Infrared-spectroscopic, dynamic near-field microscopy of living cells and nanoparticles in water

Abstract

Infrared fingerprint spectra can reveal the chemical nature of materials down to 20-nm detail, far below the diffraction limit, when probed by scattering-type scanning near-field optical microscopy (s-SNOM). But this was impossible with living cells or aqueous processes as in corrosion, due to water-related absorption and tip contamination. Here, we demonstrate infrared s-SNOM of water-suspended objects by probing them through a 10-nm thick SiN membrane. This separator stretches freely over up to 250 µm, providing an upper, stable surface to the scanning tip, while its lower surface is in contact with the liquid and localises adhering objects. We present its proof-of-principle applicability in biology by observing simply drop-casted, living E. coli in nutrient medium, as well as living A549 cancer cells, as they divide, move and develop rich sub-cellular morphology and adhesion patterns, at 150 nm resolution. Their infrared spectra reveal the local abundances of water, proteins, and lipids within a depth of ca. 100 nm below the SiN membrane, as we verify by analysing well-defined, suspended polymer spheres and through model calculations. SiN-membrane based s-SNOM thus establishes a novel tool of live cell nano-imaging that returns structure, dynamics and chemical composition. This method should benefit the nanoscale analysis of any aqueous system, from physics to medicine.

Figure 1

SiN-membrane-based infrared probing into liquids, (a) nano-FTIR spectroscopic near-field microscopy layout using light focused on a metallic AFM tip to induce an intense near-field spot which is as narrow as the tip and extends a similar distance below the tip apex (white patch)^,. A sample scanned below the tip, through the near-field spot, modifies the back-scattered light which is detected at the detector (D), via a Michelson interferometer that employs a beam splitter (BS) and a movable reference mirror (R) to form highly resolved infrared s-SNOM images. When the light is broad-banded, e.g., from a DFG (difference-frequency generation) source, a nano-FTIR spectrum can be measured at any sample position, in both scattered amplitude and phase. In addition, the tapping tip (double arrow) may sense membrane surface deformations via the AFM circuitry. (b,c) Liquid-sample cell consisting of a small metal container (black) carrying a perforated Si chip (grey) that is closed by a 10-nm thin SiN membrane (pink), loaded by drop-casting a suspension and waiting for particles or cells to settle and adhere to the membrane (top), then sealed (mid) and finally turned upside down (bottom) for microscopy so that the probing tip touches the upper surface of the membrane. When particles or cells adhere to its lower surface, the membrane can become locally distorted, as illustrated, and thus can enable a mapping of the field of adhesion forces.

Figure 2

s-SNOM-observed membrane wetting, (A) infrared amplitude s₂ and (B) topography z nanoimages of a free-standing 20-nm thick SiN membrane on its Si frame (scale bars 2 µm), as-received exhibits a partial wetting in the lower left corner of the 20 × 20 µm² hole (Norcada NX5002Y), (C) after UV hydrophilisation and complete wetting, line profiles are extracted from images (not shown) of infrared amplitude s₂ (red), topography z (black), and mechanical phase φ_mech (blue) across an edge (at x = 0) of the frame into the membrane.

Figure 3

Nano-FTIR spectra of H₂O and D₂O compared to theoretical calculations, below a free-standing, 15-nm SiN membrane stretched over a 20 × 20 µm² hole (Norcada NBPX5002YZ-HR), of infrared phase φ₃ (round data points with black average curves) exhibiting the bending-vibrational resonances of H₂O (red) or D₂O (blue), at nominally 1644 and 1209 cm⁻¹, respectively (see Supplementary Fig. S2 for simultaneously measured amplitude spectra, and for an H₂O/D₂O mixture). The theoretically predicted curves (square dots) for H₂O and D₂O under 15-nm SiN are calculated using the finite-dipole model of tip-confined near-field interaction extended to multi-layer objects (“Methods”), assuming a 100 nm tip radius and 70 nm tapping amplitude, plotted after multiplication by 1.78.

Figure 4

Calibrating the depth sensitivity of s-SNOM imaging and nano-FTIR spectroscopy by probing a 10 µm diameter PMMA sphere (PMMA-F-10.0 from micro-particles.de) adhering below a 10 nm SiN membrane (Norcada NX5025Z) in water, (A) infrared amplitude s₂ image with marked outline of the sphere, (B) sketch of taking a line scan of spectra in 48 nm sequence along a radius x of the projected sphere, resulting in (C) nano-FTIR phase φ₂ spectra showing that the PMMA resonances at 1730 and 1445 cm⁻¹ fade with x as the water thickness d increases while the 1644 cm⁻¹ resonance of water increases, (D) peak heights (data points) determined from c for the most prominent lines of PMMA and water, black and blue, respectively, vs the water depth d (see Supplementary Fig. S3b). Theoretically predicted full curves in (D) are calculated using the multi-layer model of near-field interaction (“Methods”), assuming a 100 nm tip radius and 70 nm tapping amplitude, and plotted × 1.78 in order to match the experimental PMMA peak height at sphere centre.

Figure 5

Adhesion-localised, but mobile living E. coli cells, sequential s-SNOM infrared amplitude s₂ images as cells adhere, grow and move in the medium under a 15 nm SiN membrane (Norcada NBPX5002YZ-HR), at times (min) indicated, color scale as in Fig. 4A, image size 7.5 × 5 µm².

Figure 6

Living E. coli cells characterised by s-SNOM and nano-FTIR, (A) infrared amplitude s₂ image of a preparation as in Fig. 5 exhibiting two bright, round cells which also appear in the simultaneously recorded mechanical images of (B) topography and of (C) mechanical phase φ_mech, scale bars 2 µm. (D) Nano-FTIR phase φ₃ spectra (data points) with averages (lines, offset 5° each for clarity), taken on bright round cells (black), lengthy cells (red), or between cells (blue) of (A), directly prove significantly different spectra, and thus chemical content, when probing a side vs a front location of a cell's envelope.

Figure 7

s-SNOM images and nano-FTIR spectra of living A549 cancer cells, (A) infrared amplitude s₂ image taken 5 h after drop-casting the cell suspension on the 10-nm SiN membrane (250 × 250 µm², Norcada NX5025Z), (B) simultaneously registered topography z and (C) AFM phase φ_mech images exhibiting patches interpreted as adhesion footprint, scale bars 5 µm; (D) nano-FTIR phase spectra φ₃ at a few arbitrary positions within a bright peripheral spot (green data points, taken in 4 min each), or in between such spots (red), or outside the cell's footprint (blue); average spectra (black curves) and their calculated negative 2nd derivatives (coloured curves), with frequencies stated for the peaks in the green curve together with their assignments as P-protein, L-lipid, T-tyrosine, (E) infrared amplitude s₂ images (11.6 × 15.1 µm²) of another A549 cell, corrected for sample drift, continuously acquired for 7:27 h in 22 min sequence starting 5 h after drop-casting, of which the first three and the last two are shown.