We report the imaging of sub-diffraction limited features using an optical probe generated by focusing a round spot at one wavelength, λ1 = 405nm, and a ring-shaped spot at a second wavelength, λ2 = 532nm, onto a thin photochromic layer that coats the nanostructures. Illumination at λ2 turns the photochromic layer opaque to λ1 everywhere except at the centre of the ring, where the illumination at λ1 penetrates and probes the underlying nanostructure. We confirm that this optically confined probe increases image contrast and is able to resolve features smaller than the far-field diffraction limit. Furthermore, by using an array of dual-wavelength diffractive microlenses, we demonstrate the feasibility of parallelizing this approach. Compared to previous approaches, our technique is not limited to fluorescence imaging.
© 2010 OSA
Diffraction limits the spatial resolution of optical imaging in the far-field to about half the wavelength . This diffraction barrier effectively prevents imaging of nanostructures with visible light. Traditionally, nanoscale imaging or nanoscopy was carried out via electron and X-ray microscopy. These high-energy particles impart considerable damage to fragile organic samples. Moreover, the vacuum environments within such systems are hostile to certain materials.
A major advance in far-field optical nanoscale imaging was achieved by stimulated emission depletion (STED) microscopy and related techniques [2,3]. In STED microscopy, a focused pulse of light is used to excite fluorescence, while a subsequent red-shifted pulse, shaped like a ring, is used to de-excite the fluorescence everywhere except in the near vicinity of the ring’s center. Although resolution below 20nm has been demonstrated, such techniques require relatively high light intensities in the ring-shaped pulse. This requirement arises from the metastable nature of the fluorescent (excited) state. In particular, high intensities are required to stimulate non-radiative transitions to the ground state before emission occurs. Parallelization of such approaches is therefore difficult. In this case, fluorescent markers with high photostability are required. Although recent advances allow for lower intensities , specialized switchable fluorophores are required.
Sub-diffraction-limited imaging is also attained by the temporal separation of fluorescence from closely spaced molecules [5–7]. This approach relies on the stochastic switching between a fluorescent and a non-fluorescent state. Collecting multiple photons emitted from a single fluorophore allows for nanometric localization of the molecule. Images are then generated by tens of thousands of imaging and activation cycles, followed by post-processing of the collected data. A large number of imaging cycles are required to probe all the fluorophores within the field. Most significantly, both these approaches are limited to fluorescence imaging.
General imaging modalities are enabled by near-field scanning-optical microscopy, which uses a sharp tip or a subwavelength aperture in close proximity to a sample [8,9]. Light is confined to sub-wavelength dimensions in the vicinity of the tip or aperture. The signal is exponentially dependent on the gap between the probe and the sample, and extremely precise control of the gap is required. As a result, this is a slow, serial process that is challenging to parallelize.
Photons play a key role in imaging inorganic samples [10,11]. Specifically, they provide contrast mechanisms that are unavailable with other particles. In the case of long-wavelength photons, their phase may be readily exploited to gain additional structural information [12,13]. Nevertheless, in most cases, poor-quality optical images require verification by high-resolution electron microscopy . Even for inorganic samples, this can be problematic due to the slow imaging speed, tedious and destructive sample preparation, and artefacts arising from environmental electromagnetic disturbances and charge buildup in the sample.
Here we demonstrate nanoscopy using optically confined probes via a technique that we refer to as absorbance modulation. In absorbance modulation, a thin film of photochromic material is placed on top of the substrate to be imaged. The absorbance of photochromic materials can be decreased by exposure to one wavelength, λ1, and restored by exposure to a second wavelength, λ2. When a focused spot at λ1 and a ring-shaped spot at λ2 illuminate the photochromic film, a dynamic competition between the two photo-reactions ensues. This results in a transparent region of deep sub-wavelength dimensions localized around the null of the λ2 intensity distribution. Light at λ1, subsequently transmits through this region, probing the sample underneath (see Fig. 1A ). The size of this optically confined nanoscale probe is not limited by diffraction, but by the photostationary state of the photochromic overlayer, which is dictated by its material parameters and the intensity ratio at the two wavelengths . In the past, we have successfully applied absorbance modulation to nanopatterning [16–18] and created lines whose widths were almost as narrow as one-tenth of λ1 .
This spatially-confined probe beam can excite a signal in the sample from a nanoscale volume because the optical near-field decays exponentially beyond the photochromic film. Stepping the sample relative to the optics and tabulating the signal as a function of position build up an image similar to a conventional scanning-optical microscope. The mechanism that generates the nanoscale probe, i.e., absorbance modulation, is distinct from that which generates the signal, i.e., the interaction of the probe with the sample. As a result, the two mechanisms can be independently optimized and high spatial resolution is achieved at low light intensities. As opposed to previous approaches, nanoscopy using absorbance modulation is not limited to fluorescence imaging. All results presented here are based on transmitted-light images and not fluorescence. Absorbance modulation is a surface imaging technique and may be useful for studying surface pathways, which currently use total-internal-reflection microscopy (TIRF).
In order to achieve practical throughputs, multiple probes are necessary. Absorbance modulation is relatively easy to parallelize because high intensities are required and only far-field optics is used. In order to demonstrate this, we designed and fabricated a small array of phase-diffractive microlenses, each of which simultaneously generates in its focal plane a ring-shaped spot at λ2 and a round spot at λ1 [18,20]. The spacing between the microlenses is chosen such that the signals generated from their focal regions are easily separated by the collection optics. A large array of such microlenses can provide massive parallelism and hence high throughput .
In our experimental system, a 405nm GaN diode laser (Power Technology, Inc) and a 532nm diode-pumped solid-state laser (Shanghai Dream Lasers, Inc.) illuminated the array of microlenses. Neutral density filters were used to adjust the intensities in the two beams independently. In this preliminary experiment, we used an array comprised of 10 microlenses. There were five pairs of microlenses each of NA ranging from 0.83 to 0.22. They were spaced by a distance of 140μm in the X and 160 μm in the Y direction (Fig. 2 ). The details of their fabrication and optical characterization were described earlier . The sample, which consisted of features on a glass slide, was affixed on top of a piezoelectric scanning stage (Physik Instrumente). The stage had an open aperture through which the light at λ1 scattered in the forward direction from the sample was collected by a CCD camera (Sony). The signal from the different microlenses are separated on the CCD plane and hence collected in parallel similar to a system described earlier . Scattered light at λ2 was filtered out. A schematic of our optical system is shown in Fig. 2. Before imaging, the surfaces of the microlens array and that of the sample were made parallel, and the microlens array was focused onto the sample using piezoelectric actuators following a procedure we developed previously for lithography . At each scan location, the sample was illuminated simultaneously by both wavelengths, and the position and signal data were collected. The signal was acquired after a delay of 1s at each location in order to ensure that the photostationary state was reached (see section 4). The sample was stepped in a raster fashion. A step-size of 25nm was used in the results presented here unless noted otherwise. A bilinear interpolation algorithm was used to smooth the data with a pixel size of 5nm unless noted otherwise. In this implementation, the 1s delay during signal acquisition limits the overall imaging speed. In the future, this can be avoided by increasing the incident light intensities and using thermally stable photochromic molecules. Therefore, each pixel of the image took a little more than 1s to acquire. Although these preliminary experiments were slow, they demonstrate the feasibility of parallelization and potential for faster imaging.
The intensity profiles of the beams at the two wavelengths were measured with a CCD camera. By measuring the total power within these beams, we deduced the power density incident on the microlens array. Since the area covered by the microlenses was much smaller than that of both the beams, we assumed that intensity incident on the microlenses was uniform. To compute the peak intensity in the focal plane of each microlens, we first simulated the normalized focal intensity distribution using a scalar Fresnel-Kirchoff diffraction formulation. This simulation has been verified by experiments previously . Then, the peak intensity, Ipeak, was computed as
We imaged a resolution standard containing dense line/space patterns. These gratings of period 500nm were patterned on a glass slide using interference lithography, and transferred into a 50nm-thick film of chromium using a lift-off process. We used an azobenzene polymer as the photochromic material for our experiments [see Fig. 1(B)] . This polymer was synthesized using a literature process [22,23]. The sample was spin-coated with a 220nm thick film of the polymer. Figure 3(A) shows the images of a 500nm period line/space pattern obtained using a microlens of numerical aperture (NA) = 0.83. On top is the image taken with only λ1 ( = 405nm) illumination. The bottom image is of the same region but taken with both λ1 and λ2 ( = 532nm) illumination. Even at relatively low peak intensities at λ2 (0.4W/m2), it is clear that the image contrast is improved. Figure 3(B) shows the averaged linescan of the images in Fig. 3(A). The solid line shows the scan with only λ1-illumination, whereas the dashed line depicts the scan with both λ1 and λ2 beams. Absorbance modulation clearly increases the image contrast. The contrast, K is defined as
We created an alternative resolution standard by dispersing gold nanoparticles (British Biocell International) on a glass slide. Gold nanoparticles of various nominal sizes in water (with sodium citrate as a buffer) were dispersed either by a puddle-drop method or by spin-casting onto a clean glass slide. The slide was dried in an oven at 92 degrees Celsius for about 20 minutes. Subsequently, the slides were spin-coated with a 220nm thick film of the azobenzene polymer. Figure 4(A) shows a schematic of the parallel-imaging system and the separation of the multiple signals on the camera. Figures 4(B)–4(D) show images of three different regions of one such slide on which 10nm gold nanoparticles were dispersed. The images in the top row were taken with only λ1 = 405nm. The images in the bottom row depict the corresponding regions that were imaged with both the wavelengths. The three images in each row were collected simultaneously with three different microlenses, thereby demonstrating parallelism. The images in (B) and (C) were obtained with two different microlenses, each of NA = 0.83, while those in (D) were obtained with a third microlens of NA = 0.55. The comparison of the corresponding top and bottom images in all three cases confirm that absorbance modulation is able to resolve structures that are otherwise not discernible. For example, in Fig. 4(D), structures as small as 200nm (full-width at half-maximum) are clearly identified. At NA = 0.55 and λ = 405nm, the Abbé diffraction limit is approximately 370nm. Clearly 200nm structures are not resolved without absorbance modulation. Note that the signal represents light that transmits through the gaps between clusters of nanoparticles on the sample.
Figure 5 shows images of nanoparticles on glass when the intensity ratio of λ2 to that at λ1 is large enough to overcome the far-field diffraction barrier. Figure 5(A) is a schematic of the sample being imaged. The sample consisted of 100nm gold nanoparticles (overcoated with 220nm of the azobenzene polymer). The images were taken with a microlens of NA = 0.7 with only λ1 (B), and with both λ1 and λ2 (C). Gaps between the nanoparticle clusters as small as 78nm are resolved when both wavelengths are used. Figures 5(D)–5(F) show images of the same region on a glass slide dispersed with 10nm gold nanoparticles taken at different ratios of the peak focal intensity at λ2 to that at λ1 (I2peak/I1peak). Figure 5(E) shows a magnified view of the region denoted by a white square in Fig. 5(F). Clusters of nanoparticles are evident and some of these are denoted by black dashed lines for illustration. Notably, two (possibly) single nanoparticles spaced by a distance of about 40nm are clearly resolved. Figure 5(H) shows the intensities along the white lines in Figs. 5(D)–5(F). As the ratio of intensity at λ2 to that at λ1 increases, the size of the probe decreases and finer structural details such as a 50nm gap between nanoparticle clusters are revealed.
4. Imaging with a photo-transient probe
When the photochromic overlayer is illuminated by the two wavelengths of light, the ensuing photoreactions occur at rates that increase with the incident intensities . As light propagates through the layer, the local absorbance evolves as well. Finally, a dynamic equilibrium between the two reactions is reached, which is referred to generally as a photostationary state. Once the photostationary state is reached, the absorbance of the photochromic layer (and hence, the width of the transmitted light at λ1) doesn’t change with time. This provides a robust probe, which can then be used for imaging. This phenomenon is illustrated schematically in Fig. 6 . Figure 6(A) shows the normalized intensity distribution of light at λ1 within the photochromic layer, when this layer is illuminated by a focused node at λ2 and a focused spot at λ1. As time progresses, the λ1 spot “punches” more and more through the photochromic layer. Figure 6(B) plots the normalized intensity distribution of λ1 (or the probe) at the bottom of the photochromic layer. As time progresses, the mainlobe of the probe increases and the sidelobes gain in strength. Therefore the probe is smaller at earlier times. This raises an interesting possibility of using this smaller photo-transient probe for higher-resolution imaging.
We explored this effect by varying the delay from when the stage reaches a specific position and when the signal is acquired on the CCD. Figure 7 shows the images taken with absorbance modulation of a glass slide dispersed with 100nm gold nanoparticles. When the delay increased from 1s to 5s, the contrast as well as resolution of the images became worse. This indicates that the probe size increases as the azobenzene layer evolves towards its photostationary state under these illumination conditions.
To study this effect further, we varied the intensity of the λ2 beam and acquired images with various programmed delays. It was observed that the effect of the delay is negligible at high intensities of either λ1 or λ2. This is expected since the azobenzene layer evolves towards its photostationary state at a rate that is dependent on the incident intensity . In Fig. 8 , we show images of the same region on a resolution test sample, but with different delays of 0s (no programmed delay) and 0.5s. Since the peak intensity at λ2 is high (334W/m2), the photostationary state is reached in much less than 0.5s and the images are similar.
For the images reported in Figs. 3–5, we used a delay of 1s. It was observed that at the intensities of λ2 and λ1 used in those images, the effect of the photochemical transients could be safely ignored when a delay of 1s was incorporated. Nevertheless, the results presented in this section suggest the possibility of imaging with a transient probe, whose size can be significantly smaller than the size of the corresponding probe in the photostationary state. However, such imaging at high resolution would require a relatively sensitive detector.
Absorbance modulation provides the means to perform near-field scanning optical imaging without bringing a physical nanoscale probe in close proximity to the sample. This technique does not rely on fluorescence and utilizes low light intensities. These novel attributes enable parallelization with the potential for high-speed imaging. However, it is clearly limited to surface (2D) imaging. In this article, we described preliminary results that indicate imaging of structures spaced by distances as small as λ1/10. The practically achievable resolution is limited by the thermal instability of the cis (transparent) isomer of the azobenzene. This effect was discussed earlier  and we expect thermally stable photochromic molecules to enable macro-molecular resolution.
We thank Francesco Stellacci and Timothy Swager for advice on synthesis of the photochromic molecules, Henry Smith for comments on the manuscript, James Daley for assistance with the fabrication of the resolution standards, and Mark Mondol for assistance with the electron-beam lithography. This work made use of MIT’s shared facility for scanning-electron-beam lithography in the Research Laboratory of Electronics. H-Y. T. was partially funded by an ignition grant from the MIT Deshpande Center for Technological Innovation.
References and links
1. E. Abbe, “Beiträge zur Theorie des Mikroskops und der mikroskopischen Wahrnehmung,” Archiv für Mikroskopische Anatomie 9(1), 413–418 (1873). [CrossRef]
2. S. W. Hell and J. Wichmann, “Breaking the diffraction resolution limit by stimulated emission: stimulated-emission-depletion fluorescence microscopy,” Opt. Lett. 19(11), 780–782 (1994). [CrossRef] [PubMed]
3. C. Eggeling, C. Ringemann, R. Medda, G. Schwarzmann, K. Sandhoff, S. Polyakova, V. N. Belov, B. Hein, C. von Middendorff, A. Schönle, and S. W. Hell, “Direct observation of the nanoscale dynamics of membrane lipids in a living cell,” Nature 457(7233), 1159–1162 (2009). [CrossRef]
4. M. Hofmann, C. Eggeling, S. Jakobs, and S. W. Hell, “Breaking the diffraction barrier in fluorescence microscopy at low light intensities by using reversibly photoswitchable proteins,” Proc. Natl. Acad. Sci. U.S.A. 102(49), 17565–17569 (2005). [CrossRef] [PubMed]
5. H. Shroff, C. G. Galbraith, J. A. Galbraith, and E. Betzig, “Live-cell photoactivated localization microscopy of nanoscale adhesion dynamics,” Nat. Methods 5(5), 417–423 (2008). [CrossRef] [PubMed]
10. P. Bourgeat, F. Meriaudeau, K. W. Tobin, and P. Gorria, “Content based segmentation of patterned wafers,” J. Electron. Imaging 13(3), 428–435 (2004). [CrossRef]
11. S. Tueber, A. Bzdurek, A. C. Durr, J. Heumann, and C. Holfeld, “Limitations of optical reticle inspection for 45-nm node and beyond,” Proc. SPIE 6349, 63490T/1–10 (2006).
12. K. Huang, J. Lee, Y. Wen, and J. Opsal, “Application of scatterometry for evaluation of lithographic process and OPC model generation,” Proc. SPIE 5752, 536–545 (2005). [CrossRef]
13. H. Hsu-Ting and F. L. Terry, “Spectroscopic ellipsometry and reflectometry from gratings (Scatterometry) for critical dimension measurement and in situ, real-time process monitoring,” Thin Solid Films 455–456, 828–836 (2004).
14. Y. Choi, M. Kim, S. Oh, and O. Han, “CD measurement evaluation on periodic patterns between optic tools and CD-SEM,” Proc. SPIE 6349, 6491N/1–8 (2006).
15. R. Menon and H. I. Smith, “Absorbance-modulation optical lithography,” J. Opt. Soc. Am. A 23(9), 2290–2294 (2006). [CrossRef]
17. H.-Y. Tsai, G. W. Wallraff, and R. Menon, “Spatial-frequency multiplication via absorbance modulation,” Appl. Phys. Lett. 91(9), 094103 (2007). [CrossRef]
20. R. Menon, P. Rogge, and H.-Y. Tsai, “Design of diffractive lenses that generate optical nulls without phase singularities,” J. Opt. Soc. Am. A 26(2), 297–304 (2009). [CrossRef]
21. D. Gil, R. Menon, D. J. D. Carter, and H. I. Smith, “Lithographic Patterning and Confocal Imaging with Zone Plates,” J. Vac. Sci. Technol. B 18(6), 2881–2885 (2000). [CrossRef]
22. M.-S. Ho, A. Natansohn, C. Barrett, and P. Rochon, “Azo polymers for reversible optical storage. The effect of polarity of the azobenzene groups,” Can. J. Chem. 73(11), 1773–1778 (1995). [CrossRef]
23. C. Barrett, A. Natansohn, and P. Rochon, “Cis-Trans Thermal Isomerization Rates of Bound and Doped Azobenzenes in a Series of Polymers,” Chem. Mater. 7(5), 899–903 (1995). [CrossRef]