL/min
50mL/min
P
To Mass Flow
Controller
To Mass Flow
Controller
Column
Carrier Gas
P
P
Desorb Gas
@ 50kPa
Transfer
Line/Column
Tube
Trap
Figure 11 A practical
system for automating
tube and trap impedance
measurements.
each of them. The Sulfinert
TM
deactivation process did not
significantly change the impedance of these filters. Although
the impedances of the various filters are significant when
compared against those of the tubes, they are minor when
compared against the absolute pressure applied to the tube
(150 kPa). In such cases, the relationship given in Equation
4 would apply.
f a t
p p p + (4)
where:
p
t
is the total pressure drop across the tube and the
filter connected in series
p
a
is the pressure drop across the tube when measured
alone
p
f
is the pressure drop across the filter when measured
alone
Once the impedance of the filter, p
f
, is known, it can be
subtracted from the total measurement, p
t
, made on the
instrument to give the pressure drop across just the tube,
p
a
. When installed, the filter is closely mechanically
coupled to the valve and so it will attain a temperature close
to that of the valve. The impedance of the filter will be
highly dependent on its temperature and so this needs to be
taken into account when applying Equation 4. What this
means in practice is that the filter impedance should be
recalibrated if the valve temperature is adjusted. To validate
Equation 4 when applied to tubes with different valve
temperatures, a series of impedance measurements were
conducted using a moderate impedance filter (GF/D) being
present.
February 2006
16 . . . . . . . . . . . . . . . . . . . . . . . .
The tube transducer and the particulate filter are calibrated
by passing carrier gas at 50 ml/min through an empty tube.
The instrument will record this reading and subtract it from
subsequent measurements when sample tubes are being
tested. That result will be used to calculate the equivalent
pressure drop if the tube was measured manually.

Figure 12 Effect of GF/D glass fibre filter temperature on
pressure drops with and without an Air Toxics tube, helium
at 50 kPa and 50 ml/min.
0.0
0.2
0.4
0.6
0.8
1.0
1.2
150 175 200 225 250 275 300
Valve Temperature (C)
P
r
e
s
s
u
r
e

D
r
o
p

(
k
P
a
)
(GF/D + Air Toxics) GF/D
GF/D (empty tube)
GF/D + Air Toxics
In the case of the trap, no gas flow is passed through the
trap during calibration. In this instance we are not so
concerned about obtaining a standardized measure of the
packed bed impedance but rather use the information to
track changes in prolonged use of the trap.
In use, the system is designed to measure the impedances of
both the tube and the trap immediately after both have been
leak checked but before the tube desorption is initiated. This
process does not usually add to the analysis time and results
are provided for every tube analyzed. Results of the
measurements are provided in the form of a log file; an
example of such is given in Figure 14.
Figure 12 shows that the pressure drop does increase as the
filter temperature is increased. If the pressure drop across
just the filter (lowest plot) is subtracted from the pressure
drop across the filter and the tube (highest plot), the result
(middle plot) shows a consistent result for the pressure drop
across the tube. These data validate the use of Equation 4 for
calculating the pressure drop across just the tube from a
combined result over a wide range of filter temperatures.
The impedance of the filter is calibrated at its set
temperature as shown in Figure 13.
filter
p
Rotary Gas
Valve
50mL/min
No Flow
P
To Mass Flow
Controller
Column
Carrier Gas
P
P
Desorb Gas
@ 50kPa
Transfer
Line/Column
Empty
Tube
Trap
Figure 13 Calibrating the differential pressure transducers and filters.

17 . . . . . . . . . . . . . . . . . . . . . . . .

0.0
0.5
1.0
1.5
2.0
2.5
3.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 18 19 16 17 20 21 22 23 24 25 26 27 28 29 30
Tube #
R
e
l
a
t
i
v
e

S
t
a
n
d
a
r
d

D
e
v
i
a
t
i
o
n

(
%
)
Glass/Tx
Gl/AirToxics
Steel/Tx
Gl/Empty
Gl/Gl Wool
Figure 16 Precision of tube impedance measurements on
same 30 tubes as shown in Figure 15 (Tx - Tenax; Gl -
glass).
Imprecise results may indicate movement of the
adsorbent packing inside the tube. One possible way of
checking this further is to repeat the tube impedance
measurement with the tube reversed in the flow- path.
Table 5 shows the impedance results of a batch of 10
tubes measured in both directions. In this case, excellent
agreement in the results from both directions is seen
indicating no discernable movement in the packing.
Figure 14 Example of log file content showing results
of impedance measurements. The asterisks indicate non-
compliance with user-set limits.
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 18 19 16 17 20 21 22 23 24 25 26 27 28 29 30
Tube #
I
m
p
e
d
a
n
c
e

(
k
P
a
)
Glass/Tx
Gl/Air Toxics
Steel/Tx
Gl/Empty
Gl/Gl Wool

Table 5 Impedance results from glass Air Toxics tubes
measured in both directions with helium at 50mL/min
and 50 kPa.
Tube Forward (kPa) Reverse (kPa) Ratio
Y44912 0.677 0.679 1.00
Y44913 0.612 0.612 1.00
Y44914 0.665 0.668 1.00
Y44915 0.684 0.686 1.00
Y44916 0.728 0.733 1.01
Y44917 0.672 0.675 1.00
Y44918 0.763 0.768 1.01
Y44919 0.591 0.593 1.00
Y44920 0.637 0.640 1.00
Y44940 0.735 0.738 1.00
Figure 15 Results of impedance measurements on 30
different tubes (Tx - Tenax; Gl - Glass).

Application
Figure 15 shows the impedance results obtained from a rack
of 30 different tubes. Although there is significant variation
in the measured impedances between these tubes, when the
tubes are compared against those of the same type (as
colour-coded) there is good agreement throughout. The
technique is even sensitive to some glass wool packing in a
tube rather than an adsorbent.

Figure 3 shows a glass tube with a void in the packing. Once
the void was eliminated by tapping down the packing and
relocating the retaining plugs, a very different impedance
result was observed as shown in Table 6.

Table 6 Impedance measurements of a single glass Air
Toxics tube with and without void with helium at 50mL/min
and 50kPa
Figure 16 shows the precision obtained when running the
same 30 tubes repeatedly for a total of 8 measurements. For
some of the tubes no variation in results were observed.
Most tubes had a precision of less than 1%. This indicates
the robustness of the impedance measurements.
Tube Condition Press. Drop (kPa)
With Void 0.563
Void Eliminated 0.794
February 2006
18 . . . . . . . . . . . . . . . . . . . . . . . .
The changing impedance with packing density enables the
use of the system as a means of checking for good packing
technique. Users packing their tubes typically put a small
amount of adsorbent packing in the tube, tap it down and
repeat the process until the required bed-length is filled or
the required weight of adsorbent has been added. If all the
adsorbent is added at once, then the packing density is not so
good as shown in Figure 17.
The tube was tapped about 10 times and its impedance was
measured and then the process was repeated. Figure 18
shows a plot of the impedance of this tube versus the
number of times it was subjected to a tapping session. The
plot clearly shows that a steady state is reached after about 7
such sessions indicating that the tube may now be properly
packed.
Whereas light tapping (Figure 17) can improve the packing
inside a tube (or trap) tamping with an object such as a metal
rod as shown in Figure 19 is definitely not recommended.
Some adsorbents, particularly some of the carbon blacks, are
highly friable and are easily deformed when pushed down
with a solid object.
Table 7 shows what happens to the impedance of an Air
Toxics tube when moderately tamped. In this case the
impedance has effectively doubled with consequential
effects on the tubes analytical performance.

Table 7 The effect of moderate tamping on the
impedance of an Air Toxics tube.
Condition Press. Drop (kPa)
Initial 0.831
After Tamping 1.667

So far we have looked at data from adsorbent tubes. Figure
20 shows the impedance measured on 6 different Air Toxics
traps. Very consistent results are apparent indicating that a
similar performance should be expected from each.
Note that the results from the traps, unlike the tubes, are now
dependent on the carrier gas pressure set for the GC column
and on the initial (low) temperature of the trap. Thus it is
difficult to make comparisons between traps used on
different systems with different methods but it is very
convenient to track any changes impedance on a given trap.
Figure 17 Air Toxics tube packing being tapped
down with a light metal pen.
0.59
0.60
0.61
0.62
0.63
0.64
0.65
0.66
0 1 2 3 4 5 6 7 8 9
Tapping Session
P
r
e
s
s
u
r
e

D
r
o
p

(
k
P
a
)
Figure 21 shows how the system can be used to monitor trap
impedance over long periods. In this case the impedance is
being reduced perhaps indicating a slight spreading in the
packing density. The user can set limits on this deviation to
stop the system when it is decided that the change is too
great (e.g. with a large movement of the packing).
Figure 18 The effect on measured impedance of
repeated tapping of a poorly packed Air Toxics tube.
19 . . . . . . . . . . . . . . . . . . . . . . . .

0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
A B C D E F
Trap
P
r
e
s
s
u
r
e

D
r
o
p

(
k
P
a
)
Figure 20 Impedance measurements from six different Air
Toxics traps at 30C with helium at 50 ml/min and 100 kPa.
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
0 50 100 150 200 250 300
Run #
Figure 19. Using a metal rod to tamp down
adsorbent packings not to be recommended.
Conclusions
In this paper we have described the need for tube and trap
impedance testing, the operating principles for automated
measurements and an approach to the technology needed for
a practical implementation. The technique has been shown
to be sensitive and effective and should prove to be a
valuable tool in ensuring that the packing inside thermal
desorption tubes and traps is maintained in optimum
condition. This is especially important to vendors who are
trying to provide product at a consistently high quality and
to users that want to get the best out of their tubes and traps
yet be able to identify the point at which they degrade
beyond acceptable limits. Further work is required to
establish acceptable standards for tube and trap impedances
so that all laboratories will be using such devices of
consistent quality.
Figure 21 Tracking the impedance of a given Air Toxics
trap at -30C with helium at 15 ml/min and 100 kPa
References
1. M. D. Wright, N. T. Plant and K. Gostlow, The effect of sorbent back-
pressure on apparent recovery of toluene in thermal desorption, The
Diffusive Monitor, 14, December 2003, 11-13.

February 2006
Minimising potential discrimination during calibration of thermal desorption methods
with liquid injection
Elizabeth Woolfenden, Markes International Ltd., Gwaun Elai Medi Science Campus, Llantrisant RCT, CF72 8XL, UK
Introduction
It is not uncommon for calibration standards, used during
thermal desorption (TD) analysis of vapours trapped on
sorbent tubes, to be different, in some respects, from field
samples. The ideal TD standard is prepared by introducing a
fixed volume of an accurate gas standard of the compounds
of interest onto an identical sorbent tube to those used for
field sampling of vapours. However, accurate gas standards
are notoriously difficult and expensive to generate;
particularly in the case of labile, polar or less volatile
analytes and at low concentrations. For this reason, and as
specified in most TD standard methods, routine thermal
desorption calibration is normally carried out by introducing
liquid standard solutions to the sampling end of sorbent
tubes in a stream of carrier gas. The bulk of solvent is then
generally purged from the sorbent tube during the standard
loading process.
However, it is not always possible to purge the solvent away
completely for example when dealing with volatile
compounds of interest or with strong sorbents. In this case,
standard tubes containing relatively large masses of solvent
(possibly several milligrams) will be used to calibrate
sample tubes which are solvent-free. Conversely, whether
loaded from the liquid or gas phase, standard tubes are
generally dry. However, tubes used for field sampling may
contain several milligrams of water, depending on sorbent
and prevalent atmospheric conditions.
This paper describes the issues which may arise as a result
of differences between samples and calibrants for thermal
desorption and the operating conditions under which the
effects may be significant. Guidance for minimising
discrimination and for validating that a given method is
quantitative is also presented.
The potential for split discrimination due to high
water/solvent levels
The potential impact of varying tube impedance on thermal
desorption-based analytical data has been the subject of
recent discussion [1]. However, impedance variation is
unlikely to be a significant cause of quantitative error except
when operating under unusual analytical conditions i.e.
low carrier gas head pressure (< 10 psi) and high inlet split
flow (i.e. >100 ml/min split flow during primary (tube)
desorption.)
However, residual solvent or exceptionally high humidity, in
standard or sample tubes respectively, is a much more likely
cause of significant analytical discrimination in TD methods.
The issue can arise during inlet, outlet or double split
methods. Even splitless desorption methods, though immune
to split discrimination, can suffer from adverse
chromatographic effects due to high solvent or water
content, which can effect quantification.
An example of discrimination due to variable solvent
content is shown in Tables 1 and 2 (reproduced from HSL
Report OMS/2002/15 with the kind permission of the
Health and Safety Laboratory, Buxton, UK).
The data shown in Table 1 was generated by introducing
~17 g of toluene gas standard onto tubes packed with two
different quantities of sorbent 100 mg and 530 mg of
Chromosorb 106 respectively. Despite the large difference in
impedance between these 2 groups of tubes and the fact that
an inlet split method was used (i.e. >62 ml/min split flow
was operating during primary (tube) desorption) no
significant difference in analyte recovery was observed.
Data shown in Table 2 was generated by introducing ~20 g
of toluene in 5.0 l of methanol solution (3.4 mg/ml) to both
groups of tubes (100 mg and 530 mg of Chromosorb 106) in
a stream of carrier gas (50 ml/min). The analytical
conditions were the same as those used for analysis of the
gas standard (see Table 1) in which no discrimination due to
impedance variation was observed. The 26% reduction in
data between tubes packed with 100 mg and those packed
with 530 mg of Chromosorb 106 was thus purely due to the
excess solvent retained by the tubes with the larger mass of
sorbent.
The cause of the difference in the two measurements is split
discrimination due to flash vaporisation of the retained
solvent during desorption. As the solvent vaporises and
expands this raises the gas pressure inside the TD flow path,
temporarily increasing the split flow relative to the desorb
and/or column flow. Systems with manual pressure control
of split flow will be most susceptible to this phenomenon,
but even electronic mass flow controlled (MFC) systems
will exhibit the effect as MFCs take a finite time to respond
to the changing pressure and bring the split flow back under
control. Analytes desorbing/eluting from the sorbent at the
same time as the solvent will thus be subjected to a higher
split flow/ratio than later eluting compounds.
Split discrimination of this kind has been extensively
reported for all forms of flash vaporising GC injection. The
best ways to overcome it are to slow down the heating rate
of the sorbent tube/trap (such that the solvent or water
vaporises more slowly and without causing a pressure surge)
and to minimise the mass of solvent/water retained in the
first place.
20 . . . . . . . . . . . . . . . . . . . . . . . .

Table 1 Recovery of 17 g toluene from Chromosorb 106 (loaded from the gas phase),
gas pressure 46.1 psi.

TD flows (ml/min)
.Tube
Mass
C106
(mg)
Back-pressure
(psi) (after
conditioning)
Flow during
conditioning
(ml/min)
Inlet
split
Desorb
Outlet
split
Toluene peak
(FID)
A 100 0.06 98 63 21 80 104.6
B 100 0.03 110 63 21 81 85.2
C 104 0.04 95 63 21 81 107.7
D 101 0.03 92 63 22 82 99.6
E 103 0.07 94 63 21 82 108.7
F 100 0.03 105 62 21 82 101.8

101.3 +/- 9
mean +/- sd
G 531 0.25 30 63 21 81 93
H 528 0.23 31 63 22 81 102.6
I 529 0.23 31 62 21 81 99.6
J 526 0.23 31 62 21 82 90.2
K 527 0.23 30 62 21 82 104.9
L 527 0.23 30 63 21 82 94.5

97.5 +/- 6
mean +/- sd

Table 2 Recovery of 20 g toluene from Chromosorb 106 (loaded from the liquid phase), gas
pressure 46.1 psi
TD flows (ml/min)
Tube
Mass
C106
(mg)
Back-pressure
(psi) (after
conditioning)
Flow during
conditioning
(ml/min)
Inlet
split
Desorb
Outlet
split
Toluene peak
(MS)
A 100 0.06 98 63 21 80 1086
B 100 0.03 110 63 21 81 1081
C 104 0.04 95 63 21 81 1084
D 101 0.03 92 63 22 82 1090
E 103 0.07 94 63 21 82 1087
F 100 0.03 105 62 21 82 1090

1086 +/- 4
mean +/- sd
G 531 0.25 30 63 21 81 822
H 528 0.23 31 63 22 81 842
I 529 0.23 31 62 21 81 792
J 526 0.23 31 62 21 82 792
K 527 0.23 30 62 21 82 784
L 527 0.23 30 63 21 82 770

801 +/- 27
mean +/- sd
21 . . . . . . . . . . . . . . . . . . . . . . . .

February 2006
Key thermal desorption features for minimising and/or
validating split discrimination due to differences in humidity
/solvent content between standards and sample tubes
include:
1. Dry purging sample tubes, in the sampling direction,
before or during an automated desorption sequence. This
helps reduce any differences in humidity content between
the sorbent tubes in a sequence.
2. Gentle (< 5C/sec) heating of the primary tube
desorption oven from near ambient at the start of tube
desorption to ensure gradual solvent/water vaporisation and
to eliminate any risk of a pressure surge which might change
the inlet split ratio. Note that a pre-heated tube desorption
oven was used to generate the data shown in Table 2. This is
more prone to causing flash vaporisation and inlet split
discrimination during tube desorption with samples
containing high levels of solvent/water.
3. Quantitative re-collection of both inlet and outlet split
for validation of split consistency at all stages of thermal
desorption operation.
4. Use of an extended (60 mm) bed length of sorbent in the
focusing trap (allowing trap sorbent selection to match that
of sample tubes) ensures quantitative retention of target
analytes without sub-ambient cooling, which in turn
facilitates automatic and selective purging of
water/solvent from the focusing trap during and after
primary (tube) desorption. The total maximum sorbent bed
length used in standard sample tubes is 60 mm, comprising
up to 4 sorbents arranged in order of increasing sorbent
strength. If the type and bed length of each sorbent used in
the focusing trap can replicate what is used in the sample
tube for air monitoring, this ensures quantitative recovery of
target analytes without sub-ambient cooling. While the I.D.
of a TD focusing trap is much smaller than that of a sample
tube (to ensure optimum desorption / efficiency during
capillary GC analysis) the volume of gas passing from the
sample tube to the focusing trap during primary (tube)
desorption (typically <500 ml) is invariably lower than that
used for tube sampling (typically >5l). Therefore any
compound which can be quantitatively retained by the
sample tube during air monitoring at ambient temperatures,
can also be quantitatively retained by the focusing trap at
ambient temperature. The advantage of this is that most
volatile solvents and water can be selectively purged from
sorbent tubes/traps at ambient temperature. In effect the
extended sorbent bed of the focusing trap thus introduces a
second automated dry purge / solvent purge step allowing
the mass of water/solvent to be reduced, typically by another
order of magnitude.
References
1. M D Wright, N T Plant and K Gostlow, The effect of flow resistance
on the recovery of volatile compounds from thermal desorption air
sampling tubes, HSL Report OMS/2002/15, 2002.

22 . . . . . . . . . . . . . . . . . . . . . . . .
23 . . . . . . . . . . . . . . . . . . . . . . . .

THE DIFFUSIVE MONITOR
The Diffusive Monitor is distributed electronically by E-mail.

It is also available on the HSL web site at: http://www.hsl.gov.uk/publications/diffuse-monitor.htm

If you wish to add your name to the electronic mailing list or delete your name from the list, or have any other
comments, send the reply slip below (or a photocopy) with details of your name and address, tel., fax and E-mail to:
M D Wright, Health and Safety Laboratory, Harpur Hill, Buxton, SK17 9JN, UK. fax +44 1298 218570, E-mail
contact mike.wright@hsl.gov.uk.

THE DIFFUSIVE MONITOR

Response form for fax

... Please add my name to the electronic mailing list

... Please delete my name from the mailing list

E-mail...

Other comments....

Correct address as follows:

Name and Title....

Organisation....

Address....

Post code....

Country....

Telephone....

FAX....

SIGNED....

Return to: M D Wright, Health & Safety Laboratory, Harpur Hill, Buxton SK17 9JN UK, fax +44 1298 218570,
e:mail mike.wright@hsl.gov.uk

Data Protection Act: The above details will be collated on a computer database for the purpose of distributing The Diffusive Monitor only.

Diffusive Monitor 1

Diunggah oleh

Informasi Dokumen

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Diffusive Monitor 1

Diunggah oleh

Hak Cipta:

Format Tersedia

The Diffusive

Anda mungkin juga menyukai