Biomass and Bioenergy 108 (2018) 1–6

Contents lists available at ScienceDirect

Biomass and Bioenergy

journal homepage: www.elsevier.com/locate/biombioe

Research paper

Evaluation of sampling techniques for gas-phase siloxanes in biogas T

a a,b a,∗ a a
Clara M.A. Eichler , Yaoxing Wu , Steven S. Cox , Stephanie Klaus , Gregory D. Boardman
a
Department of Civil and Environmental Engineering, Virginia Tech, Blacksburg, VA 24061, USA
b
Department of Environmental Engineering, Texas A & M University-Kingsville, Kingsville, TX 78363, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Biogas from landfills and wastewater treatment facilities typically contain siloxane contaminants that can cause
Siloxanes severe operational problems in engines and boilers when biogas is used as fuel. Likely due to their relatively low
Biogas volatility, reliable siloxane sampling and analysis has proven challenging, and no standard sampling technique
Sampling for gas-phase siloxanes exists, leading to high variability in analytical results. This study evaluates four tech-
D5
niques commonly used for sampling gas-phase siloxanes. Samples of a reference gas containing dec-
Thermal desorption
amethylcyclopentasiloxane (D5) were taken using these techniques and measured D5 concentrations were
compared to the reference gas D5 concentration. Methanol impingers and thermal desorption tubes proved to be
most accurate and reliable, whereas Tedlar® bags and SUMMA canisters yielded lower D5 recovery rates due to
adsorption to container surfaces. Based on these results, the methanol impinger and thermal desorption tube
sampling methods appear to be more suitable for the quantitative analysis of gas-phase siloxanes in biogas.

1. Introduction is the very commonly found decamethylcyclopentasiloxane (D5, CAS

Biogas produced by anaerobic digestion of sludge from wastewater
treatment plants (WWTPs) and of organic waste in landfills is fre-
quently used for energy production because it contains large amounts of
methane. Recently, organosilicon compounds, often referred to as vo-
latile methyl siloxanes, have become of great concern because they can
cause severe operational problems when combusted in energy produc-
tion processes which use biogas as fuel. When oxidized in combustion
engines or boilers, siloxanes form microcrystalline silicon dioxide,
which deposits as a sand-like residue on engine parts and can result in
catastrophic engine failure. Silicon dioxide can also deposit on boiler
tubes, causing inhibition of heat transfer, thereby resulting in reduced
process efficiency [1,2].
Siloxanes are typically added to consumer products such as cos-
metics, soaps, shampoos, paints and sealants to improve lubricity. The
use of siloxanes in these products has grown rapidly in recent years [3].
Siloxanes find their way into waste streams when personal care pro-
ducts are rinsed to drains and when containers containing product re-
sidue are disposed in landfills. During anaerobic digestion processes,
siloxanes can then be volatilized into the biogas [2].
Although siloxanes are often referred to as volatile organic com-
pounds (VOCs), they possess significantly different mass-transfer
properties as compared with common VOCs due to their relatively high
molecular weight and low vapor pressure. In particular, low volatility
makes siloxane sampling and analysis more challenging. One example
541-02-6) whose vapor pressure is cited to be in the range of 20 Pa to
53 Pa at 25 °C [1,2,4–11].
Because siloxanes have highly negative impacts on operational costs
and the overall efficiency of biogas-to-energy projects, removal strate-
gies are required. To facilitate development and evaluation of effective
siloxane removal processes, sensitive and reliable methods for the
identification and quantification of siloxanes need to be established. In
a recent review on siloxanes in biogas, it was concluded that GC-MS is
the most established method for the analytical determination of silox-
anes in biogas, although other detectors such as FIDs are sometimes
used successfully [12]. More problematic is the selection of a sampling
method for gaseous siloxanes. Although the need for an approved
method is great, no standard method has yet been agreed upon [12,13].
For many sampling techniques, advantages and disadvantages have
been described. Ajhar et al. [13] examined the suitability of Tedlar®
bags for siloxane sampling in landfill gas and concluded that the bags
were a viable option, as long as they had non-adsorptive fittings such as
polypropylene. However, they also reported that losses were highest for
D5 using the bags, which is among the siloxanes most often found in
biogas [13]. The advantages of using Tedlar® bags cited in this work
were ease of handling and low risk of leakage [13]. In a comparison of
Tedlar® bags, adsorbent tubes, and impingers, Raich-Montiu et al. [14]
found no significant differences between the methods, although the
standard deviations of recoveries for octamethylcyclotetrasiloxane (D4)
and D5 with the Tedlar® bag were higher than for the two other

∗
Corresponding author. 1145 Perry Street, 418 Durham Hall, Virginia Tech, Blacksburg, VA 24061, USA.
E-mail address: stcox2@vt.edu (S.S. Cox).

http://dx.doi.org/10.1016/j.biombioe.2017.10.049
Received 19 February 2017; Received in revised form 30 October 2017; Accepted 31 October 2017
Available online 07 November 2017
0961-9534/ © 2017 Elsevier Ltd. All rights reserved.
C.M.A. Eichler et al.
methods. It was concluded that adsorbent tubes filled with an activated
carbon charcoal matrix were most reliable [14]. However, because the
adsorbent tube method required removal of the adsorbent, extraction,
and replacement of the adsorbent prior to the next sampling event, it
was considered to be rather impractical for monitoring of siloxane re-
moval in WWTPs. Tansel and Surita [15] investigated the suitability of
activated carbon as sorbent and found that the affinities of different
siloxanes for the carbons varied strongly. Rasi et al. [2] sampled si-
loxanes with Tenax GR-filled sorbent tubes, but had difficulty com-
paring their results with studies in which other techniques were used.
Schweigkofler and Niessner [16] used stainless steel canisters for the
analysis of siloxanes in landfill and digester gases. After introducing a
pure, 30-compound standard mixture for an equilibration time of 2 h,
the recovery rates were above 90% for all common siloxanes except for
D5 (85-90%) and dodecamethylpentasiloxane (L5) (35%). Longer sto-
rage times lowered the average recovery rates to 85% [16]. This con-
firmed that analyte loss due to surface adsorption increases with in-
creasing boiling point and decreasing vapor pressure, and that the
storage stability of high molecular weight gas-phase analytes must be
taken into account when selecting an appropriate sampling method.
Saeed et al. [17] reached a similar conclusion when comparing siloxane
capture with impingers and canisters. Although the canister method is
simpler and faster than impinger sampling, D4, D5, and D6 could not
sufficiently be recovered from the canister and the impinger proved to
be more reliable and sensitive [17].
A noticeable feature of available literature on siloxane sampling
techniques is the lack of a siloxane-containing reference gas with which
to compare measurements. Instead, past studies relate to an environ-
mental sample with an unknown background matrix and use the
highest concentration found as the “true” concentration [2,13,14,16],
or to liquid standard mixtures which were introduced into bags or
canisters without complete knowledge of the mass transfer processes
involved [13,16,17].
For this study, D5 was chosen for the reference gas because it is
among the most common siloxanes, generally present in high con-
centrations in biogas from WWTPs and at lower concentrations in
landfill gas [2,13,16]. In addition, it is also one of the less volatile si-
loxanes with a vapor pressure of 20.4 Pa (25 °C) and more likely to
adsorb to surfaces [13,14,16,17]. The sampling techniques tested were
thermal desorption (TD) tubes, methanol impingers, SUMMA canister,
and Tedlar® bags because they are commonly used in biogas sampling
(Fig. 1) [14,16]. TD tubes and methanol impingers are direct sampling
techniques in which analytes are removed from the gas stream during
sampling, whereas for indirect sampling techniques, like SUMMA can-
isters and Tedlar® bags, gas samples are captured and stored in con-
tainers prior to analysis.
The TD method can be used for many air monitoring applications,
with a wide variety of commercial sorbents, such as Tenax and gra-
phitized carbon black. In the sampling process, a known volume of gas
sample is drawn through the TD tube. The tube is then sealed and can
theoretically be stored for up to several months at room temperature
[18,19]. During the TD process, analytes are extracted from the sorbent
and introduced to the analytical instrument in an inert gas stream.
An impinger is a solvent-filled container through which a known
volume of analyte-containing gas stream is passed. Within the
Fig. 1. The siloxane sampling techniques considered, from left to right: Thermal desorption (TD) tubes, methanol impingers, Tedlar® bag, SUMMA canister.
2
Biomass and Bioenergy 108 (2018) 1–6
impinger, analytes are absorbed by the solvent, and the solvent is then
analyzed. The solvent is chosen based on the solubility characteristics of
the analytes. Using the known solvent volume, gas flowrate, and sam-
pling time, the gas-phase concentration can be calculated from the
measured concentration of analyte in the solvent. The liquid sample can
be stored in a sealed vial prior to analysis, depending on the properties
of the target compounds.
Tedlar® bags and SUMMA canisters are mainly used for collecting,
storing, and transporting a complete gas sample prior to analysis,
whereas TD tubes and impingers capture and store the analyte. A cri-
tical requirement for the containers is that the loss of gas-phase analytes
during storage and transportation, through leakage or surface sorption,
is minimal. For analysis, a gas phase sample is transferred from the
container to the analytical instrument. Due to surface adsorption,
analyte concentrations in the air could be underestimated. Surface
sorption tends to increase with decreasing volatility of the analyte.
Storage conditions and duration could also impact analyte adsorption
behavior, as well.
Accurate measurement of siloxane concentrations in biogas is cri-
tical to increasing renewable energy production and advancing sus-
tainability in wastewater treatment processes because underestimating
siloxane concentrations in biogas can lead to increased operating costs,
equipment damage, and improper biogas treatment process operation.
This study was conducted in order to identify the most effective biogas
sampling methods by comparing gas-phase siloxane concentration es-
timates obtained using four commonly utilized biogas sampling tech-
niques. This study is novel in that analytical siloxane concentration
estimates were compared to a standard siloxane-containing gas stream
in which the siloxane concentration was known through primary
standards. The most commonly used, state-of-science sampling techni-
ques were evaluated in this study.
2. Materials and methods
2.1. Chemicals
D5 (97% pure) was obtained from Sigma-Aldrich (St. Louis, MO,
USA). Its physicochemical properties can be found in Table 1. D4-bis(2-
ethylhexyl) phthalate-3,4,5,6 (d4-DEHP, 98 atom% D), used as surro-
gate for the determination of the recovery rate, and d4-dibutyl phtha-
late-3,4,5,6 (d4-DiBP, 98 atom% D), used as internal standard (IS),
were also obtained from Sigma-Aldrich. Reagent-grade methanol, used
as solvent, and acetone, used for cleaning, were obtained from Fisher
Scientific (Waltham, MA, USA). Nitrogen used as the gas-stream carrier
was obtained from Airgas (Radnor, PA, USA).
2.2. Calibration gas generation
The calibration gas generator (CGG) used was a Dynacalibrator
Model 190 (VICI Metronics Inc., Santa Clara, CA, USA). A glass diffu-
sion vial filled with liquid D5 was placed in the Dynacalibrator at an
oven temperature of 90 °C to ensure a steady mass emission rate of
gaseous D5 into the nitrogen carrier gas stream (Fig. 2). The nitrogen
carrier gas was controlled at a flowrate of 640 cm3/min. The D5
emission rate was measured gravimetrically by weighing the vial
C.M.A. Eichler et al.
Table 1
Physicochemical properties of D5 siloxane frequently detected in biogas [14].
CAS Chemical formula
Decamethylcyclopentasiloxane (D5) 541-02-6 C10H30O5Si5
a
Used in this study, as cited in Refs. [2,4,5,8].
several times over a 3-month period. The D5 gas-phase concentration
was calculated based on the D5 emission rate and the carrier gas
flowrate. All gas volumes were measured at room temperature (20 °C)
and atmospheric pressure (101.325 kPa).
2.3. Apparatus
The TD tubes (89 mm long, 6.35 mm OD, 5 mm ID, Supelco, St.
Louis, MO, USA) were packed with 200 mg of Tenax TA (40-60 mesh,
MARKES International, Llantrisant, Wales, UK). The three, 25 cm3,
fritted impingers (SKC) were standard laboratory equipment. The 6 L
stainless steel SUMMA canister (“TO-Can”) was obtained from Restek
(Bellefonte, PA, USA) and modified by adding a stainless steel Swagelok
fitting to the bottom so that a gas stream could be passed through the
canister. The 10 L Tedlar® bags (30.48 cm × 48.26 cm) were obtained
from ESS Environmental Sampling Supply (San Leandro, CA, USA). The
bags came equipped with a single polypropylene combination (2-N-1)
fitting with integrated septum. All necessary connections were made
with 6.35 mm Teflon tubing and 6.35 mm Teflon Swagelok fittings.
Prior to usage, the TD tubes were conditioned in an oven at 325 °C
by passing a helium carrier gas through each tube at 100 cm3/min for
30 min. The impingers were cleaned by rinsing three times with me-
thanol and three times with acetone. The canister was rinsed thor-
oughly with distilled water, methanol, and acetone, and then dried with
clean air. The Tedlar® bags were used as received from the supplier. The
tubing was rinsed with methanol and dried with clean air.
2.4. Sampling
For sampling with TD tubes, a stainless-steel Swagelok tee fitting
was installed at the outlet of the CGG. A TD tube was connected to one
port of the fitting, while the other port served as ventilation. An air
Biomass and Bioenergy 108 (2018) 1–6
Molar mass (g/mol) Boiling point (°C) Melting point (°C) Vapor pressure (Pa at 25 °C)
370.77 205 −44 23.2a
sampling pump (AirChek 2000, SKC, Pittsburgh, Pennsylvania) was
used to draw gas through the tube. A mini-BUCK™ primary flow cali-
brator (M-5, apbuck, Orlando, Florida) was used to set the sampling
flowrate at 20 cm3/min. Sampling time for the TD tubes was 1 min.
Duplicate samples were taken at each sampling event. Previous work
using a two-stage TD tube sampling system consistently showed no
detectable concentrations of D5 in the second TD tube. Hence, a one-
stage approach was used in this study.
For impinger sampling, a 25 cm3 methanol-filled impinger was
placed in an ice bath and connected directly to the gas outlet of the
CGG. Sampling times were set at 3, 5, and 8 min. At each sampling
time, three samples were taken from the impinger. For the same reason
as for TD tube sampling, a one-stage impinger sampling system was
used.
For Tedlar® bag sampling, each bag was connected directly to the
outlet of the CGG and filled for 6 or 8 min at a flowrate of 640 cm3/min.
After each sampling event, the bags were stored at conditions sum-
marized in Table 2. Gas from each bag was then sampled with TD tubes
at a flowrate of 20 cm3/min for 1 min. After gas sampling and evacu-
ating the bags completely, 100 cm3 methanol spiked with 100 mm3 d4-
DEHP solution (200 μg/L) was added to each bag, and the bags were
vigorously shaken to ensure dissolution of any D5 adsorbed to the bag
surface. The solvent was then captured by cutting a corner of each bag
and decanting the solvent into a sample vial. The extract was the
concentrated using a rotary evaporator (Brinkmann Buchi Rotavapor
R110, Switzerland) to 500-1000 mm3. 10 mm3 of IS was added prior to
analysis for volume correction and signal intensity normalization.
The fate of D5 in a SUMMA canister was tested in two phases, ad-
sorption and desorption. In the adsorption phase, the D5-containing
reference gas was introduced into a clean canister at a flowrate of
640 cm3/min. Outlet samples were collected with TD tubes every
15 min for 1.5 h, and every 30 min for another 2.5 h. In the desorption
Fig. 2. Calibration gas generator (CGG) and diffusion vial used for
reference gas generation. Impinger and TD tube samples were taken
from the sampling port at the bottom right.
3
C.M.A. Eichler et al.
Table 2
Tedlar® bag storage conditions. Variables were storage duration and storage temperature.
Duration of storage Temperature Number of bags
tested
1 day Room temperature (20-25 °C) 4
1 day Fridge (4 °C) 4
1 day Fridge (4 °C), then reheated for 4
40 min to 30 °C
5 days Room temperature (20-25 °C) 2
phase, the siloxane source was removed from the carrier gas stream and
pure nitrogen was allowed to flow through the canister at the same flow
rate. Outlet samples were taken with TD tubes every 5 min for the first
half hour after disconnecting the D5 source, every 10 min for the fol-
lowing half hour, and then every 30 min for another hour. The total
sampling time of the desorption phase was 2 h. After desorption, the D5
remaining on the interior surfaces of the canister was solvent extracted
using the method similar to that of the Tedlar® bag.
2.5. Instrumentation
Analysis of the samples was either performed with gas chromato-
graphy coupled with mass spectrometry (GC-MS) or with gas chroma-
tography with a flame ionization detector (GC-FID). GC-MS analysis of
all liquid samples (impinger, liquid extraction) was performed with a
Finnigan TraceGC ultra with Finnigan Trace DSQ (Thermo Electron
Corporation, Waltham, MA) in SIM mode. The TD tubes were analyzed
by thermal desorption (TurboMatrix TD, PerkinElmer) coupled with a
GC-FID (Agilent 6890 GC/FID). Further information is given in Table S1
in the Supplementary Information.
For quantification with GC-MS, a 5-point calibration curve was
applied (1 mg/L to 20 mg/L, R2 = 0.999). For the TD tubes, two 5-point
calibration curves were used, 10-200 mg/L and 200-4000 mg/L, with
R2 values of 0.995 and 0.998, respectively. The lower range calibration
curve was only used for the SUMMA canister desorption test samples.
3. Results and discussion
3.1. Calibration gas concentrations
The assumption of a constant mass emission rate can be made. On
average, the evaporation rate was 48.86 μg/min. This is sufficiently
close a calculated evaporation rate under the given conditions of
47.86 μg/min, assuming a D5 vapor pressure of 23.2 Pa at 25 °C. At a
carrier gas flowrate of 640 cm3/min, the average D5 gas phase con-
centration was calculated as 76.3 ng/cm3 ± 4.2 ng/cm3 (N = 10).
Assuming a normal distribution of the data, the 95% confidence in-
terval lies between 73.7 ng/cm3 and 79.0 ng/cm3, so the margin of
error is 76.3 ng/cm3 ± 2.6 ng/cm3. The decrease of the mass of liquid
D5 in the diffusion vial over time is shown in Table S2 and Fig. S1 in the
Supplementary Information. The D5 gas phase concentration in the
reference gas is in the range of D5 concentrations found in biogas from
anaerobic digesters at WWTPs, which is usually 10-100 ng/cm3 [4,7,9].
3.2. Sampling results
The average D5 recovery rates from the gas phase for all sampling
techniques are given in Fig. 3. Further information is given in Table S3
and Fig. S2 in the Supplementary Information. Two impinger sampling
trials were conducted. During each trial, triplicate samples were taken
from the impingers (N = 6). When using TD tubes, a total of six sets of
duplicate samples were collected; on one additional occasion, five
samples were taken within a 1-hr time period, resulting in N = 17
samples. The Tedlar® bag samples were collected in two trials. After
introducing the calibration gas into the bags, samples were taken from
4
Biomass and Bioenergy 108 (2018) 1–6
Fig. 3. Means of the D5 recovery rates for all sampling techniques. The results for the
Tedlar® bags are listed separately for each storage condition. The error bars indicate the
standard error of the means. The letters give the results of a Tukey-Kramer test (5%
significance level) for pairwise comparison of group means. Bars with the same letters
indicate that the respective group means are not significantly different.
the bags after storing them under four different conditions. Gas-phase
sampling from the bags was done with TD tube duplicates; five of a total
of 28 samples had to be discarded due to problems during the analysis.
The SUMMA canister adsorption/desorption test was performed on one
day. At steady state, eight duplicate gas-phase samples were taken with
TD tubes from the canister outlet, of which one was discarded
(N = 15). Recovery rates were calculated to compare the performance
of the different sampling techniques. The recovery rate was defined as
measured D5 concentration obtained with the respective sampling
technique divided by the gravimetrically measured D5 concentration in
the reference gas. The gravimetric gas-phase concentration used for this
calculation was 76.3 ng/cm3.
D5 recovery rates for sampling with the TD tubes varied between
88% and 122%, with an average of 109% and a standard deviation of
11%. These values are acceptable according to the National Institute for
Occupational Safety and Health (NIOSH) criterion, which suggests that
uncertainty should be no greater than 25% [20]. Overall, the results
demonstrated that siloxanes are retained and extracted efficiently from
TD tubes.
The impinger sampling method yielded reasonable results, since the
reported concentration was very close to those measured gravime-
trically. The D5 recovery rate was on average 104% with a standard
deviation of 7%. Overall deviations were small and most likely due to
random analytical errors. The downside to impinger sampling is the
time and effort required.
Gas sampling with TD tubes is simpler than use of impingers,
making the method easier to use in the field. Another advantage of the
tubes is their stability; once sealed they can be stored and shipped for
analysis. A concern associated with this method is the relatively higher
detection limit when TD is used with GC-FID; then the detection limit is
about 10 ng. However, the limit can be improved by coupling TD with a
more advanced analytical technique, such as GC-MS, where the detec-
tion of 0.1 ng becomes possible.
When using Tedlar® bags, D5 recovery rates were significantly lower
C.M.A. Eichler et al. Biomass and Bioenergy 108 (2018) 1–6

than those associated with impinger or TD tubes. On average, only

75% ± 5% of the introduced amount of D5 was recovered after being
stored in the bags. Given that no significant differences regarding the
recovery after storage under varying conditions were observed, further
improvements in the recovery rates appear to be unlikely through
manipulation of storage conditions.
The D5 gas-phase concentration at the outlet of the SUMMA canister
at steady state was 72.4 ng/cm3 on average; about 95% ± 4.0% of the
inlet D5 gas-phase concentration. It has to be noted that this recovery
rate was only achieved after 60 min of constant flow of the reference
gas through the canister. During this time, the inner canister surface
served as a sink for the introduced D5. These results will be further
discussed in section 3.4.
A single-factor ANOVA test with a significance level of 5% was
performed to determine if the means of the recovery rates obtained with
different sampling techniques were significantly different from each
other, which is the case for at least two of them
(p = 5.2·10−13 < 0.05). Therefore, a post-hoc Tukey-Kramer test was
conducted to describe which techniques actually differ. The results are
displayed in Fig. 3. All results from sampling with Tedlar® bags were
significantly different than the results from sampling with the impinger,
the SUMMA canister, or the TD tubes. The impinger and the TD
achieved similar results, as did the impinger and the SUMMA canister. Fig. 4. Averaged D5 mass balance with regard to the gravimetrically measured input
Detailed results of the ANOVA and the Tukey-Kramer tests can be found concentration for the Tedlar® bags.
in Tables S4 and S5 in the Supplementary Information.

3.3. Surface adsorption to Tedlar® bags

D5 was extracted from the inner surfaces of the empty bags and the
canister by means of liquid extraction to assess the amount of D5 ad-
sorbed. The surface area-to-volume ratio was 0.77 cm2/cm3 and
therefore rather high, the bags were filled to about 40% of their theo-
retical capacity. The surrogate d4-DEHP was recovered within the ac-
ceptable range of 71% and 115% for most samples, confirming the
overall performance of the extraction method. There was one exception
(239%) that was likely due to analytical error and/or background
contamination.
An average of 8.9% of total D5 collected from the bag was retained
on the bag's surface. With standard sampling procedures, the adsorbed
D5 cannot be extracted from the Tedlar® bags, leading to significant
underestimation of the siloxane introduced. Fig. 4 shows a mass balance
for D5 in the Tedlar® bags based on recovery from the gas phase and the
inner bag surface. Losses ranged from 5.9% to 29.3% with an average of
16.4%. Both the highest and the lowest losses were observed under the
same storing condition (1 day, 4 °C). A t-test (p = 5%) revealed that
there was no significant difference between the recoveries at different Fig. 5. Results of the adsorption test and the desorption test with the SUMMA canister.
storage conditions. Losses were probably due to an untight fitting or
leakage from tubing when connected to the calibration gas chamber or decreased to 1.5 ng/cm3 and remained around this value for another
to the TD tubes. Furthermore, some adsorption to tubing might have hour before the trial was completed. This was confirmed by the very
occurred, although the tubing was made as short as possible. Alto- small amount (3.11 ng/cm2) of D5 recovered from the canister surface
gether, siloxane sampling and sample storage with Tedlar® bags ex- by liquid extraction.
hibited several limitations which need to be accounted for when the Based on the data from the desorption test shown in Fig. 5 and using
technique is used. a method established by Cox et al. [21] for VOCs, the surface/air par-
tition coefficient K S for D5 on stainless steel was calculated. A linear and
3.4. SUMMA canister adsorption/desorption test results instantaneously reversible equilibrium relationship between the inner
canister surface and the canister gas phase is assumed [21]. The re-
The adsorption curve provided in Fig. 5 reveals the sink effect of the sulting equation used for this purpose is
SUMMA canister. About 45 to 60 min were required to reach a steady
state outlet concentration. During this period, D5 continued to adsorb y Q
ln ⎛⎜ ⎞⎟ = −⎛⎜
⎞t ⎟
to the inner canister surface until equilibrium between the gas phase
⎝ y0 ⎠ ⎝ V + K S AS ⎠ (1)
and surface concentration was achieved. After 15 min of introducing
D5, the D5 recovery rate was 74%. At steady state, the D5 gas-phase where, y is the D5 gas phase concentration at the outlet, y0 is the steady
concentration at the outlet of the canister was 72.4 ng/cm3 on average; state D5 gas phase concentration, Q is the gas flowrate (640 cm3/min),
about 95% ± 4.0% of the inlet D5 gas-phase concentration. After V is the canister volume (6795 cm3), AS is the inner surface area of the
disconnecting the siloxane source, D5 was almost completely desorbed canister (1735 cm3), and t is the time in minutes. It was assumed that
within 60 min (Fig. 5). The outlet D5 gas-phase concentration the canister was spherical with a diameter of 23.5 cm, as given by the

5
C.M.A. Eichler et al.
Fig. 6. Graphic solution for KS. The slope of the curve corresponds to − ( Q
V+K SA S
which KS can be extracted.
manufacturer. The equation was solved graphically by displaying
ln ( ) over t . The graphic solution can be found in Fig. 6.
y
y0
The slope of the resulting linear curve yielded the value for
− ( Q
V + K S AS ). Inserting the known data for Q, V , and A S allows for a
calculation of K S :
Q biogas, Water Sci. Technol. 61 (8) (2010) 2003–2012.
−⎛ ⎜
⎞ t = −0.0656t
⎟
V + K S AS ⎠
⎝
ng ng
K S = 0.0171 / Approaches concerning siloxane removal from biogas - a review, Can. Biosyst. Eng.
m2 m3
The calculated surface/air partition coefficient indicates that per ng
of D5 in 1 m3 gas phase, 0.0171 ng of D5 adsorb to 1 m2 of stainless
steel surface. Therefore, stainless steel is less than desirable for storing
or sampling siloxanes, as reported by Ajhar et al. [13] for sampling bags
with a stainless steel fitting. Based on this calculation and an actual
sampling volume of 5 L, 37% of the mass of D5 adsorbed to the inner
canister surface and 63% remained in the gas phase. If an aliquot of the
sample was taken from the canister, complete desorption of the D5 from
the surface may be unlikely, and the actual level of D5 in the gas phase
of the sample tested would be greatly underestimated. Additional tests
under different storing conditions, similar to those performed with the
Tedlar® bags, should be performed to validate this observation.
4. Conclusions
Indirect sampling techniques were shown to underestimate the
amount of D5 originally present in the gas phase. Adsorption to gas
container surfaces is very likely to occur due to the low volatility of
siloxanes, and complete desorption cannot always be achieved. Losses
due to leakage from gas containers can also be an issue. Direct sampling
techniques, such as the impinger and TD methods, exhibited improved
recovery of D5 and are therefore considered more suitable.
A limitation of this study is that only D5 was used and the effect of
the presence of other siloxanes was not included. Competitive sorption
and desorption could be an issue limiting the accuracy of sampling with
TD tubes. Further testing, using actual or multi-siloxane containing
Biomass and Bioenergy 108 (2018) 1–6
reference biogas, would therefore be worthwhile to determine if more
complex gas matrices affect the efficiency of the sampling techniques.
Acknowledgements
This work was supported by the Water Environment Research
Foundation (WERF) [grant number LIFT7T14].
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://dx.
doi.org/10.1016/j.biombioe.2017.10.049.
References
[1] M. Ajhar, M. Travesset, S. Yüce, T. Melin, Siloxane removal from landfill and di-
gester gas – a technology overview, Bioresour. Technol. 101 (9) (2010) 2913–2923.
[2] S. Rasi, J. Lehtinen, J. Rintala, Determination of organic silicon compounds in
biogas from wastewater treatments plants, landfills, and co-digestion plants,
), from Renew. Energy 35 (12) (2010) 2666–2673.
[3] B. Tansel, S.C. Surita, Differences in volatile methyl siloxane (VMS) profiles in
biogas from landfills and anaerobic digesters and energetics of VMS transforma-
tions, Waste Manag. 34 (2014) 2271–2277.
[4] R. Dewil, L. Appels, J. Baeyens, Energy use of biogas hampered by the presence of
siloxanes, Energy Convers. Manag. 47 (2006) 1711–1722.
[5] R.L. Gao, S.K. Cheng, Z.F. Li, Research progress of siloxane removal from biogas,
Int. J. Agric. Biol. Eng. 10 (1) (2017) 30–39.
[6] S. Nam, W. Namkoong, J.-H. Kang, J.-K. Park, N. Lee, Adsorption characteristics of
siloxanes in landfill gas by the adsorption equilibrium test, Waste Manag. 33 (2013)
2091–2098.
[7] K. Oshita, Y. Ishihara, M. Takaoka, N. Takeda, T. Matsumoto, S. Morisawa,
A. Kitayama, Behaviour and adsorptive removal of siloxanes in sewage sludge
[8] M. Schweigkofler, R. Niessner, Removal of siloxanes in biogases, J. Hazard. Mater.
83 (3) (2001) 183–196.
[9] G. Soreanu, M. Béland, P. Falletta, K. Edmonson, L. Svoboda, M. Al-Jamal, P. Seto,
J. 53 (2011) 8.1–8.18.
[10] E. Wheless, J. Pierce, Siloxanes in landfill and digester gas Update, 27th Annual
SWANA LFG Symposium, 2004, pp. 1–9.
[11] Environmental Risk Assessment Report: Decamethylcyclopentasiloxane,
Environment Agency, Almondsbury, Bristol, UK, 2009.
[12] N. de Arespacochaga, C. Valderrama, J. Raich-Montiu, M. Crest, S. Mehta,
J.L. Cortina, Understanding the effects of the origin, occurrence, monitoring, con-
trol, fate and removal of siloxanes on the energetic valorization of sewage biogas—a
review, Renew. Sustain. Energy Rev. 52 (2015) 366–381.
[13] M. Ajhar, B. Wens, K.H. Stollenwerk, G. Spalding, S. Yüce, T. Melin, Suitability of
Tedlar® gas sampling bags for siloxane quantification in landfill gas, Talanta 82
(2010) 92–98.
[14] J. Raich-Montiu, C. Ribas-Font, N. de Arespacochaga, E. Roig-Torres, F. Broto-Puig,
M. Crest, L. Bouchy, J.L. Cortina, Analytical methodology for sampling and ana-
lysing eight siloxanes and trimethylsilanol in biogas from different wastewater
treatment plants in Europe, Anal. Chim. Acta 812 (2014) 83–91.
[15] B. Tansel, S.C. Surita, Selectivity and limitations of carbon sorption tubes for cap-
turing siloxanes in biogas during field sampling, Waste Manag. 52 (2016) 122–129.
[16] M. Schweigkofler, R. Niessner, Determination of siloxanes and VOC in landfill gas
and sewage gas by canister sampling and GC-MS/AES analysis, Environ. Sci.
Technol. 33 (20) (1999) 3680–3685.
[17] S. Saeed, S.F. Kao, G.J. Graening, Comparison of impinger and canister methods for
the determination of siloxanes in air, AWMA Symposium on Air Quality
Measurement Methods and Technology, San Francisco, CA, 2002, pp. 1–13.
[18] E. Woolfenden, Sorbent-based sampling methods for volatile and semi-volatile or-
ganic compounds in air. Part 2. Sorbent selection and other aspects of optimizing air
monitoring methods, J. Chromatogr. A 1217 (2010) 2685–2694.
[19] J. Volden, Y. Thomassen, T. Greibrokk, S. Thorud, P. Molander, Stability of work-
room air volatile organic compounds on solid adsorbents for thermal desorption gas
chromatography, Anal. Chim. Acta 530 (2004) 263–271.
[20] NIOSH, Guidelines for Air Sampling and Analytical Method Development and
Evaluation, National Institute for Occupational Safety and Health (NIOSH),
Cincinnati, OH, 1995.
[21] S.S. Cox, J.C. Little, A.T. Hodgson, Predicting the emission rate of volatile organic
compounds from vinyl flooring, Environ. Sci. Technol. 36 (4) (2002) 709–714.