Drumming up single-molecule beats

Single-particle and single-molecule techniques have become invaluable tools to unravel complex chemical phenomena and biological processes. While fluorescence-based techniques have flourished due to advances in wavelength-dependent optical filtering and highly sensitive detectors (1), other methodologies (Fig. 1) have only recently achieved single-molecule sensitivity limits. A technique based on an oscillating nanodrum offers a promising new direction (2).

Fig. 1.

Various photothermal transducers for single-particle and single-molecule detection. (A) Nanomechanical resonator. A driven high-Q mechanical resonator oscillates at its stress-dependent frequency (red arrow). The photothermal plume produces thermal expansion, relieving stress and altering the resonance frequency measured by a Doppler vibrometer (green). (B) Surrounding medium. The heat plume produces a thermal plume (depicted by black circles) that a probe beam (green) scatters off of (dashed green arrows). The use of a high thermooptic coefficient medium (purple) enables sensitive detection. (C) Optical microresonator. A high-Q optical microresonator is photothermally heated, shifting the resonance condition monitored by a frequency-locked probe laser (green). (D) Atomic force microscope. A photothermal plume introduces thermal expansion in the sample, measured by deflection of the cantilever of the atomic force microscope (green arrow). (E) Surface plasmon resonance. Surface plasmons (green) are excited in a wide-field configuration. The thermal plume alters the local refractive index, shifting the plasmon.

Room-temperature absorption-based techniques, while commonplace at the bulk level, offer significant challenges at the single-molecule level. A major obstacle arises when applying a typical transmission-type experimental geometry to single molecules from the extremely small single-molecule absorption cross-section, with a molecule yielding an almost imperceptible change in a large transmitted signal. At low temperature, cross-sections are substantially larger, a feature which paved the way for the first optical single-molecule measurement (3)—but what to do at room temperature?

Retrieving this minute single-molecule signal from technical noise sources such as laser intensity fluctuations can be accomplished with fast modulation of the signal itself, as is the case with ground-state depletion microscopy (4) or through use of microscopy with balanced detection (5, 6). Spatial modulation microscopy (7, 8) provides a convenient way to encode the transmitted signal at a frequency away from noise sources and has demonstrated robust single-particle sensitivity, although it has yet to reach single-molecule levels.

A subtly different approach is to use a transduction element for the absorbed signal. A particularly successful version takes advantage of the thermal signature of nonradiatively dissipated absorbed light: photothermal microscopy. In general, most photothermal transduction elements can be imagined as micro- or nanoscopic thermometers. The nature and design of photothermal transduction schemes have evolved in search of the most sensitive platform. Most versions rely on the temperature sensitivity of the refractive index of the particle’s or molecule’s surrounding medium. One way of detecting that shifted refractive index is to detect light scattered by the thermal plume enveloping the photothermally heated particle (9). Increased responsivity can be achieved with materials that possess a higher thermooptic coefficient (10). Single-molecule detection was achieved with glycerol immersion (11), and substantially higher responsivities have been demonstrated through the use of thermotropic liquid crystals (12, 13) and supercritical Xe (14, 15) while probing individual particles and polymers, respectively. Another way of detecting the shifted refractive index is to monitor the response of an optical resonator. In this vein, optical microresonators have been used as exquisite photothermal transduction elements, reaching both the single-particle (16) and single-polymer detection limits (17) without the use of a medium with a high thermooptic coefficient. Alternatively, shifted refractive index can be monitored through its effect on the surface plasmon resonance of a gold film in a wide-field imaging mode (18). In all these cases, the thermal response is dependent on the thermooptic coefficient, but recently, other methodologies for photothermal transduction have been used.

Nanomechanical systems have recently been developed for efficient photothermal transduction. In one version, atomic force microscopy is harnessed to measure the thermal expansion of an illuminated sample (19, 20). Although only demonstrated in the infrared while probing vibrational transitions with very small absorption cross-sections, the technique has demonstrated sensitivity to extremely small thermal fluxes. In a different mode, pioneering work by Boisen and coworkers (21, 22) enabled single-particle photothermal sensing via nanomechanical resonators, whereby the stress in the resonator is relieved by thermal expansion driven by pump light absorbed and thermalized at the particle, producing a mechanical resonance frequency shift. In PNAS, Chien et al. (2) advance this technique, thereby expanding the single-molecule absorption microscopy toolkit.

Chien et al. (2) use the high-Q mechanical resonances (∼10⁶) of a silicon nitride drum under vacuum as the thermal sensor. As above, the resonances are shifted by the thermal-expansion stress relief after photothermal excitation. Here, a piezoelectric actuator is used to excite a mechanical resonance, which is monitored with a Doppler vibrometer and lock-in amplifier, with the measured frequency reapplied to the piezoelectric actuator in a phase-locked loop. The impressive sensitivity of their device stems from three major sources. First, the high-Q resonances provide robust frequency resolution of ∼46 ppm, measured at an Allan deviation optimized integration time constant of 40 ms. Second, the thin (50 nm) drum reduces the device heat capacity, increasing the temperature rise from a small-input thermal power. The surrounding vacuum further plays a vital role by mitigating heat losses, allowing for heat to build across the entire resonator. Together, these two points allowed for the robust detection of individual Au nanoparticles down to 10 nm, even at wavelengths far detuned from the plasmon absorption peak. Third, seeing that the thermal responsivity is inversely proportional to stress, the stress of the silicon nitride is reduced from 30 to 0.8 MPa by the creation of a thin layer of silicon oxynitride, thereby increasing the thermal response by ∼70 times. This increased sensitivity puts Chien et al. well within range to image single molecules.

Lastly, Chien et al. (2) demonstrate single-molecule sensitivity by imaging a dye molecule, Atto 633. Upon identifying photothermal signals consistent with expected single-molecule values, they separately acquired fluorescence data in ambient conditions and observed single-step photobleaching and blinking, hallmarks of single-molecule behavior. They achieved this single-molecule sensitivity with a high signal-to-noise ratio of ∼70, especially impressive considering that only ∼38% of the absorbed light is thermalized to produce signal.

This technique demonstrates several exciting features beyond single-molecule sensitivity. One of the most striking features of this experiment is that most of the critical components, such as the Doppler vibrometer and silicon nitride drums, are commercially available. Further, their long-working-distance objective and modest N.A. are attractive attributes compared with the high-N.A. oil-immersion objectives and consequent small working distances used in all other single-molecule absorption techniques to date. Lastly, their setup circumvents many experimental difficulties associated with other absorption-based techniques by involving only a single objective and beam, by offering a relatively large sensing area (∼160 × 160 μm²), and by enabling an easy way to deposit molecules via spin-coating. However, like all techniques, certain limitations also exist. Most important for this methodology, the need for vacuum (though not high vacuum) limits applicability. Ambient conditions will significantly reduce signal strength due to the considerable lowering of mechanical quality factors due to dampening by air, which may in part be overcome (23). Thermal losses to air may also be significant. Thus, ambient and condensed-phase experiments will be challenges to achieve, but this promising method has ample room to grow.

Overall, the work by Chien et al. (2) is a creative addition to the single-molecule absorption toolkit. Their unique and simple experimental design adds to the diverse and growing geometries for sensitive absorption microscopy and spectroscopy. Significant obstacles still remain before single-molecule absorption can be a robust and widely applicable tool, but the boundaries are continually being eroded.