Research at Harvard

Single Molecule Spectroscopy

Studying a large collection of molecules spectroscopically is a lot like trying to track a single conversation in a crowded room. Instead of hearing exactly what a single person is saying, we find that all we are able to detect is a Gaussian distribution of what appears to be noise. One way to converse with such a mob is to ask everyone to say the same thing at the same time, which is the coherence approach that has enjoyed so much success since the advent of the laser (bull horn). In practice, however, not everyone responds exactly the same and what you end up with is an average over all the people's similar, but not exactly the same, response (inhomogeneous broadening). In the optical portion of this analogy there are things we can do to see through the fog of inhomogeneity, but nothing that relates conveniently to the other side of the analogy. Another approach is to find a way to filter through all the background noise and query people one at a time. This is the essence of single molecule detection—sensitive hearing. If we can do that we find that molecules, like people, have interesting and somewhat unique personalities that are interesting to study in their own right.

So this benefit partially answers the question posed above. Aside from learning more about single molecules themselves and how they behave in a variety of settings, single molecule detection can also be used to study the environment or system in which they reside. Imagine being able to detect, with high specificity, the presence of a single protein in a cell. Let's say that when we start watching the cell the protein is not yet present. And then suddenly it is. How could that be? Well, among other things, cells make proteins. One or more proteins are encoded by a single gene, and when that gene is expressed, a protein is made. The details of how, why, when, and where a certain gene is expressed is still somewhat of a mystery, but single molecule fluorescence is presently being used to demystify gene expression. Other applications include resolving viral pathways, studying transport through porous membranes, and ultra-sensitive chemical and biological sensing, which is crucial for homeland security. This work took place at Harvard in Sunney Xie’s group.

Optical detection of single molecules, both microscopy and spectroscopy, has become a robust technique for investigations in biology, chemistry, physics, and materials science. Whereas, for example, many of the details about the dynamics of local environments in solids, liquids, and biological systems are obscured by ensemble averaging, this is not the case with optical detection of individual molecules. Single molecule spectroscopy began in 1989 when the absorption signal due to a single dye molecule in a low temperature solid was detected. Soon afterwards, single molecule fluorescence was demonstrated and shown to be superior to absorption due to the background-free nature of fluorescence. Since then, advances such as near-field optical detection, far-field confocal microscopy, and wide-field and total-internal reflection microscopy have enabled the detection of single molecules at room temperature.

Single Molecule Detection Beyond Fluorophores

One of the chief limitations to room-temperature single molecule techniques today is that it is limited mainly to fluorescence detection. The number of fluorescent molecules with sufficiently high quantum yield constitutes a small subset of molecular species. My research will focus on extending single molecule detection to a larger subset -- chromophores -- that may be detected by absorption, scattering, and nonlinear scattering.

Single molecule absorption is essentially the same as its ensemble version -- the attenuation in the forward direction as given by Beer’s Law -- only the attenuation is much smaller than shown on left (image taken by R. Thom and D.W. Ward, MIT). The difficulty of single molecule detection by absorption can be seen in considering the required detection sensitivity of an absorption measurement as follows: ΔI/I = σ/A. where ∆I is the intensity absorbed by a single molecule, I is the incident intensity, σabs is the absorption cross-section, and A is the focal area. Typical absorption cross-sections have an upper bound of the order of one or two bond lengths, 1-5 Å2 (1-5x10-16 cm2). For example, a Rhodamine dye molecule at room temperature has a cross-section of 4x10-16 cm2. When excited on resonance with a laser beam focused to the diffraction limit, the expected attenuation would be on the order of 10-7. Presently, the highest sensitivity achieved in an absorption measurement is 10-6. Thus, the chief difficulty in single molecule absorption is overcoming noise limitations that prevent resolution of such small changes in intensity.


My attempts to measure the attenuation of the incident intensity directly, using lock-in detection and low-noise electronics, have shown to be limited by a noise floor that results in a minimum detectable attenuation of 10 -6 of the incident light. This means that the minimum detectable cross-section is on the order of 1x10 -15 cm 2, which corresponds to approximately 5-10 strongly absorbing molecules. Similar detection limits have been observed through attempts at single molecule absorption by other investigators. Research in the area of noise reduction and signal amplification may make room temperature single molecule absorption possible in the near future.

Not every chromophore is a fluorophore. This is the motivation behind optical techniques other than fluorescence. The success of fluorescence single molecule detection is attributable to its background-free nature, but it is not the only source of a background-free signal. Rayleigh scattering and hyper-Rayleigh scattering are also virtually background-free.

Resonant Rayleigh Scattering

Interference of Rayleigh scattering with the unscattered incident light is the source of the attenuation experienced in an absorption measurement. Forward Rayleigh scattered light mixes coherently with the unscattered incident light. As the absorption resonance is approached, the scattered light becomes increasingly phase shifted, resulting in destructive interference -- hence attenuation. There is, however, scattered light in other directions. The radiated field has the shape of a Hertzian dipole. Measuring either perpendicular to the incident light or backwards results in near background-free resonance Rayleigh scattered signal. The downside is that the resonance enhanced scattering cross-section is very small for molecules (4-5 orders of magnitude less than absorption cross-section). For nanoparticles, however, it can be quite large. The absorption cross-section is proportional to the polarizability, whereas the Rayleigh scattering cross-section is proportional to the polarizability squared. In general, scattering begins to dominate over absorption for particle diameters on the order of 10 nanometers.

a) Illustration of the magnitude of Resonance Rayleigh scattering (inset) to Fluorescence in Rhodamine 6G. b) Crystal violet has substantially more Rayleigh scattering than flourescence (inset). Data taken by E. Karp, W. Min, and D.W. Ward, Harvard University.

Hyper-Rayleigh Scattering

Hyper-Rayleigh scattering is scattering by the second order polarization and is proportional to the intensity of the incident light. Scattering, in this case, occurs at twice the optical frequency. Hyper-Rayleigh is performed with detection perpendicular to the incident light, while second harmonic is scattered in the forward and backwards directions. Hyper-Rayleigh scattering is more promising than linear scattering for three reasons. First, the Hyper-Rayleigh signal is more background free than linear Rayleigh scattering since it is not at the same frequency as the incident excitation light (fundamental). Second, the scattering cross-section is proportional to the intensity of the fundamental, so it can be increased by increasing the intensity in order to bring the Hyper-Rayleigh signal above detection limits, provided material damage does not result. Third, liquid environments such as water do not generate any second harmonic in bulk due to inversion symmetry requirements. At first, one might think that SHG is limited because non-inversion symmetry is required, but single molecule detection breaks this symmetry. Symmetry-breaking has been observed in ensemble measurements of surface SHG from centro-symmetric media and in quantum dots, where the spherical shape of the dots establishes inversion symmetry and thus should not radiate second harmonic.

a) Hyperpolarizability as a function of CdSe quantum dot radius. Symmetry breaking results in an increase in hyperpolarizability with decreasing quantum dot radius. Reprinted from M. Jacobsohn et. al., J.Phys.Chem.B,104, 1, 2000. b) Hyperpolarizabilities for several chromophores. Data from W. Min, E. Karp, and D.W. Ward.
Multi-photon emission from single gold nanoparticles (10 nm) in water. Emission peak is around 530 nm. Signal disappears when laser is not mode-locked. Autocorrelation indicates single particle diffusion through detection volume. Gold particles were excited by 800 nm, 150 fs (FWHM), 800 mW average power, 74 MHz ultrafast laser, focussed with a N.A. 1.0, water dipping objective. Data taken by W. Min and D.W. Ward, Harvard University.
Comparison of resonant Rayleigh scattering (RRS) (red) and fluorescence in quantum dots (blue). a) Scanning micrograph of 800 nm fluorescent quantum dots spin-coated on a glass microscope cover-slip. Imaging is slightly out of focus, revealing the quadrature component of the RRS heterodyne image. b) RRS and fluorescence scanning micrographs of single 655 nm fluorescent quantum dots. Digital two-level blinking is evident in the fluorescence image, but not in the RRS image. c) Two-level blinking is clearly demonstrated for the fluorescence time trajectory and histogram (far right of figure) acquired by parking on the single quantum dot imaged in b). Blinking is not observed in the corresponding RRS time trajectory or histogram, which was acquired simultaneously with the fluorescence. The histogram indicates that most of the time the quantum dot was in the dark state. This is not evident in the time trajectory shown because the histogram is constructed with a much higher temporal resolution (1 microsecond) than the time trajectory shown (3 seconds). Data taken by W. Min and D.W Ward, Harvard University.

Pattern Formation

An interesting application of single chromophore detection without intermittent signal is to monitor pattern formation of nanoparticles and molecules in real-time. The organization of a loose collection of molecules into the highly structured, heterogeneous, and transitory environment in a living cell is the motivation behind my interest in pattern formation. One interesting area that is amenable to investigation by single particle detection is drying-mediated pattern formation in quantum dot solutions. The figure below illustrates some of the structures attainable through this process. Under the right conditions, it should be possible to image the structures being formed one particle at a time in real-time. Eventually, it should also be possible to study crystal formation of molecules in real-time.

Pattern formation in drying-mediated quantum dots resolved with DIC microscopy. a) (10x10 microns) Single CdSe nanocrystal formed by rapid evaporation. b) (200x200 microns) Nanocrystals surrounded by CdSe clusters. Crystals only formed under conditions of rapid evaporation, while large scale structures like the quantum ropes shown in c) (10x10 microns) and d) (60x60 microns) formed under conditions of slow evaporation rates. Images taken by W. Min and D.W. Ward, Harvard University.