When the narrow junction between a sharp metallic tip and a sample is illuminated by light, the electric field is tightly confined and highly enhanced near the tip end. If a material attached to the tip has an absorption resonance at the optical frequency, the material can absorb the enhanced field near the tip apex and dissipate heat. The consequent thermal expansion modulates the tip-sample gap distance (H), leading to a thermal expansion force via the van der Waals interaction, which can easily reach a few tens to a few hundreds of pN14, resulting in the generation of a force gradient signal in PiFM17. For a metallic tip with an organic material attached, light absorption followed by thermal expansion occurs at two locations: at the tip and at the sample, as sketched in Fig. 1a. If the material on the tip has no absorption near the sample resonance, the sample solely expands by an amount ΔLs by absorbing the tip-enhanced field, which penetrates into the sample. Since the dipole force for organic molecules is typically one to two orders smaller than the thermal expansion force14, 18, PiFM mapping for organic samples at their absorption resonance corresponds primarily to local absorption mapping for the sample. Conversely, if the sample has no absorption near the resonance of the organic material on the tip, the tip solely expands by ΔLt by absorbing the enhanced field, the strength of which depends on the local refractive index of the sample. This implies that PiFM mapping corresponds to a dipole contrast (local refractive index contrast) for the sample via tip-enhanced thermal expansion of the organic material on the tip. Note that the photoinduced force spectrum at a certain position on the sample is still the absorption spectrum of the organic material on the tip. Lastly, if the absorption resonance of the organic material on the tip has the same frequency as that of the sample, the thermal force signal for the sample is convolved with the response of the tip because the tip-sample distance modulation is summed as ΔLt + ΔLs.
Schematics for tip-enhanced thermal expansion and the IR-PiFM system.
a Diagram showing thermal expansion of the tip (ΔLt) and of the sample (ΔLs). The orange dashed layer on the tip is the PDMS contamination. The incident quantum cascade laser (QCL) illuminates the tip-sample junction at an angle of θ. Vabs is the absorption volume for the tip-enhanced field inside the sample. Vheat is the heated volume due to the absorbed heat over a single pulse width τp. d is the thickness of the sample. b Sketch of the IR-PiFM experimental setup
We first examine the tip expansion behavior using the PDMS contamination layer on a commercial tip, which can be a very sensitive thermal probe for PiFM. Despite the small thickness of several nm for the contaminant layer, due to its high thermal expansion and high extinction coefficient in the mid-infrared range, the tip-enhanced thermal expansion easily reaches a few tens of pm, which results in several tens of pN via the van der Waals force [Section 2 of Supplementary Information]. In Fig. 2a, the PiFM spectra measured by an Au-coated tip with PDMS contaminant are plotted with respect to different substrate materials. The red solid line shows the data measured for a Si substrate whose index of refraction is 3.42 at 1268 cm−1 19, and the black solid line shows data for a ZnSe substrate with an index of refraction of 2.41 at 1268 cm−1 20. In the measured spectral range, no significant absorption is found for the Si wafer or ZnSe window except for the PDMS on the tip. The measured spectra correspond well to the FTIR spectrum obtained for PDMS (blue dashed line) from the library20. The peaks for PDMS (805, 1020, 1097, and 1268 cm−1) on the Si wafer are larger than those on the ZnSe window because of a larger enhanced field on the Si substrate due to the larger refractive index.
Measured PiFM spectra due to the PDMS contamination on a commercial tip.
PiFM spectra due to the thermal expansion of PDMS on a commercial Au-coated tip measured over a clean Si wafer (red solid line) and over a ZnSe window (black solid line). The blue dashed line shows the FTIR result for bulk PDMS obtained from the library20
PiFM spectra due to tip expansion measured over a ZnSe window with respect to different tip-coating materials
The field enhancement also depends on the radius of the tip and not just on the refractive index, due to the lightening-rod effect. In Fig. 2b, the measured PiFM spectra are plotted with respect to different commercial tips. The refractive indices for an Au-coated tip (PPP-NCHAu), PtSi-coated tip (PtSi-NCH), and bare-Si tip (PPP-NCHR) from Nanosensors Inc. are n ~ 7.7021, 8.2722, and 3.4219, respectively, at 1268 cm−1, with their tip radii guaranteed to be ~30, 25, and 10 nm, respectively. Note that the bulk refractive indices are still applicable until the coating thickness exceeds ten nm23, 24. We can assume that the thickness of the PDMS layer is 1–2 nm for all the tips in this experiment because the three tip materials have a similar hydrophobicity16, 25, 26. Since the PtSi-coated tip has the highest refractive index and a relatively small tip radius compared to the other tips, the PiFM spectrum measured with a PtSi-coated tip shows the strongest signal, as expected. Note that even though the refractive index of the bare-Si tip is far smaller than the other metallic tips, an appreciable signal is still measured due to the small tip radius. Although the refractive index difference between the two metals (Au and PtSi) is relatively small, the measured signals show a clear difference because the tip radius of the Au-coated tip is relatively larger than that of the PtSi-coated tip.
The tip expansion is more sensitive to the substructure of the sample than the sample expansion. The electric field distribution for a layered system of 10 nm polystyrene (PS) film on a Si (or ZnSe) substrate is analytically calculated by implementing the finite dipole method, as plotted in Fig. 3a. The red and blue solid lines show data for the Si substrate and ZnSe substrate, respectively. The field distribution normalized by the incident field E0 shows a monotonic decrease away from the end of the tip followed by a discontinuous drop at the surface of each layer. The degree of the sudden drop depends on the refractive index of each layer. Since the field intensity near the tip end depends more sensitively on the substrate materials, Si or ZnSe substrate in the present case, than does the field inside the substrates, the tip expansion probes the substructure better than does the sample expansion. The normalized field intensity, |E/E0|2, inside the substrate is more rapidly decreased with respect to the PS thickness than at the tip end. The calculated |E/E0|2 at the tip end (red) and just underneath the Si surface (black) is plotted as a function of PS film thickness in Fig. 3b. The calculation details are given in Section 1 of the Supporting Information. The |E/E0|2 at the tip end is gradually decreased and never reaches zero with respect to the PS thickness. However, the normalized field intensity inside the substrate is approximately two orders of magnitude smaller than at the tip end and is rapidly decreased being close to zero for large PS film thickness.
Tip expansion over a layered sample.
a Electric field distributions along the depth direction from the Au-coated tip end to the 10 nm polystyrene (PS) film on the Si wafer (red solid line) and on the ZnSe window (blue solid line). b Normalized field intensities as a function of the PS thickness calculated at the Au-coated tip end (red solid line) and at the Si subsurface (black solid line). c Measured tip expansion spectra with respect to polystyrene film thickness (offset for clarity). d Comparison between experiment and theory for tip expansion as a function of PS thickness at the molecular vibrational peak (1268 cm−1) of the PDMS
Figure 3c plots the measured tip-expansion force spectra for the PDMS contaminant with respect to the PS thickness with offsets for clarity. In the measured spectral range, no significant absorption for the PS film and Si substrate is observed, unlike the results obtained for the PDMS. The refractive index of PS (nPS ~ 1.59 at 1268 cm−1)27 is smaller than that of Si (nSi ~ 3.42 at 1268 cm−1)19. The PiFM force spectra show a monotonic decrease as the PS film thickness increases because the field enhancement decreases with respect to the PS thickness, as shown in Fig. 3b. This leads to a decreased local absorption on the PDMS at the tip end. By gathering the peak values at 1268 cm−1, the change in PiFM signal due to the tip expansion is plotted as a function of the PS thickness along with the calculation result in Fig. 3d. The experimental results correspond remarkably well to the theoretically expected tip expansion given by Eq. (S8). Note that our noise level is ~10 μV, which is far below the measured PiFM signal shown in Fig. 3d.
The enhanced sensitivity due to the tip expansion enables one to characterize the substructure of complex heterogeneous nano/biomaterials, such as polyacrylonitrile-nanocrystalline cellulose (PAN-NCC) nanofiber, which shows a small difference in the refractive indices for the different compositions (nPAN ~ 1.3228, nNCC ~ 1.3929 at 1268 cm−1). It is well known that polyacrylonitrile (PAN)-based carbon nanofibers are low-cost materials with good environmental sustainability, high efficiency and low energy consumption30-32. Compositing nanocrystalline cellulose (NCC) into PAN nanofibers enhances the tensile strength and elastic modulus of the fibers, leading to an enhanced thermomechanical performance. The density and distribution of the NCCs in the PAN are crucial factors that determine the quality of the nanofibers. It is known that the solvent exchange method enables one to distribute the NCCs more uniformly inside the PAN compared to the use of a simple mixing method due to the avoidance of NCC precipitation in the organic solvent32. However, interestingly, the tensile strength is actually decreased while the elastic modulus is increased by the solvent exchange method. This finding might be related to the alignment of the NCCs inside the PAN nanofiber; however, so far, it has been very challenging to visualize the substructure of the PAN-NCC nanofiber because they are organic chemicals with only a small difference in refractive indices.
The PiFM successfully visualizes the NCCs deposited inside the PAN nanofiber with nanoscale resolution, as shown in Fig. 4. The PAN-NCC structure is sketched in Fig. 4a. The PAN (11%)-NCC (3%) in N, N-dimethylformamide (DMF) solution is used to produce electrospun nanofibers, wherein the PAN polymer eventually surrounds the NCC grains32. We used the solvent exchange method to mix the NCC into the PAN-DMF solution, which prevents the precipitation of NCC to maintain a uniform distribution in the solution. Figure 4b presents the FTIR absorption spectra for each material (PAN32, NCC32, glass33, and PDMS20) as obtained from the library data (shown with offsets for clarity). The two strands of PAN-NCC nanofibers on the glass substrate show strong thermal expansion at 1462 cm−1 (CH group vibrations of different modes for the PAN fiber) in Fig. 4d, where the bare NCC debris is not visible. In Fig. 4e, the overall contrast is reversed because of the strong glass absorption in the range 1000–1200 cm−1. The bare NCC debris in the red dash box show a stronger thermal expansion signal compared to the NCCs inside the PAN fiber at 1027 cm−1 (C–OH deformation mode of the NCC). This is because the field on the PAN-NCC nanofiber should penetrate the PAN to reach the NCCs, while the bare NCC debris directly absorb the field. Note that the PDMS on the tip also shows a strong absorption near the NCC resonance such that the NCC grains might show a slightly enhanced contrast due to the tip expansion contrast. The PiFM force spectra in Fig. 4i clearly show how the PDMS on the tip interferes with the sample spectrum. Normalizing the measured spectra on I (bare NCCs) and Ⅱ (PAN-NCC) with the glass signal on Ⅲ in Fig. 4c, one can see that the sample expansion spectra (bare NCC and PAN-NCC) are convolved with the tip expansion (PDMS). Since the glass substrate shows a strong broadband absorption below 1200 cm−1, the peaks for the PDMS at 1020 and 1097 cm−1 are below the glass substrate signal, as expected. Above 1200 cm−1, the main chemical bonds for PAN are well manifested by the representative peaks at 1250, 1360, 1462 and 1640 cm−1. The tip expansion is also evident with a sharp peak observed at 1268 cm−1. Thus, it is not surprising that there is no signal at 1555 cm−1 in Fig. 4f where all the materials (PAN, NCC, glass, and PDMS) are out of resonance, which also implies that the signal due to the dipole force between the tip and the sample is below the noise level.
Hyperspectral PiFM images and spectra for PAN-NCC nanofiber in the mid-IR.
Sketch of the PAN-NCC nanofibers. b
FTIR spectra for bulk PAN, NCC, glass, and PDMS obtained from the library. c
Topography of the PAN-NCC nanofibers. d
PiFM images d
at the PAN absorption resonance (1462 cm−1
near the NCC absorption resonance (1027 cm−1
away from the absorption resonance (1555 cm−1
) of any material, g
at the PAN absorption resonance (1250 cm−1
at the PDMS absorption resonance (1268 cm−1
), which corresponds to a local refractive index contrast for the PAN-NCC nanofiber. i
Measured PiFM spectra convolved by the tip expansion at position (Ⅱ) for the PAN-NCC (red solid line) and at position (Ⅰ) for the bare NCC (blue solid line) in the topography. The data are normalized by the glass spectrum at position (Ⅲ) in the topography. j
Spatial frequency plots for three images measured at the line-cut (Ⅳ) shown in Fig. 4c
As shown in Fig. 3, the refractive index contrast between NCC and PAN can be greatly enhanced through tip expansion and clearly visualized at 1268 cm−1, where no significant absorption is observed due to the sample and the substrate, except for the PDMS on the tip. In this case, the PiFM image represents the local refractive index map of the sample. Since the peak at 1268 cm−1 is a very sharp resonance line due to the Si-CH3 deformation mode in the PDMS20, the refractive index contrast at 1268 cm−1 in Fig. 4h is sharply distinguished from the broad absorption for PAN near 1250 cm−1, where the bare NCC debris are not visible in Fig. 4g. Even though the refractive indices for PAN and NCC are comparable to each other, one can clearly distinguish the NCC grains inside the PAN nanofiber in Fig. 4h because of the significantly different field enhancement at the tip end. The NCCs deposited in the PAN nanofiber show a signal level that is comparable to that of the bare NCC debris in the red dashed box, which is in contrast to the PiFM image at 1027 cm−1 shown in Fig. 4e. This occurs because the PiFM contrast in Fig. 4h solely depends on the refractive index, rather than the resonant absorption of the sample. The NCCs are expected to be deposited inside the PAN approximately 40 nm below the surface with an effective local refractive index of approximately 1.35 on the deposited NCC, which is determined by the calibration of the measured signal [Section 3 of Supplementary Information]. Through the line-cut on the PAN-NCC strand designated by the white solid line (Ⅳ) in Fig. 4c, the alignment of the NCC grains inside the PAN-NCC fiber is clearly manifested. The Fourier transform for the line-cut profile is shown in Fig. 4j. The main spatial frequency of the NCC arrangement is 4.7 × 104 cm−1, which corresponds to a spacing of 213 nm. The periodicity is slightly distinct in topography (black solid line), while it is barely discernable at 1462 cm−1 (blue solid line); it is the most apparent at 1268 cm−1 (red solid line). This result clearly demonstrates the substructure imaging capability of the PDMS-coated tip in PiFM.
The reported tensile strength and elastic modulus for the bare PAN nanofiber and PAN-NCC, which is prepared by using the solvent exchange method, is 17.65 MPa (13.52 MPa) and 0.43 GPa (0.64 GPa), respectively32. The periodic alignment of the uniform NCCs may decrease the tensile strength of the PAN nanofiber because the small volume of the PAN polymer between the adjacent NCCs may be easily broken up against external tension. Conversely, the random size and orientation of the NCCs may increase the tensile strength of the nanofiber. The tensile strength of the PAN-NCC fiber prepared using a simple mix method, which leads to a random size and orientation for the NCCs due to precipitation, is reported to lie in the range of 18.63–23.62 MPa32, which is the highest value obtained compared to samples prepared using the other methods, as expected. Note that the elastic modulus for the sample prepared using the simple mix method (0.34–0.61 GPa) is slightly smaller than that of the sample prepared by the solvent exchange method (0.64 GPa). Since the elastic modulus depends on the quantity of the deposited NCC in the PAN nanofiber, which may be decreased by random distribution, the hardness of the nanofiber is further increased by use of the solvent exchange method.