Atomic Force Microscopy-Infrared Spectroscopy (AFM-IR) for the analysis of atmospheric aerosol particles



Amy L. Bondy1, Rachel M. Kirpes1, Rachel L. Merzel1, Kerri A. Pratt1, Mark M. Banaszak Holl1, Andrew P. Ault1,2
1 Department of Chemistry, University of Michigan, Ann Arbor, MI 48109, United States
2 Department of Environmental Health Sciences, University of Michigan, Ann Arbor, MI 48109, United States
Amy Bondy defended her PhD in chemistry from the University of Michigan in November 2017 with her research involving the single particle microscopic and spectroscopic chemical analysis of aerosols. She graduated in April 2018. Continuing to pursue her passion for microscopy she now works as a research specialist in the cryo-EM facility within the Life Sciences Institute at the University of Michigan.
Chemical analysis of atmospheric aerosols is an analytical challenge, as aerosol particles are complex chemical mixtures that can contain hundreds to thousands of species in attoliter volumes at the most abundant sizes in the atmosphere (~100 nm). These particles have global impacts on climate and health, but there are few methods available that combine imaging and the detailed molecular information from vibrational spectroscopy for individual particles <500 nm. Herein, we show the first application of atomic force microscopy with infrared spectroscopy (AFM-IR) to detect trace organic and inorganic species and probe intraparticle chemical variation in individual particles down to 150 nm. Combining strengths of AFM (ambient pressure, height, morphology, and phase measurements) with photothermal IR spectroscopy, the potential of AFM-IR is shown for a diverse set of single-component particles, liquid-liquid phase separated particles (core-shell morphology), and ambient atmospheric particles. The sub-diffraction limit capability of AFM-IR has the potential to advance understanding of particle impacts on climate and health by improving analytical capabilities to study water uptake, heterogeneous reactivity, and viscosity.
Reprinted (adapted) with permission from Amy L. Bondy, Rachel M. Kirpes, Rachel L. Merzel, Mark M. Banaszak Holl, Andrew P. Ault. Anal. Chem., 2017, 89 (17), pp 8594–8598. Copyright 2017 American Chemical Society. This project was supported by NSF CAREER Award CHE-1654149 and startup funds from the University of Michigan.
Corresponding author
Amy Bondy, Life Sciences Institute, University of Michigan, Ann Arbor, MI 48109, United States
Atmospheric aerosol particles <1 µm in diameter impact climate by scattering and absorbing solar radiation, nucleating cloud droplets and ice crystals, and acting as surfaces for heterogeneous reactions in the atmosphere.1 Additionally, through inhalation, these particles can penetrate deeply into the lungs, depositing in the alveoli.2 This has large consequences as air pollution accounts for 10% of global deaths annually.3
The size, chemical composition, and physical structure (eg well-mixed, core-shell, partially engulfed) of individual particles is critical for determining their climate and health impacts.4-6 However methods that can provide detailed molecular information at ambient pressure, allowing detection of volatile components for individual particles near the mode of the atmospheric number size distribution (~100 nm), are limited.7
Vibrational spectroscopy has great potential to provide insight into chemical processes within aerosols,7 as it has been used to provide detail on functional groups, such as ν(SO42-), ν(NO3-), ν(C-H) and ν(O-H).8 Vibrational methods provide molecular information that complements elemental information and electronic transitions from energy dispersive X-ray spectroscopy and near edge X-ray absorption fine structure spectroscopy, respectively.9-10 Additionally, vibrational modes are sensitive to surrounding molecular environments, such as whether ions are free or bound to specific cations (NO3-(aq) vs. NO3-(s)). These peak shifts can probe processes like phase separation and hydrogen bonding.8
The greatest challenge of using optical microscopy-based vibrational spectroscopy for analysis of submicron particles has been the diffraction limit of visible and infrared (IR) light. Both Raman microspectroscopy and micro-IR spectroscopy of individual atmospheric particles provide insights for particles >1 µm,7, 11-13 but have limited ability to probe accumulation mode particles (<1 µm) and the atmospheric surface area mode.14
To probe smaller particles, surface-enhance Raman spectroscopy (SERS)15 and tip-enhanced Raman spectroscopy (TERS)16 have been used, but uneven enhancements require further development for quantification. To understand the relationship between chemical composition and morphology in a critical size range for aerosol impacts, vibrational methods are needed that chemically analyze particles <500 nm at atmospheric pressure and probe intra-particle compositional variability.
Atomic force microscopy with infrared spectroscopy (AFM-IR) has the potential to overcome size limitations with imaging capabilities on the scale of nanometers, and ~50 nm chemical resolution.17 This method combines simultaneous single particle measurements of physical properties (hygroscopicity,18 surface tension19, 20, phase, morphology21 by AFM) with chemical composition (functional groups by IR absorption). In this study, AFM-IR was used to analyze accumulation mode aerosol particles (>150 nm) for the first time. Inorganic and organic functional groups were characterized for laboratory-generated standards and ambient particles. Phase separation and spatial variation of chemical species were observed on spatial scales of 100 nm, demonstrating that AFM-IR can analyze particles below the diffraction limit.
Materials and Methods
Laboratory Standard Aerosol Generation and Impaction
Laboratory-generated aerosol samples were created by atomizing and impacting particles from 0.05 M standard solutions of ammonium sulfate ((NH4)2SO4, Alfa Aesar, 99%), sodium nitrate (NaNO3, Sigma Aldrich 99.0%), succinic acid ((CH2)2(CO2H)2 Alfa Aesar, 99%), and D-sucrose (C12H22O11, Fisher Scientific 99.9%) onto substrates using a microanalysis particle sampler (MPS-3, California Measurements, Inc.). Particles were impacted on stage 3 of the MPS (70-400 nm equivalent aerodynamic diameter). (CH2)2(CO2H)2 was generated via dissolution of succinic anhydride in Millipore water. Generated aerosols were passed through two diffusion dryers (drying to ~15% relative humidity) prior to impaction. Core-shell particles were generated by atomizing a 1% by weight solution of (NH4)2SO4 and polyethylene glycol 400 (Fluka) in a 1:1 ratio. Before impaction, (NH4)2SO4/polyethylene glycol (PEG) particles were not passed through diffusion dryers, resulting in liquid-liquid phase separation. Particles with AFM heights <250 nm were used in the subsequent analysis. 
AFM-IR Imaging and Spectral Acquisition
A nanoIR2 system (Anasys Instruments, Santa Barbara, CA) was used to test single-component, atmospherically-relevant standards in contact mode. The principle of AFM-IR, shown in Figure 1, is explained in detail by Dazzi et al.17 Briefly, the sample is pulsed with a tunable IR source (2.5-12 µm, 1 kHz) over a frequency range of 900-3600 cm-1. Upon absorption of IR radiation, thermal expansion of the sample occurs, causing the AFM probe in contact with the surface to oscillate at its resonant frequency. 
Figure 1. Schematic of AFM-IR operation. Local thermal expansion from the IR laser is detected by the cantilever, allowing IR spectra with ~50 nm resolution to be collected. IR spectra were collected from individual (NH4)2SO4 particles using AFM-IR (this study) and micro-FTIR (Liu et al.25)
The oscillations are detected by the deflection laser’s position on the photodiode, and the amplitude of oscillation is proportional to IR absorbance, yielding an IR spectrum as a function of wavelength with 4 cm-1/point resolution (instrument limit is 4 cm-1).17
AFM height/deflection images and IR spectra of (NH4)2SO4, NaNO3, succinic acid, and sucrose particles were collected in contact mode (IR power 21.27%, filter in) at a scan rate of 1 Hz using a gold-coated contact mode silicon nitride probe (Anasys Instruments, 13 ± 4 kHz resonant frequency, 0.07-0.4 N/m spring constant). AFM height/phase images of phase separated (NH4)2SO4 and PEG particles were collected in tapping mode at a scan rate of 0.4 Hz using a dual-purpose gold-coated silicon nitride tapping probe for NIR2 (Anasys Instruments, 75 ± 15 kHz resonant frequency, 1-7 N/m spring constant).
To collect IR spectra, the dual-purpose tapping probe was put in contact mode (since on the nanoIR2 this is required to detect the sample’s photothermal expansion). The amplitude of cantilever oscillations was mapped using 128 co-averages, with a resolution of 4 cm-1/point. Eight spectra with an IR power of 38.46% (filter in) were averaged for the PEG reference spectrum, and four spectra collected with an IR power of 21.27% (filter in) were averaged for IR spectra collected from both the core and shell of the phase separated (NH4)2SO4/PEG particle.
For the IR spectral maps of an ambient aerosol particle, maps were collected in contact mode using a gold-coated contact mode probe at 1476 cm-1 and 1580 cm-1 with a trace rate of 0.1 Hz and retrace rate of 1 Hz so that the update of the IR-peak, IR-amplitude and frequency data approximated the pixel rate of the image. The amplitude of cantilever oscillations was mapped using 16 co-averages, 300 pt. resolution for X and Y, and IR power at 1.03%.
The IR ratio map (Figure 4f) most clearly shows differences in spatial distribution of the two vibrational modes (1476 and 1580 cm-1). This map, showing the ratio between IR intensity at 1476 cm-1 vs. 1580 cm-1, was generated in Analysis Studio (Anasys analysis software) by cross-correlating the spatial distribution of the two AFM height images and calculating the ratio between the two IR intensities at each point.
Since the ratio map correlates the IR maps with the height image, changes in intensity due to topography are normalized. Furthermore, thermal drift is accounted for in this analysis, hence the resulting ratio map is an accurate representation of the particles. Thermal drift of the sample between IR frequency maps is noted in Figure 4f by the solid purple regions along the top and right edge of the image. 
Results / Discussion
Single Component Aerosol Particles
Aerosol particles were generated from single-component solutions to evaluate AFM-IR for model aerosol particles. Two inorganic salts (NaNO3 and (NH4)2SO4), and two organic compounds (succinic acid and sucrose), were aerosolized, dried, and impacted onto Si substrates using a microanalysis particle sampler (MPS-3, California Measurements Inc.). Particles with volume equivalent diameters (Dve) ~200-350 nm were selected for IR analysis, where Dve corresponds to the diameter of a sphere with volume equivalent to the impacted particle. Measured IR peak positions in submicron particles were compared to conventional FTIR mode frequencies, with good agreement observed. (NH4)2SO4 was chosen as it is ubiquitous in aerosols and frequently used as a benchmark compound in aerosol studies,14 while NaNO3 is readily formed in the atmosphere from reactions of HNO3 with sea spray aerosol and mineral dust.22 Succinic acid, (CH2)2(CO2H)2, was chosen as dicarboxylic acids are the most abundant organic species in aerosols,23 while sucrose (C12H22O11) is used as a model compound for glassy secondary organic aerosol in lab studies.24
In Figure 2, representative AFM height and deflection images and IR spectra are shown for individual particles generated from the four standards. The spectrum for (NH4)2SO4 particles showed ν(SO42-) at 1091 cm-1 and δ(NH4+) at 1422 cm-1, which agree with spectra of ~1 µm particles observed by Liu et al.25 in a micro-FT-IR study. Although less intense, a weak mode at 3139 cm-1 consistent with ν(NH4+)25-26 was also observed. In the NaNO3 particles, the sharp ν(NO3-) at 1356 cm-1 aligns well with literature νa(NO3-) for particles at low relative humidity.25
For the more complex organic particle spectra, IR mode assignments also matched prior IR studies, particularly in the fingerprint region. Peaks at 1201, 1308, and 3049 cm-1 consistent with ν(C-C), ν(C-O), and ν(CH2), were observed in (CH2)2(CO2H)2 particles.27-28 A strong vibrational mode at 1691 cm-1, ν(C=O), and a very strong vibration at 1404 cm-1, δ(CH2), were also observed.27-28 For the C12H22O11 particles, a variety of functional groups were observed including ν(C-O) at 1057 cm-1, δ(C-O-H) at 1439 cm-1, ν(CH2) at 2913 cm-1, and an intense ν(O-H) mode at 3345 cm-1.29 Aside from the ν(O-H) stretch in C12H22O11, the IR intensity of the vibrational modes in the fingerprint region were most intense for all compounds. Although many vibrational modes are possible in this region, these results indicate that the fingerprint region could be used to make tentative assignments in multi-component particles. 
Figure 2. AFM height, deflection, and IR spectra of single-component particles: a) (NH4)2SO4, b) NaNO3, c) succinic acid, and d) sucrose. Dve of analyzed particle from each standard: 346 nm, 303 nm, 335 nm, and 202 nm
Core-Shell Aerosol Particles
To examine capabilities for more complex morphologies, core-shell particles were generated similar to prior microscopy studies.31-32 Particles with an (NH4)2SO4 core and a PEG shell were impacted onto Si and analyzed by AFM (tapping mode). Although the height images appear similar to the single-component particles, the phase images clearly show two distinct phases (Figure 3a-b). AFM phase imaging has been used previously to detect phase separation of submicron particles including partially engulfed and core-shell morphologies,5, 33 however chemical characterization of each phase in this size range has been limited due to the optical diffraction limit. Since AFM-IR has spatial resolution on the scale of 50-100 nm, dual-purpose AFM tips were used to collect images (tapping mode) and IR spectra (tapping mode; probe in contact with sample) from each phase. Spectra collected from the core and shell of the 550 nm particle (Dve) show two distinct compositions (Figure 3c). The core has an intense vibrational mode at 1090 cm-1, ν(SO42-) of (NH4)2SO4,25-26, 34 while the shell shows two modes at 1105 cm-1 and 1256 cm-1, the ν(C-O-C) and CH2 twisting modes in PEG.27, 35 Since the PEG shell covers the particle, a small mode at 1266 cm-1 in the ‘core’ spectrum is observed from PEG located on the top of the impacted particle. Similarly, the shoulder at 1090 cm-1 is likely due to a less intense PEG vibration ~1105 cm-1. Thus, AFM-IR chemically distinguished the core and shell for a 550 nm particle.
Figure 3. a) AFM height and b) phase images, as well as c) IR spectra of a submicron core-shell morphology particle consisting of (NH4)2SO4 coated with polyethylene glycol (PEG) The blue traces are AFM-IR spectra for (NH4)2SO4 and PEG. d) Optical image, e) Raman spectral map, and f) Raman spectra of a supermicron (NH4)2SO4/PEG particle. Black and red traces were collected from the core and shell, respectively. (NH4)2SO4 modes (yellow), and PEG modes (red) are highlighted
To compare AFM-IR results to more traditional vibrational spectroscopy techniques, Raman microspectroscopy was used to collect spectra and chemical maps of core-shell (NH4)2SO4/PEG particles larger than one micron (Figure 3d-f). The resulting Raman spectra agree with the AFM-IR results, with the shell containing solely PEG, while the “center” of the particles has both (NH4)2SO4 and PEG. The Raman map (Figure 3e), with regions corresponding to the PEG modes at 1465 cm-1 and 2874 cm-1, and the SO42- and NH4+ modes at 974 cm-1 and 3150 cm-1, clearly depict a core-shell morphology, similar to the AFM phase image. However while Raman microspectroscopy can analyze particles >1 µm, the greatest advantage of AFM-IR is that it can investigate submicron particles. One limitation of the (NH4)2SO4/PEG system studied here is that an IR spectral map could not be collected because PEG is a liquid, necessitating AFM analysis in tapping mode. As contact mode is currently needed for collecting IR spectra with the nanoIR2, only discreet point spectra could be collected. 
Ambient Aerosol Particles
To demonstrate spatial resolution for spectral mapping of ambient particles with numerous chemical components, ambient aerosol particles were collected on Si substrates in Ann Arbor, MI (August 2016). IR spectral maps were collected for particles with Dve <800 nm (Figure 4).
Figure 4. AFM height (a) and deflection images (b), IR spectrum (c), as well as IR spectral maps at 1476 cm-1 (d) 1580 cm-1, and the ratio of 1476/1580 cm-1 shown for an ambient aerosol particle from Ann Arbor, MI. Four particles (P1-P4) exhibiting chemical heterogeneity are identified (e). Purple edge on (e) represents thermal drift between each map
Two modes, 1476 cm-1 and 1580 cm-1 suggestive of δ(CH2)27 and ν(C=C),27-28 respectively, were observed for these particles, with different spatial distributions (Figure 4d-f). The ratio map (Figure 4f) most clearly highlights differences in spatial distribution of these two modes, as areas enriched in CH2 (1476 cm-1) appear red, while those enriched with C=C (1580cm-1) appear blue. Within the smaller particles (P1-P2), approximately half the particle contains significant IR intensity from 1476 cm-1, while the other half contains signatures from the 1580 cm-1 mode.
Additionally, the two large agglomerate particles (P3-P4) exhibit chemical heterogeneity within localized regions. These results show that AFM-IR can effectively and simultaneously determine particle physical parameters, chemical composition, and distribution of chemical species within individual atmospheric particles, with a focus on organic functional groups.
Summary and conclusions
The simultaneous spectroscopic and morphological analysis of accumulation mode aerosol particles (<1 µm) is challenging since techniques that are currently available either do not provide the detailed vibrational analysis necessary to identify distinct moieties, such as organic functional groups, or are diffraction limited and cannot investigate particles in this size range. 
Traditionally, AFM has been limited by its lack of chemical information, and micro-FTIR is limited by the diffraction of light to >3 µm particles. AFM-IR however, has great potential to analyze submicron aerosol particles by imaging and providing vibrational information for species within <500 nm particles at ambient pressure.
As shown in this study, AFM-IR was applied to the study of single-component model systems, phase-separated particles, and ambient aerosol particles for the first time, detecting functional groups in particles concurrently imaged, so that particle diameter, height, morphology, phase, and chemical composition were all discerned.
The novel application of this analytical method to atmospheric particles enabled detection of organic and inorganic vibrational modes in standards and ambient particles, as well as identified the composition of phase-separated components within a particle size range that has previously been unstudied by vibrational spectroscopy. The enhanced spatial scale for analysis of atmospheric particles using AFM-IR has the potential to provide key insights regarding size-dependent phase-separated atmospheric particles within an atmospherically critical size range.
1. Pöschl, U., Angew. Chem.-Int. Edit. 2005, 44 (46), 7520-7540.
2. Hinds, W. C., 2nd ed.; John Wiley & Sons: New York, 1999; p 483. 
3. Kennedy, I. M., Proc. Combust. Inst. 2007, 31, 2757-2770.
4. Zhang, Q.; Thompson, J. E., GeoResJ 2014, 3-4, 9-18.
5. Laskina, O.; Morris, H. S.; Grandquist, J. R.; Qin, Z.; Stone, E. A.; Tivanski, A. V.; Grassian, V. H., J. Phys. Chem. A 2015, 119 (19), 4489-4497.
6. Fierce, L.; Bond, T. C.; Bauer, S. E.; Mena, F.; Riemer, N., Nat. Comm. 2016, 7.
7. Ault, A. P.; Axson, J. L., Anal. Chem. 2017, 89 (1), 430-452.
8. Ault, A. P.; Zhao, D.; Ebben, C. J.; Tauber, M. J.; Geiger, F. M.; Prather, K. A.; Grassian, V. H., Phys. Chem. Chem. Phys. 2013, 15 (17), 6206-6214.
9. Laskin, A.; Gilles, M. K.; Knopf, D. A.; Wang, B.; China, S., Progress in the Analysis of Complex Atmospheric Particles. In Annual Review of Analytical Chemistry, Vol 9, Bohn, P. W.; Pemberton, J. E., Eds. 2016; Vol. 9, pp 117-143.
10. Moffet, R. C.; Henn, T.; Laskin, A.; Gilles, M. K., Anal. Chem. 2010, 82 (19), 7906-7914.
11. Craig, R. L.; Bondy, A. L.; Ault, A. P., Aerosol Sci. Technol. 2017, 00-00.
12. Ghorai, S.; Wang, B.; Tivanski, A.; Laskin, A., Env. Sci. & Tech. 2014, 48 (4), 2234-2241.
13. Baustian, K. J.; Cziczo, D. J.; Wise, M. E.; Pratt, K. A.; Kulkarni, G.; Hallar, A. G.; Tolbert, M. A., J. Geophys. Res.: Atmos. 2012, 117 (D6), n/a-n/a.
14. Seinfeld, J. H.; Pandis, S. N., John Wiley & Sons: 2016.
15. Craig, R. L.; Bondy, A. L.; Ault, A. P., Anal. Chem. 2015, 87 (15), 7510-7514.
16. Ofner, J.; Deckert-Gaudig, T.; Kamilli, K. A.; Held, A.; Lohninger, H.; Deckert, V.; Lendl, B., Anal. Chem. 2016, 88 (19), 9766-9772.
17. Dazzi, A.; Prater, C. B.; Hu, Q.; Chase, D. B.; Rabolt, J. F.; Marcott, C., Appl. Spectrosc. 2012, 66 (12), 1365-1384.
18. Morris, H. S.; Estillore, A. D.; Laskina, O.; Grassian, V. H.; Tivanski, A. V., Anal. Chem. 2016.
19. Morris, H. S.; Grassian, V. H.; Tivanski, A. V., Chemical Science 2015, 6 (5), 3242-3247.
20. Hritz, A. D.; Raymond, T. M.; Dutcher, D. D., Atmos. Chem. Phys. 2016, 16 (15), 9761-9769.
21. Krueger, B. J.; Ross, J. L.; Grassian, V. H., Langmuir 2005, 21 (19), 8793-8801.
22. Weis, D. D.; Ewing, G. E., J. Phys. Chem. A 1999, 103 (25), 4865-4873.
23. Kawamura, K.; Bikkina, S., Atmos. Res. 2016, 170, 140-160.
24. Zobrist, B.; Soonsin, V.; Luo, B. P.; Krieger, U. K.; Marcolli, C.; Peter, T.; Koop, T., PCCP 2011, 13 (8), 3514-3526.
25. Liu, Y.; Yang, Z.; Desyaterik, Y.; Gassman, P. L.; Wang, H.; Laskin, A., Anal. Chem. 2008, 80 (3), 633-642.
26. Weis, D. D.; Ewing, G. E., J. Geophys. Res.: Atmos. 1996, 101 (D13), 18709-18720.
27. Larkin, P., Chapter 8 - Illustrated IR and Raman Spectra Demonstrating Important Functional Groups. In Infrared and Raman Spectroscopy, Elsevier: Oxford, 2011; pp 135-176.
28. Miñambres, L.; Sánchez, M. N.; Castaño, F.; Basterretxea, F. J., J. Phys. Chem. A 2010, 114 (20), 6124-6130.
29. Max, J.-J.; Chapados, C., J. Phys. Chem. A 2001, 105 (47), 10681-10688.
30. Hossain, U. H.; Seidl, T.; Ensinger, W., Polym. Chem. 2014, 5 (3), 1001-1012.
31. You, Y.; Smith, M. L.; Song, M.; Martin, S. T.; Bertram, A. K., Int. Rev. Phys. Chem. 2014, 33 (1), 43-77.
32. Veghte, D. P.; Altaf, M. B.; Freedman, M. A., J. Am. Chem. Soc. 2013, 135 (43), 16046-16049.
33. Freedman, M. A.; Baustian, K. J.; Wise, M. E.; Tolbert, M. A., Anal. Chem. 2010, 82 (19), 7965-7972.
34. Cziczo, D. J.; Nowak, J. B.; Hu, J. H.; Abbatt, J. P. D., J. Geophys. Res.: Atmos. 1997, 102 (D15), 18843-18850.
35. Lu, F.; Jin, M. Z.; Belkin, M. A., Nat. Photonics 2014, 8 (4), 307-312.


Website developed by S8080 Digital Media