ARTICLE IN PRESS

Contents lists available at ScienceDirect

Journal of Quantitative Spectroscopy &Radiative Transfer

Journal of Quantitative Spectroscopy & Radiative Transfer 110 (2009) 766–782

0022-40

doi:10.1

� Cor

E-m

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

UV Fourier transform absorption cross sections of benzene, toluene,meta-, ortho-, and para-xylene

Sophie Fally a, Michel Carleer a, Ann C. Vandaele b,�

a Universite Libre de Bruxelles, Faculte des Sciences, Service de Chimie Quantique et Photophysique, C.P. 160/09, 50 Avenue F.D. Roosevelt, B-1050 Brussels, Belgiumb Institut d’Aeronomie Spatiale de Belgique, 3 Av Circulaire, B-1180 Brussels, Belgium

a r t i c l e i n f o

Article history:

Received 12 August 2008

Received in revised form

21 November 2008

Accepted 30 November 2008

Keywords:

Laboratory measurement

Molecular spectroscopy

FTS

Photoabsorption

73/$ - see front matter & 2008 Elsevier Ltd. A

016/j.jqsrt.2008.11.014

responding author. Tel.: +32 2 3730367.

ail address: A-C.Vandaele@aeronomie.be (A.C

a b s t r a c t

New measurements of the absorption cross sections of gaseous benzene, toluene, meta-,

ortho-, and para-xylene have been performed with a Fourier transform spectrometer

Bruker IFS120 M at the resolution of 1 cm�1 over the 30 000–42 000 cm�1 spectral range.

The recordings were carried out under different pressure and temperature conditions

with pure samples. The effect of the temperature on the absorption cross sections is

investigated. Comparison with the literature shows large differences, largely attributed

to the experimental difficulties encountered during these previous measurements and

to a resolution effect. To our knowledge, it is the first time that such a dataset of UV

absorption cross sections with temperature dependence is reported in the literature.

Such data should be useful for upcoming remote sensing applications, such as

atmospheric studies both on Earth and on other planets.

& 2008 Elsevier Ltd. All rights reserved.

1. Introduction

The aromatic species such as benzene (C6H6), toluene (C7H8), and xylenes (or the three isomers of dimethyl-benzeneC6H4(CH3)2) present a growing interest, because of their role in the chemistry of the tropospheric ozone [1], but alsobecause of their noxious effects on human health. Typical annual average concentrations in urban locations rangefrom 1 to 100 ppbv but exhibit considerable diurnal and seasonal variations [2–8]. Daily mean values in a urbanenvironment have been observed to lie in the ranges 1–12, 1–30, 0–4, 1–18, and 0–4 ppb for benzene, toluene, m-, o-,and p-xylenes, respectively, but higher peak values are also measured [2,6,8]. Spectroscopic measurements ofaromatic compounds in the atmosphere have been reported by several authors [3,6,7,9]. However, the quality of thosedetections is poor as comparisons with gas chromatography measurements show discrepancies up to a factor of 10 [10–12].Volkamer et al. [13] attributed the weak correlation to the use of a spectrometer with insufficient spectral resolution, theoverlap with the Hertzberg O2 bands absorbing in the same spectral region, and the use of poor quality absorption crosssections.

Benzene is one of the heaviest hydrocarbons ever identified in the giant planets’ atmospheres. It has been detected inthe north polar auroral region of Jupiter [14] and in the stratospheres of Jupiter and Saturn [15] as well as in Titan’satmosphere [16]. Benzene and substituted benzenes as well as toluene were observed in laboratory experimentsreproducing the photochemistry of Titan [17]. Models of the atmospheres of Jupiter, Saturn, and Titan now include benzenein their photochemistry scheme [18–21]. The accurate knowledge of the photoabsorption cross section of such species is

ll rights reserved.

. Vandaele).

ARTICLE IN PRESS

S. Fally et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 110 (2009) 766–782 767

required for those modeling studies. The need for laboratory spectroscopic data to study organic chemistry in planetaryatmospheres was highlighted in [22–24]. In particular, low temperature dependence in the UV range is an importantconcern.

The UV absorption spectrum of benzene is attributed to the S1 (1B2u)’S0 (1A1g) electronic transition. It is thelowest singlet system of the molecule and is electronically forbidden but vibrationally induced. The dominantvibronic progression is designated 61

0 and involves n1 the symmetrical ring-breathing vibration. This band becomes moreallowed and therefore more intense as the D6h symmetry is broken in methyl and dimethyl-substituted benzene, i.e.,toluene and xylene, respectively. The rotational structure of the S1’S0 61

0 transition in benzene has been completelyanalyzed by [25].

Existing data in the literature for remote sensing applications [26–31] have been recorded at low resolutions and atroom temperature (see Table 1 for the description of the existing literature data). The absolute absorption cross section ofbenzene was measured by Pantos et al. [28] at room temperature in the 135–270 nm (37 100–74 200 cm�1) regionwith a resolution of 0.08 and 0.25 nm and by Hiraya and Shobatake [26] in the 130–260 nm (38 500–76 900 cm�1) at 185 K.Suto et al. [30] measured the photoabsorption cross section of benzene and m-, o-, and p-xylenes in the 106–295 nm(33 900–94 300 cm�1) spectral region using synchrotron radiation as the light source and at a resolution of 0.1 nm.Milton et al. [27] report measurements of the absorption cross section of benzene and toluene in the near-ultravioletperformed at a resolution of 0.2 nm. However these data are no longer available in digital format. Trost et al. [31] reportcross section of several aromatic hydrocarbons, including those of benzene, toluene, and the xylenes, measured at aresolution of 0.11 nm in view of DOAS (differential optical absorption spectroscopy) measurements of hydrocarbons in thefield. These data were upgraded by measurements performed by Etzkorn et al. [29], who obtained UV cross sections at aresolution of 0.15 nm and IR spectra at 1 cm�1 resolution. Recently, interest was shown in the measurements of theabsorption cross sections of hydrocarbons in the mid-IR region: Klingbeil et al. [32] investigated a series of gaseoushydrocarbons, and in particular derived the temperature dependency of the cross sections of toluene and m-xylene;Rinsland et al. [33] measured the absorption of benzene vapor and analyzed the integrated band intensities as function ofthe temperature.

Several authors [13,34–36] have shown that, in the frame of atmospheric studies, absorption cross sections have to bemeasured at high resolution, even if their final use only requires low resolution, such as in DOAS techniques. Otherwisenon-linearities are introduced during the convolution process and therefore in the retrieval process. The importance of theresolution, of the temperature dependence, but also of the wavelength accuracy has been addressed in the recent literaturein several similar laboratory studies on other molecules like ozone [37], BrO [38], OClO [39], formaldehyde [40], NO2

[41,42], and NO3 [43]. Some of them also deal with pressure effects. By taking into account these experimental aspects,these studies achieve to report accurate absolute absorption cross sections aiming at improvements of reference data forspectroscopic databases.

Similarly to such studies, the aim of the present work is to provide absorption cross sections at higher resolution thanpreviously reported and to investigate the temperature effect in support of tropospheric and astrophysical applications. Thepresent unique systematic study of five organic molecules includes an error estimate and comparisons with availableliterature data. The complete dataset comprising the absorption cross sections at different temperatures is available inelectronic format from the website of the Belgian Institute for Space Aeronomy (http://www.aeronomie.be/spectrolab/) orupon request to the corresponding author.

2. Experimental description

All measurements were carried out with a Fourier transform spectrometer Bruker IFS120M (see Fig. 1 for aschematic of the experimental configuration). The combination of a Xenon high pressure lamp (450 W) and a UV-diodedetector allows the covering of the 30 000–42 000 cm�1 spectral region. Spectra have been recorded at theresolution of 1 cm�1 (MOPD ¼ 0.9 cm) with no apodization. The choice of this resolution is a compromise betweenhaving the best resolution and the highest signal to noise ratio and limiting the measurement time thus reducingpressure variations in the cell. The latter may be due to photodissociation processes under UV illuminationor polymerization on the cell windows and walls. To evaluate and limit those processes, (1) the cell was notilluminated, when it was filled, preventing any photodissociation to occur while the pressure was stabilizing;(2) spectra were recorded in successive blocks, allowing the detailed study of the evolution of the signal during thewhole duration of the experiment. The recording of one block of scans (see Table 2) lasted about 4–8 min dependingon the number of co-added scans (64 and 128, respectively). Once the cell was illuminated, the total pressure insidethe cell would in general increase by 2–10% while the absorption amplitude measured in the successive blockswould decrease, indicating that photodissociation was the main process taking place inside the cell. To correct for thispressure variation, the absorption signature observed in the successive blocks was scaled to the absorption present in thefirst block.

Two types of 10 cm long cells were used during this study. The first series of cells were 5 cm of diameter with the bodymade of Pyrex and the windows of Quartz. These cells could not be temperature stabilized and were only used for thedetermination of the absorption cross sections at room temperature. The cell was replaced each time the molecular species

ARTIC

LEIN

PRESS

Table 1Description of the literature data. For each reference, the following information is given: temperature of the recording, spectral region investigated, technique used, spectral resolution, uncertainty e on the cross

section, and some general comment.

Author Speciesa T (K) Region Techniqueb Resolutionc Dispersiond ewnd exs

(%)d

Comment

Pantos et al. [28] B Room UV

(135–270 nm ¼ 37 100–74 200 cm�1)

H-G-PM 0.08–0.25 nm n N 10 No data in digitized

format

C ¼ 1 cm (11–34 cm�1)

P ¼ 0.2–25 Torr

Hiraya and

Shobatake [26]

B 185e UV

(130–260 nm ¼ 38 500–76 900 cm�1)

S�J 0.065 nm n N n

(9 cm�1)

Milton et al. [27] BT+N2 Room Near-UV

(220–270 nm ¼ 37 000–45 500 cm�1)

G 0.2 nm n 1 nm n Data from plots

(27 cm�1)

Suto et al. [30] B3X 297 UV

(106–295 nm ¼ 33 900–94 300 cm�1)

S�G-PM

C ¼ 40.7 cm

0.1 nm (14 cm�1) n N 10 Data from plots

Trost et al. [31] BT3X+others Room UV

(230–290 nm ¼ 34 500–43 500 cm�1)

D-G-PD

C ¼ 1–20–50 mm

0.11 nm (15 cm�1) 0.038 nm N 4

(T3X)

12

(B)f

Etzkorn et al. [29] BT3X+others 293 UV

(230–290 nm ¼ 34500–43 500 cm�1)

UV:X-G-PD UV 0.146 nm

(20 cm�1)

0.0385 nm N 7

(3X)

Upgrade of Trost

IR (1000–4000 cm�1) IR:FT IR 1 cm�1 11 (B)

C ¼ 6.22 m 16

(T)g

Klingbeil et al. [32] TmX+N2+others 278–773 Mid-IR (2500–3400 cm�1) FT 1 cm�1 n N 2 (T) Tdep

C ¼ 15.9 cm 5

(mX)

P ¼ 1–50 Torr

Rinsland et al. [33] B 278, 298,

323

Mid-IR (615–6100 cm�1) FT 0.112 cm�1 n N 3 Tdep, integrated band

intensities

C ¼ 20 cm

P ¼ 0.1–22 Torr

a B ¼ benzene, T ¼ toluene, mX ¼ m-xylene, 3X ¼ m-, o-, p-xylenes.b H ¼ H2 discharge lamp; D ¼ deuterium lamp; S ¼ synchrotron radiation; X ¼ xenon arc lamp; G ¼ grating; FT ¼ Fourier transform spectrometer; J ¼ supersonic jet; PM ¼ photomultiplier;

PD ¼ photodiode array; C ¼ cell length; P ¼ pressure.c Conversion to wavenumber (cm�1) performed at 37 000 cm�1.d n ¼ not specified.e Vibrational temperature.f Uncertainty calculated from uncertainty on the cell length, temperature and concentration reported in Table 3 of Trost et al. [31].g Uncertainty calculated from Table 1 in Etzkorn et al. [29].

S.Fa

llyet

al.

/Jo

urn

al

of

Qu

an

titativ

eSp

ectrosco

py

&R

ad

iative

Tra

nsfer

110(2

00

9)

76

6–

78

27

68

ARTICLE IN PRESS

10 cm cell

FourierTransform

Spectrometer

BrukerIFS120M

to pressure gauges

togas

handlingsystem

T°Reading

Xe arcfrom / to coolingsystem

Fig. 1. Experimental configuration.

S. Fally et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 110 (2009) 766–782 769

was changed, i.e., between two experiments. Another cell, made of stainless steel with double CaF2 windows,was temperature stabilized by the circulation of a cooling fluid (methanol) through the foam insulated jacket ofthe cell and by use of a cryostat (Haake DC10). This cell was used for the experiments at room and low temperatures.Temperature was measured using a thermocouple type T sensor placed inside the cell (accuracy of 2 K). The windows werecontinuously flushed by dry N2 to prevent any condensation at cold temperatures. During one complete measurement(about 6 h), the temperature was observed to vary by less than 0.5 K. The cell windows were replaced by clean onesbetween two experiments. For all cells, pressure was measured using two Baratron pressure gauges with, respectively, 1000and 10 Torr full range (accuracy of 0.08% reading). Both gauges were calibrated (0 and full scale) before the measurementcampaign.

The gaseous samples were prepared from liquid products (Fluka and Riedel-De Haen with purity of ‘‘pro analysis’’ type,499% for xylenes and 499.9% for benzene and toluene). The liquid sample was first purified by repeatedly pumping on thesample frozen at liquid nitrogen temperature. Because of the large volatility of those compounds, vapor was then obtainedby degassing.

One complete experiment at a given temperature consisted of the following steps:

(i)

before a set of measurements, the light source was left in operation for several hours to stabilize its intensity; (ii) a blank measurement was recorded with the cell empty;

(iii)

the light beam was deviated from the entrance of the cell while the species was introduced. After a time long enough(a few minutes) to allow for the stabilization of the pressure inside the cell, the light was admitted into the cell andspectra were recorded;

(iv)

the gas was pumped out and a new blank spectrum was recorded.Such a procedure was repeated at least two times for each species and each temperature. Spectra were recorded at253, 263, 273, 283, and 293 K. Some pressure–temperature combinations were however impossible to achieve due tothe decrease of the saturation vapor pressure with temperature. In particular, for the three xylene molecules, nomeasurements at 263 and 253 K could be performed. Table 2 presents the experimental conditions under which all thespectra used in this study were obtained. By performing numerous recordings at several pressures and temperatures,we could validate each individual measurement, check their repeatability, and average them in order to obtain a bettersignal-to-noise ratio.

ARTICLE IN PRESS

Table 2Experimental conditions for recording the BTX spectra.

Species T (K) Partial pressure (Torr) (Nb blocks)� (Nb scans)

Benzene 253 0.702, 0.750, 0.895 10� 64

263 0.790, 0.8025, 0.824 10� 64

273 0.807, 0.996, 1.009 10� 64

283 0.801, 0.824, 0.866, 0.917, 1.008 10� 64

293 0.999, 1.004 10� 64

0.987, 1.693 8�128

2.439 6�128

3.079 4�128

Toluene 273 2.556, 3.091 10� 64

283 2.244, 2.363 10� 64

293 2.288, 2.535 10� 64

1.265 16�128

3.427 8�128

o-Xylene 273 1.028, 1.123 10� 64

283 1.139, 1.268 10� 64

293 5.062, 5.530, 6.100 8�128

p-Xylene 273 0.992, 1.250 10� 64

283 1.578, 1.595 10� 64

293 1.562, 1.727 10� 64

2.001, 3.054 8�128

m-Xylene 273 0.902, 0.993 10� 64

283 2.033, 2.118 10� 64

293 3.087, 3.662 10� 64

3.043, 4.617 8�128

S. Fally et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 110 (2009) 766–782770

3. Data analysis

Absorption cross sections were deduced from the different measurements using the following equation:

sðnÞ ¼ 1

nBTXd� ln

I0ðnÞIðnÞ

� �(1)

with

I0ðnÞ ¼ 12½I0;beforeðnÞ þ I0;afterðnÞ� (2)

where nBTX is the concentration of the species inside the cell calculated from its partial pressure, d is the cell length (10 cm),I0,before and I0,after are, respectively, the blank spectra recorded before and after the BTX measurement, and I is the intensitymeasured when the cell contained the studied species. Measured pressures were corrected for the thermal transpiration ofthe capacitance manometer using the formalism described in Poulter et al. [44].The final absorption cross section at a giventemperature is the average of all cross sections obtained independently.

The temperature effect was determined by taking into account all the absorption cross sections of a species at all theavailable temperatures. When the dataset was large enough (spectra available for at least three different temperatures), alinear dependency with respect to temperature was used to simulate this effect defining a temperature coefficient c(n). Thechoice to use this simple description for the temperature effect results from the small number of different temperaturesinvestigated in this work. As will be shown later for benzene and toluene, the absorption cross section values at the peakmaxima are well reproduced by a linear temperature dependence. In the troughs, the temperature effect is very small,generally within the uncertainty on the absorption cross section themselves. The absorption cross section at temperature T

can be calculated using the following relation:

sT ðnÞ ¼ s293 KðnÞ þ cðnÞ T½K� � 293ð Þ (3)

The systematic uncertainties on the absorption cross sections have been estimated by considering different error sources:uncertainties on the concentration, taking into account the uncertainties on the temperature (2 K) and the pressure, as well

ARTICLE IN PRESS

S. Fally et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 110 (2009) 766–782 771

as uncertainties on the length of the cell (2%). The uncertainties on the pressure come from the reading accuracy and thepressure variation during the measurement. As already mentioned, the total pressure could vary by 2–10% inside the cellbut this variation was not a direct reading of the evolution of the partial pressure of the species of interest. Spectra weretherefore corrected by scaling the absorption measured in each individual block of scans to the absorption observed in thefirst block, and inaccuracy on the pressure was estimated to be 5%. Taking into account all these error sources, the totalsystematic uncertainty on the cross section has been estimated to be of the order of 8% if we consider a simple errorpropagation calculation from Eq. (1).

Non-systematic uncertainties on the absorption cross sections were also estimated. Those mostly are due to the noiselevel on the absorbance (i.e., the noise levels on I and I0) and to the lamp drift with time. Noise levels were evaluated as therms standard deviation of the measured spectra where no absorption features are located. Lamp drift was estimated bylooking at the evolution of the intensity of the lamp spectra in the 35 000–36 000 cm�1 interval, where there is noabsorption due to the investigated species. This simple correction assumes that the evolution of the lamp spectrum in timedoes not vary with wavenumbers. It was found that the lamp intensity would be, for the worst cases, between 0.95 and 1.05that of the first block of the I0,before, leading to a upper limit of 10% for the lamp intensity variation. This variation influencethe accuracy on the absorption cross section through the definition of the absorbance (ln(I0/I)). At lower temperature, thenon-systematic error is larger because the averaging had to be performed on a smaller number of individual cross sections:typically two measurements were carried out at 253 K instead of 5 at 293 K (see Table 2). Non-systematic errors are given inthe files which are available from the website of the Belgian Institute for Space Aeronomy (http://www.aeronomie.be/spectrolab/).

In the next sections, those results are presented in more details for each species separately.

4. Results and discussion

4.1. Absorption cross section at room temperature

4.1.1. Resolution effect and wavenumber calibration

Fig. 2, which compares the absorption cross section of benzene obtained in this work with some recent data from theliterature [27,29–31], illustrates the effect of the resolution: at the resolution used in this work, structures of high intensityappear, which are completely smoothed out when operating at a coarser resolution.

Another problem encountered with data from the literature is the poor wavelength calibration of these data. Forexample, from the lower panels of Fig. 4, the importance of the wavenumber discrepancy may be inferred. Althoughbenzene data of Trost et al. [31] have been shifted by 20.2 cm�1 to make the absorption structures coincide at best, therestill is a difference in the positions of some of the absorption structures. This means that the discrepancy between bothwavenumber scales is not reproduced by only a shift but that a stretching of the scale should also be introduced. The sameis observed for the other species, as for example in Fig. 6(b) for toluene where small shifts are still visible even after havingshifted the data of Etzkorn et al. by a constant value of 24.6 cm�1. In this study, we have however limited the correction ofthe wavelength scale of the different literature datasets to a shift, as the introduction of a stretch would not change theconclusions of the comparison. The values of the applied shifts are summarized in Table 3 (and reminded in the legend ofthe figures). It is seen that the wavenumber shifts vary between �48.8 and +42.7 cm�1, and that the best agreement isobserved with the values of Etzkorn et al. [25] except for toluene. For comparison purposes those shifts have also beenconverted in nanometers (conversion performed at 37 000 cm�1) and have to be compared to the announced accuracy onthe wavelength scales of the literature data. For the data of Etzkorn et al. [25] no wavelength accuracy is given, however thevalues of the shifts are of the same order of their wavelength sampling (0.038 nm) except for toluene. Milton et al. [27]announced an accuracy of 1 nm on their wavelength scale which is far higher than the shifts found in this work. Shiftsintroduced in the data of Trost et al. [31] are a factor of 10 higher than their sampling (0.038 nm) and are also higher thanthe stated resolution. In Suto et al. [30], there is no discussion of the wavelength calibration and the shifts are differentfrom one species to the other. The use of a FTS greatly improves the accuracy on the determination of the wavenumberscale, as the latter is derived from the intervals at which the interferogram is sampled, which are defined by the fringes of astabilised He–Ne laser, that acts thus as an internal reference. The wavenumber of this laser is known very accurately and isvery stable. As a result, the wavenumber calibration of interferometers is much more accurate and has much better longterm stability than the calibration of dispersive instruments (Connes advantage). The estimated accuracy on thewavenumbers, based on I2 calibration measurements, is of the order of 0.02 cm�1. The spectrometer was not operatedunder vacuum so that all wavenumbers given in this work are expressed in air and are therefore defined as 1/lair.

In order to perform reliable comparisons between absorption cross sections obtained at different resolutions, the data ofthe present work have been convolved with a Gaussian function to correspond to the lower resolution of the literature data.It should be reminded that a Gaussian function is not the exact representation of the instrumental (ILS) function of agrating spectrometer. It is however deemed a close enough approximation for the purpose of this paper. Of course, whenadapting a cross section to a specific instrument, care should be taken to use the exact ILS function, and to even take intoaccount the fact that this function might vary with wavelength. Those remarks explain some of the spikes seen in the plotscomparing two datasets. Some of those spikes can also be attributed to the interpolation procedure. In the next sections of

ARTICLE IN PRESS

Fig. 2. Comparison of the data of this work (black) with the data available in the literature: Trost et al. [31]: red; Etzkorn et al. [29]: green; Milton et al.

[27]: blue; Suto et al. [30]: cyan. Values from the literature have been shifted in accordance to Table 3.

S. Fally et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 110 (2009) 766–782772

this paper, all the comparisons are therefore performed at equivalent resolution and with the best possible coincidentwavenumber scale.

4.1.2. Benzene

The absorption cross section of benzene at 293 K obtained in this work corresponds to the average of all the spectra ofthe six experiments described in Table 2, with pressure of benzene ranging from 0.987 to 3.079 Torr (1.3–4.1 hPa). Fig. 3shows the absorption cross section on the 37 000–42 000 cm�1 spectral range, as well as the non-systematic error on thecross sections.

The detailed comparison of the data obtained in this work with data of the literature is presented in Fig. 4, where theabsolute differences are shown in the lower panels.

The best agreement is obtained for the data of Etzkorn et al. [29], with a maximum absolute difference of0.5�10�18 cm2/molecules located on the peak of absorption. Data from Milton et al. [27] also compare favorably with ournew measurements. It should be noted that Milton et al. [27] reported measurements of BTX using a differential-absorptionLIDAR and showed the absorption cross sections of toluene and benzene as examples to better locate the wavelengths to bechosen for the LIDAR detection. No data are available and we obtained the absorption cross section values by digitalizingthe two graphics displayed in the paper. This should be kept in mind when looking at the comparison presented in Fig. 4.Suto et al. [30] report values for the absorption cross section of benzene which are systematically higher than our newvalues by a constant value of 0.2�1018 cm2/molecules. Moreover the discrepancy is larger at the peak maxima. Values ofTrost et al. (1997), shifted by 20.3 cm�1, are systematically larger.

ARTICLE IN PRESS

Table 3Summary of the comparison of the absorption cross sections obtained in this work with data from the literature.

Species Reference s(Lit) ¼ a+b�s(This work) Shift

a (�1018 cm2/molecules) b (cm�1) (nm)

Benzene Etzkorn et al. [29] 0.00 1.11 �1.5 0.01

Milton et al. [27] 0.04 1.10 9.7 0.07

Suto et al. [30] 0.17 1.09 �27.6 0.20

Trost et al. [31] 0.11 1.14 20.2 0.15

Toluene Etzkorn et al. [29] 0.02a 1.40a 24.6 0.18

Milton et al. [27] 0.02 1.47 10.1 0.07

Trost et al. [31] 0.11 1.41 42.7 0.31

m-Xylene Etzkorn et al. [29] 0.01 0.97 3.6 0.03

Suto et al. [30] 0.04 1.03 �15.7 0.12

Trost et al. [31] 0.31 1.26 42.2 0.31

o-Xylene Etzkorn et al. [29] 0.01 1.03 3.3 0.02

Suto et al. [30] 0.05 1.16 �48.8 0.35

Trost et al. [31] 0.14 1.18 36.1 0.25

p-Xylene Etzkorn et al. [29] 0.00 1.01 4.2 0.03

Suto et al. [30] 0.01 1.18 �6.0 0.04

Trost et al. [31] 0.17 1.14 40.9 0.30

a and b are the linear parameters expressing, respectively, the offset and the multiplicative factor existing between the values of this work and the

literature data. The last column indicates the wavenumber shift n(Literature)�n(This work). Values are given in cm�1 and nm (conversion performed at

37 000 cm�1).a Comparison limited to the range 36 000–39 600.0 cm�1.

Fig. 3. Benzene absorption cross section obtained at 293 K. The bottom panel illustrates the non-systematic error on the absorption cross section.

S. Fally et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 110 (2009) 766–782 773

ARTICLE IN PRESS

Fig. 4. Comparison of the absorption cross section of benzene at 293 K of this work (black) with data from the literature (red): (a) Etzkorn et al. [29]

shifted by �1.5 cm�1, (b) Milton et al. [27] shifted by 9.7 cm�1, (c) Suto et al. [30] shifted by �27.6 cm�1, and (d) Trost et al. [31] shifted by 20.2 cm�1. For

each comparison, the differences s(Literature)�s(This work) are indicated in the bottom panel, as well as their mean value (solid line) and standard

deviation (dashed line).

S. Fally et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 110 (2009) 766–782774

Results of the systematic comparison between the literature data and the results from the present study aresummarized in Table 3. Each dataset from the literature has been fitted by the linear expression

sLitðlþ shiftÞ ¼ aþ b� sThis workðlþ convolutionÞ (4)

where the parameters a and b represent, respectively the offset and multiplicative factor. These numbers confirm theconclusions drawn before, i.e., that in the case of the data from Etzkorn et al. [29] and Milton et al. [27] the difference withthe values of the present work can be essentially described by a multiplicative factor, whereas the data of Suto et al. [30]differ with ours essentially by an offset. Also, the relation with the data of Trost et al. [24] is well described by both amultiplicative factor and an offset.

4.1.3. Toluene

The absorption cross section of toluene has been determined from spectra recorded with pure samples at pressuresranging from 1.265 to 3.427 Torr (1.7–4.6 hPa) and is shown for the 36 000–42 000 cm�1 spectral range in Fig. 5.These data have been compared to data available in the literature in Fig. 6. The values of the three literaturedataset are larger than the values of this work, with values from Etzkorn et al. and of Trost et al. being 1.5 times higherthan ours for example. This is confirmed by the high and similar (�1.4) multiplicative factors reported in Table 3. Fig. 6 andTable 3 show that the differences between our values and those of Etzkorn et al. [25] and Milton et al. [22] can bothbe attributed to a multiplicative factor, whereas the relation with Trost et al. [24] needs both an offset and a multiplicativefactor to be described. In the case of the data of Etzkorn et al. the comparison has been limited to below 39 600.0 cm�1, asfor higher wavenumbers, Etzkorn’s data decrease very rapidly. The same behavior of the three literature datasetswith respect to our values was observed for benzene, although the multiplicative factors were closer to 1.0. The systematicbias observed between our absorption cross sections of toluene and three independent literature dataset is difficult toexplain. Even if photolysis and/or polymerization occurred in the cell during our experiment, the fact that the recordingswere performed in successive blocks of scans allowed us to correct for the evolution of the signal in time, as alreadyexplained.

ARTICLE IN PRESS

Fig. 5. Toluene absorption cross section at 293 K and at the resolution of 1 cm�1. The bottom panel illustrates the non-systematic error on the absorption

cross section.

S. Fally et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 110 (2009) 766–782 775

4.1.4. Xylenes

The absorption cross sections of m-, o-, and p-xylene determined at room temperature are shown in Fig. 7 and detailedcomparisons with data available in the literature [29–31] are detailed in Fig. 8–10. Similar amplitude differences as forbenzene and toluene are observed. Data of Etzkorn et al. [29] do compare very well with the new datasets for the isomersof dimethyl-benzene: within 3%, 3% and 1% for m-, o- and p-xylene, respectively. The agreement with the m-xylene datafrom Suto et al. [30] is within 3%, once a small offset has been removed, but reaches 16% and 18% for o- and p-xylene.Comparison with data from Trost et al. [31] show the presence of a large offset in each case combined to an amplitudefactor up to 26%, as also mentioned in Table 3.

4.2. Temperature effect

4.2.1. Benzene

Absorption cross sections of benzene at 253, 263, 273, 283, and 293 K have been determined from the spectra recordedduring this study (see Table 2). Fig. 11 shows the evolution of the absorption cross section for four temperatures (253, 263,283, and 293 K). The effect is small but still visible. It is not straightforward to explain the temperature effect on such broadstructures, which are composed of a lot of individual lines, each of them having its particular temperature dependency asdescribed by the Boltzmann population. However the net effect is that a decrease in temperature implies a decrease ofintensity in the broad structures and an increase of the peak intensities.

Fig. 12 shows trends in the absorption cross section for some selected wavenumbers with respect to the temperature.The selected frequencies correspond to peaks of absorption. The graphics reveals that for those frequencies the variation ofabsorption is linear with temperature over the temperature range investigated in this study and considering the error(systematic and non-systematic) on the cross section values.

The temperature coefficient deduced from a linear dependency assumption (see Eq. (4)) is presented in Fig. 13. It isslightly positive in the region corresponding to the large structures and negative at the localizations of the peaks; it is nullin between the absorption structures. By using this linear approximation, the absorption cross section at any temperature(within the temperature range investigated) may be determined. Such a temperature effect was also observed in the mid-IRregion by Rinsland et al. [33] and Klingbeil et al. [32], who moreover verified that the integrated band intensity wasconstant within the accuracy of their measurements for the different temperatures investigated. In the spectral regionobserved in the present work, it is not as easy as in the IR region to isolate completely some absorption band. We thereforearbitrarily selected four intervals: [37 800–38 650], [38 650–39 660], [40 100–40 600] cm�1, and a last interval covering theentire spectrum of benzene [37 800–40 600] cm�1 with the exception of the first band which is just outside the noise level.

ARTICLE IN PRESS

Fig. 6. Comparison of the absorption cross section of toluene at 293 K of this work (black) with data from the literature (red): (a) Etzkorn et al. [29] shifted

by 24.6 cm�1, (b) Milton et al. [27] shifted by 10.1 cm�1, and (c) Trost et al. [31] shifted by 42.7 cm�1. For each comparison, the differences

s(Literature)�s(This work) are indicated in the bottom panel, as well as their mean value (solid line) and standard deviation (dashed line).

S. Fally et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 110 (2009) 766–782776

ARTICLE IN PRESS

Fig. 7. Absorption cross sections at 293 K of (a) m-xylene, (b) o-xylene, and (c) p-xylene. For each of the species the non-systematic error is also given.

S. Fally et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 110 (2009) 766–782 777

In each of these regions we determined the integrated absorption cross section. Ratios of those values relative to the valueobtained at 293 K are listed in Table 4. These values seem to indicate that the integrated absorption cross sections areconstant at any of the investigated temperature and within the experimental accuracy.

4.2.2. Toluene

Absorption cross sections of toluene at 273, 283, and 293 K have been determined from the spectra recorded during thisstudy (see Table 2). Fig. 14 shows the evolution of the absorption cross section at the three temperatures. The intensity atthe peak maxima increases for decreasing temperatures. This is also illustrated in Fig. 15, where the evolution of theabsorption cross section for some selected frequencies has been plotted against the temperature. The hypothesis of a linearregression describing the temperature effect seems to be validated by this graphics. The effect of the temperature outside

ARTICLE IN PRESS

Fig. 8. Comparison of the absorption cross section of m-xylene at 293 K of this work (black) with data from the literature (red): (a) Etzkorn et al. [29]

shifted by 3.6 cm�1, (b) Suto et al. [30] shifted by �15.7 cm�1, and (c) Trost et al. [31] shifted by 42.2 cm�1. For each comparison, the differences

s(Literature)�s(This work) are indicated in the bottom panel, as well as their mean value (solid line) and standard deviation (dashed line).

Fig. 9. Comparison of the absorption cross section of o-xylene at 293 K of this work (black) with data from the literature (red): (a) Etzkorn et al. [29]

shifted by 3.3 cm�1, (b) Suto et al. [30] shifted by �48.8 cm�1, and (c) Trost et al. [31] shifted by 36.1 cm�1. For each comparison, the differences

s(Literature)�s(This work) are indicated in the bottom panel, as well as their mean value (solid line) and standard deviation (dashed line).

Fig. 10. Comparison of the absorption cross section of p-xylene at 293 K of this work (black) with data from the literature (red): (a) Etzkorn et al. [29]

shifted by 4.2 cm�1, (b) Suto et al. [30] shifted by �6.0 cm�1, and (c) Trost et al. [31] shifted by 40.9 cm�1. For each comparison, the differences

s(Literature)�s(This work) are indicated in the bottom panel, as well as their mean value (solid line) and standard deviation (dashed line).

S. Fally et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 110 (2009) 766–782778

ARTICLE IN PRESS

Fig. 12. Temperature effect on the absorption cross section of benzene. Evolution of the cross section for selected wavenumbers in function of the

temperature: (�) 37492.6 cm�1, (’) 38 458.0 cm�1, (m) 38 531.8 cm�1, (.) 38 620.8 cm�1, (~) 39 544.4 cm�1 and ( ) 40 467.0 cm�1 which correspond to

peak maxima.

Fig. 11. Temperature effect on the absorption cross section of benzene: at 253 K (cyan), 263 K (blue), 283 K (green), and 293 K (red).

S. Fally et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 110 (2009) 766–782 779

the peak locations is very small. This temperature coefficient is reproduced in the bottom panel of Fig. 13. It is negative atthe peaks; slightly positive or null in between the absorption structures.

4.2.3. Xylenes

Due to the low vapor pressures of the xylenes, it was not possible to decrease the temperature below 273 K.Measurements performed with m-xylene were moreover rejected because the partial pressure of the gas was found not to

ARTICLE IN PRESS

Fig. 13. Temperature coefficient for the absorption cross section of benzene (top panel) and toluene (bottom panel).

Table 4Integrated absorption cross section of benzene for selected wavenumber intervals in function of the temperature.

Wavenumber interval (cm�1) 293 K 283 K 273 K 263 K 253 K

37 800–38 650 1.0 1.00 1.07 0.89 1.01

38 650–39 660 1.0 0.99 1.13 0.90 1.01

40 100–40 600 1.0 0.99 1.23 0.96 1.05

37 800–40 600 1.0 0.98 1.16 0.91 1.02

The numbers are the ratio of the integrated values relative to the value at 293 K.

Fig. 14. Temperature effect on the absorption cross section of toluene: at 263 K (cyan), 273 K (blue), 283 K (green), and 293 K (red).

S. Fally et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 110 (2009) 766–782780

ARTICLE IN PRESS

Fig. 15. Temperature effect on the absorption cross section of toluene. Evolution of the cross section for selected wavenumbers in function of the

temperature: (�) 37483.4 cm�1, (n) 38 017.4 cm�1, (’) 38 419.8 cm�1, (~) 38 453.8 cm�1, (,) 38 672.2 cm�1, (J) 38 748.4 cm�1, and (m) 38 953.0 cm�1

which correspond to peak maxima.

S. Fally et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 110 (2009) 766–782 781

be sufficiently stable during the experiment, as it varied a lot from one block of scans to the other. The signal to noise ratiosassociated to the measurements performed at the lower temperature are very low, impending the accurate determinationof the temperature coefficients for those compounds. However, qualitative results can still be drawn. It is observed againthat cooling the gas strengthens the absorption cross section at the peaks and weakens it in the large structures away fromthe peaks.

5. Conclusions

Absorption cross sections of benzene, toluene, and m-, o-, and p-xylenes have been measured with a Fourier transformspectrometer at the resolution of 1 cm�1 in the 30 000–42 000 cm�1 spectral range. Spectra were recorded with samples ofpure gas at room and low temperature (253, 263, 273, 283, and 293 K).

The effect of low temperature on the absorption cross sections was investigated. The influence of temperature wasalways the largest at the maximum of the peaks. For benzene and toluene, a linear parameterization of the temperaturedependence is proposed.

Comparison with literature data have shown that the overall closest agreement with our values is obtained with thedata of Etzkorn et al. [24]. But, in general, it was found that literature data lacked the sufficient resolution to be compareddirectly, that their wavelength scales presented problems of shifts combined to stretches, and that the absolute values ofthe absorption cross sections present sometimes large discrepancy between the different datasets. In particular, thesystematic bias of 40% between our values and three literature datasets observed for toluene needs to be more investigatedby performing independent new measurements. The data from the literature were obtained at room temperaturepreventing the comparison of our new values at low temperature, and therefore forbidding the confirmation of thetemperature effect.

The absorption cross sections with their temperature dependency presented in this paper should be helpful forupcoming atmospheric studies, on Earth but also on other planets. All data are available in digital form on the website ofthe Belgian Institute for Space Aeronomy (http://www.aeronomie.be/spectrolab/).

Acknowledgments

This project was supported by the National Fund for Scientific Research (FNRS FRFC convention no. 2.4536.01), by theBelgian Federal Science Policy Office (SSD program) and by the Communaute Franc-aise de Belgique (Actions de RechercheConcertees).

ARTICLE IN PRESS

S. Fally et al. / Journal of Quantitative Spectroscopy & Radiative Transfer 110 (2009) 766–782782

References

[1] Finlayson-Pitts BJ, Pitts JN. Atmospheric chemistry: fundamentals and experimental techniques. New York: Wiley; 1986.[2] Clarkson TS, Martin RJ, Rudolph J, Graham BWL. Benzene and toluene in New Zealand air. Atmos Environ 1996;30:569–77.[3] Barrefors G. Monitoring of benzene, toluene and p-xylene in urban air with differential optical absorption spectroscopy technique. The Sci Total

Environ 1996;189/190:287–92.[4] Bravo H, Sosa R, Sanchez P, Bueno E, Gonzalez L. Concentrations of benzene and toluene in the atmosphere of the southwestern area at the Mexico

City Metropolitan Zone. Atmos Environ 2002;36:3843–9.[5] Hellen H, Kukkonen J, Kauhaniemi M, Hakola H, Laurila T, Pietarila H. Evaluation of atmospheric benzene concentrations in the Helsinki Metropolitan

Area in 2000–2003 using diffusive sampling and atmospheric dispersion modeling. Atmos Environ 2005;39:4003–14.[6] Kourtidis K, Ziomas I, Zerefos C, Gousopoulos A, Balis D, Tzoumaka P. Benzene and toluene levels measured with a commercial DOAS system in

Thessaloniki, Greece. Atmos Environ 2000;34:1471–80.[7] Kourtidis K, Ziomas I, Zerefos C, Kosmidis E, Symeonidis P, Christophilopoulos E, et al. Benzene, toluene, ozone, NO2, and SO2 measurements in an

urban street canyon in Thessaloniki, Greece. Atmos Environ 2002;36:5355–64.[8] Singh HB, Salas LJ, Cantrell BK, Redmond RM. Distribution of aromatic hydrocarbons in the ambient air. Atmos Environ 1985;19:1911–9.[9] Patterson B, Lenney J, Sibbett W, Hirst B, Hedges N, Padgett M. Detection of benzene and other gases with an open-path, static Fourier-transform UV

spectrometer. Appl Opt 1998;37:3172–5.[10] Brocco D, Fratarcangeli R, Lepore L, Petricca M, Ventrone I. Determination of aromatic hydrocarbons in urban air of Rome. Atmos Environ

1997;31:557–66.[11] Kim K. Comparison of BTX measurements using a commercial differential absorption spectroscopy and an on-line gas chromatography system.

Environ Eng Sci 2004;21:181–94.[12] Lee C, Choi Y, Jung J, Lee J, Kim K, Kim Y. Measurement of atmospheric monochromatic hydrocarbons using differential optical absorption

spectroscopy: comparison with on-line gas chromatography measurements in urban air. Atmos Environ 2005;39:2225–34.[13] Volkamer R, Etzkorn T, Geyer A, Platt U. Correction of the oxygen interference with UV spectroscopic (DOAS) measurements of monocyclic aromatic

hydrocarbons in the atmosphere. Atmos Environ 1998;32:3731–47.[14] Kim SJ, Caldwell J, Rivolo AR, Wagener R, Orton GS. Infrared polar brightening on Jupiter III. Spectrometry from the Voyager 1 IRIS experiment. Icarus

1985;64:233–48.[15] Bezard B, Drossart P, Encrenaz T, Feuchtgruber H. Benzene on the giant planets. Icarus 2001;154:492–500.[16] Coustenis A, Salama A, Schulz B, Ott B, Lellouch E, Encrenaz T, et al. Titan’s atmosphere from ISO mid-infrared spectroscopy. Icarus

2003;161:383–403.[17] Tran B, Joseph J, Force M, Briggs R, Vuitton V, Ferris J. Photochemical processes on Titan. Irradiation of mixtures of gases that simulate Titan’s

atmosphere. Icarus 2005;177:106–15.[18] Lebonnois S. Benzene and aerosol production in Titan and Jupiter’s atmospheres: a sensitivity study. Planet Space Sci 2005;53:486–97.[19] Moses J, Bezard B, Lellouch E, Gladstone G, Feuchtgruber H, Allen M. Photochemistry of Saturn’s atmosphere, I, hydrocarbon chemistry and

comparison with ISO observations. Icarus 2000;143:244–98.[20] Wilson E, Atreya S, Coustenis A. Mechanisms for the formation of benzene in the atmosphere of Titan. J Geophys Res 2003;108.[21] Wong A, Yung Y, Friedson A. Benzene and haze formation in the polar atmosphere of Jupiter. Geophys Res Lett 2003;30.[22] Bruston P, Khlifi M, Benilan Y, Raulin F. Laboratory studies of organic chemistry in planetary atmospheres: from simulation experiments to

spectroscopic determinations. J Geophys Res 1994;99:19047–62.[23] Shindo F, Benilan Y, Guillemin J-C, Chaquin P, Jolly A, Raulin F. Bull Am Astron Soc 2001;33.[24] Wu R, Chen F, Judge D. Cross section measurements of gaseous and liquid H2O, D2O, and C6H6. Bull Am Astron Soc 2000;32:1646.[25] Okruss M, Muller R, Hese A. High resolution UV laser spectroscopy of jet-cooled benzene molecules: complete rotational analysis of the S1o�S0 61

0

(l ¼71) band. J Mol Spectrosc 1999;193:293–305.[26] Hiraya A, Shobatake K. Direct absorption spectra of jet-cooled benzene in 130–260 nm. J Chem Phys 1991;94:7700–6.[27] Milton MJT, Woods PT, Jolliffe BW, Swann N, McIlveen T. Measurements of toluene and other aromatic hydrocarbons by differential-absorption LIDAR

in the near-ultraviolet. Appl Phys B 1992;55:41–5.[28] Pantos E, Phillis J, Bolovinos A. The extinction coefficient of benzene vapor in the region 4.6 to 36 eV. J Mol Spectrosc 1978;72:36–43.[29] Etzkorn T, Klotz B, Sorensen S, Patroescu IV, Barnes I, Becker KH, et al. Gas-phase absorption cross sections of 24 monocyclic aromatic hydrocarbons

in the UV and IR spectral ranges. Atmos Environ 1999;33:525–40.[30] Suto M, Wang X, Shan J, Lee LC. Quantitative photoabsorption and fluorescence spectroscopy of benzene, naphthalene, and some derivatives at

106–295 nm. J Quant Spectrosc Radiat Transfer 1992;48:79–89.[31] Trost B, Stutz J, Platt U. UV-absorption cross sections of a series of monocyclic aromatic compounds. Atmos Environ 1997;31:3999–4008.[32] Klingbeil A, Jeffries J, Hanson RK. Temperature-dependent mid-IR absorption spectra of gaseous hydrocarbons. J Quant Spectrosc Radiat Transfer

2007;107:407–20.[33] Rinsland CP, Malathy Devi V, Blake T, Sams RL, Sharpe SW, Chiou L. Quantitative measurement of integrated band intensities of benzene vapor in the

mid-infrared at 278, 298, and 323 K. J Quant Spectrosc Radiat Transfer 2008;105:2511–22.[34] Orphal J, Bogumil K, Dehn A, Deters B, Dreher S, Fleischmann OC, et al. Laboratory spectroscopy in support of UV–visible remote-sensing of the

atmosphere. In: Recent Research Development in Physical Chemistry, vol. 6. Trivandrum: Transworld Research, 2002. p. 15–34.[35] Vandaele AC, Carleer M. Development of Fourier transform spectrometry for UV–visible DOAS measurements of tropospheric minor constituents.

Appl Opt 1999;38:2630–9.[36] Hermans C, Vandaele AC, Carleer M, Fally S, Colin R, Jenouvrier A, et al. Absorption cross-sections of atmospheric constituents: NO2, O2, and H2O.

Environ Sci Pollut Res 1999;6:151–8.[37] Voigt S, Orphal J, Bogumil K, Burrows JP. The temperature dependence (203–293 K) of the absorption cross sections of O3 in the 230–850 nm region

measured by Fourier-transform spectroscopy. J Photochem Photobiol A 2001;143:1–9.[38] Fleischmann OC, Hartmann M, Burrows JP, Orphal J. New ultraviolet absorption cross-sections of BrO at atmospheric temperatures measured by

time-windowing Fourier transform spectroscopy. J Photochem Photobiol A 2004;168:117–32.[39] Kromminga H, Orphal J, Spietz P, Voigt S, Burrows JP. New measurements of OClO absorption cross-sections in the 325–435 nm region and their

temperature dependence between 213 and 293 K. J Photochem Photobiol A 2003;157:149–60.[40] Meller R, Moortgat GK. Temperature dependence of the absorption cross sections of formaldehyde between 223 and 323 K in the wavelength range

225–375 nm. J Geophys Res 2000;105:7089–101.[41] Vandaele AC, Hermans C, Fally S, Carleer M, Merienne M-F, Jenouvrier A, et al. Absorption cross-section of NO2: simulation of the temperature and

pressure effects. J Quant Spectrosc Radiat Transfer 2003;76:373–91.[42] Vandaele AC, Hermans C, Fally S, Carleer M, Colin R, Merienne M-F, et al. High-resolution Fourier transform measurement of the NO2 visible and

near-infrared absorption cross-sections: temperature and pressure effects. J Geophys Res 2002;107.[43] Orphal J, Fellows CE, Flaud P-M. The visible absorption spectrum of NO3 measured by high-resolution Fourier transform spectroscopy. J Geophys Res

2003;108:4077.[44] Poulter KF, Rodgers M-J, Nash PJ, Thompson TJ, Perkin MP. Thermal transpiration correction in capacitance manometers. Vacuum 1983;33:311–6.