Single-Molecule Sensitivity in Optical Absorption at Room Temperature

Sensitive detection of condensed matter is of utmost importance in fundamental research as well as cutting-edge applications such as molecular analytics and diagnostics. Until very recently, all existingmethods for the detection of single molecules at room temperature have required highly efficient fluorophores. Here, we demonstrate, for the first time, that single molecules can also be detected via standard modulation-free absorption measurements. Our work extends single-molecule detection to a huge class of materials that absorb light but do not fluoresce efficiently. SECTION Kinetics, Spectroscopy A bsorption measurements are common place in the laboratory because they provide a convenient and label-freemeans for characterizing an unknownmaterial via its absorption spectrum. In a typical experiment, a macroscopic sample is held in a beam of light and attenuates it according to the Beer-Lambert law. To perform such a measurement at the single-molecule level, one would need to compare thepowerof the light beamwith andwithout a single molecule in its path, which can be achieved by translating the molecule in and out of the incident light. Despite its apparent simplicity, such an approachhas provedunsuccessful over the past decades. As a result, alternative techniques based on cryogenic line narrowing, Sagnac interferometry, thin-film interferometry, thephotothermal effect, stimulatedemission, and high-frequency modulation have been explored for sensitive detection of optical absorption. Themajor difficulty in performing single-molecule absorption detection at room temperature stems from the discrepancy between the absorption cross section (∼10-10 cm) and the minimum size of a light beam dictated by the diffraction limit (∼10 cm). A simple estimate suggests that to detect the expected absorption effect directly, one would have to suppress any intensity fluctuations in the detected light beam below the parts-per-million level. However, laser intensity fluctuations are invariably orders of magnitude larger than the desired signal on the time scale of translating the molecule. Furthermore, any background scattering caused by sample inhomogeneities or nearby molecules easily modifies the signal, overwhelming the tiny absorption signature. Here, we show that both of these noise sources can be successfully suppressed by combining balanced detection and an indexmatched sample geometry. We thus achieve single-molecule absorption sensitivity in a simple and direct manner, which can be readily implemented in various applications. Ourexperimental setup resembles a single-moleculevariant of the standard Beer-Lambert type experiment, as depicted in Figure 1a. We split the linearly polarized output of a fiber-coupled helium-neon laser at a wavelength of 633 nm into probe and reference beams. Fiber-coupling was essential to reduce beam-pointing fluctuations aswell as optimizing the mode to achieve the smallest possible beam area. The probe was injected into a home-built invertedmicroscope equipped with a closed-loop piezoelectric stage for sample scanning. Two matched oil-immersion microscope objectives with numerical apertures of 1.4 focused the incident laser beam onto the sample and collected it in transmission. Probe and reference beams were focused onto a balanced photodetector (Nirvana, NewFocus) and adjusted to an intensity ratio of 1:2, respectively. In this way, we were able to reduce laser intensity fluctuations by more than 50 dB down to the shot noise limit. Probe and reference beam intensities then only differed by the presence or absence of a molecule in the focus as the sample was translated. Achieving shot-noise-limited detection below the partsper-million level is a nontrivial task, even with sophisticated laser noise cancellation systems such as balanced photoreceivers, because a typical level of laser intensity noise has to be reduced by about 5 orders of magnitude to ensure singlemolecule sensitivity in absorption. In our particular case, where the log output of the Nirvana detector at maximum loop bandwidth was used, shot-noise-limited fluctuations in the electronic output of the detector were much too small to be detected by a standard data acquisition card. To understand this, it is useful to consider our experimental conditions; 100 μWincident power is accompanied by shot-noise-induced photocurrent fluctuations of about 10 A rms for ameasurement bandwidth of 1 kHz and a photodiode responsivity of 0.45 A/W. Taking into account the shot noise of the reference (power Pref) and the signal (power Psig) beams and that the Nirvana output corresponds to V = -ln(Pref/Psig 1), the resulting fluctuations amount to about 7 10 V. Received Date: October 18, 2010 Accepted Date: October 27, 2010