Near-field assisted white-light interferometry for 3D nanoscale imaging

In general, existing microscopy methods can be separated into three categories, namely, optical microscopy, electron microscopy, and scanning probe microscopy (SPM). Optical microscopy provides a non-invasive, high-speed, relatively low-cost, and environmentally compatible technique for the observation of objects down to the sub-microscale. The imaging resolution of optical microscopy, however, is confined to about half of the illuminating wavelength (in the lateral dimensions), and is several times worse in the axial dimension because of the diffraction limit. Alternative, high-resolution microscopy methods in which the diffraction limit can be overcome are therefore required.

In recent work, super-resolution fluorescence microscopes have been successfully advanced to nanoscale resolutions in life science studies.^1–3 With these microscopes, however, there is a loss of structural information that restricts their potential applications in semiconductor wafer inspection, the exploration of nanoscale materials/structures, and the non-invasive study of dynamic processes.^{4, 5} Furthermore, there are few microscopy methods (except for SPM) with which super-resolution imaging can be achieved in all three dimensions, without the use of fluorescent dyes. Indeed, there are also some limitations that prevent the application of SPM in more research fields. Such limitations of SPM include the invasiveness (because of scanning tips), poor time efficiency, and tip (broadening) effects. In other studies, efforts have been made to overcome the lateral diffraction limit of microscopy techniques in a more convenient way. It has been demonstrated that dielectrics with micro- or wavelength-scale spherical structures can be used to realize sub-diffraction-limited resolution (from about 200–50nm).^{6, 7} Observations of sub-cellular structures⁸ and 75nm adenoviruses⁹ have also been successfully demonstrated with microsphere superlenses. When these lenses are combined with a scanning laser confocal microscope, they can be used to resolve structures with feature sizes of about 25nm.¹⁰

In our previous work, we showed that a microsphere superlens can be used to enhance the resolution—to the nanoscale—of a microendoscopy approach that was based on a graded-index lens.¹² In our more recent work, we have demonstrated a new 3D super-resolution microscopy technique. This method—known as near-field assisted white-light interferometry (NFWLI)—is illustrated in Figure 1. For this approach, we take advantage of the topography acquisition capability of white-light interferometry and the lateral near-field imaging principles of the microsphere superlens.¹¹

Figure 1. Schematic illustration of the near-field assisted white-light interferometry (NFWLI) technique. (a) The NFWLI system is constructed by integrating a microsphere superlens into a Linnik white-light interferometer. The system consists of an analyzer (A), quarter-waveplate (Q), beam splitter (BS), polarizing beam splitter (PBS), polarizer (P), Köhler illumination (KI), mirror (M), piezoelectric ceramic scanner (PZT), and two objectives (O1 and O2). (b) Image frames are recorded during PZT linear scanning along the light path, as indicated by the red arrow. (c) Analyses of the recorded interference images are used to construct the 3D morphology of the measured objects. (d) The final constructed (super-resolution) 3D morphology.¹¹

In most situations, the vertical information for a sample cannot be constructed directly by analyzing interference images of evanescent waves (because of the lack of phase information along the vertical direction). In our NFWLI technique, however, we use a microsphere superlens to convert the evanescent waves into propagating waves. We also reconstruct the phase information along the vertical direction from the evanescent waves. By analyzing the recorded images that are generated by the interference of the converted propagating waves and the reference light beams, we can construct the morphology of a sample (with sub-diffraction-limited resolution) in all three dimensions.

We have tested our NFWLI method by observing structures within central processing unit chips. These chips have lateral dimensions of about 50nm and a vertical dimension of about 10nm. We conducted these observations (see Figure 2) without the use of fluorescent dyes and within 25s, i.e., 40 times faster than for atomic force microscope (AFM) measurements. We find—see Figure 2(d–f)—that the profiles we obtained with our NFWLI technique are consistent, and that they are similar to those we obtained via AFM scanning. Our results also demonstrate that the NFWLI method can operate well in both air and in water.

Figure 2. Examples of using NFWLI for the inspection of semiconductor chips. Observations are made of (a) 90nm features, (b) an array of nanodots, and (c) 65nm structures. Scanning electron microscope images (i), virtual images generated with a microsphere superlens (ii), atomic force microscope (AFM) scanning images (iii), and the 3D morphology constructed by NFWLI (iv) are shown in each case. (d–f) Comparisons of the cross sections—marked by the blue and red lines in (a–c)—obtained via AFM scanning and NFWLI.¹¹

In summary, we have developed a new technique—near-field assisted white-light interferometry—that provides significant advantages over standard scanning probe microscopy for 3D nanoscale morphology profiling. For example, our NFWLI approach has a much improved time efficiency for large-area nanoscale morphology acquisition (i.e., with a field of view that is on the order of micrometers). In addition, our method is non-invasive because the microsphere superlens does not need to be in direct contact with the samples, and it produces reduced repeatability errors (i.e., it does not require scanning probe tips that may wear out during repeated scans and thus cause surface profiling errors). We believe that NFWLI therefore has great potential as a practical 3D super-resolution microscopy method for label-free, high-speed, non-invasive, and environmentally compatible imaging. Indeed, all these properties should facilitate 3D nanoscale imaging of integrated circuit devices, nanomaterials, and biological entities (e.g., cell membranes, viruses, or biomolecules) in the future. The observable area of NFWLI, however, is currently confined by the microsphere's field of view, and we are therefore currently exploring several effective methods to overcome this limitation.