1355 Research Article Received: 23 January 2009 Accepted: 1 June 2009 Published online in Wiley Interscience: 31 July 2009 (www.interscience.wiley.com) DOI 10.1002/jrs.2382 Optimising tip-enhanced optical microscopy Andrew Downes, * Rabah Mouras, Meropi Mari and Alistair Elfick We discuss the range of tip-enhanced optical microscopies – namely mapping scattered light, fluorescence, Raman and coherent anti-Stokes Raman scattering (CARS) signals on the scale of 10– 20 nm. Through the use of measurements and finite-element simulations, we explain what the optimum tip materials and sizes are and illumination wavelengths to use for all these modes of imaging. We also show that the observed limit of 50–100 μW of illumination power is due to a temperature rise of tens of degrees beneath the tip apex, and is most likely linked to the evaporation of water from the tip apex region. This has two important implications – that the tip is surrounded by water even when imaging ‘in air’, and that the tip and sample are both heated when imaging. With emphasis on Raman and CARS, we discuss how both these types of tip-enhanced optical microscopy can be made about as fast as their diffraction-limited counterparts. Copyright c  2009 Her Majesty the Queen in right of Canada. Published by John Wiley & Sons. Ltd. Keywords: tip-enhanced Raman spectroscopy; SNOM; coherent anti-Stokes Raman spectroscopy Introduction Standard optical microscopy is limited in its spatial resolution to around half the wavelength of light, due to diffraction. One way to circumvent this limit, realised in the 1980s, is the scanning near- field optical microscope (SNOM). [1] An optical fibre is drawn down to a much smaller diameter, or coated with metal and a nano-scale aperture drilled in the film, and scanned above a surface. A resolu- tion of 50 nm could be routinely achieved, which is limited by the achievable size of the light source at the tip apex. Improvements to this resolution by superior manufacturing techniques would be hampered by the transmission through an aperture having a fourth power dependence on the aperture diameter. [2] A later opti- cal variant of scanning probe microscopy was the photon emission scanning tunnelling microscope. [3] Applying a voltage to a metal tip, light emission was observed from samples (which need to be electrically conducting), because a localised surface plasmon was excited at the end of the tip. This oscillation of electrons within the metal tip at optical frequencies produces an electric field in a tiny spot directly beneath the tip apex – effectively a nano-sized light source. Material contrast on samples was observed with ∼2-nm resolution, [4] and atomic resolution imaging was also achieved. [4] To extend this to non-conducting samples, the same plasmon in the metal tip can be excited by focussed light, instead of by tun- nelling electrons. Named ‘apertureless’ scanning near-field optical microscopy, [5,6] the light scattered from an atomic force micro- scope (AFM) tip apex placed within a diffraction-limited focal spot is measured when a sample is scanned in close proximity to the tip apex. This method can give material contrast in images, with approximately the same spatial resolution as topographic images. Variations in surface topography also dramatically affect the to- tal amount of light scattered – a convex surface curvature scatters more light, so chemical characterisation by mapping the amount of scattered light is limited to atomically flat surfaces. Contrast is nor- mally weak, unless there is large change in refractive index between materials for the wavelength of light used. [6] In order to achieve a good degree of contrast between materials, whether in standard diffraction-limited microscopy or tip-enhanced, some kind of spec- troscopy is required to identify the type of material beneath the tip. In this paper, for each form of spectroscopic optical imaging, we will discuss how best it can be achieved on the nano-scale with tip-enhanced optical microscopy. Using our finite-element modelling and experimental results, and considering published results from leading groups, we show what the limiting factors to each imaging method are, and which illumination conditions and type of tip should yield the best signal levels. The amount of light scattered from a 10-nm spot beneath a sharp tip can be greatly improved by the choice of tip materials and wavelengths. In the best case, enhancements in sample signal levels per square nanometre of order 10 8 have been observed by close proximity to a silver-coated AFM tip. [7] Imaging in contact mode AFM, a large enhancement is required so that the signal from beneath the tip apex dominates over that from the diffraction- limited spot. The sharpest AFM tips can have a radius as good as 1 nm, but they are made of silicon dioxide, which is a material that does not give a useable enhancement. Silicon tips will generally have a radius of around 10 nm, while coating with gold or silver will increase that radius to 20 – 50 nm. This is governed by the minimum thickness of metal that is required to make a robust tip – sustained or intermittent contact with a hard surface will blunt tips of smaller radii. Any form of polarised light microscopy is not applicable to tip-enhanced microscopy, as the electric field beneath the tip apex is highly polarised in the tip axis (perpendicular to the surface of interest). So, in order to achieve sensitive material contrast, tip-enhanced microscopy techniques must involve spectroscopy. Fluorescence imaging is the most widely used form of microscopy in the biomedical sciences. The large enhancement for fluorescent materials and molecules is to some extent cancelled out by the quenching of fluorescence due to close proximity to a gold tip. Nevertheless, a net fluorescence enhancement factor of ∼10 by the tip is achievable. [8,9] This means that the background ∗ Correspondence to: Andrew Downes, Institute for Materials and Processes, The University of Edinburgh, Mayfield Rd, Edinburgh EH9 3JL, United Kingdom. E-mail: andy.downes@ed.ac.uk Institute for Materials and Processes, The University of Edinburgh, Edinburgh EH9 3JL, United Kingdom J. Raman Spectrosc. 2009, 40, 1355–1360 Copyright c  2009 Her Majesty the Queen in right of Canada. Published by John Wiley & Sons. Ltd.