Peculiarities in quantification of airborne particulate matter by means of Total Reflection X-ray Fluorescence
PPeculiarities in quantification of airborneparticulate matter by means of TotalReflection X-ray Fluorescence
Yves Kayser, ∗ , † János Osán, ‡ Philipp Hönicke, † and Burkhard Beckhoff † † Physikalisch-Technische Bundesanstalt, Abbestraße 2-12, 10587 Berlin, Germany ‡ Environmental Physics Department, Centre for Energy Research, Konkoly-Thege M. út29-33., 1121 Budapest, Hungary
E-mail: [email protected]
Abstract
Knowledge on the temporal and size distribution of particulate matter (PM) in airas well as on its elemental composition is a key information for source appointment,for the investigation of their influence on environmental processes and for providingvalid data for climate models. A prerequisite is that size fractionated sampling times offew hours must be achieved such that anthropogenic and natural emissions can be cor-rectly identified. While cascade impactors allow for time- and size-resolved collectionof airborne PM, total reflection X-ray fluorescence (TXRF) allows for element-sensitiveinvestigation of low sample amounts thanks to its detection sensitivity. However, duringquantification by means of TXRF it is crucial to be aware of the limits of TXRF in orderto identify situations where collection times or pollution levels were exceedingly longor high. It will be shown by means of grazing incidence X-ray fluorescence (GIXRF),where different reflection conditions are probed, that a self consistent quantification ofelemental mass depositions can be performed in order to validate or identify issues in a r X i v : . [ phy s i c s . a o - ph ] J a n uantification by means of TXRF. Furthermore, monitors of validity for a reliable quan-tification of the elemental composition of PM by means of TXRF will be introduced.The methodological approach presented can be transferred to tabletop instrumenta-tion in order to guarantee a reliable quantification on an element sensitive basis of thePM collected. This aspect is highly relevant for defining appropriate legislation andmeasures for health and climate protection and for supporting their enforcement andmonitoring. Introduction
Aerosols present in the environment affect our daily life at multiple levels. For example, air-borne particulate matter (PM) in air can impact health due to inhalation or can influenceatmospheric processes, more precisely the climate and environmental ecosystems throughinfluencing cloud formation or reflecting and scattering sunlight. The chemical composi-tion, which requires element-sensitive analytical methods, as well as the chemical speciationof the PM is of interest for a correct comprehension of the physical and chemical propertiesof the individual particles. With regard to health concerns, the smallest particles, so calledfine and ultrafine particles with sizes in the sub-micrometer and sub-100 nanometer range,are the most concerning for epidemiology as they can penetrate into the airways of the lungsand may be held responsible for health-averse effects on the respiratory and cardio-vascularsystem upon long-term exposure.
In particular anthropogenic emissions result in a no-ticeably higher generation of ultrafine particles. While toxicological studies have to assesspossible health risks of different nanomaterials, parallel efforts have to be undertaken toquantify the presence of the different elements in air and consider their dilution under dif-ferent weather conditions, at best in a time-resolved and size-fractionated fashion. Thisinformation is relevant for accurate modeling of climate changes, for regulatory bodies toimpose preventive measures and for legal entities to enforce regulations on air quality andto correctly pinpoint anthropogenic or natural sources. Adding to this the requirement to2ot only detect but to quantify reliably trace levels of aerosols contained in air in order toachieve good time resolution during environmental monitoring campaigns, highly sensitiveand accurate techniques need to be used.Among different available techniques X-ray fluorescence (XRF) based methods haveemerged as a promising contributor to the field by delivering ensemble information on thechemical composition. In contrast to other analytical techniques, XRF based investigationscan be performed without imposing sample consumption or altering the chemical composi-tion, which makes cross-validation measurements of other techniques feasible, and withoutrequiring high sample concentration.
In combination with impactors, where the par-ticulate matter is collected in a time- and size-resolved manner, all relevant informationfor quantifying the mass of different elements in a defined volume of air, even at in therange of few ng/m , are at hand and can be applied to a wide range of elements andparticle sizes. Synchrotron radiation-based approaches and total reflection XRF(TXRF) were used to achieve the best possible detection limits.It will be shown hereafter that for elemental mass depositions above trace level contam-ination, which can occur when sampling airborne PM without further knowledge on the airpollution levels, quantification by means of TXRF needs verification. Since TXRF is a single-point measurement technique, cross-validation and indicators are required to identify whenquantification could be compromised. Grazing incidence X-ray fluorescence (GIXRF) andX-ray reflectivity (XRR) will be used since these techniques are able to deliver informationbeyond the horizon of TXRF. Indeed, during a GIXRF measurement the incidence angle isvaried such that the excitation conditions are varied from excitation under total reflectionconditions to excitation under shallow incidence angle. Using this information, GIXRF al-lows for a robust assessment of the validity of quantification under TXRF conditions andfor self consistent quantification where quantification by means of TXRF is comprised by tohigh mass deposition of airborne PM. Furthermore, XRR will be introduced as a monitor ofthe validity of TXRF quantification results. 3 uantification by means of TXRF
TXRF employs a specific geometry in which the X-ray beam used for the excitation of theXRF signal impinges the sample at a very shallow angle beneath the critical angle for totalexternal reflection. Hence, TXRF demands for collimated and monochromatic excitationconditions but offers advantages such as the illumination of large sample areas and a largesolid angle of detection. Further benefits offered by TXRF are twofold. On the one side,the penetration of the incident beam into the substrate is reduced to an evanescent waveand any background signal, XRF or scattering, originating from it is suppressed. On theother side, the reflection of the incident X-ray beam at the substrate surface leads to thecreation of a X-ray standing wavefield (XSW) due to interference between incident andreflected X-rays. Consequently, enhanced excitation conditions for XRF originating fromthe particulate matter deposited on the top of the substrate can be achieved. A prerequisiteto profit from total external reflection is to use substrates which are flat on a macroscopicscale and characterized on a microscopic scale by a roughness smaller than the wavelengthof the incident X-rays.The knowledge of the XSW is of importance for quantitative measurements independentlyif external standardization, internal standardization or reference-free quantification schemesare applied. When using internal standardization the XSW created needs to be identicalthroughout the sample area illuminated, while in case of external standardization the sameXSW needs to be created in a reproducible manner for all samples investigated. Externalstandardization means that for each element of interest in an experimental campaign acalibration curve is established by means of a set of reference samples with different massdepositions of the selected elements. This approach requires adequate reference sampleswhich are sufficiently representative of the samples investigated.
For samples collectedduring outdoor campaigns the criteria include elemental composition (sample matrix), massdeposition (concentration), particle size range and morphology as well as deposition pattern.Hence, the production and selection of adequate calibration samples for outdoor sampling4ampaigns requires additional a priori information. The characterization of actual samplesfrom the measurement campaign via complementary techniques in order to use these samplesas a kind of standards is often impeded by the sensitivity, i.e., the amount of sample required,of these techniques.For this reason approaches based on internal standardization were developed as an alter-native that can be used with digested samples or in conjunction with substrates prepared forsampling.
Internal standardization means that on each sample to be analyzed a knownquantity of a reference element is added beforehand of the measurement while assuming thatthe excitation and detection conditions at the position where the standard is deposited isrepresentative for the whole sample. However, the standard which is added needs to fulfillthe requirements of non-toxicity, not being ubiquitous and having XRF energies which do notoverlap with the XRF lines to be contained within the sample. The goal of both approaches,external and internal standardization, is to extract combined information on instrumentalfactors in order to allow quantifying the elemental content of the material deposited on thetop of the substrates.In the reference-free XRF quantification scheme, information on the different exper-imental and fundamental parameters is used to calculate the mass deposition of differentelements from the measured count rate of the corresponding fluorescence line. This approachrequires the use of (radiometrically) calibrated instrumentation, e.g. apertures, diodes andsilicon drift detector (SDD) for an accurate knowledge of the solid angle of detection, theincident photon flux and the detected XRF intensity, and the knowledge of atomic funda-mental parameters (FPs), such as ionization cross-sections and fluorescence factors, whichare element-dependent and in a large part also energy-dependent, is also required.5 amples The size-fractionated sampling of PM was realized by means of a 9-stage extension of theMay-type cascade impactor. The aerodynamic cut-off diameters of the stages 1 to 9 arerespectively 17.9 µ m, 8.9 µ m, 4.5 µ m, 2.25 µ m, 1.13 µ m, 0.57 µ m, 0.29 µ m, 0.18 µ m and0.07 µ m. The cut-off diameter is defined as the dimension of the PM which is collectedwith 50 % efficiency, smaller particles escaping with a higher probability. A well-known andconstant airflow is required during the collection of airborne PM. The first two stages withthe coarsest particle sizes were disregarded and for the 7 further stages × mm Si wafersare used as substrates. As Si wafers have very low background contamination and very lowsurface roughness, they are ideally suited for TXRF and GIXRF experiments. Measurementswith good signal-to-background ratio can be expected even though other substrates mightbe more suitable for the collection of PM. Indeed, the collection efficiency depends not onlyon the design of the airflow, where losses of particles should be minimized, but also onthe substrate surface which can be pre-treated to minimize bounce-off effects for example.This requires, however, additional a posteriori treatment prior TXRF investigation and isdetrimental if other, complementary analytical techniques shall be used as well. Hence, inorder to preserve the capabilities offered by TXRF, no pre-treatment was used. Sample setsselected for the present study were collected from two campaigns at two cities, Budapest,Hungary, 24 – 31 May 2018; and Cassino, Central Italy, 20–27 September 2018, with samplingduration ranging from 20 min to 5 h. In total 19 Si substrates collected on the 3 stages withthe finest particle distributions were used in this survey. More information on 6 of these 19samples is provided in Table 1.The deposition area from the May-type cascade impactor corresponds to a stripe of 20mm length and, depending on the stage, of 0.1 to 1 mm width (fine to coarser PM). The widthis determined by the width of the slits used as nozzles for the different stages. This type ofdeposition pattern presents the advantage of being highly suitable for investigation by meansof TXRF and GIXRF once the stripe is aligned along the incidence direction. Thus, the6ay-type cascade impactor is ideal to demonstrate the capability offered by the combinationof cascade impactors and TXRF, respectively GIXRF analysis to provide element, size- andtime-resolved information on the PM collected.Table 1: Description of the 6 selected Si substrates with aerosol particles collected in Bu-dapest and at Cassino which are discussed in more details in Figure 2 and 3. Samples arelisted in the order of deposited particulate mass and cover the full range of elemental massdepositions quantified on the total of 19 samples investigated.Sample Duration Stage Diameter range / nmA 30 min 7 300 – 600B 20 min 9 70 – 180C 20 min 8 180 – 300D 1 h 7 300 – 600E 5 h 9 70 – 180F 5 h 7 300 – 600 Experimental
The reference-free GIXRF measurements for quantification of elemental mass deposi-tions present in the particulate matter collected were realized at the plane grating monochro-mator (PGM) beamline in the PTB laboratory at the BESSY II electron storage ring. Theexperiments were conducted at an incident photon energy of 1620 eV which is below the SiK-edge in order to suppress the contribution of the Si K XRF lines. An ultrahigh-vacuumchamber equipped with a 9-axis manipulator was used. The instrument allows for preciselytuning the incident angle θ between the incidence direction of the synchrotron radiation andthe sample surface (Fig. 1). The fluorescence radiation emitted from the sample was de-tected by means of a silicon drift detector (SDD) calibrated in terms of response function and detection efficiency, which is placed in the polarization plane and perpendicular tothe propagation direction of the linearly polarized incident X-ray beam in order to minimizescattered radiation. The SDD allows for an energy-dispersive detection of the XRF emit-ted from the sample such that the information from different elements can be discriminated7nd processed in parallel during quantification. The incident photon flux is determined byusing a calibrated photodiode. The spectra were deconvoluted using the known detectorresponse functions for the relevant fluorescence lines and background contributions, whichwas mainly resonant Raman scattering (RRS) from the Si K shell and to a lower extentBremsstrahlung from L shell electrons from the Si substrate. The resulting count rate I foreach fluorescence line of interest is normalized with respect to the sine of the incident angle θ , the incident photon flux I , the effective solid angle of detection Ω4 π and the energy depen-dent detection efficiency (cid:15) ( E ) of the SDD for the respective fluorescence photons in orderto derive the emitted fluorescence intensity. It has to be emphasized that the calculation ofthe incident angle dependent solid angle of detection requires an accurate knowledge of thedetection geometry but also of the incident beam profile.Figure 1: Illustration of the experimental setup with the Si wafer and the collected airbornePM on the top of it, a diode for measuring the reflectivity and a calibrated SDD for recordingthe XRF emitted for different incidence angles θ (top panel). Typical XRF spectra recordedfor an incidence angle beneath and above the critical angle of total external reflection illus-trate the lower background contributions from the Si wafer at the smaller incidence angle(bottom panels).From the absolute XRF intensity the elemental mass deposition m can be extracted foreach position of the GIXRF measurements where the incident angle θ was varied in variablesteps from 0 ◦ to 10 ◦ (Figure 2). Hence, the measurement conditions on each sample were8odified from excitation under grazing incidence conditions, where an XSW needs to beconsidered, to excitation under shallow incidence angles, where no XSW is present. Thecalculation of the XSW requires the knowledge of the optical properties of the substrate forthe incident photon energy used during the experiment. However, even if the incident photonenergy dependent optical properties are measured beforehand from a blank Si substrate tonot rely on tabulated data, the presence of the PM on the top of the Si wafer will impact thereflectivity to a certain extent as the contrast in electronic density at the interface separatingthe bulk Si from the vacuum or PM is changing. Therefore, the reflectivity R ( θ ) for eachwafer was measured by means of a photodiode positioned in a θ - θ configuration duringthe GIXRF measurement. This approach allows for a direct calculation of the incident angledependent XSW for each sample.Figure 2: GIXRF data for the 6 selected different samples (labelled A to F and described inTable 1). The changes in the angular intensity profiles for each element indicate differencesin the excitation of the XRF signal. The typical particle-like signature of the main elementsdetected in the GIXRF measurement gradually vanishes which is a clear indicator that theXSW on the top of the substrates significantly differs between the samples. It can be notedas well that for sample A, the angular evolution of O contains both particle- and layer-likesignatures. The latter contribution arises from the surface oxide of the Si wafers used, butthe relative contribution vanishes with increasing mass of collected airborne PM.9 eference-free GIXRF Quantification In a TXRF measurement the XRF intensity is usually recorded at a single incidence anglecorresponding to √ ( ≈ ) of the critical angle for total external reflection θ c , whichdepends on the incident photon energy and the substrate density and which was about 1 ◦ for the Si wafers used. The relative intensity distribution within the XSW is given by, XSW ( θ, z ) = 1 + R ( θ ) + 2 (cid:112) R ( θ ) cos (cid:18) arccos(2 θ θ c − − π sin θ z E hc (cid:19) (1)with E the energy of incident photons, z the height above the reflecting substrate and R ( θ ) the measured reflectivity. For experimental reasons the reflectivity could only be accuratelymeasured for incidence angles above 0.6 ◦ . In the following the mean intensity of the XSW,labelled XSW ( θ ) , over the direction vertical z (Figure 1) to the substrate surface is consid-ered. The assumptions of a laterally and vertically homogeneous chemical composition ofthe collected PM and of PM dimensions extending over several periods hc E sin θ of the XSWare made thereby. The averaging of the XSW by integration is further backed up by thefact that different particles sizes and compositions are intermixed on each stage and that thedeposition pattern is homogeneous along the direction of the incident radiation (Supp. Fig.1). A more intricate calculation would require knowledge on the particle size and relativeparticle size distribution, as well as on the surface coverage. Under TXRF conditions,the mass deposition m k of element k for each incidence angle θ can then be determined fromthe respective measured XRF count rate m k = − µ eff ( E , E k ) ln (cid:32) − I k ( θ ) sin θ µ eff ( E , E k ) Ω4 π I XSW ( θ ) ω k τ k ( E ) (cid:15) ( E k ) (cid:33) (2)where ω k corresponds to the fluorescence factor and τ k ( E ) to the photoionization cross-section of the element being quantified. The values of atomic fundamental parameters canbe found in literature databases or selected parameters are determined in dedicated ex-10eriments as for the fluorescence yield for C or O. The factor µ eff ( E , E k ) accounts forthe effective absorption cross-section of incident and emitted X-ray photons labelled µ in ( E ) and µ out ( E k ) respectively, within the PM investigated µ eff ( E , E k ) = (cid:88) j c j (cid:18) µ in,j ( E )sin θ + µ out,j ( E k )sin π − θ (cid:19) (3)and requires hence knowledge on the mass deposition of the different elements present inorder to take correctly into account the relative contributions via the factor c k = m k (cid:80) j m j with m k being the mean quantified mass deposition for element k at incidence angles above thecritical angle for total external reflection (more precisely from 6 ◦ to 10 ◦ ) where no XSW ispresent ( XSW ( θ ) = 1 for θ > θ c ).A consideration which is usually made at larger incidence angles during the quantificationis the correction for absorption of X-rays on the incidence and emission paths M k ( E , E k ) = (cid:80) j m j ( µ in,j ( E )sin θ + µ out,j ( E k )sin π − θ )1 − exp( − (cid:80) j m j ( µ in,j ( E )sin θ + µ out,j ( E k )sin π − θ )) (4)It was found that for incidence angles in the range from 6° to 10° this factor accounts for atmost a few percent only (< 5 % ) for most of the samples. Only for samples with very highmass depositions a relative correction of 25 % to 30 % was introduced in this iterative correc-tion scheme. For a most accurate correction factor and quantification a complete knowledgeof the matrix composition is required. For the investigated samples, minor contributions ofFe and Cu were detected as well. However, Fe showed a rather layer-like angular intensityprofile, such that a contamination of the substrate from a different source than airborne PMmust be assumed and Cu was too low in mass deposition to allow for a reliable quantification.Furthermore, secondary fluorescence due to photoelectrons or fluorescence is neglected. Thisintroduces only a minor error for low mass depositions but, depending on the matrix compo-sition, should not be disregarded for high mass depositions where errors of up to 20 % -30 % Finally, the GIXRF measurement allows to compare the quantification results of the el-emental mass deposition m k between TXRF conditions and XRF conditions under shallowincidence angles. The uncertainty made in the quantification depends on the uncertaintieson the incident flux ( ), the XSW factor ( ) the atomic fundamental parameters (fluo-rescence yield, for light elements, and photoionization cross-section, . ), the detectorefficiency and spectral deconvolution ( . ), the counting statistics and the solid angle ofdetection (about for the smallest incidence angles to about for the largest incidenceangles used for quantification). Note, that the mass deposition in terms of mass (or likewise number of atoms) for eachelement per unit area is quantified. A conversion to mass, which is a more commonly usedmetric in the aerosol community, can be straightforwardly realized if the area on which theairborne PM is collected and its lateral distribution are known.
Results & Discussion
Given the uniform distribution of the PM, quantification results by means of Eq. 2 canbe expect to be constant for a GIXRF measurement on a given sample. Indeed, the inci-dence angle should only affect the lowest limit of detection achievable but not impact thequantification result.However, the quantification result for each sample and each element under TXRF condi-tions ( θ ≈ . ◦ ) and under XRF conditions ( θ >
61 + x. ◦ ), only part of the samples presenta good agreement (Fig. 3). The horizontal bar indicates the average mass deposition quan-tified at the largest incidence angles used and the vertical bar the incidence angle at whicha TXRF quantification is typically performed for Si wafers and the incident photon energyused. For the lowest mass depositions used, the quantification results are independent ofthe incidence angle and agree reasonably well with each other (Fig. 3, upper panels). A12pecificity can be observed for samples A and B where an imperfect deconvolution of theXRF spectra recorded at larger incidence angles affects the quantification results because ofthe underling Si-RRS which is not perfectly described by the model used. In particular forAl, whose main characteristic line is close to the high energy cut-off of the Si RRS at 1520eV, this results as well in a larger scattering of the quantification results at larger incidenceangles. This illustrates perfectly the main benefit of TXRF for low mass depositions sinceit allows suppressing background contributions from the substrate.For higher mass depositions, a discrepancy between the quantified mass depositions ap-pears (Fig. 3, lower panels) in the sense that the under TXRF conditions the mass deposi-tions are underestimated. This deviation occurs despite the fact that the XSW is calculatedon the basis of reflectivity measured in parallel to the XRF intensity. In Fig. 4 it is shownthat the reflectivity at a typical angle used for TXRF measurements drops significantly withincreasing mass deposition of airborne PM. This observation means that the XSW is sig-nificantly different for each sample and different compared to the case of a blank substrate(Supp. Fig. 2). By not taking into account the how increasing mass deposition affectsthe contrast in optical density at the interface defined by the surface of the substrate thediscrepancy between quantification results for the different incidence angles used would evenbe more important. This insight emphasizes the benefit of monitoring in parallel to a TXRFor GIXRF measurement the reflectivity from the sample. The dependence of the XSW onthe surface coverage must to be taken into account when quantifying the mass deposition, astatement which is not only valid when using the reference-free quantification approach butalso when applying an internal or external standard.The relationship between the mass deposition and the XSW is noteworthy (Eq. 2) withregard to the need for using representative specimen, besides applying comparable exper-imental conditions, when applying external standards for quantification purposes. In caseof internal standards, reliable results are only obtained if the collected mass deposition ofthe airborne PM is within the range of mass depositions covered by the standard under13igure 3: Quantified mass deposition for the different incidence angles covered when varyingthe excitation conditions during the GIXRF measurement from the TXRF regime to theXRF regime under shallow incidence angles. The vertical bar indicates the position typicallyselected for a TXRF measurement for a Si substrate and the incident photon energy used,while the horizontal bar indicate the mass deposition quantified at the largest incidenceangles for each element. For low mass depositions (upper 3 panels) a satisfyingly goodagreement can be observed, but for increasing mass deposition a growing discrepancy appearsfor all the elements. This indicates that not all physical effects due to attenuation of X-rays in the collected airborne PM are accounted for in the quantification scheme. Undershallow incidence angles attenuation is less important and has therefore a lesser impact onthe quantification scheme as can be seen from the results approaching a constant value.14he premise that a homogeneous intermixing is realized. In other words, the dynamic rangewithin which the calibration is valid needs to be considered. Finally, in the reference-basedapproaches the XSW needs to be comparable between the calibration material and inves-tigated sample material, be it locally when using an internal standard or between sampleswhen using an external standard, in order to avoid a calibration bias. An upper limit forreliable TXRF quantification is discussed in literature in terms of critical thickness andsaturation effect. Figure 4: The reflectivity from the Si substrate for the same samples than displayed inFigures 2 and 4 (left panel) indicates the growing impact on the attenuation of X-rays withinthe collected PM, resulting in significant differences in the
XSW ( θ ) between the differentsamples. The vertical bar in the left panel indicates the position typically selected for aTXRF measurement for a Si substrate and the incident photon energy used. This positionwas used for the calculation of XSW ( θ ) (Supp. Fig. 2).Besides the differences in the XSW a further reason for the deviation in the quantificationof the mass deposition at different incidence angles is that the full volume of the PM collectedis not illuminated homogeneously in its depth direction because of the attenuation of theincident and reflected X-ray radiation. For increasing incidence angles and high surfacecoverage the effective path length is reduced as θ such that the X-ray attenuation withinthe PM volume becomes less pronounced. This argument becomes even more evident whenconsidering that under conditions where an XSW is expected and for high surface density ofPM, the photons need to travel twice through the airborne PM collected.This insight impacts directly the reliability of TXRF quantification results in the sense15igure 5: Comparison of the quantified mass deposition for TXRF and XRF (under shallowincidence angles). In the case of low mass depositions, the expected 1:1 ratio indicated bythe dashed line is matched but for increasing mass deposition a deviation is observed whichindicates that under TXRF conditions the quantified mass deposition underestimates theactually deposited mass deposition. This does not become apparent from the TXRF mea-surement itself. Therefore more exhaustive data as collected during a GIXRF measurementis required.that TXRF underestimates the mass of PM collected for the samples with the highest load-ing which, in general, is due to exceedingly long collection times or high pollution levels. Thestrategy to collect by means of GIXRF more exhaustive data to quantify the mass depositionof the collected PM under different incidence angles and excitation conditions allows assess-ing the reliability and validity of the quantification performed. A self-consistent verificationof the quantification can be performed by means of GIXRF by either having an agreementthroughout the full angular range , i.e. with and without XSW, respectively virtually con-stant quantification results in case no XSW needs to be considered ( θ > θ c ). In the firstcase low mass depositions are investigated and the better sensitivity in terms of detectionlimit offered by TXRF can be profited from to perform quantification. In the second case,where deviations between GIXRF and TXRF quantification, are observed higher mass de-positions are investigated. GIXRF allows nevertheless for robust and reliable quantificationsince in this angular regime the quantification is less prone to attenuation of the incidentX-ray photons. Data at larger incidence angles where no XSW needs to be considered allowsassessing the impact of X-ray attenuation within the airborne PM collected on the results. Inview of the demands on quantitative techniques for regulatory purposes this critical assess-16ent of the validity of the results is mandatory. On a further positive note, the requirementfor sufficiently low attenuation of X-rays, i.e., adequately short collection times of PM, forreliable TXRF quantification makes it mandatory to use effectively the sensitivity in termsof detection limits offered by TXRF and is beneficial for the time resolution which can beachieved during field campaigns.For the extreme case of the two samples with the highest PM loads (panels E and F inFigs. 2 and 3) the variation of the quantified mass deposition with the incident angle indicatesthat an accurate quantification is tedious since here the attenuation of the incident radiationwithin the PM collected would need to be considered. This aspect introduces considerableuncertainties in the final result. Hence, a GIXRF measurement allows to discard these typesof samples from further use in measurement campaigns.When comparing for all 19 samples investigated the quantified mass deposition underTXRF conditions and when using the largest incidence angles such that no XSW is created,the deviation from the expected ratio with increasing elemental mass deposition becomeseven more evident (Fig. 5). A better representation would be using the total mass, butdue to the soft X-ray radiation used it can not be ascertained that the XRF radiation forall elements present in the matrix was excited. In general, this deviation indicates thatfor reliable quantification under TXRF conditions the attenuation of X-rays within the col-lected airborne PM and the differences in the XSW created should remain below a thresholdvalue. This criteria cannot be specified generically since it depends on the incident photonenergy and the absolute elemental composition of the PM. Hence for field campaigns wherehigh-throughput TXRF quantification of PM collected on suitable substrates by means oflaboratory instrumentation is conducted at least a subset of the samples should be investi-gated in more details. It has to be noted, that the GIXRF measurement can be restrictedto few points since only data for an incidence angle typically chosen for a TXRF measure-ment and for a series of larger incidence angles ( θ > θ c ) is needed for cross-checking thequantification results. 17n addition, simple control monitors can be used to identify possible issues with thequantification performed, regardless if standards are used or a reference-free quantificationscheme is being applied. One indicator is the reflectivity from the substrate which can evenbe used when samples are investigated under TXRF conditions solely. An other indicatorduring a GIXRF measurement is that the X-ray radiation from the bulk volume of thesubstrate can be used to validate the quantification results (Supp. Fig. 3). For increasingmass deposition of airborne PM and larger incidence angles significant attenuation comparedto a blank Si substrate can be observed. For both control monitors, a direct comparison toblank substrates (or calibration samples) allows elucidating if the amount of airborne PMcollected might be too important for quantification by means of TXRF alone. Conclusion & Outlook
It was shown that GIXRF allows for applying robust quantification scheme and hence forassessing the validity of quantification under TXRF conditions. While increasing mass depo-sition results in different XSW created on the top of the substrates and pronounced attenu-ation of the incident and reflected X-rays within the airborne PM collected, the comparisonof quantification results between TXRF and GIXRF for a subset of samples covering thefull range of mass depositions of the airborne PM collected during a measurement campaignallows assessing the limits for reliable quantification results and identifying the range of sam-ples where quantification should be realized by means of GIXRF. This information on thevalidity of the quantification results depends on the matrix composition and the incidentphoton energy, but cannot be assessed from TXRF measurements alone. It has to be notedthat the presented experiment with its emphasis on light elements was realized in the softX-ray regime but the conclusions made can also be applied for higher X-ray energies.While advanced instrumentation was applied for allowing for a physically traceable quan-tification which does not rely on the use of standards, it can be emphasized that this approach18s transferable to laboratory instrumentation. Indeed, manufacturers of TXRF instrumen-tation start offering advanced instrumentation where the incidence angle of the monochroma-tized exciting radiation can be tuned. Thus, possibilities are offered to transfer the approachpresented to instrumentation used for high-throughput measurements in the field or in thelaboratory. Even without tuning the incident angle the implementation of a diode to mea-sure the reflectivity would allow a rough but straightforward indication whether the quantityof airborne PM collected presents an issue for the quantification by means of TXRF. Fur-thermore, differences in the XSW between samples with different quantities of airborne PMcould be accounted for.
Acknowledgement
Parts of this research was performed within the EMPIR project 19ENV08 AEROMET II.This project has received funding from the EMPIR programme co-financed by the Partic-ipating States and from the European Union’s Horizon 2020 research and innovation pro-gramme. The support of the European Structural and Investment Funds jointly financedby the European Commission and the Hungarian Government through grant no. VEKOP-2.3.2-16-2016-00011 (on behalf of J. Osán) is also appreciated.19 eferences (1) Apte, J. S.; Marshall, J. D.; Cohen, A. J.; Brauer, M. Addressing Global Mortalityfrom Ambient PM2.5.
Environ. Sci. Technol. , , 8057–8066.(2) Lelieveld, J.; Evans, J. S.; Fnais, M.; Giannadaki, D.; Pozzer, A. The contribution ofoutdoor air pollution sources to premature mortality on a global scale. Nature , , 367–371.(3) Leni, Z.; Cassagnes, L. E.; Daellenbach, K. R.; El Haddad, I.; Vlachou, A.; Uzu, G.;Prévôt, A. S. H.; Jaffrezo, J.-L.; Baumlin, N.; Salathe, M.; Baltensperger, U.; Dom-men, J.; Geiser, M. Oxidative stress-induced inflammation in susceptible airways byanthropogenic aerosol. PLOS ONE , , 1–17.(4) Mellouki, A.; Wallington, T. J.; Chen, J. Atmospheric Chemistry of OxygenatedVolatile Organic Compounds: Impacts on Air Quality and Climate. Chem. Rev. , , 3984–4014.(5) McNeill, V. F. Aqueous Organic Chemistry in the Atmosphere: Sources and ChemicalProcessing of Organic Aerosols. Environ. Sci. Technol. , , 1237–1244.(6) Zhang, R.; Wang, G.; Guo, S.; Zamora, M. L.; Ying, Q.; Lin, Y.; Wang, W.; Hu, M.;Wang, Y. Formation of Urban Fine Particulate Matter. Chem. Rev. , , 3803 –3855.(7) Prather, K. A.; Hatch, C. D.; Grassian, V. H. Analysis of Atmospheric Aerosols. Annu.Rev. Anal. Chem. , , 485–514.(8) Møller, P.; Folkmann, J. K.; Forchhammer, L.; Bräuner, E. V.; Danielsen, P. H.; Ri-som, L.; Loft, S. Air pollution, oxidative damage to DNA, and carcinogenesis. CancerLett. , , 84 – 97. 209) Stone, V.; Johnston, H.; Clift, M. J. D. Air Pollution, Ultrafine and Nanoparticle Tox-icology: Cellular and Molecular Interactions. IEEE Transactions on NanoBioscience , , 331–340.(10) Valavanidis, A.; Fiotakis, K.; Vlachogianni, T. Airborne Particulate Matter and HumanHealth: Toxicological Assessment and Importance of Size and Composition of Particlesfor Oxidative Damage and Carcinogenic Mechanisms. J. Environ. Sci. Health Pt. C , , 339–362.(11) Brook, R. D.; Rajagopalan, S.; Pope, C. A.; Brook, J. R.; Bhatnagar, A.; Diez-Roux, A. V.; Holguin, F.; Hong, Y.; Luepker, R. V.; Mittleman, M. A.; Peters, A.;Siscovick, D.; Smith, S. C.; Whitsel, L.; Kaufman, J. D. Particulate Matter Air Pollu-tion and Cardiovascular Disease. Circulation , , 2331–2378.(12) Rönkkö, T.; Timonen, H. Overview of Sources and Characteristics of Nanoparticles inUrban Traffic-Influenced Areas. J Alzheimers Dis. , , 15–28.(13) Daellenbach, K. R. et al. Sources of particulate-matter air pollution and its oxidativepotential in Europe. Nature , , 414–419.(14) Bulska, E.; Ruszczyńska, A. Analytical Techniques for Trace Element Determination. Phys. Sci. Rev. , , 20178002.(15) Ault, A. P.; Axson, J. L. Atmospheric Aerosol Chemistry: Spectroscopic and Micro-scopic Advances. Anal. Chem. , , 430–452, PMID: 28105816.(16) Furger, M.; Minguillón, M. C.; Yadav, V.; Slowik, J. G.; Hüglin, C.; Fröhlich, R.;Petterson, K.; Baltensperger, U.; Prévôt, A. S. H. Elemental composition of ambientaerosols measured with high temporal resolution using an online XRF spectrometer. Atmos. Meas. Tech. , , 2061–2076.2117) Bukowiecki, N.; Lienemann, P.; Zwicky, C. N.; Furger, M.; Richard, A.; Falkenberg, G.;Rickers, K.; Grolimund, D.; Borca, C.; Hill, M.; Gehrig, R.; Baltensperger, U. X-rayfluorescence spectrometry for high throughput analysis of atmospheric aerosol samples:The benefits of synchrotron X-rays. Spectrochim. Acta B , , 929 – 938.(18) Richard, A.; Bukowiecki, N.; Lienemann, P.; Furger, M.; Fierz, M.; Minguillón, M. C.;Weideli, B.; Figi, R.; Flechsig, U.; Appel, K.; Prévôt, A. S. H.; Baltensperger, U.Quantitative sampling and analysis of trace elements in atmospheric aerosols: impactorcharacterization and Synchrotron-XRF mass calibration. Atmos. Meas. Tech. , ,1473–1485.(19) Klockenkämper, R.; Bayer, H.; von Bohlen, A.; Schmeling, M.; Klockow, D. Collectionof Airborne Particulate Matter for a Subsequent Analysis by Total Reflection X-RayFluorescence. Anal. Sci. , , 495–498.(20) Bontempi, E.; Zacco, A.; Benedetti, D.; Borgese, L.; Colombi, P.; Stosnach, H.;Finzi, G.; Apostoli, P.; Buttini, P.; Depero, L. Total reflection X-ray fluorescence(TXRF) for direct analysis of aerosol particle samples. Environ. Technol. , ,467–477.(21) Prost, J.; Wobrauschek, P.; Streli, C. Quantitative total reflection X-ray fluorescenceanalysis of directly collected aerosol samples. X-Ray Spectrom. , , 454–460.(22) Osán, J.; Börcsök, E.; Czömpöly, O.; Dian, C.; Groma, V.; Stabile, L.; Török, S. Exper-imental evaluation of the in-the-field capabilities of total-reflection X-ray fluorescenceanalysis to trace fine and ultrafine aerosol particles in populated areas. Spectrochim.Acta B , , 105852.(23) Fomba, K. W.; Deabji, N.; Barcha, S. E. I.; Ouchen, I.; Elbaramoussi, E. M.;El Moursli, R. C.; Harnafi, M.; El Hajjaji, S.; Mellouki, A.; Herrmann, H. Application22f TXRF in monitoring trace metals in particulate matter and cloud water. Atmos.Meas. Tech. , , 4773–4790.(24) Bukowiecki, N.; Hill, M.; Gehrig, R.; Zwicky, C. N.; Lienemann, P.; Hegedüs, F.; Falken-berg, G.; Weingartner, E.; Baltensperger, U. Trace Metals in Ambient Air: HourlySize-Segregated Mass Concentrations Determined by Synchrotron-XRF. Environ. Sci.Technol. , , 5754–5762.(25) Beckhoff, B.; Fliegauf, R.; Kolbe, M.; Müller, M.; Weser, J.; Ulm, G. Reference-Free To-tal Reflection X-ray Fluorescence Analysis of Semiconductor Surfaces with SynchrotronRadiation. Anal. Chem. , , 7873–7882.(26) Streli, C.; Wobrauschek, P.; Meirer, F.; Pepponi, G. Synchrotron radiation inducedTXRF. J. Anal. At. Spectrom. , , 792–798.(27) Borgese, L.; Bilo, F.; Zacco, A.; Federici, S.; Mutahi, A. W.; Bontempi, E.; Trzepla, K.;Hyslop, N.; Yatkin, S.; Wobrauschek, P.; Prost, J.; Ingerle, D.; Depero, L. E. Theassessment of a method for measurements and lead quantification in air particulatematter using total reflection X-ray fluorescence spectrometers. Spectrochim. Acta B , , 105840.(28) Klockenkämper, R.; von Bohlen, A. Total Reflection X-Ray Fluorescence Analysis andRelated Methods ; John Wiley & Sons, Ltd, 2014.(29) Hönicke, P.; Krämer, M.; Lühl, L.; Andrianov, K.; Beckhoff, B.; Dietsch, R.; Holz, T.;Kanngießer, B.; Weißbach, D.; Wilhein, T. Development and characterization of sub-monolayer coatings as novel calibration samples for X-ray spectroscopy.
Spectrochim.Acta B , , 36 – 42.(30) Horntrich, C.; Kregsamer, P.; Prost, J.; Stadlbauer, F.; Wobrauschek, P.; Streli, C.Production of the ideal sample shape for Total Reflection X-ray Fluorescence analysis. Spectrochim. Acta B , , 31 – 34.2331) Böttger, S.; Tyssebotn, I. M. B.; Jansen, W.; Fittschen, U. E. Evaluating internalstandards for the determination of gas phase mercury using silver nanoparticle assistedtotal reflection X-ray fluorescence. Spectrochim. Acta B , , 93 – 99.(32) May, K. An “ultimate” cascade impactor for aerosol assessment. J. Aerosol Sci. , , 413 – 419.(33) Marple, V. A. History of Impactors—The First 110 Years. Aerosol Sci. Technol. , , 247–292.(34) Müller, M.; Hönicke, P.; Detlefs, B.; Fleischmann, C. Characterization of High-kNanolayers by Grazing Incidence X-ray Spectrometry. Materials , , 3147–3159.(35) Hönicke, P.; Detlefs, B.; Nolot, E.; Kayser, Y.; Mühle, U.; Pollakowski, B.; Beckhoff, B.Reference-free grazing incidence x-ray fluorescence and reflectometry as a methodologyfor independent validation of x-ray reflectometry on ultrathin layer stacks and a depth-dependent characterization. Journal of Vacuum Science & Technology A , ,041502.(36) Senf, F.; Flechsig, U.; Eggenstein, F.; Gudat, W.; Klein, R.; Rabus, H.; Ulm, G. Aplane-grating monochromator beamline for the PTB undulators at BESSY II. Journalof Synchrotron Radiation , , 780–782.(37) Lubeck, J.; Beckhoff, B.; Fliegauf, R.; Holfelder, I.; Hönicke, P.; Müller, M.; Pol-lakowski, B.; Reinhardt, F.; Weser, J. A novel instrument for quantitative nanoanalyticsinvolving complementary X-ray methodologies. Rev. Sci. Instrum. , , 045106.(38) Scholze, F.; Procop, M. Modelling the response function of energy dispersive X-rayspectrometers with silicon detectors. X-Ray Spectrom. , , 312–321.(39) Scholze, F.; Procop, M. Detection efficiency of energy-dispersive detectors with low-energy windows. X-Ray Spectrom. , , 473–476.2440) Müller, M.; Beckhoff, B.; Ulm, G.; Kanngießer, B. Absolute determination of crosssections for resonant Raman scattering on silicon. Phys. Rev. A , , 012702.(41) Unterumsberger, R.; Hönicke, P.; Kayser, Y.; Pollakowski-Herrmann, B.; Gholhaki, S.;Guo, Q.; Palmer, R. E.; Beckhoff, B. Interaction of nanoparticle properties and X-rayanalytical techniques. J. Anal. At. Spectrom. , , 1022–1033.(42) Kayser, Y.; Sá, J.; Szlachetko, J. Nanoparticle characterization by means of scanningfree grazing emission X-ray fluorescence. Nanoscale , , 9320–9330.(43) Schoonjans, T.; Brunetti, A.; Golosio, B.; Sanchez del Rio, M.; Solé, V. A.; Ferrero, C.;Vincze, L. The xraylib library for X-ray–matter interactions. Recent developments. Spectrochim. Acta B , , 776 – 784.(44) Beckhoff, B.; Ulm, G.; Bundesanstalt, P.-T. Determination of fluorescence yields usingmonochromized undulator radiation of high spectral purity and well known flux. Adv.X-Ray Anal. , , 349–354.(45) Hönicke, P.; Kolbe, M.; Krumrey, M.; Unterumsberger, R.; Beckhoff, B. Experimentaldetermination of the oxygen K-shell fluorescence yield using thin SiO2 and Al2O3 foils. Spectrochim. Acta B , , 94 – 98.(46) Wählisch, A.; Streeck, C.; Hönicke, P.; Beckhoff, B. Validation of secondary fluorescenceexcitation in quantitative X-ray fluorescence analysis of thin alloy films. J. Anal. At.Spectrom. , , 1664–1670.(47) Klockenkämper, R.; von Bohlen, A. Determination of the critical thickness and thesensitivity for thin-film analysis by total reflection X-ray fluorescence spectrometry. Spectrochim. Acta B , , 461 – 469.(48) Hellin, D.; Fyen, W.; Rip, J.; Delande, T.; Mertens, P. W.; De Gendt, S.; Vinckier, C.25aturation effects in TXRF on micro-droplet residue samples. J. Anal. At. Spectrom. , , 1517–1523.(49) Hönicke, P.; Waldschläger, U.; Wiesner, T.; Krämer, M.; Beckhoff, B. Towards a cali-bration of laboratory setups for grazing incidence and total-reflection X-ray fluorescenceanalysis. Spectrochim. Acta B , , 106009.26 upplementary information - Peculiarities in quantification of airborne par-ticulate matter by means of Total Reflection X-ray Fluorescence Supp. Fig. 1: Lateral scan along the line of the deposition pattern exhibiting that in thecentral area a quite uniform distribution has been achieved. Due to the shallow incidenceangle of 5 ◦ , the position close to the boarder of the sample could not be investigated but nodecreased collection efficiency is expected from design consideration of the May-type cascadeimpactor. The sample is the one for which the GIXRF data is displayed in Fig. 1.27upp. Fig. 2: An increasing mass deposition affects the contrast in optical density at theinterface defined by the surface of the substrate: while for a low mass deposition the surfacecoverage is low enough to assume that the airborne PM does not alter the optical propertiesin the area above the substrate surface, this assumption does not hold for high mass depo-sition and the optical properties of the airborne PM for the incident X-ray radiation needsto be considered in addition to optical properties of the substrate. based on the measuredreflectivity at an incidence angle of 0.7 ◦ (as shown in Fig. 4, vertical bar), significant dif-ferences in the XSW ( θ ))