Determination of recovery rates of adsorbents for sampling very volatile organic compounds (C1–C6) in dry and humid air in the sub-ppb range by use of thermal desorption gas

The reliable measurement of very volatile organic compounds (VVOC) in indoor air by use of thermal desorption gas chromatography (TD-GC) in order to include them into evaluation schemes for building products even nowadays is a great challenge.

Contents lists available at ScienceDirect

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

volatile organic compounds (C 1 –C 6 ) in dry and humid air in the

spectrometry

Matthias Richter∗, Elevtheria Juritsch, Oliver Jann

Bundesanstalt für Materialforschung und -prüfung (BAM), Unter den Eichen 87, 12205 Berlin, Germany

Article history:

Received 13 March 2020

Revised 2 July 2020

Accepted 3 July 2020

Available online 4 July 2020

Keywords:

VVOC

Indoor air: Adsorbent performance

Recovery rate

Thermal desorption

Gas chromatography

Thereliablemeasurementofveryvolatileorganiccompounds(VVOC)inindoorair byuseofthermal desorptiongaschromatography(TD-GC)inordertoincludethemintoevaluationschemes forbuilding productseven nowadays isa great challenge.Forcapturing thesesmall molecules withcarbon num-bersrangingfromC1–C6,strongadsorbentsareneeded.Inthepresentstudy,recoveryratesofnine suit-ableadsorbentsofthegroupsofporouspolymers,graphitisedcarbonblacks(GCB)andcarbonmolecular sieves(CMS)aretestedagainstacomplextestgasstandardcontaining29VVOC.Byconsiderationofthe recoveryand therelativehumidity(50% RH),combinations oftheGCBCarbograph5TD, thetwo CMS Carboxen1003and CarbosieveSIIas wellas theporouspolymerTenax® GRwereidentifiedtobe po-tentiallysuitableforsamplingthemajorityoftheVVOCoutofthegasmix.Theresultsrevealabetter performanceoftheadsorbentsincombinationthanbeingusedalone,particularlyunderhumidsampling conditions.Therecoveryratesofthechosencompoundsoneachadsorbentshouldbeinthe rangeof 80–120%

© 2021 The Authors Published by Elsevier B.V ThisisanopenaccessarticleundertheCCBY-NC-NDlicense (http://creativecommons.org/licenses/by-nc-nd/4.0/)

1 Introduction

In the indoor environment, residents are exposed to a large

number of various chemical pollutants originating from both the

ingress from the outside and emissions from permanent sources

indoors like building materials, furniture, electronic devices or

non-permanent sources like household chemicals, etc Most of

them are organic compounds, which are classified into very

volatile, volatile and semi-volatile organic compounds (VVOC, VOC,

SVOC) In the last decades, many studies have shown that these

substances are responsible for health complaints often referred to

as the Sick Building Syndrome (SBS) [ 1, 2] The study discussed in

this paper is focusing on the group of the VVOC, and follows the

definition of the European testing standard EN 16516, in which

VVOC are defined as “…volatile organic compounds eluting before

n-hexane on the gas chromatographic column specified as a 5%

phenyl / 95% methyl polysiloxane capillary column, …” (non-polar

column) [3]

∗ Corresponding author

E-mail address: matthias.richter@bam.de (M Richter)

In Europe, the Construction Products Regulation (CPR, 2011/305/EU) sets basic requirements (BR) on how construc- tion works must be designed and built BR 3 “hygiene, health and the environment” states low emissions of toxic gases, VOC, particles, etc from building materials The relevant procedures for the determination of chemical emissions from materials used indoors in emission test chambers are described in the interna- tional standard series ISO 160 0 0 [4–7] and are specified in the harmonized European testing standard EN 16516 [3] This standard focuses on the analysis of pollutants in the VOC range, which it defines as all compounds eluting between C 6 and C 16on a slightly polar capillary column with a 5%phenyl-/95%methyl-polysiloxane phase using thermal desorption gas chromatography coupled with

a mass selective detector (TD-GC/MS) Measurement and analysis procedures are described in one document, yet it lacks an evalua- tion of the results To account for this gap, an expert group from

EU member states has developed a roadmap towards an EU-wide harmonised framework for the health-based evaluation of indoor emissions from construction products published in the ECA-reports

No 24, 27 and 29 [8–10] Relevant target compounds to be identi- fied and traceably quantified in the test chamber air are listed on https://doi.org/10.1016/j.chroma.2020.461389

0021-9673/© 2021 The Authors Published by Elsevier B.V This is an open access article under the CC BY-NC-ND license ( http://creativecommons.org/licenses/by-nc-nd/4.0/ )

2 M Richter, E Juritsch and O Jann / Journal of Chromatography A 1626 (2020) 461389

the EU-LCI list Currently, this list is limited to only a few VVOCs

( < C 6), such as formaldehyde, acetaldehyde, butanal, pentanal and

2-butanone [11], since these analytes are measurable with HPLC

using 2,4-dinitrophenylhydrazine (DNPH) as sorbent according to

the ISO 160 0 0-3 procedure The porous polymer type adsorbent

Tenax® TA is very well suited for the sampling of compounds in

the VOC and SVOC range However, what poses a challenge on

the TD-GC/MS method required by the testing standards EN 16516

and ISO 160 0 0-6 is the decrease of the retention volume of the

stipulated adsorbent Tenax® TA with increasing volatility [ 12, 13]

Therefore, other adsorbents need to be selected to solve this

problem A selection of suitable adsorbents can be found in the

literature, e.g in Dettmer and Engewald [13] Woolfenden [14],

ISO 16017-1 [15] or in manufacturers’/suppliers’ information, e.g

Camsco [16]

A standardised method for the analysis of VVOC is currently not

available In his review, Salthammer [17]gives a good overview of

approaches that have been published to date Only few are dedi-

cated to a systematic validation of adsorbents and combinations of

adsorbents to cover a wide VVOC range from carbon number C 1

to C 6 Schieweck, Gunschera, et al [18]went into this direction by

systematically testing six different graphitized carbon blacks (GCB)

and carbon molecular sieves (CMS) adsorbents for covering the

compounds range of C 3–C 6 For testing the suitability of the ad-

sorbents, a recovery rate was determined by referring the arith-

metic peak areas of the target compounds for each adsorbent to

the arithmetic mean of the areas obtained by measurements on

Tenax TA This procedure enables a rating of potentially suitable

adsorbents but is neglecting matrix effects affecting measurement uncertainty On the one hand the test standards the adsorbents are spiked with are solutions of methanol, which is beyond sam- pling practice, and on the other hand the use of an adsorbent serv- ing as reference is improper Pech, Wilke, et al [19]compared the three adsorbents Tenax TA, Carbograph 5TD and Carbopack X as to their suitability to retain a VVOC mix of 20 components in the gas phase However, they used Carbograph 5TD as reference In both studies, the performance of the adsorbents in the presence of wa- ter vapour in the sample air was excluded

The aim of the present study is to determine the recovery rates of commercially available adsorbents suitable for the sam- pling of VVOC including compounds with carbon numbers C 1 to

C 6 Nine adsorbents involving porous polymers, GCB and CMS were checked under consideration of relative humidity of the sampled air and loaded with a complex gas standard mixture composed of

29 VVOC and 3 VOC around the C 6 limit in the sub-ppb range Finally, based on the values obtained, possible combinations of ad- sorbents should be tested to get indication if this will lead to im- proved recovery

2.1 Test gas preparation

The gas mixture listed in Table 1 was prepared in a gas col- lecting tube (GCT) with a volume of 500 mL and equipped with

a septum and a valve for additional tightness Benzene, pentanal

Table 1

Analytes in gas mixes used for experiments Compound properties, such as retention time (RT), molecular weight (MW) and boiling point (b.p.) are given as well as the absolute mass loaded on adsorbent tubes for injection volumes 60 and 100 μL Substances printed in italic do not belong to the group of VVOC according to the definition

of ISO 160 0 0-6 and EN 16516

Carbon No Compound CAS No Formula RT (min) MW (g mol −1 ) b.p ( °C) Loaded mass (ng) Stability 1 (%) Note

60 μL 100 μL

C 1 Dichlorodifluoromethane 75–71–8 CCl 2 F 2 9.845 120.9 8.9 30 50 12 customised gas cylinder

Isopropyl alcohol 67–63–0 C 3 H 8 O 18.709 60.1 82.3 15 25 11 customised gas cylinder Methyl acetate 79–20–9 C 3 H 6 O 2 19.801 74.1 56.8 18 31 4 customised gas cylinder 2-Chloro propane 75–29–6 C 3 H 7 Cl 20.272 78.5 35.0 33 20 4 customised gas cylinder

Vinyl acetate 108–05–4 C 4 H 6 O 2 23.671 86.1 71.6 21 36 2 customised gas cylinder

Ethyl acetate 141–78–6 C 4 H 8 O 2 25.333 88.1 77.1 22 37 2 customised gas cylinder

Pentanal 110–62–3 C 5 H 10 O 29.962 86.1 103.1 47 79 47 pure compound

n-Hexane 110–54–3 C 6 H 14 27.767 86.2 69.0 21 36 3 customised gas cylinder

1 relative standard deviation of samplings out of the gas collecting tubes over a period of 14 days and calculated relative to the ISTD benzene-d 6 Direct injection via split/splitless injector 2 internal standard

Table 2

Adsorbents used for the study Data provided by Woolfenden, manufacturer/supplier and Schieweck, Gunschera et al [ 14 , 16 , 18 ] T des corresponds to the desorption temperature used in this study ( Section 2.2 )

Physical properties Adsorbent type Name

Surface area (m ² g −1 ) Packing density (g cm −3 ) T max

T cond ( °C) T des Mesh size

Volatility range Features

hydrophobic

than Tenax TA Graphitized carbon

black (GCB)

Carbograph 5TD 560 n/a > 400 350 350 40/60 C 3 –C 8 High thermal stability, low

artifacts, hydrophobic Carbopack B 100–200 0.35 > 400 350 325 60/80 C 5 –C 12 High thermal stability, low

artifacts, hydrophobic

Carbon molecular

sieve (CMS)

Carbosieve SII 1060 0.61 > 400 350 330 60/80 C 1 –C 2 Different data available:

some hydrophilicity to significant water retention, low artifacts

Carboxen 569 485 0.61 > 400 350 330 20/45 C 2 –C 5 Different data available:

hydrophobic to some hydrophilicity Carboxen 1003 1000 0.46 > 400 350 330 40/60 C 2 –C 5 Different data available:

hydrophobic to some hydrophilicity, inert

hydrophobic to some hydrophilicity, inert

1 mixture of Tenax TA and a GCB type adsorbent

and n-hexane do not belong to the group of the VVOC but were

chosen as compounds of the transition region between the VVOC

and VOC range The mixture contained 23 compounds taken from a

pressurised gas cylinder, custom-made by Linde AG, Germany The

remaining 10 compounds were mixed in equal proportions with-

out solvent to two solutions Aliquots were spiked with a gas-tight

syringe through the septum of the GCT that was already filled with

the gas mix of the pressurized cylinder The temperature was kept

at 23 °C For the tests, volumes of 60 or 100 μL of the test gas

mix were taken with a gas-tight syringe and injected either di-

rectly into the split/splitless injector of the GC or onto the adsor-

bent to be tested as described in Section 2.4 Resulting amounts

are given in Table 1 To compensate measurement-related varia-

tions, benzene-d 6and ethanol-d 6were added as internal standards

(ISTD)

Prior to the experiments, the GCT was thoroughly checked for

tightness and the generated test gas mixture for its stability Fol-

lowing a test gas mix injection into the GCT, constant amounts

of the mixes were directly injected on a daily basis into the GC’s

split/splitless injector over a period of 14 days with the relative

standard deviation (RSD) being calculated

2.2 Analysis

All test series were carried out on a gas chromatograph

equipped with a split/splitless injector (Agilent 7890 N), an au-

tomated thermal desorption system (TDS 3/TDS A, Gerstel) us-

ing liquefied nitrogen cooling (CIS 4) and a mass selective de-

tector (Agilent MSD 5975 C inert XL) A PLOT column (PoraBond

Q, 50 m × 0.32 mm × 5 μm, Agilent) with a polystyrene-

divinylbenzene phase suitable for the separation of low boiling

compounds was installed flushed with helium (ALPHAGAS, Air Liq-

uide) as carrier gas Additionally, a particle trap was installed be-

tween column and MSD The m/z scan range was between 25 and

131

During the analyses, the test gas mix was injected in two ways:

a) Directly with a gas-tight syringe via the split/splitless injector

(splitless mode) to obtain an unaffected analysis signal (refer-

ence value): The oven programme started at 35 °C for 1 min,

then heating with 8 °C min −1 to 80 °C for 1 min, further heat- ing with 5 °C min −1 to 230 °C A carrier gas pressure of 0.97 bar was adjusted

b) Via thermal desorption of the loaded adsorbent Since sampling

of humidified air may have an impact on the analysis, two dif- ferent thermal desorption modes were applied: b1) the split- less mode when dry air was used and b2) the solvent vent- ing dry purge mode at humid conditions to prevent icing in the cold injection system (CIS) The TDS in both cases was pro- grammed to start at 35 °C for 1 min, then heating with a rate of

60 °C − 1to 300–350 °C depending on the used adsorbent ( T des

in Table 2) for 5 min The CIS programme started at −150 °C, heating at 12 °C − 1 to 30 °C for 1 min followed by further heating at 12 °C − 1 to 150 °C held for 1 min A quartz wool filled liner was installed For the measurements of the adsor- bents the GC oven was programmed to start at 35 °C for 1 min, then heating at 6 °C min −1to 80 °C for 1 min, further heating at 4.8 °C min −1to 200 °C immediately followed by further heating

at 5 °C min −1 to 230 °C The carrier gas pressure was adjusted

to 1.4 bar

Fig.1depicts a chromatogram of the VVOC test gas mixture af- ter injection via the split/splitless injector

2.3 Selection of adsorbents

Sampling air always contains water that potentially affects sam- pling and analysis Helmig, Schwarzer, et al [20] report injected water can cause peak shifting due to restricted flow of carrier gas through the column, changes in carrier gas viscosity, and changes

in the stationary phase polarity and split ratios Moreover, water vapour is able to condense in the small pores of molecular sieves [21] Other authors report on competition between analytes and water for active adsorbent sites [ 14, 22], which may impact break- through volumes of analytes Vallecillos, Maceira, et al [23]report

on significantly decreased breakthrough volumes for 1,3-butadiene

on a multi-sorbent bed (Carbotrap B/Carbopack X/Carboxen 569) of 66% at an RH of 56–68%

For the present study, mainly hydrophobic or slightly hy- drophilic common adsorbents were selected ( Table 2) However,

4 M Richter, E Juritsch and O Jann / Journal of Chromatography A 1626 (2020) 461389

Fig 1 Chromatogram of the VVOC test gas mixture analysed after direct injection into the split/splitless injector on a PoraBond Q (50 m × 0.32 mm × 5 μm)

the data provided for this parameter diverge in the literature

Tenax® TA was used as benchmark

Glass tubes (Gerstel, Germany) with an outer diameter of 6 mm

and a length of 176 mm were filled with the selected adsorbents

Using the manufacturer’s marking, equal volumes of each adsor-

bent were filled into the tubes This resulted in the exact same bed

lengths (60 mm) but in different absolute masses depending on the

materials’ densities ( Table 3) Tube conditioning was carried out

according to the manufacturer’s recommendations ( Table2) Prior

to the analysis, blank measurements were carried out

2.4 Determination of recovery

The recovery is affected by the sorption behaviour, the desorp-

tion temperature and the relative humidity at the time of sam-

pling Generally, for a distinct indication of the recovery of com-

pounds from each adsorbent type, a reference value is required

that represents 100% of the loaded amount (without losses) The

reference value will then be related to the amount of substance

desorbed from the adsorbent All effects of above discussed influ-

ences can be evaluated with this single value

In some studies, clean adsorbent tubes are loaded with a test

mixture of known composition and concentration and compared

with the performance of other adsorbent types or the same ad-

sorbent type impacted by variations of test parameters [ 18, 24–

28] The adsorbent retaining the highest amounts of the target

molecules is then taken as reference These procedures disregard

any effects on the reference value obtained that might be resulting

from interactions of the test sample molecules with the adsorbent,

e.g breakthrough phenomena, insufficient desorption or chemical

reactions

Similar to the procedure reports by Dettmer, Knobloch, et al

[29], the recovery in this study was determined with a test set-

up depicted in Fig.2 The TD injector as well as the split/splitless

injector were connected with the column via a Y-splitter Disacti- vated pre-columns were used to connect the injector with the Y- splitter This set-up enabled switching between both injectors and allowing a direct comparison of the amount of substance directly injected over the split/splitless injector with the amount that was desorbed from the tested adsorbent

The recovery R i was calculated according to Eq.(1)

R i = A i ,T D ,rel

A i ,re f,rel × 100%=A i ,T D × A IST D,re f

A IST D ,T D × A i ,re f × 100%, (1)

with

R i Recovery of component i in%

A i,TD Peak area of component i obtained by thermal desorption (TD) of adsorbent tube

A i,TD,rel A i,TD in relation to the area of ISTD

A i,ref Peak area of component i obtained by direct injection onto

GC column via split/splitless injector (reference)

A i,ref,rel A i,refin relation to the area of ISTD

A ISTD,TD Peak area of ISTD obtained by thermal desorption of ad- sorbent tube

A ISTD,ref Peak area of internal standard obtained by direct injec- tion onto GC column via split/splitless injector

For any experiment as described in this section, the reference value was determined by injection (n =6) of an aliquot of the test gas mix directly into the split/splitless injector of the GC (route

A in Fig.2) by use of a gas-tight syringe The average of the ob- tained peak areas was taken as A i,ref and A ISTD,ref , respectively The adsorbent tubes from Table2were spiked with the same volume

of test gas mix by injection into a carrier gas flow ( V= 1 L) passing through the adsorbent This spiking took place in the same room

as the determination of the reference value to ensure the same ambient conditions The analysis of the adsorbent tubes, also given

as peak areas, resulted in the values for A i,TD and A ISTD,TD respec-

Recovery rates of tested adsorbents under dry (0% RH) and humid (50% RH) sampling conditions in order of their elution from the column The values are related to the internal standard (ISTD) benzene-d 6 Recovery rates between 80% and 120% were allowed ( bold numbers ) Water retention at 50% RH is given as well Compounds in italic do not belong to the group of VVOC as to definition in ISO 160 0 0-6 or EN 16516

Adsorbent (mass per tube)

Compound

Tenax TA (200 mg)

Tenax GR (240 mg)

Carbograph 5TD (300 mg)

Carbopack B (275 mg)

Carbopack Z (140 mg)

Carbosieve S

II (500 mg)

Carboxen 569 (440 mg)

Carboxen

1003 (365 mg)

Carboxen

1018 (570 mg)

Tx GR/Cx 1003/Cs SII 1

(85/105/115)

mg

Cg 5TD/Cx 1003/Cs SII 1

(95/95/140)

mg

Carbotrap

300 (n a.)

RH (%) Chlorodifluoromethane 0 n d n d n d n d n d (107 ± 5)% (107 ± 2)% (106 ± 3)% (108 ± 3)% (88 ± 1)% (82 ± 3)% (93 ± 2)%

Methanol 0 (5 ± 5)% (35 ± 17)% (73 ± 7)% (49 ± 13)% (57 ± 12)% (95 ± 17)% (95 ± 17)% (85 ± 1)% (80 ± 9)% (114 ± 5)% (116 ± 4)% (87 ± 1)%

Propene 0 (1 ± 2)% (3 ± 4)% (24 ± 22)% (8 ± 6)% (8 ± 5)% (169 ± 43)% (112 ± 4)% (113 ± 29)% (131 ± 28)% (225 ± 4)% (136 ± 5)% (189 ± 11)%

n-Propane 0 (1 ± 1)% n d (7 ± 5)% n d n d (104 ± 6)% (105 ± 3)% (103 ± 3)% (102 ± 4)% (153 ± 6)% (129 ± 3)% (134 ± 6)%

Dichlorodifluoromethane 0 n d n d (91 ± 15)% n d n d (113 ± 6)% (114 ± 0)% (113 ± 3)% (110 ± 4)% (86 ± 1)% (86 ± 4)% (97 ± 1)%

Vinyl chloride 0 n d n d (80 ± 15)% n d n d (105 ± 6)% (104 ± 4)% (105 ± 4)% (103 ± 5)% (113 ± 3)% (108 ± 3)% (92 ± 4)%

Ethanol 0 (17 ± 6)% (73 ± 5)% (66 ± 13)% (68 ± 4)% (41 ± 3)% (83 ± 4)% (80 ± 6)% (79 ± 2)% (77 ± 7)% (115 ± 10)% (133 ± 0)% (77 ± 2)%

1,3-Butadiene 0 (2 ± 0)% (2 ± 2)% (100 ± 3)% (41 ± 36)% (103 ± 1)% (72 ± 20)% (79 ± 19)% (81 ± 11)% (66 ± 17)% (85 ± 5)% (108 ± 2)% (35 ± 2)%

Acetonitrile 0 (25 ± 12)% (71 ± 4)% (75 ± 3)% (62 ± 15)% (67 ± 5)% (80 ± 12)% (76 ± 6)% (68 ± 13)% (74 ± 1)% (117 ± 2)% (108 ± 5)% (99 ± 2)%

trans-2-Butene 0 (2 ± 0)% (1 ± 1)% (103 ± 4)% (62 ± 31)% (103 ± 2)% (103 ± 7)% (103 ± 2)% (102 ± 4)% (103 ± 3)% (110 ± 2)% (105 ± 0)% (85 ± 2)%

n-Butane 0 (2 ± 1)% (1 ± 1)% (99 ± 2)% (15 ± 18)% (100 ± 2)% (102 ± 5)% (105 ± 2)% (101 ± 4)% (101 ± 3)% (138 ± 4)% (129 ± 5)% (114 ± 3)%

cis-2-Butene 0 (2 ± 1)% (1 ± 1)% (101 ± 1)% (9 ± 9)% (102 ± 2)% (97 ± 4)% (97 ± 6)% (99 ± 4)% (97 ± 4)% (102 ± 4)% (105 ± 3)% (83 ± 3)%

Acrolein 0 (28 ± 5)% (83 ± 5)% (85 ± 14)% (46 ± 1)% (25 ± 7)% (90 ± 7)% (82 ± 10)% (82 ± 11)% (82 ± 4)% (100 ± 3)% (100 ± 1)% (37 ± 10)%

Furan 0 (9 ± 5)% (12 ± 8)% (105 ± 1)% (10 ± 9)% (105 ± 1)% (96 ± 8)% (105 ± 3)% (104 ± 4)% (102 ± 4)% (91 ± 6)% (91 ± 1)% (79 ± 2)%

Propanal 0 (33 ± 4)% (114 ± 12)% (91 ± 19)% (90 ± 5)% (69 ± 4)% (45 ± 30)% (74 ± 21)% (40 ± 18)% (46 ± 13)% (92 ± 15)% (145 ± 6)% (86 ± 6)%

Acetone 0 (38 ± 2)% (101 ± 4)% (100 ± 4)% (104 ± 5)% (91 ± 18)% (99 ± 3)% (101 ± 3)% (96 ± 4)% (98 ± 3)% (113 ± 12)% (134 ± 2)% (124 ± 5)%

Carbon disulfide 0 (10 ± 6)% (7 ± 6)% (95 ± 4)% (3 ± 1)% (91 ± 5)% (103 ± 4)% (100 ± 5)% (104 ± 4)% (102 ± 5)% (106 ± 1)% (118 ± 1)% (80 ± 1)%

Isopropyl Alcohol 0 (60 ± 38)% (120 ± 69)% (83 ± 22)% (62 ± 24)% (48 ± 31)% (65 ± 36)% (117 ± 66)% (90 ± 23)% (71 ± 28)% (109 ± 9)% (125 ± 5)% (94 ± 1)%

Methyl acetate 0 (61 ± 5)% (89 ± 2)% (72 ± 13)% (46 ± 21)% (20 ± 9)% (92 ± 2)% (92 ± 1)% (91 ± 3)% (92 ± 2)% (130 ± 5)% (130 ± 1)% (110 ± 3)%

( continued on next page )

6

Table 3 ( continued )

Adsorbent (mass per tube)

(200 mg)

Tenax GR (240 mg)

Carbograph 5TD (300 mg)

Carbopack B (275 mg)

Carbopack Z (140 mg)

Carbosieve S

II (500 mg)

Carboxen 569 (440 mg)

Carboxen

1003 (365 mg)

Carboxen

1018 (570 mg)

Tx GR/Cx 1003/Cs SII 1

(85/105/115)

mg

Cg 5TD/Cx 1003/Cs SII 1

(95/95/140)

mg

Carbotrap

300 (n a.)

2-Chloro propane 0 (21 ± 1)% (29 ± 9)% (87 ± 11)% (87 ± 4)% (100 ± 4)% (100 ± 5)% (92 ± 4)% (83 ± 6)% (92 ± 5)% (86 ± 28)% (125 ± 1)% (51 ± 17)%

1-Propanol 0 (54 ± 3)% (67 ± 5)% (58 ± 9)% (50 ± 17)% (31 ± 9)% (57 ± 7)% (61 ± 5)% (55 ± 4)% (51 ± 5)% (104 ± 8)% (117 ± 2)% (102 ± 2)%

Diethyl ether 0 (55 ± 6)% ( 93 ± 3)% (98 ± 1)% (98 ± 0)% (99 ± 2)% (96 ± 1)% (96 ± 3)% (97 ± 3)% (95 ± 3)% (114 ± 3)% (115 ± 1)% (112 ± 1)%

Isoprene 0 (32 ± 2)% (56 ± 8)% (102 ± 1)% (103 ± 1)% (104 ± 2)% (57 ± 19)% (87 ± 10)% (86 ± 11)% (69 ± 18)% (66 ± 10)% (94 ± 2)% (96 ± 1)%

n-Pentane 0 (24 ± 1)% (42 ± 9)% (97 ± 1)% (98 ± 0)% (98 ± 2)% (98 ± 3)% (98 ± 1)% (97 ± 2)% (97 ± 3)% (133 ± 4)% (129 ± 2)% (125 ± 3)%

Vinyl acetate 0 (75 ± 4)% (75 ± 13)% (28 ± 18)% (24 ± 6)% (3 ± 3)% (35 ± 24)% (66 ± 22)% (40 ± 24)% (66 ± 30)% (69 ± 29)% (94 ± 7)% (32 ± 8)%

Chloroform 0 (86 ± 7)% (101 ± 2)% (97 ± 5)% (103 ± 0)% (103 ± 1)% (104 ± 2)% (101 ± 4)% (95 ± 3)% (97 ± 6)% (78 ± 8)% (91 ± 1)% (72 ± 4)%

2-Butanone 0 (85 ± 3)% (85 ± 3)% (77 ± 11)% (81 ± 5)% (69 ± 6)% (46 ± 23)% (77 ± 5)% (64 ± 15)% (63 ± 21)% (79 ± 3)% (84 ± 2)% (83 ± 3)%

Ethyl acetate 0 (88 ± 3)% (87 ± 1)% (76 ± 10)% (68 ± 15)% (28 ± 12)% (81 ± 7)% (86 ± 1)% (84 ± 4)% (84 ± 3)% (98 ± 2)% (97 ± 1)% (92 ± 1)%

2-Methylpentane 0 (40 ± 2)% (75 ± 4)% (96 ± 1)% (98 ± 0)% (97 ± 2)% (93 ± 4)% (97 ± 0)% (95 ± 3)% (94 ± 1)% (129 ± 3)% (127 ± 1)% (123 ± 2)%

Benzene 0 (105 ± 19)% (94 ± 87)% (119 ± 5)% (103 ± 23)% (109 ± 9)% (107 ± 10)% (120 ± 10)% (133 ± 43)% (104 ± 12)% (88 ± 7)% (92 ± 1)% (97 ± 1)%

Number of retained

compounds in the range

of 80–120% recovery

Water uptake at 50% RH

per sampling volume

(mg H 2 O/g adsorbent)

n d.: not detectable

1 measurements under humid conditions carried out without repetition

Fig 2 Set-up for the determination of the VVOC recovery of the adsorbent tubes

tively The calculation of the recovery in% was carried out accord-

ing to Eq.(1) A tolerance of the recovery of ± 20% around 100%

was permitted as this variation might be resulting from other ef-

fects not necessarily related to the sampling, e g measurement-

related variations

The recovery rate was firstly determined under dry conditions

(0% RH of carrier gas) with an injection of 60 μL of the test gas mix

leading to a first selection of potentially suitable adsorbents These

were then investigated with a humidified carrier gas flow adjusted

at 50% RH, since this degree of humidity is required in the rele-

vant testing standards mentioned in the introduction section As

the analysis is impacted by humidity, the TD method was adjusted

by switching to the solvent venting dry purge mode This in turn

led to a decrease of the sensitivity of the analysis, which was com-

pensated by an increase of the injected amount of the test gas mix

to 100 μL The resulting loading amounts are given in Table1

3 Results and discussion

3.1 Stability of gas standards and GCT tightness

As shown in Table1, for the majority of the compounds the sta-

bility expressed by the RSD determined by single injections over 14

days was better than 10%, the maximum was obtained for pentanal

with 47% Based on the analyses carried out for the experiments,

a satisfying explanation for this result cannot be given However,

tightness of the GCT and compound stability could be regarded as

sufficient for use of the gas standard for at least 14 days

3.2 Determination of recovery under dry and humid sampling

conditions

In Table3, the recovery rates determined for dry and humid air

sampling on single and combined adsorbents are listed The mean

values and standard deviations of four (dry air) and seven (hu-

midified air) repetitions are given, except for the multi-bed tubes

Here, the loadings were repeated only three times For the ref-

erence values A i,ref and A ISTD,ref , relative standard deviations (RSD)

between 1 and 8% throughout both measurement series were ob-

tained From the two ISTD only for benzene-d 6 recovery rates

near 100% were obtained on all tested adsorbents Benzene-d 6was

hence used to compensate variation of measurement performance

Chromatograms of the analysis of the adsorbents under both sam-

pling conditions are provided in the supplementary material (S1–

S12)

3.2.1 Dry sampling conditions (0% RH)

In view of the amount of retained compounds within the av-

erage recovery range of 80–120%, the CMS Carbosieve S II, Car-

boxen 569 and Carboxen 1003 showed the best retention ability

for the majority of the VVOCs at 0% RH followed by the GCB Carbo- graph 5TD The weaker Carbopack B and Tenax GR performed well for the polar compounds 2-butanone [(85 ± 2)% and (81 ± 5)%], propanal [(114 ± 12)% and (90 ± 5)%] and Carbopack B for the less polar compound isoprene (103 ± 1)% compared to the others Fi- nally, these six adsorbents were selected for further tests under humid sampling conditions Although Carboxen 1018 showed as good recovery rates as the other CMS it was not selected, since sul- phur dioxide (SO 2) is produced in the adsorbent (cf Section3.2.5)

3.2.2 Humid sampling conditions (50% RH)

The repetition of the recovery tests under humid sampling con- ditions revealed a significant impact of water vapour From the two Carboxens, the number of retained compounds with average recov- eries between 80 and 120% decreased from 25 to 5 for Carboxen

569 and from 24 to 11 for Carboxen 1003 As reported by Valle- cillos, Maceira, et al [23] this may also be linked with the active sites on the adsorbents’ surfaces covered by water molecules and

is correlating with the relatively high water uptake compared to the other adsorbents Moreover, breakthrough volumes of the tar- get compounds can also be affected by the presence of humidity during the sampling [30] Carbosieve S II as well showed decreased retention capacity for some VVOC, however, to a much lower ex- tent (from 23 to 16 compounds) and at a significantly higher water uptake as observed for the Carboxens The GCBs Carbopack B, Car- bograph 5TD and Tenax GR, which is a mixture of Tenax TA and a graphitised carbon, are only slightly affected by air humidity cor- responding to their low water uptake

As could be observed, the recovery of some – mainly polar – compounds increased in presence of water vapour in the supply air These are methanol on Carboxen 1003 [increase from (85 ± 1)%

to (120 ± 20)%] and ethyl acetate on Carbopack B [(68 ± 15)% to (92 ± 8)%] and Carbograph 5TD [(76 ± 10)% to (95 ± 2)%]. For pen- tanal, which is less polar and not a VVOC the recovery in- creased significantly on Tenax GR [(56 ± 8)% to (80 ± 3)%], Car- bopack B [(48 ± 1)% to (81 ± 3)%] and Carbograph 5TD [(39 ± 11)%

to (79 ± 4)%]. Generally, for all adsorbents, dissatisfying recoveries ( < 80%) under humid conditions were observed for chlorodifluo- romethane, n-propane, 1,3-butadiene, isopropyl alcohol and vinyl acetate

3.2.3 Testing of multi-bed tubes

The high standard deviations of the recovery for a few com- pounds can either be explained by analytical reasons or by in- complete desorption or breakthrough Therefore, combinations of adsorbents should be taken into consideration Based on the re- coveries in Table 3 and under consideration of a relative humid- ity of 50%, the combinations Carbograph 5TD/Carboxen 1003/Car- bosieve SII and Tenax GR/Carboxen 1003/Carbosieve SII were con- sidered for further testing following the procedure described in

8 M Richter, E Juritsch and O Jann / Journal of Chromatography A 1626 (2020) 461389

Section2.4and compared to the commercial multi-bed tube Car-

botrap 300 (Gerstel, Germany) containing Carbotrap C/Carbotrap

B/Carbosieve SIII

The results in Table3show an improvement of the performance

of the multi-bed tubes compared to the single adsorbents How-

ever, there is no significant difference between the combinations

identified in this study compared to the commercial tube Further-

more, the results are comparable to the recoveries determined for

Carbograph 5TD, which is part of one multi-sorbent tube tested It

is noticeable that the polar VVOC methanol is not retained apart

from Carbotrap 300, although its very good recovery determined

on both Carbosieve S II and Carboxen 1003 Since the assumption

can be made that the adsorbents in combination will complement

each other, optimisation might be obtained by adapting the bed

lengths

3.2.4 Water management

For an efficient measurement method, water management is

highly recommended Some authors propose the use of pre-tubes

filled with drying agents, e.g CaCl 2or Nafion® [ 14, 23, 24, 26] How-

ever, since these might serve as adsorbents themselves, losses at

non-targeted analysis might be the result Pollmann, Helmig, et al

[31]used a Peltier-cooled, regenerable water trap inserted into the

sample flow to condensate water prior to analysis Dry-purging of

the adsorbents would also be an option [14] During the research

for this study, good experiences were made with the solvent vent-

ing dry-purge mode of the thermal desorption system, which in-

deed led to reduced sensitivity of measurement, but which could

be compensated with an enhanced sample amount ( Table1) How-

ever, to obtain a reliable measurement method, more efforts must

be made to solve the humidity issue

3.2.5 Chemical reactions

Although Carboxen 1018 showed as good recovery rates as the

other CMS it was not selected, since sulphur dioxide (SO 2) is pro-

duced in the adsorbent giving a large peak at the beginning of the

chromatogram impacting the analysis The same was also observed

in the other Carboxen type sorbents but to a much lower extent

The SO 2peak disappeared or reduced at least to a negligible area

after the tube was thermally handled prior to use ( ∼ 20°C above

recommended desorption temperature) Although the test gas mix-

ture was containing CS 2, there was no significant indication for it

to trigger any reaction, since SO 2 was also occurring in the blank

measurements However, Brown and Shirey [32]reported that the

formation of SO 2 or CO 2 is common to most carbon molecular

sieves, and does not pose a problem unless the user is trying to

sample for these two analytes They do not explain why the for-

mation of these molecules takes place but Boehm [33]reports that

surface oxides inherent to carbon materials decompose to CO 2and

CO on heating to high temperatures and that highly reactive sites

remain on the carbon surface After cooling to room temperature,

they can react with oxygen (air) or even water vapour, giving new

surface oxides It can be assumed that this mechanism is also re-

sponsible for the oxidation of sulphur, inherently occurring in car-

boxen type adsorbents, which are produced from sulfonated poly-

mers [34] Since the group of the VVOC contains highly reactive

compounds, a close look into the occurrence of chemical reac-

tions in the employed adsorbents must be taken Some insight into

this already is given in the literature, e.g in Schieweck, Gunschera,

et al [18]

Moreover, for some single adsorbents but particularly for the

multi-bed combinations recoveries greater than the tolerated 120%

for some compounds were determined These observations can

only partially be explained as the blank measurements carried

out prior to loading revealed blank values for some components

that even did not decrease after repeated desorption Artefacts or

residues of propene and n-propane were found on Carbosieve S II

as well as on Carbotrap 300 together with n-butane Tenax GR and Carbograph 5 TD showed high benzene blanks, whereupon arte- fact formation of benzene in Tenax adsorbents is well known Arte- fact formation might furthermore be promoted by the presence of water However, detailed investigations on this issue are necessary and objective of ongoing work A suitable method for this might be the standard elevation method to get indication on matrix effects Chromatograms of blank measurements of each adsorbent are added to the supplementary material (S13–S24)

4 Conclusions

The recovery rates of 29 VVOC and three VOC in nine different adsorbent materials (porous polymers, GCB and CMS) were deter- mined The recovery calculation was obtained by direct and, hence, unaffected measurement of the gas standard mixture This way, any effects that might be resulting from interactions of the test sample molecules with the adsorbent, e.g breakthrough phenom- ena, insufficient desorption or chemical reactions are considered and evaluated

Sampling performance is strongly affected by water vapour in the sample air A comparison between dry (0% RH) and humid (50% RH) sampling conditions revealed that the number of retained VVOC with average recoveries between 80 and 120% dropped sig- nificantly for the CMS Carboxen 569 and Carboxen 1003 compared

to the recoveries under dry sampling conditions This was further- more well correlating with the relatively high water uptake com- pared to the other adsorbents Water management measures are therefore highly recommended In this context, the common prac- tice of calibration with liquid standard solutions followed by flush- ing with a dry inert gas flow should be rethought Due to the ob- vious impact of air humidity leading to lower adsorption capacity particularly of the GCB and CMS, underestimations during analysis are likely

Chemical reactions in the carbon-based adsorbents themselves

or surface reactions with analytes might be a problem In this study, the generation of SO 2 in the CMS and particularly in Car- boxen 1018 was observed This can be a problem when analytes of interest elute with the same retention time or close to it

For the measurement of complex gas samples, combinations of adsorbents should be used With the procedure described here, the combinations Carbograph 5TD/Carboxen 1003/Carbosieve SII and Tenax GR/Carboxen 1003/Carbosieve SII were identified to be po- tentially suitable The improvement of the performance compared

to the single adsorbents particularly under humid sample condi- tions could be shown The comparison with a commercial tube revealed no significant difference However, one third of the tar- get analytes could not be satisfyingly retained so that potential for optimisation can be seen in the adaptation of the adsorbent bed lengths

Future research should focus on investigations on the optimum composition of multi-bed sampling tubes, the recovery under re- alistic sampling conditions, also including the always present VOC and SVOC, possible chemical reactions, storage effects (compounds migration between sorbent beds) and the loss-free water man- agement These items are objectives of a research project recently started and funded by the German Environment Agency (UBA) Its outcome will be published in a forthcoming paper

Declaration of Competing Interest

The authors declare that they have no known competing finan- cial interests or personal relationships that could have appeared to influence the work reported in this paper

CRediT authorship contribution statement

analysis, Supervision, Writing original draft Elevtheria Juritsch:

Methodology, Investigation, Formal analysis, Writing review &

editing Oliver Jann: Conceptualization, Resources, Writing re-

view & editing

Acknowledgements

This research did not receive any specific grant from funding

agencies in the public, commercial, or not-for-profit sectors The

authors would like to thank Timo Juritsch for proofreading the

manuscript

Supplementary material associated with this article can be

found, in the online version, at doi: 10.1016/j.chroma.2020.461389

References

[1] P Wolkoff, Indoor air pollutants in office environments: assessment of comfort,

health, and performance, Int J Hyg Envir Heal 216 (2013) 371–394

[2] P Burge , Sick building syndrome, Occup Environ Med 61 (2004) 185–190

[3] ISO, Indoor air - Part 6: determination of volatile organic compounds in indoor

and test chamber air by active sampling on Tenax TA® sorbent, thermal des-

orption and gas chromatography using MS/FID (ISO 160 0 0-6:2011)International

Organization for Standardization, Beuth, Berlin, 2011

[4] ISO, Indoor air – Part 9: determination of the emission of volatile organic

compounds from building products and furnishing - emission test cham-

ber method (ISO 160 0 0-9:20 06)International Organization for Standardization,

Beuth, Berlin, 2006

[5] ISO, Indoor air - Part 11: determination of the emission of volatile organic

compounds from building products and furnishing - sampling, storage of sam-

ples and preparation of test specimens (ISO 160 0 0-11:20 06)International Or-

ganization for Standardization, Beuth, Berlin, 2006

[6] ISO, Indoor Air - Part 3: determination of formaldehyde and other carbonyl

compounds - active sampling method (ISO 160 0 0-3:2011)International Organi-

zation for Standardization, Beuth, Berlin, 2011

[7] ENConstruction Products - Assessment of Release of Dangerous Substances -

Determination of Emissions Into Indoor Air (EN 16516:2017), Beuth, Berlin,

2017

[8] ECAECA report no 27 - Harmonisation framework for indoor products labelling

schemes in the EU, European Collaborative Action, Publications Office of the

European Union, Luxembourg, 2012

[9] ECAECA report no 29 - Harmonisation framework for health based evaluation

of indoor emissions from construction products in the European Union using

the EU-LCI concept, Publications Office of the European Union, Luxembourg,

2013

[10] ECAReport No 24 - Harmonisation of indoor material emissions labelling sys-

tems in the EU, European Collaborative Action, Publications Office of the Euro-

pean Union, Luxembourg, 2005

[11] EU, Agreed EU-LCI values, in, 2018

[12] E Gallego , F.J Roca , J.F Perales , X Guardino , Comparative study of the adsorp-

tion performance of a multi-sorbent bed (Carbotrap, Carbopack X, Carboxen

569) and a Tenax TA adsorbent tube for the analysis of volatile organic com-

pounds (VOCs), Talanta 81 (2010) 916–924

[13] K Dettmer , W Engewald , Adsorbent materials commonly used in air analy- sis for adsorptive enrichment and thermal desorption of volatile organic com- pounds, Anal Bioanal Chem 373 (2002) 490–500

[14] E Woolfenden , Sorbent-based sampling methods for volatile and semi-volatile organic compounds in air Part 2 Sorbent selection and other aspects of opti- mizing air monitoring methods, J Chromatogr A 1217 (2010) 2685–2694 [15] ISO, Indoor, ambient and workplace air - Sampling and analysis of volatile or- ganic compounds by sorbent tube/thermal desorption/capillary gas chromatog- raphy - Part 1: pumped sampling (ISO 16017-1:20 0 0)International Organiza- tion for Standardization, Beuth, Berlin, 20 0 0

[16] I Camsco, Camsco Sorbent Selection Chart, in, 2009

[17] T Salthammer , Very volatile organic compounds: an understudied class of in- door air pollutants, Indoor Air 26 (2016) 25–38

[18] A Schieweck , J Gunschera , D Varol , T Salthammer , Analytical procedure for the determination of very volatile organic compounds (C3–C6) in indoor air, Anal Bioanal Chem 410 (2018) 3171–3183

[19] A Pech , O Wilke , W Horn , O Jann , Analysis of very volatile organic com- pounds (VVOC) with thermal desorption GC–MS, in: Indoor Air 2016 - The 14th International Conference on Indoor Air Quality and Climate, Ghent, Bel- gium, 2016, p 2

[20] D Helmig , N Schwarzer , J Steinhanses , Flame control accessory for GC-FID op- eration with autosampler injection, J High Res Chromatog 13 (1990) 849–851 [21] F Stoeckli , T Jakubov , A Lavanchy , Water adsorption in active carbons de- scribed by the Dubinin–Astakhov equation, J Chem Soc Faraday Trans 90 (1994) 783–786

[22] J Koziel , M.Y Jia , J Pawliszyn , Air sampling with porous solid-phase microex- traction fibers, Anal Chem 72 (20 0 0) 5178–5186

[23] L Vallecillos , A Maceira , R.M Marce , F Borrull ,Evaluation of active sampling strategies for the determination of 1,3-butadiene in air, Atmos Environ 176 (2018) 21–29

[24] A Maceira , L Vallecillos , F Borrull , R.M Marce , New approach to resolve the humidity problem in VOC determination in outdoor air samples using solid ad- sorbent tubes followed by TD-GC-MS, Sci Total Environ 599 (2017) 1718–1727 [25] G.K.S Wong , S.J Ng , R.D Webster , Quantitative analysis of atmospheric volatile organic pollutants by thermal desorption gas chromatography mass spectrom- etry, Anal Methods-UK 5 (2013) 219–230

[26] P Ciccioli , A Cecinato , E Brancaleoni , M Frattoni , A Liberti , Use of carbon adsorption traps combined with high resolution gas chromatography – mass spectrometry for the analysis of polar and non-polar C4-C14 hydrocarbons in- volved in photochemical smog formation, Separ Sci 15 (1992) 75–84 [27] V.M Brown , D.R Crump , An investigation into the performance of a multi-sor- bent sampling tube for the measurement of VVOC and VOC emissions from products used indoors, Anal Methods-UK 5 (2013) 2746–2756

[28] A Pech , O Wilke , B Mull , W Horn , O Jann , Development of a TDS-GC-FID method for the determination of methanol and ethanol in air, Gefahrst Rein- halt L 73 (2013) 477–482

[29] K Dettmer , T Knobloch , W Engewald , Stability of reactive low boiling hydro- carbons on carbon based adsorbents typically used for adsorptive enrichment and thermal desorption, Fresen J Anal Chem 366 (20 0 0) 70–78

[30] D Helmig , L Vierling , Water adsorption capacity of the solid adsorbents Tenax

TA, Tenax GR, Carbotrap, Carbotrap C, Carbosieve SIII, and Carboxen 569 and water management techniques for the atmospheric sampling of volatile or- ganic trace gases, Anal Chem 67 (1995) 4380–4386

[31] J Pollmann , D Helmig , J Hueber , D Tanner , P.P Tans , Evaluation of solid ad- sorbent materials for cryogen-free trapping - gas chromatographic analysis of atmospheric C2-C6 non-methane hydrocarbons, J Chromatogr A 1134 (2006) 1–15

[32] J Brown , B Shirey , Technical Report - A Tool for Selecting an Adsorbent for Thermal Desorption Applications, Supelco (2001) 36

[33] H.P Boehm , Surface oxides on carbon and their analysis: a critical assessment, Carbon N Y 40 (2002) 145–149

[34] Y Juillet , S Le Moullec , A Begos , B Bellier , Optimisation of sorbent trap- ping and thermal desorption-gas chromatography-mass spectrometric condi- tions for sampling and analysis of hydrogen cyanide in air, Analyst 130 (2005) 977–982