Surface Chemical Characterizationof 2.5-µm Particulates (PM2.5) fromAir Pollution in Salt Lake City UsingTOF-SIMS, XPS, and FTIRY I N G - J I E Z H U , † N E A L O L S O N , ‡ A N DT H O M A S P . B E E B E , J R . * , †

Department of Chemistry and Surface Analysis Facility,University of Utah, Salt Lake City, Utah 84112, andDepartment of Environmental Quality, Utah Air MonitoringCenter, West Valley, Utah 84119

Particulate matter with a diameter of 2.5 µm collected inSalt Lake City (SLC PM2.5) was studied using TOF-SIMS (time-of-flight secondary-ion mass spectrometry), XPS (X-rayphotoelectron spectroscopy), and FTIR (Fourier transforminfrared spectroscopy). The high spatial resolution and highsurface sensitivity of TOF-SIMS allow the surfaces ofindividual particulates to be analyzed. The high mass-resolution of TOF-SIMS provides good separation of signalsfrom different chemical species at the same nominalmass, and the extremely high detection sensitivity of TOF-SIMS makes the detection of trace elements possible.Metallic elements such as Li, Na, Mg, Al, K, Ca, Cr, Mn,Fe, Cu, Zn, Cs, and Bi were detected by TOF-SIMS on thesurface of SLC PM2.5. The uranium ion U+ together withits oxide ions UO+ and UO2

+ were also found. Inorganiccompounds detected include oxides, hydroxides, nitrates,sulfates, silicates, borates, chlorides, etc. Organiccompounds detected include hydrocarbons, alcohols,aldehydes, ethers, carboxylic acids, amines, amides, nitriles,etc. A number of polycyclic aromatic hydrocarbons(PAH) and nitrated polycyclic aromatic hydrocarbonswere detected by TOF-SIMS. High-resolution XPS C1sspectrum shows functional groups such as C-O, CO2,C-CO2, C-C, and C-H and aromatic π-π* shake-up transitions. High-resolution XPS O 1s spectrum indicatesthe coexistence of different oxygen compounds on thesurface of PM2.5. FTIR results confirm the presence of variousorganic compounds in SLC PM2.5 detected by TOF-SIMSand XPS.

IntroductionA major form of modern-day pollution is microscopicairborne solid material called particulates or particulatematter (also called aerosols). Particulates have seriousenvironmental and health-related consequences becausethey contain a wide variety of toxic metals (1, 2) and toxicorganic compounds such as polycyclic aromatic hydrocar-bons (PAH) (3, 4). Many PAH species are mutagenic and/orcarcinogenic (3, 5) and are thought to damage DNA. Thesespecies undergo a wide variety of transformations via

photolysis and reactions with atmospheric pollutants (i.e.O3, NO, NO2) which can modify and possibly increase thetoxicity of these compounds (6-8). Nitrated polycyclicaromatic hydrocarbons (nitro-PAH) have also caused muchenvironmental concern because of their potential mutagenicactivity (9) and carcinogenicity (9, 10). They are found in theparticulates emitted from diesel and gasoline engines (11,12) and are most likely to be formed during the process offuel combustion as well as by the reactions of parenthydrocarbons with nitrogen oxides in ambient air (13-18).Another important source of these mutagens is car emissions.

Fine particulates have a diameter of less than 2.5 µm, andthe atmospheric lifetime of such particulates is on the orderof days or weeks. The adsorption of polycyclic aromatichydrocarbons onto surfaces of airborne fine particulates withgreater relative surface area is a fundamental pathway forthe distribution of these toxic compounds in the environment.The U.S. Environmental Protection Agency (EPA) set aparticulate matter standard of diameter < 10 µm (PM10) (19)in 1982. Recent research (20, 21) suggests that there is nosafe level of PM10 in the atmosphere to which we may beexposed. The increasing interest in 2.5-micron diameterparticulates (PM2.5) results from the hypothesis that fineparticles are of greater concern to human health (22). Forthese reasons, the EPA has extended its regulations forparticulate matter emissions from PM10 down to a morestringent PM2.5 (23, 24). This drives the need for understandingof the chemistry associated with smaller particulates.

The early applications of mass spectrometry to environ-mental research can be retrieved back to the 1960s and 1970s.Mass spectrometry has played an important role in aidingour understanding of environmental chemistry and pollutionprocesses since then. A recent comprehensive review of massspectrometry in environmental sciences was given byRichardson (25). Among different kinds of mass spectrometry,TOF-SIMS is a surface analysis technique that has verypromising applications in environmental sciences as well asmany other research fields (26). The high spatial resolution(∼100 nm) and high surface sensitivity (topmost one or twomonolayers) of TOF-SIMS allow the surfaces of individualparticulates to be analyzed. The high mass-resolution (m/∆m ≈ 10 000) of TOF-SIMS provides good separation ofsignals from different chemical species at the same nominalmass. The extremely high detection sensitivity (ppm or ppb)of TOF-SIMS makes the detection of trace elements possible.A recent successful application of the TOF-SIMS techniquefor analysis of individual aerosol particles was demonstratedby Bentz et al. (27). FTIR has also been used to characterizeaerosol particles (28-31). To our knowledge, analysis ofairborne PM2.5 particulates by XPS has not been reported.

Like other Western urban centers located in mountainvalleys, the Salt Lake Valley is frequently plagued by airpollution. Air monitoring networks typically set a thresholdfor total mass per air volume sampled in a 24-h period. If thethreshold is exceeded, then a battery of tests is triggered todetermine metals, elemental, and organic carbon, andpossibly other determinations. In all of these cases anaggregate analysis is performed after pooling, through somesort of extraction procedure, all particulates into one sample.Information about surfaces of individual particles that mayhave originated from different sources or processes is lost asthe entire time-integrated sample is pooled into onehomogeneous (or homogeneously mixed) sample. In thispaper, we report on the studies of PM2.5 collected in SaltLake City, UT by the combined surface analysis techniquesof TOF-SIMS, XPS, and FTIR. The comprehensive study of

* Corresponding author fax: (801)581-8433; e-mail: beebe@chem.utah.edu.

† University of Utah.‡ Utah Air Monitoring Center.

Environ. Sci. Technol. 2001, 35, 3113-3121

10.1021/es0019530 CCC: $20.00 2001 American Chemical Society VOL. 35, NO. 15, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 3113Published on Web 06/20/2001

PM2.5 by the combined surface analysis techniques of TOF-SIMS, XPS, and FTIR has not yet been reported in the literatureto the authors’ knowledge. We believe that combined surfacetechniques can offer more advantages and can provideabundant surface chemical information of PM2.5 that comple-ments other established analysis methods.

Experimental SectionAll PM2.5 specimens were sampled in Salt Lake City by theState of Utah, Department of Environmental Quality, Divisionof Air Quality Monitoring Center according to the federalreference method (FRM) for PM2.5. The Partisol Model 2025Sequential Air Sampler (Rupprecht & Patashnick Co. NY)was used for PM2.5 sampling. Polytetrafluoethylene (PTFE,Teflon) with an integral support ring (about 4 cm in diameter)was used as the filter. The filter thickness was 30-50 µm, andthe pore size was about 2 µm. PM2.5 collection was achievedusing a Well Impactor Ninty-Six (WINS) impactor. Thesampling time was 6 h from 6:00 am to 12:00 pm on October5, 1999 with an air flow rate of 16.7 L min-1. Figure 1 showsthe PM2.5 Tapered Element Oscillating Microbalance (TEOM)data on October 5, 1999 in Salt Lake City. The total massdeposited onto the PTFE substrate was 17.7 µg m-3, deter-mined gravimetrically. The PTFE filters supporting the PM2.5

samples were cut into a number of smaller pieces (about 1× 1 cm2) for analyses by TOF-SIMS, XPS, and FTIR,respectively. For comparison, the gold-coated PTFE filter wasalso used for PM2.5 sampling. The coating of gold on thePTFE filter (10-nm thick) was achieved prior to PM2.5 samplingby evaporation of gold metal using a commercial vacuumevaporator (BOC Edwards Auto 306). Reasons for the goldcoating of the PTFE filter are related to TOF-SIMS analysisperformance and will be discussed below.

TOF-SIMS measurements were performed on a time-of-flight secondary ion mass spectrometer (TOF-SIMS IV, ION-TOF, Μunster, Germany) using a 25-kV monoisotopic 69Gaprimary ion beam. The operating pressure in the mainchamber was typically 10-10 mbar. A low-energy electronbeam that was pulsed between periods of secondary-ioncollection compensated the sample charging. Secondary-ion mass spectra were collected in both positive and negativemodes using a primary Ga+ ion dose which did not exceedthe dose threshold of 1013 ions cm-2 for static SIMS (32, 33).“Bunched mode” was used to achieve highest mass resolution(m/∆m ≈ 10 000) in the mass spectra. The typical targetcurrent of the primary Ga+ ion beam in the bunched modewas 3 pA. Each TOF-SIMS spectrum was acquired from anarea of 500 × 500 µm2 (on a new spot for each spectrum) andacquisition time was about 3 min. For acquisition of ionimages, “burst mode” of the Ga+ ion gun was used to achieve

both high spatial resolution (∼0.3 µm) and suitably high massresolution (m/∆m ≈ 4000).

XPS data were obtained using a VG ESCALAB 220i-XLelectron spectrometer (VG Scientific Ltd., U.K.). Monochro-matic Al KR X-rays (1486.7 eV) were employed. Typicaloperating conditions for the X-ray source were 400-µmnominal X-ray spot size (fwhm) operating at 15 kV, 8.9 mA,and 120 W for both survey and high-resolution spectra. A2-micron-thick aluminum window was used to isolate theX-ray chamber from the sample analysis chamber, to preventhigh-energy electrons from impinging on the sample. Surveyspectra, from 0 to 1200-eV binding energy, were collected at100-eV pass energy with an energy resolution of ∼1.0 eV, adwell time of 100 ms, and a total of two scans averaged.High-resolution spectra were collected at a pass energy of20 eV, energy resolution of ∼0.1 eV, dwell time of 100 ms,and a total of 10 scans averaged in the respective bindingenergy range. The data acquisition and data processing weredone using the Eclipse data system software. The operatingpressure of the spectrometer was typically 10-9 mbar. Theelectron flood gun was used for sample charging compensa-tion.

Infrared spectra were acquired with a Bruker Model 66FTIR spectrometer (Germany) configured with a mercury-cadmium-tellurium detector and Glowbar (MIR) source. Thespectra were recorded in a vacuum with an average of 100scans and a resolution of 4 cm-1. The spectra were back-ground-corrected by acquiring a spectrum of the blank PTFEsubstrate. This background spectrum was subtracted fromthe spectrum containing particulates to remove PTFE bandsfrom the particulate spectrum.

Results and DiscussionTOF-SIMS Analysis. Due to TOF-SIMS’s high mass resolution,high detection sensitivity and high surface sensitivity, surfacechemical information can be obtained from particulates.Blank PTFE filters were used for the control experiments toexamine the contamination on blank PTFE filters. (CxFy)(

fragment ions were dominant in both positive- and negative-ion TOF-SIMS spectra. In the positive-ion mass spectra, themost intense peak was at m/z 31 (CF+), followed by otherintense (CxFy)+ ions. (CxHy)+ ions were also observed at muchlower intensity, indicating some contamination of organiccompounds such as hydrocarbons on blank PTFE filters. Noevidence of siloxane contamination was found. Na+ wasdetected on the blank PTFE filters. In the negative-ion massspectra, the most intense peak was at m/z ) 19 (F-), followedby F2

- (m/z ) 38), CF- (m/z ) 31), and C- (m/z ) 12). (CxHy)-

as well as O- (m/z ) 16) and OH- (m/z ) 17) ions fromcontaminants were also observed at low intensity on theblank PTFE filter.

Figure 2 shows the positive-ion images for some metallicions detected by TOF-SIMS in SLC PM2.5. These images showthe lateral distribution of specific chemical species on thesurface of the particulates, within a randomly chosen 49 ×49 µm area on the sample. One can see that the particulatesare 2-3 µm in diameter. Some particulates may appear largerdue to the aggregation of several smaller particulates. Thesechemically and spatially resolved ion images also providechemical composition information about each individualparticulate. For example, particle 1 (see arrow) is about 2.5µm in diameter and contains Na, Mg, Al, Si, K, and Ca. Particle2 with a diameter of about 2 µm also contains Na, Mg, Al,Si, K, and Ca. The particle indicated by arrow 3 appears tobe a small iron particle with relatively low intensity for allother displayed masses. Other differences between theparticulates can be ascertained by careful inspection. Frompostacquisition analysis of the TOF-SIMS raw data (contain-ing all the chemical, spatial, and time information in eachpixel) we were able to obtain the mass spectrum for each

FIGURE 1. TEOM data of PM2.5 in Salt Lake City on October 5, 1999,and the date and time of the collection of PM2.5 that was analyzed.

3114 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 15, 2001

individual particulate. Examples of TOF-SIMS spectra ofindividual particulates are shown in Figure 3. One can seethat the chemical composition of each individual particle isconsistent with that from the ion image. In addition to themetallic elements in these particles, there are also CxHy

+ ionspresent in the mass spectra (Figure 3), indicating the presenceof organic compounds on the surface of these particles. FromFigure 3 one can also see that the intensity of secondary ionsis different for different particles, indicating differing amountsof these chemical species on the surfaces of these particles,or differing influence of the SIMS matrix effect, or both. Suchchemically and spatially resolved information for individualparticles may be related to the source and formationmechanism of particulates. This will be one of our futurestudies. We are also exploring more sophisticated multivariatedata processing and image analysis tools for images such asFigure 2, and these will be discussed in a forthcomingpublication.

Figure 4 shows the positive- and negative-ion TOF-SIMSspectra of SLC PM2.5 on a PTFE filter, collected in a large-area mode (500 × 500 µm2), and therefore integrating overmany particulates. Peaks from particulates are present inthe spectrum in addition to the peaks from PTFE. Acomparison of the normalized intensity of CxHy ions (both

positive and negative) from the PTFE blank filter and fromthe PM2.5 sample is shown in Table 1. One can see that bothpositive and negative CxHy ions from the PM2.5 sample aremuch more intense than those from the blank filter. Thisindicates that both positive and negative CxHy ion signalsfrom the particulate sample come overwhelmingly from theparticulates rather than from the PTFE filter and that thecontamination on the blank filter is negligible.

The metallic ions detected by TOF-SIMS from SLC PM2.5

include Li+, Na+, Mg+, Al+, K+, Ca+, Cr+, Mn+, Fe+, Cu+, Zn+,Cs+, Bi+, and U+. The most intense ion is K+, followed byNa+, Ca+, Mg+, and Fe+. The intensity of transition metalions such as Cr+, Mn+, Cu+, and Zn+ is relatively low.Nonmetallic ions such as B+, O+, Si+, S+, and Cl+ are alsoobserved in the positive-ion mass spectra with a low intensity.The NH4

+ ion is detected with a high intensity.The uranium ion U+ (m/z ) 238) together with its oxide

ions, UO+ (m/z ) 254) and UO2+ (m/z ) 270), were found

in SLC PM2.5. Some uranium-containing positive ions de-tected by TOF-SIMS are shown in Table 2. These ions likelyoriginated from the mining and refining processes. Figure 5shows the high-resolution mass spectra of U+, UO2

+, UV+,

FIGURE 2. TOF-SIMS positive-ion images of some metallic ions from PM2.5 on a PTFE filter. Field of view: 49 × 49 µm2, five pulses perpixel, and 120 scans.

FIGURE 3. The positive ion TOF-SIMS spectra of three individualparticles (particles 1, 2, and 3) as labeled in Figure 2.

FIGURE 4. High mass-resolution TOF-SIMS spectra of SLC PM2.5 ona PTFE filter. They were acquired from an area of 500 × 500 µm2.A pulsed electron flood gun was used for sample chargingcompensation: (a) positive-ion mass spectrum and (b) negative-ion mass spectrum.

VOL. 35, NO. 15, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 3115

and UVO+ examined on an expanded scale. In addition tothese species, other hazardous species found by TOF-SIMSare thallium-containing ions such as Tl2O+ (m/z 426), Tl2O2

+

(m/z 442), Tl2O3+ (m/z 458), Tl3O+ (m/z 629), etc., indicating

the oxide nature of Tl on the surface of SLC PM2.5.

Molecular ions are formed by ionization of moleculeswithout losing any atoms (no fragmentation). These mo-lecular ions have great importance in providing detailedchemical and structural information. Therefore, we put ouremphasis on the identification of the molecular ions ofaromatic organic compounds in TOF-SIMS spectra. A numberof positive aromatic ions are found in SLC PM2.5. The massspectra of some selected positive aromatic ions on anexpanded scale are shown in Figure 6. Table 3 summarizesa number of observed peaks in the high-resolution massspectra and their assigned chemical compositions. All positiveions listed in Table 3 have an integer saturation index,implying that they are possible molecular ions.

As shown in Table 3, C5H6+ and C9H8

+ ions were detectedby TOF-SIMS in the mass spectra. These two ions have an

integer saturation index, implying that they are possiblemolecular ions. If they indeed are molecular ions, then theymust come from the adsorption from the vapor phase.However, we cannot rule out the possibility that they arefragment ions formed by the fragmentation of other parentmolecules during ion bombardment. In this case, C5H6

+ andC9H8

+ ions do not represent the presence of cyclopentadieneand indene, respectively.

1-Nitropyrene and 3-nitrobenzanthrone may be presentin SLC PM2.5 samples, as shown in Table 3. However, as themass of the secondary ion increases, the number of possiblecompounds for the peak assignment also increases, and sothe uncertainty in the peak assignments likewise increases.3-Nitrobenzanthrone has been found to be a powerfulbacterial mutagen and is a suspected human carcinogen (34).Enya et al. (34) suggested that 3-nitrobenzanthrone is mostlikely to be formed not only during the combustion processof fuels but also from the atmospheric reaction betweenbenzanthrone and lower oxides of nitrogen. Benzanthronewas found to be nitrated quite easily under an artificialatmosphere containing gaseous NO2 and O3 to produce thepowerfully mutagenic 3-nitro derivative as the major product,along with several other isomeric mononitrobenzanthronesand dinitrobenzanthrones as minor products (34).

The intense fluorine-containing ions from the PTFE filterand possible reactions of fluorine in the filter with compo-nents of particulates during ion bombardment makesfluorinated materials among the most difficult to identify.We observed F+ in the positive-ion mass spectrum. F+ ionmay originate from the PTFE filter, and its presence does notnecessarily indicate that there are F-containing compoundson the surface of PM2.5. The HF+ (m/z 20), H2F+ (m/z 21), andCH2F+ (m/z 33) ion were also observed. These ions mayoriginate from reactions of the PTFE with hydrocarbonsduring ion bombardment. However, we cannot rule out thepossibility of the presence of fluorine-containing compounds.The Cl+ (m/z 35) and CH2Cl+ (m/z 49) ions are evidence ofthe presence of inorganic chlorides and chlorinated hydro-carbons. CH2Cl+ forms due to the R-cleavage of chloro-compounds, forming CH2dCl+ and an alkyl radical. Therewere other positive ions of chloro-compounds observed,including C4H8OCl+ (m/z 107), C6H7O2Cl+ (m/z 146), andC5H5O3Cl+ (m/z 148). The presence of bromine-containingcompounds is supported by Br- in the negative-ion massspectrum (see below). No evidence was found for thepresence of iodine-containing compounds.

CH3OH+ ion (m/z 32, saturation index ) 0) is a possiblemolecular ion of methanol, and a possible fragment ion fromother organic compounds. C2H5O+ ion (m/z 45, saturationindex ) 0.5) is a fragment ion rather than a molecular ion.Positive ions with the formula RCHdOH+, for example,HCHdOH+ (m/z 31), CH3CHdOH+ (m/z 45), and CH3CH2-CHdOH+ (m/z 59), may originate from ethers as well as fromethanols.

The expected fragment ions from esters RCOOR′ wouldbe R+, RCO+, RCOO+, R′+, R′O+, and R′OCO+. RCO+ ionsobserved in the mass spectrum are HCO+ (RdH, m/z 29),

TABLE 1. Comparison (I ) Is/Ib) of Secondary Ion Intensity between the PM2.5 Sample and the Blank PTFE Filtera

positive ion C+ CH+ CH2+ CH3

+ C2+ C2H+ C2H2

+ C2H3+ C2H4

+ C2H5+ Na+

ratio Is/Ib 1.9 14.6 109.1 67.3 2.4 43.4 67.3 72.5 54.4 56.9 323.5

negative ion C- CH- CH2- O- OH- C2

- C2H-

ratio Is/Ib 12.3 1011.3 2013.1 1144.2 4339.6 17.9 94.8a Is is the normalized intensity of secondary ions (both positive and negative) from the PM2.5 sample. Ib is the normalized intensity of secondary

ions from the blank PTFE filter. All positive-ion peaks were normalized to the CF+ intensity, and all negative-ion peaks were normalized to the F-

intensity for comparison.

TABLE 2. Some Uranium-Containing Species Detected in SLCPM2.5 by TOF-SIMS

radioactivespecies mass (exptl) mass (theor) ∆(exptl - theor)

238U+ 238.0416 amu 238.0508 amu -0.0092 amu238U16O+ 254.0430 254.0457 -0.0027238U16O2

+ 270.0190 270.0406 -0.0216238U51V+ 288.9817 288.9947 -0.0130238U51V16O+ 304.9991 304.9897 +0.0094238U51V16O2

+ 320.9927 320.9846 +0.0081238U51V16O3

+ 337.0029 336.9795 +0.0234238U51V16O4

+ 352.9723 352.9744 -0.0021238U51V2

16O+ 355.9946 355.9336 +0.0609238U51V2

16O2+ 371.9768 371.9286 +0.0482

FIGURE 5. TOF-SIMS spectra in an expanded scale of uranium-containing positive ions U+, UO2

+, UV+, and UVO+.

3116 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 15, 2001

CH3CO+ (m/z 43), C2H5CO+ (m/z 55), C3H7CO+ (m/z 71), etc.McLafferty ion CH2dCHOH+ was also observed, indicatingthe presence of aldehydes. C3H6O+ ion at m/z 58 formed bythe McLafferty rearrangement is indicative of the presenceof methyl ketones. The peak at m/z 149 is assigned to C8H5O3

+,an ion that probably originates from dialkyl phthalates whichare commonly used as plasticizers.

For nitrogen-containing organic compounds, severalpositive ions such as CH2N+ (m/z 28), CH3N+ (m/z 29), andCH4N+ (m/z 30) are characteristic of amines. Other detectedpositive ions containing nitrogen are as follows: C2H4N+ (m/z42), C2H6N+ (m/z 44), C2H7N+ (m/z 45), C2H8N+ (m/z 46),C3H6N+ (m/z 56), C3H8N+ (m/z 58), C3H9N+ (m/z 59), C3H10N+

(m/z 60), C4H8N+ (m/z 70), C4H10N+ (m/z 72), and C3H9N2+

(m/z 73), etc. Positive fragment ions from aromatic aminessuch as C8H8N+ (m/z 126) and C9H8N+ (m/z 130) are alsodetected in the SLC PM2.5. Thus, both aliphatic and aromaticamines are present on the surface of SLC PM2.5.

The ion at m/z 44 which is assigned to OdCdNH2+ is

indicative of the presence of amides. Additional evidence forthe presence of amides is indicated by the positive ions NH3

+

(m/z 17), NH2+ (m/z 16), and NH+ (m/z 15) and acylium ions

RCO+ such as CH3CO+ (m/z 43), C2H5CO+ (m/z 57), and C3H7-CO+ (m/z 71). The McLafferty ions from amides such as CH2dC(OH)NH2

+ at m/z 59 are also present in the mass spectra.The fragment ion +CH2CH2CONH2 at m/z 72 originates fromprimary amides by cleavage between the γ- and â-carbonsin the chain. The fragment ion CH4N+ at m/z 30 is due to lossof ketene (CH2dCdO) from this ion. The positive ions suchas NO+ (m/z 30), C13H6NO2

+ (m/z 208), C13H9NO2+ (m/z 211),

and C20H11NO2+ (m/z 297) may originate from nitro-

compounds (RNO2).

The presence of sulfur-containing compounds is indicatedby the observation of positive sulfur-containing ions. Theseions include S+ (m/z 32), HS+ (m/z 33), H2S+ (m/z 34), H3S+

(m/z 35), CS+ (m/z 44), CHS+ (m/z 45), SO+ (m/z 48), SOH+

(m/z 49), C2H2S+ (m/z 58), C2H3S+ (m/z 59), C3H5SN+ (m/z87), C5H10SN+ (m/z 116), C4H6SO4

+ (m/z 150), C7H6SO3+ (m/z

170), C14H8S+ (m/z 208), and C15H12S+ (m/z 224), etc. C4H6-SO4

+ has a saturation index of 2, implying it is a possiblemolecular ion. The possible structure of the molecule is asfollows: HOOC-CH2-S-CH2-COOH (thiodiglycolic acid)

FIGURE 6. TOF-SIMS spectra in an expanded scale of selected positive aromatic ions detected on the surface of SLC PM2.5.

TABLE 3. Some Aromatic Positive Ions detected in SLC PM2.5 by TOF-SIMSa

mass measd mass (theor) positive ion common name of possible compds

66.0480 66.0469 C5H6 cyclopentadiene78.0475 78.0470 C6H6 benzene92.0621 92.0625 C7H8 toluene

104.0638 104.0626 C8H8 styrene116.0624 116.0626 C9H8 indene128.0623 128.0626 C10H8 naphthalene, azulene178.0651 178.0783 C14H10 anthracene, phenanthrene, tolan192.0937 192.0939 C15H12 methylphenanthrene194.0763 194.0732 C14H10O anthranol, anthrone, diphenylketene196.0601 196.0524 C13H8O2 xanthone202.0599 202.0630 C12H10O3 2-naphthoxyacetic acid, spizofurone206.1047 206.1095 C16H14 distyryl208.0918 208.0889 C15H12O chalcone234.1105 234.1045 C17H14O dibenzalacetone247.0663 247.0633 C16H9NO2 (also possibly 1-nitropyrene

C5OF9, C10O2F5)257.0911 257.0347 C14H11NO4 benzoylpas, diphenylamine-2,2′-dicarboxylic acid275.0609 275.0582 C17H9NO3 (also possibly 3-nitrobenzanthrone

C6O2F9)276.0922 276.0938 C22H12 indenopyrene

a All ions are possible molecular ions.

VOL. 35, NO. 15, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 3117

or HOOC-CH(SH)-CH2-COOH (thiomalic acid). Thioetherswith a formula R-S-R′ will produce RS+ and R′S+ ions. TheRS+ ions are more stable than the corresponding RO+ ions.The observation of CS+ ion is an indication of the presenceof arylthioethers. Arylthioethers will tend to lose the alkylgroup, producing ArS+ which will lose CS to form thecyclopentadienyl ion.

Figure 4b shows the large-area (500 × 500 µm2) negativeTOF-SIMS spectrum of SLC PM2.5 on the PTFE filter. Themost intense ion is F- (m/z 19) that originates from the PTFEfilter. Other intense peaks are O-, CH-, and OH-. The highintensity of O- and OH- may be due to oxides and hydroxideswith relatively high concentrations on the surface of SLCPM2.5. Adsorbed H2O is an additional possible source for O-

and OH-. The negative-ion mass spectra of some selectednegative ions are shown on an expanded scale in Figure 7.The S- and O2

- ions can be distinguished at the same nominalmass 32. The same situation occurs between inorganic ionsand organic ions at the same nominal mass. For example,37Cl- is well separated from C3H- at the nominal mass 37.The detection of Cl-, Br-, and S- negative ions support theresults from positive ion mass spectra that compoundscontaining these elements are present. Other inorganicconstituents of PM2.5 are borates, nitrates, sulfides, sulfates,silicon hydrides, silicides, etc. The possible assignments fornegative ions observed in the mass spectra of SLC PM2.5 aregiven in Table 4 and are not discussed further.

PM2.5 was collected according to the official FRM on aPTFE filter which is nonconductive. Therefore, samplecharging could occur when acquiring TOF-SIMS data. Pulsedlow-energy electron flooding onto the sample during mea-surements compensated this charging. To lower even furtherthe extent of sample charging, we coated a thin layer of gold(about 10 nm thick) onto the surface of the PTFE filter priorto use in the FRM PM2.5 collection protocol. Because thethickness of the gold film was small compared with the poresize of the PTFE filter, gold coating did not block the poresof the PTFE filter. TOF-SIMS spectra collected from PM2.5 onthe gold-coated PTFE filters show a high similarity to thoseon uncoated PTFE filters. However, the relative intensity ofsecondary ions changes on these different filters, as does themass resolution. The intensity of some secondary ions fromPM2.5 on the gold-coated filters increased compared withthose on noncoated filters. This result is due to the combinedeffect of improved sample charging conditions and the SIMSmatrix effect.

An example of enhanced intensity and mass resolutionof some positive ions from PM2.5 on the gold-coated PTFEfilter is demonstrated in Figure 8. It shows the mass spectraat the nominal mass m/z 28 (Figure 8a,b) and 29 (Figure8c,d) of PM2.5 on Au-coated (Figure 8b,d) and uncoated(Figure 8a,c) PTFE filters. At nominal mass m/z 28, Si+, CO+,and C2H4

+ are detected from PM2.5 on the uncoated PTFEfilter. However, in addition to these species, two more species,AlH+ and CH2N+, are well resolved and detected from PM2.5

on the gold-coated PTFE filter. Similarly, at nominal massm/z 29, SiH+, CHO+, and C2H5

+ ions are present in the massspectrum of PM2.5 on an uncoated sample (Figure 8c).However, the additional species 29Si+, CH3N+, and B2H7

+ aredetected from PM2.5 on the gold-coated PTFE filter (Figure8d). Note the enhanced mass resolution that results fromuse of a gold-coated PTFE filter. For example, the mass

TABLE 4. Possible Compounds Leading to Negative Secondary Ions Detected in SLC PM2.5 by TOF-SIMS

negative ion possible assignments

CxHy- organic compds, for example, hydrocarbons

O-, O2- oxides, water

NH-, NH2-, NH3

- amines, amidesOH- hydroxides, alcohols, carboxylic acids, waterCN- nitrilesBO-, BO2

- boratesSi-, SiH- silicon hydrides, silicidesCO- ketones, aldehydes, carboxylic acidsCHO- aldehydesNO-, NO2

-, NO3- nitrates, nitrites, nitro-compds, amine oxides

S-, HS- sulfidesCl- chlorinated hydrocarbons, chloridesC2O- ethersCO2

-, CO2H- carboxylic acidsCONH2

- amidesSiO-, SiO2

-, SiO3- silicates, silicon oxides

PO-, PO2-, PO3

- phosphatesSO-, SO2

-, SO3-, SO4

-, HSO3-, HSO4

- sulfates, sulfites, and sulfuric acidClO- hypochloritesBr- brominated hydrocarbons, bromidesVO4

- vanadates (orthovanadates)

FIGURE 7. TOF-SIMS spectra in an expanded scale of selectednegative ions detected on the surface of SLC PM2.5.

3118 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 15, 2001

resolution for Si+ (m/z 28) is m/∆m ) 5411 and 3250 for thegold-coated filter and the uncoated substrate, respectively.The intensity of the negative ions from PM2.5 on the gold-coated filter also increased compared with those on thenoncoated filter. The magnitude of the increase in intensityis more than that in the case of positive ions. Therefore,additional negative species such as NO3

-, SiO3-, and PO3

-

with low intensity are detected from PM2.5 on the gold-coatedfilter. Although these results in Figure 8 demonstrate anexpected enhanced TOF-SIMS performance on the conduc-tive filter, it is important to point out that the conclusionsare essentially the same for PM2.5 on the unmodified PTFEfilters.

XPS Measurements. XPS measurements were carried outto study the organic functional groups on the same PM2.5

samples as studied by TOF-SIMS. Spectra were collected fromSLC PM2.5 samples on the uncoated PTFE filter from an areaof approximately 500 µm in diameter. Figure 9 shows thesurvey spectrum and high-resolution C 1s and O 1s XPSspectra. From the survey spectrum (Figure 9a), one can seethe presence of C, O, and F. For the high-resolution C 1s XPSspectrum (Figure 9b), the overlapping peaks extend over abroad energy range from 295 to 280 eV, indicating thepresence of a complex mixture of organic and inorganiccarbon compounds. We have performed peak fitting andquantification of the C 1s envelope, and the results are alsoshown in Figure 9b. Based on the general classes of carbonspecies detected by TOF-SIMS, we find that seven peaks canbe used to fit the C 1s XPS curve. The most intense peak at∼293 eV originates from the C-F bonds of the PTFE filter.The low intensity peak at a binding energy of ∼292 eV isindicative of the aromatic organic compounds on the PM2.5

surface and is associated with low-energy π-π* shake-uptransitions (35, 36). The presence of aromatic compounds issupported by TOF-SIMS results (Table 3). The quantificationof aromatic compounds on PM2.5 surface is 4.1% (notincluding carbon from PTFE) according to XPS peak fitting,where error bars are typically < ( 10% of the given values.The peak at a binding energy of ∼290 eV is associated withC in the -CO2 functional group, indicative of the presenceof carboxylic acids and carboxylates. The peak located at∼286.6 eV is an indication of the existence of the C-Ofunctional group, implying the presence of alcohols, ethers,and carboxylic acids. The peak at a binding energy of ∼285.7eV is due to the R-C in the C-CO2 functional group, indicative

of the existence of carboxylic acids and carboxylates. TheC-C and C-H bonds are represented by the peak at ∼285eV. The last peak is at a binding energy of ∼281.6 eV. Thecarbides are likely to be the source of this peak. The particulardetailed distribution and relative intensities of this peak fitare by no means implied to be a unique fit. Nevertheless, thenumber of peaks employed and their binding energies areconsistent with the functional groups known to be presentby the other techniques employed here.

The O 1s XPS curve in Figure 9c also shows broad signalsranging from ∼535 to 526 eV. This indicates the coexistenceof different oxygen chemical environments on the surface ofSLC PM2.5. The peak-fitting and quantification for O 1s arealso shown in Figure 9c. The peak located at the binding

FIGURE 8. Comparison of positive TOF-SIMS spectra at nominal mass m/z ) 28 and 29: (a, c) from PM2.5 on a blank PTFE filter and (b,d) from PM2.5 on a gold-coated PTFE filter.

FIGURE 9. XPS spectra of SLC PM2.5 on a PTFE filter: (a) surveyspectrum, (b) high-resolution C 1s, and (c) high-resolution O 1s.

VOL. 35, NO. 15, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 3119

energy of ∼529.9 eV is assigned to the oxygen from metaloxides. The O 1s from sulfates, phosphates, and hydroxides(also organics) is at the binding energy of 531.6 eV. The peakat about 533.2 eV is indicative of the presence of nitrates.The overall O 1s XPS curve covers all these binding energyranges, indicative of many contributions to the O 1s signalsfrom these inorganic compounds. These XPS results areconsistent with those from TOF-SIMS. Like the XPS C 1speak, it is confirmatory although not particularly diagnostic.

FTIR Measurements. FTIR measurements in an ATR(attenuated total reflectance) mode were performed to studythe functional groups of organic compounds in SLC PM2.5

(Figure 10). C-F bond signals from absorbance by the PTFEfilter is beyond the range shown in this figure. The PTFEfilter therefore provides a convenient IR-transparent windowfor these IR experiments. These FTIR results are consistentwith the results obtained from TOF-SIMS and XPS whichhave been discussed above.

The narrow peak located at ∼2926 cm-1 is due to the CH2

asymmetric stretch, and the peak at 2856 cm-1 originatesfrom the CH2 symmetric stretch. CH3 stretches located at2961 and 2872 cm-1 are weaker than the CH2 stretches andappear as shoulders. These peaks are associated withhydrocarbons present in particulates. The unsaturated CdCbond stretch is observed at 1630-1680 cm-1, similar to thearomatic CdO stretch (1640-1700 cm-1). The relativelyintense peak at 1715 cm-1 is due to the saturated CdO stretch.The presence of aldehydes is indicated by the relatively strongpeak at 1390 cm-1 which is assigned to the C-H bend inaldehydes.

The existence of carboxylates is indicated by the mostintense peak located at 1414 cm-1 which is assigned to thesymmetric COO stretch. Additional evidence of carboxylatesis indicated by a less intense peak at 1610 cm-1 which is dueto the asymmetric COO stretch and overlapped by otherabsorbances. The peak at 2815 cm-1 is an indication of thepresence of N-CH3 aromatic compounds. Two peaks at 1290and 1326 cm-1 could be assigned to the C-N stretch inaromatic amines, although they appear only as a shoulderon another broad peak here. The intense peak at 1330-1390cm-1, overlapping the aldehydes and carboxylates, evidencesthe nitrogen-containing compounds. Like the area-averagedXPS spectra, these area-averaged FTIR data are confirmatoryalthough not particularly diagnostic or unique.

In summary, we have demonstrated that the combinationof three surface analysis techniques (TOF-SIMS, XPS, andFTIR) complements each other and provide abundant surfacechemical information of PM2.5. The high spatial resolution(∼100 nm) and high surface sensitivity (topmost one or twomonolayers) of TOF-SIMS enable the surfaces of individual

particulates to be analyzed. The high mass-resolution (m/∆m ≈ 10 000) of TOF-SIMS provides good separation ofsignals from different chemical species at the same nominalmass, and the extremely high detection sensitivity (ppm orppb) of TOF-SIMS makes the detection of trace elementspossible. All above-mentioned advantages of TOF-SIMS makeit a perfect technique for qualitative surface analysis ofindividual particulates. However, the quantification of TOF-SIMS data is difficult due to the so-called matrix effect, andit is only possible by using standard samples with knownchemical compositions and concentrations that match thematrix of the unknown samples. This makes it more difficultwhen the complex samples with mixtures of chemicalcompositions such as PM2.5 samples are analyzed. XPS is aquantitative surface analysis technique that can complementTOF-SIMS in terms of quantification as demonstrated in thispaper. XPS also gives us information about the chemicalstates of each of the compositional elements. Peak fitting ofXPS data provides quantitative information for specificfunctional groups. However, its less sensitive detection ability(>0.1% of a monolayer) makes the analysis of trace chemicalspecies impossible. FTIR provides detailed information aboutthe presence of specific functional groups of compositionalcompounds, complementing both TOF-SIMS and XPS.However, elemental species cannot be detected by FTIR. Inthis paper, we have shown that these complementary piecesof information that help to fully characterize complex PM2.5

samples can be obtained by the combination of TOF-SIMS,XPS, and FTIR techniques.

In the future we hope to use TOF-SIMS in conjunctionwith XPS and FTIR to explore the chemical reactivity ofindividual PM2.5 particulates when they are exposed tocontrolled doses of gas-phase reactants.

AcknowledgmentsFinancial support from the University of Utah FundingIncentive Seed Grant Program and the National ScienceFoundation (DMR-9724307; CHE-9814477) are gratefullyacknowledged. The authors thank the State of Utah Depart-ment of Environmental Quality, Division of Air QualityMonitoring Center for PM2.5 sampling. We also thank PaulDryden for his help with TOF-SIMS and XPS experimentsand Joel Harris and Dion Rivera for help in FTIR measure-ments.

Literature Cited(1) Linton, R. W.; Williams, P.; Evans C. A., Jr.; Natusch, D. F. S.

Anal. Chem. 1977, 49, 1514-1521.(2) Fisher, G. L. Environ. Health Perspect. 1983, 47, 189-199.(3) Nikolaou, K.; Masclet, P.; Mouvier, G. Sci. Total Environ. 1984,

32, 103-132(4) Eiceman, G. A.; Vandiver, V. J. Atmos. Environ. 1983, 17, 461-

465.(5) Chrisp, C. E.; Fisher, G. L.; Lammert, J. E. Science 1978, 199,

73-75.(6) Pitts, J. N., Jr.; Van Cauwenberghe, K. A.; Grosjean, D.; Schmid,

J. P.; Fritz, D. R.; Belser, W. L., Jr.; Knudson, G. B.; Hynds, P. M.Science 1978, 202, 515-519.

(7) Finlayson-Pitts, B. J.; Pitts, J. N., Jr. Atmospheric Chemistry:Fundamentals and Experimental Techniques; John Wiley &Sons: New York, 1986.

(8) Lao, R. C.; Thomas, R. S. In Polynuclear Aromatic Hydrocarbons;Bjorseth, A., Dennis, A. J., Eds.; Battelle, Columbus, OH, 1978.

(9) Tokiwa, H.; Ohnishi, Y. CRC Crit. Rev. Toxicol. 1986, 17, 23-60.(10) WHO. World Health Statistics Annual; WHO: Geneva, 1995.(11) Paputa-Peck, M. C.; Marano, R. S.; Schuetzle, D.; Riley, T. L.;

Hampton, C. V.; Prater, T. J.; Skewes, L. M.; Jensen, T. E.; Ruehle,P. H.; Bosch, L. C.; Duncan, W. P. Anal. Chem. 1983, 55, 1946-1954.

(12) Kamiura, T.; Kawaraya, T.; Tanaka, M.; Nakadoi, T. Anal. Chim.Acta 1991, 254, 27-31.

(13) Pitts, J. N., Jr.; Sweetman, J. A.; Zielinska, B.; Atkinson, R.; Winter,A. M.; Harger, W. P. Environ. Sci. Technol. 1985, 19, 1115-1121.

FIGURE 10. FTIR spectrum of SLC PM2.5 on a PTFE filter.

3120 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 15, 2001

(14) Pitts, J. N., Jr.; Zielinska, B.; Sweetman, J. A.; Atkinson, R.; Winter,A. M. Atmos. Environ. 1985, 19, 911-915.

(15) Arey, J.; Zielinska, B.; Atkinson, R.; Winter, A. M.; Ramdahl, T.;Pitts, J. N., Jr. Atmos. Environ. 1986, 20, 2339-2345.

(16) Zielinska, B.; Arey, J.; Atkinson, R.; Ramdahl, T.; Winter, A. M.;Pitts, J. N., Jr. J. Am. Chem. Soc. 1986, 108, 4126-4132.

(17) Zielinska, B.; Arey, J.; Atkinson, R.; Winter, A. M. Atmos. Environ.1989, 23, 223-229.

(18) Ciccioli, P.; Cecinato, A.; Cabella, R.; Brancaleoni, E.; Buttni, P.Atmos. Environ. 1993, 27A, 1261-1270.

(19) Air Quality Criteria for Particulate Matter and Sulfur Oxides.EPA Report 600/8-82-029aF; U.S. Environmental ProtectionAgency: Washington, DC, 1982.

(20) Bates, D. V. Environ. Res. 1992, 59, 336-349.(21) Schwartz, J. Environ. Res. 1994, 64, 36-52.(22) Pope, C. A.; Dockery, D. W.; Schwartz, J. Inhalation Toxicol.

1995, 7, 1-18.(23) U.S. Environmental Protection Agency. Revised Requirements

for Designation of Reference and Equivalent Methods for PM2.5 and Ambient Air Quality Surveillance for Particulate Matter-Final Rule. 40 CFR part 58. Federal Register 1997, 62 (138):38830-38854.

(24) U.S. Environmental Protection Agency. Revised Requirementsfor Designation of Reference and Equivalent Methods for PM2.5 and Ambient Air Quality Surveillance for Particulate Matter-Final Rule. 40 CFR part 53. Federal Register 1997, 62 (138):38736-38830.

(25) Richardson, S. D. Chem. Rev. 2001, 101, 211-254.

(26) Benninghoven, A. Angew. Chem., Int. Ed. Engl. 1994, 33, 1023-1043.

(27) Bentz, J. W. G.; Goschnick, J.; Schuricht, J.; Ache, H. J.;Zehnpfennig, J.; Benninghoven, A. Fresenius J. Anal. Chem. 1995,353, 603-608.

(28) Jambers, W.; Bock, L. D.; Grieken, R. V. Analyst 1995, 120, 681-692.

(29) Xhoffer, C.; Wouters, L.; Artaxo, P.; Van Put. A.; Van Grieken. R.In Environmental Particles; Buffle, J., Van Leeuwen, H. P., Eds.;Lewis: Chelsea, MI, 1992; Vol. 1, pp 107-143.

(30) Kellner, R.; Malissa, H. Aerosol Sci. Technol. 1989, 10, 397-407.(31) Dangler, M.; Burke, S.; Hering, S. V.; Allen, D. T. Atmos. Environ.

1987, 21, 1001-1004.(32) Briggs, D.; Hearn, M. J. Vacuum 1986, 36, 1005-1010.(33) Hearn, M. J.; Briggs, D. Surf. Interface Anal. 1986, 9, 411-417.(34) Enya, T.; Suzuki, H.; Watanabe, T.; Hirayama, T.; Hisamatsu, Y.

Environ. Sci. Technol. 1997, 31, 2772-2776.(35) Clark, D. T.; Adams, D. B.; Dilks, A.; Peeling, J.; Thomas, H. R.

J. Electron Spectrosc. Relat. Phenom. 1976, 8, 51-60.(36) Clark, D. T.; Dilks, A. J. Polym. Sci., Polym. Chem. Ed. 1977, 15,

15-30.

Received for review December 7, 2000. Revised manuscriptreceived May 1, 2001. Accepted May 7, 2001.

ES0019530

VOL. 35, NO. 15, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 3121