An assessment of the sensitivity of the technique is performed by detecting small microspheres and thin films of IR-absorbing material (polymeric) through an identifiable spectroscopic response. The effect of the substrate on the strength of the recorded spectra is also investigated.
Polystyrene microspheres 5 µm in diameter, suspended in a solution of surfactant and distilled water at 1% solids (from PSI supplies), were deposited on a rock salt substrate. Salt was chosen as a substrate because it is transparent to IR radiation. Spectral contamination is therefore eliminated. After water evaporation, the microspheres were embedded on the surface of the substrate by pressing with a glass slide. Using a conventional silicon nitride AFM probe, the microspheres were observed to be dispersed on the surface in small groups of a few spheres as well as isolated. Their diameter was observed to be ∼5 µm as expected. A Wollaston wire probe was then used to scan the sample. However, some spheres were observed to be dislodged and dragged along by the scanning tip. To improve their adherence to the surface, the sample was heated at 100 °C, above the polystyrene glass transition (90 °C), for a few minutes. Further scanning with the Wollaston wire probe did not dislodge any spheres. The size of the spheres was now observed to be approximately three times the actual 5 µm diameter. This broadening effect is an artefact resulting from the large Wollaston wire probe (5 µm wire diameter, ∼45° loop angle), as seen in Fig. 3(a). Some pits can be observed on the surface of the substrate previously occupied by the dislodged microspheres. These were occupied by polystyrene microspheres, which were pushed away by the scanning probe. The probe was then positioned at locations (i) and (ii). The spectrometer was run at 2.2 kHz. One thousand scans were co-added to produce spectra (i) and (ii) shown in Fig. 3(c) from the corresponding locations (i) and (ii). When a measurement is performed at a region where a few spheres have agglomerated, location (i), a strong photothermal signal results in a spectrum with good signal-to-noise ratio and in which the main absorption bands are identifiable. Where measurement is performed near a single isolated sphere, location (ii), poor signal-to-noise ratio is observed.
Figure 3. (a) Polystyrene spheres 5 µm in diameter embedded on the surface of a rock salt substrate and imaged with a Wollaston wire probe. (b) Photograph of a few polystyrene spheres picked up by the Wollaston wire probe. (c) Photothermal spectra obtained at positions (i) and (ii) shown in (a), from the spheres (iii) shown in (b) and an absorbance spectrum (iv) of polystyrene taken from a database.
Download figure to PowerPoint
A second sample was prepared by depositing the microspheres on a glass slide. After water evaporation, a few spheres were picked up using the probe and a spectrum taken with probe and spheres free standing in the air, the glass substrate having been removed. A much stronger signal is recorded, resulting in a very clean spectrum as shown in Fig. 3(c), spectrum (iii). The substrate acts as a heat sink lowering the temperature of the spheres. When the spheres are attached to the probe tip, heat loss is only to the surrounding air, which has a much lower thermal conductivity than the solid rock salt. Spectrum (iv) is that of polystyrene as given in the Bruker database.
The sensitivity of the techniques was also assessed by recording the IR spectra of a series of poly(ethylene terephthalate) (PET) films of various thickness: 0.9, 3, 6, 9, 12, 23, 36, 50, 75, 100, 125 and 250 µm. The films were mounted on hollow stubs. During measurement, the probe was placed on top of the part of the sample that was surrounded on both sides by air. The probe was also placed in contact with the sample in such a way that only the tip was in contact. The position of the probe, with regard to the location of IR focus, was exactly the same for each film. Measurement conditions can therefore be considered to be very similar for each film.
Spectra were recorded with the spectrometer operated at a 2.2 kHz mirror speed. Each spectrum results from the averaging of 1000 scans. Spectral resolution was set to 8 cm−1. Electronic amplification was the same for all measurements. Resulting spectra are plotted in groups with the same scaling factor. A direct comparison of the strength of the spectra is therefore straightforward.
Figure 4(a) shows spectra recorded from films with thickness ranging from 0.9 µm to 23 µm. It is clear that the strongest temperature rise and associated spectrum is recorded from the thinnest film. Surface temperature is a function of optical absorption, thermal diffusion length and film thickness. The thicker the film, the more heat diffuses into the bulk, resulting in a lowering of surface temperature. However, it can be expected that below a certain thickness, smaller than the optical absorption length, the reduction in deposited electromagnetic radiation energy will result in a lowering of surface temperature. From these measurements on thin films, the critical value is less than 0.9 µm. On the basis of transmission measurements (using the internal IR optical detector) through the 0.9-µm, 9-µm, 6-µm and 12-µm films, optical absorption lengths are estimated to be 1.6 µm, 1.5 µm and 1.9 µm for the 1720-cm−1, 1260-cm−1 and 1110-cm−1 peaks, respectively.
Figure 4. Photothermal spectra obtained from thin poly(ethylene terephthalate) (PET) films: (a) free-standing films of thickness ranging from 0.9 to 23 µm, (b) free-standing films of thickness ranging from 23 to 250 µm, (c) 0.9-µm film free standing (i) and on a steel stub (ii).
Download figure to PowerPoint
Figure 4(b) shows spectra recorded from films 23, 100 and 250 µm thick. The spectra are observed to have similar strength. For films thicker than the thermal diffusion length, the damped thermal waves will fully diffuse into the bulk. Surface temperature then remains the same. With the spectrometer operated at 2.2 kHz interferometer speed, the one-dimensional thermal diffusion length is around 25 µm at 2 µm wavelength and 10 µm at 20 µm wavelength.
The effect of substrate is also illustrated in Fig. 4(c). The spectra shown were recorded from a free-standing 0.9-µm film and the same film mounted on a steel stub. The 0.9-µm film being much thinner than the thermal diffusion length, the steel substrate, because of its higher thermal conductivity, acts as heat sink. This results in a lowering of the overall temperature of the film, in particular on the surface.
An illustration of a practical application of the technique is given in Fig. 5. The 3-µm coating of a copper wire is identified as polyimide from its photothermal IR spectrum.
Figure 5. Photothermal spectrum of a 3-µm coating film on copper wire revealing the material to be polyimide. Inset: the probe in contact with the wire during measurement.
Download figure to PowerPoint
AFM-type probes functioning with a temperature sensor will in theory have their spatial resolution limited by the size of the sensor. Here we explore the use of a conventional silicon AFM probe as a thermal sensor. Measurements were carried out using two modes of operation: (i) a contact mode of operation in which sample surface expansion resulting from heat generated from IR absorption causes the cantilever to deflect; (ii) a non-contact mode of operation in which IR absorption-generated heat diffuses from the surface of the irradiated sample surface into the adjacent layer of air, resulting in vibration of the cantilever. Conventional laser beam deflection and photodiode detection is used in both cases. The contact mode of operation is a localized measurement. The non-contact mode would be suitable for macroscopic measurements. Note that a cantilever without a pyramid tip would be sufficient in the latter case.
The spectra shown in Fig. 6 are obtained from a polypropylene sample, with the probe at the IR beam focal point, as when the thermal probe is used, and the sample surface (i) far away, (ii) in near contact (estimated at a few micrometres from probe jump to contact and scanner retraction) and (iii) in contact with a high force feedback time constant. In all cases the spectra are obtained from the photodiode top–bottom signal. In the latter case, Z feedback does not respond to changes in top–bottom input signal but an average force is maintained. Spectra (i) and (ii) result from 1000 co-added scans. Spectrum (iii) results from 100 000 co-added scans. The resultant signal-to-noise ratio is then similar. Spectrum (iv) results from a weighted subtraction of spectrum (i) from (ii). Characteristic absorption bands of polypropylene (3000–2750 cm−1 and 1500–1200 cm−1) are clearly identifiable. When the tip is in contact, the free end of the cantilever is pinned to the surface. Deflection was expected to result from surface expansion only at the contact point. However, self-heating and most likely also heat diffusing from the surface through the air and reaching the cantilever results in some deformation, thus contributing to the recorded spectrum. The first effect is evident from the presence of the absorption band between 1400 cm−1 and 700 cm−1, owing to presence of silicon nitride. The overall energy profile of the IR source is also apparent. However, the absence of the dip in the spectrum around 2300 cm−1, associated with energy absorbed by CO2 in the path of the beam, reveals that direct heating of the surrounding air does not contribute to total bending of the cantilever. In this contact mode poor signal-to-noise ration requires long integration time. At present, the sample rests on a holder at the end of a relatively long and thin rod. This makes the mechanical loop susceptible to mechanical vibration. Improvement can in principle be made to minimize this source of noise.
Figure 6. Photothermomechanical spectra obtained with a conventional AFM probe: (i) the probe free standing in air, (ii) the probe in near contact with the surface of a polypropylene sample, (iii) the probe in contact with the surface of the sample, and (iv) the photothermal spectrum obtained by subtracting spectrum (i) from (ii).
Download figure to PowerPoint