Optical condensers formed in wet-mounting setup - OSA Publishing

2 downloads 0 Views 879KB Size Report
Apr 20, 2015 - in biomedical imaging laboratories is a microscope condenser. ... scope condenser is formed when the object under observation immersed in a ...
Optical condensers formed in wet-mounting setup Darshan B. Desai1,2,* and Luis Grave de Peralta1,2 1

Department of Physics, Texas Tech University, Lubbock, Texas 79409, USA 2

NanoTech Center, Texas Tech University, Lubbock, Texas 79409, USA *Corresponding author: [email protected]

Received 9 December 2014; revised 27 January 2015; accepted 6 February 2015; posted 6 February 2015 (Doc. ID 228966); published 14 April 2015

We reveal that a practical and simple sample arrangement for optical microscopy that is commonly used in biomedical imaging laboratories is a microscope condenser. This unnoticed but high quality microscope condenser is formed when the object under observation immersed in a liquid is sandwiched between two glass coverslips, and is observed using an oil-immersion objective lens. We demonstrate that the advantages in image resolution and contrast provided by this imaging arrangement come from the resulting microscope condenser. We also demonstrate that this overlooked condenser can be reconfigured as a variable numerical aperture microscope condenser by depositing a drop of low boiling point liquid on top of it. We present and discuss several experiments suggesting that the condenser-like rings observed in the Fourier plane images are formed when the incoming light bounces off, or gets scattered by the inner edge of the top aperture of the metal cage of the oil-immersion objective lens toward the top surface of the sample arrangement, and is either reflected, or totally-internally reflected back at a highly inclined angle toward the object under observation. © 2015 Optical Society of America OCIS codes: (110.0180) Microscopy; (350.5730) Resolution; (230.0230) Optical devices. http://dx.doi.org/10.1364/AO.54.003580

1. Introduction

Ernst Abbe proposed for the first time in 1873 that using a microscope condenser to illuminate the sample under observation at inclined angles enhances the resolution of the optical microscope [1–3]. Condensers have now become an integral part of optical microscopes that are commonly found in laboratories needing high resolution images for research and applications. When a microscope condenser is used to illuminate a periodic-patterned sample, the minimum resolvable period, pmin , can be reduced from ∼λo ∕NAo for perpendicular illumination [4–6] to ∼λo ∕2 NAo , which is the well-known Rayleigh resolution limit [1–3,7]. Here λo is the vacuum wavelength of the illumination, and NAo  n sinθ is the numerical aperture of the light-collecting objective lens. This means that, in order to achieve subwavelength resolution that corresponds to λo ∕NAo < 1559-128X/15/123580-08$15.00/0 © 2015 Optical Society of America 3580

APPLIED OPTICS / Vol. 54, No. 12 / 20 April 2015

pmin < λo ∕2 NAo , the sample could be immersed in a medium with higher refractive index, n, or the light source could be modified to illuminate the sample directly at highly inclined angles, θ. Traditional microscope condensers have several optical components such as lenses and diaphragms that make them bulky and requiring mechanical movement of some parts [3]. Several approaches have been implemented for developing more advanced microscope condensers; for instance, recently developed ultrathin condensers (UTCs) [7–10] and digital condensers [11–14] provide near-field [7–10] and far-field [11–14] radiation to illuminate the sample at highly inclined angles. Nevertheless, research quality microscope condensers are, in general, elaborate devices that are not inexpensive. In this work we present an amazingly overlooked microscope condenser. We show here that a microscope condenser is formed when the object under observation immersed in a liquid is sandwiched between two glass coverslips, and is observed using an oil-immersion objective lens. This coverslip-sandwich arrangement,

Fig. 1. [Not to scale] Schematic illustration of microscope condensers based on scattering of the perpendicular-to-the-sample beam from the microscope’s built-in white-light illumination source. (a) The coverslip-sandwich condenser is formed using the wet-mounting setup that is commonly found in biomedical imaging labs, (b) and the same sample arrangement with a drop of liquid on the top. The square array of Cr pillars fabricated over the bottom coverslip is the object under observation for both sample arrangements.

which is sketched in Fig. 1(a), has been extensively used in biomedical imaging laboratories to observe the sample slides that are prepared by the wet mounting method, but in this work we demonstrate for the first time to our knowledge that this sample arrangement in fact constitutes a simple and practical microscope condenser. We demonstrate and analyze the dependence of the numerical aperture of the condenser (NAc ) on the thickness of the liquid layer in between the coverslips. As sketched in Fig. 1(b), we also present a way to make a reconfigurable condenser by depositing a drop of liquid with a low boiling point on top of such a coverslip-sandwich arrangement. Finally, we present some experimental facts that suggest the origin of the inclined illumination in these overlooked microscope condensers. As illustrated by the arrows in Figs. 1(a) and 1(b), we hypothesize that the incident light bounces off, or is scattered by, the inner edges of the metal cage of the oil-immersion objective lens toward the top surface of the sample arrangement, where it is reflected back at a highly inclined angle toward the object under observation. It is worth noting that in this work we use the term microscope condenser for any device or structure that is designed to illuminate the sample with a highly inclined illumination. In reality, no source of illumination provides perfectly perpendicular illumination (NAc  0). But being practical, we will refer here to (almost) perpendicular illumination when NAc ≪ NAo . The work presented here is organized in the following way. In Section 2, we describe the experimental setup used in this work and we compare the coverslip-sandwich microscope condenser described above with plasmonic UTCs [7–10], and with the simplest microscope condenser ever, to our knowledge, demonstrated [15]. In Section 3, we characterize the coverslip-sandwich microscope condenser. The feasibility of biomedical imaging applications of the coverslipsandwich arrangement is discussed in Section 4. In Section 5, we present experimental evidence related to the physical phenomena producing the discovered microscope condenser. Finally, we present the conclusions of this work in Section 6.

Fig. 2. [Not to scale] (a) Schematic illustration of the experimental setup; (b) schematic illustration of the cross section of a microscope condenser formed by liquid sandwiched between two glass coverslips, containing a square array of Cr pillars with diameter d  150 nm and lattice period p  300 nm that was fabricated on top of the bottom coverslip, and serves as the object for observation; (c) schematic illustration of the cross section of a hemispherical digital condenser that was used as a source for direct inclined illumination in some experiments.

2. Experimental Setup

Figure 2(a) shows the schematic diagram of the experimental setup used in this work. For the experiments described in this paper, we used a commercial Nikon Ti Eclipse inverted microscope and an oilimmersion objective lens with numerical aperture NAo  1.49 and 100× magnification. The light collected by the objective lens was bandpass filtered at λo  570 nm using a filter with a bandwidth of ∼10 nm, and was then imaged at the microscope’s real plane (RP) and Fourier plane (FP) using two charge-coupled device (CCD) cameras. We used two types of sources for far-field illumination of the object under observation in this work. The first was the microscope’s built-in white light illumination source that is capable of producing a partially coherent light beam arriving almost perpendicularly to the object under observation. The second source of illumination was a hemispherical digital condenser (HDC) [sketched in Fig. 2(c)] that provides partially coherent illumination at various inclined angles [11,16]. Therefore, the monochromatic light used for imaging in the experiments described in this work was the light coherently diffracted by the sample. Perpendicular incidence occurred when the built-in white light illumination source was used. Inclined illumination was provided by the HDC. The schematic diagram of the transverse cross section of the coverslip-sandwich arrangement containing a periodic-patterned object used for quantitative evaluation of resolution in this work is sketched in Fig. 2(b). The periodically-patterned object that was used in our experiments consists of ∼15 nm-high cylindrical chromium pillars, with 20 April 2015 / Vol. 54, No. 12 / APPLIED OPTICS

3581

diameter d  150 nm, that are arranged in a square array with lattice period p  300 nm. These chromium pillars were fabricated on a ∼150 μm thick glass coverslip using a combination of electron beam lithography and lift-off techniques as follows: a ∼100 nm-thick PMMA patterning resist layer for electron beam lithography was spin-coated on a ∼150 μm-thick glass coverslip. A ∼10 nm-thick layer of aluminum was then deposited on top using thermal evaporation so that the charges were grounded during lithography. After lithography, hydrofluoric acid was used to etch the aluminum layer. A mixture of methyl-isobutyl ketone and isopropyl alcohol was then used to develop the resist layer, which results in a template with a patterned array of holes. A ∼15 nm-thick layer of chromium was then deposited on this resist with holes, and then the resist layer was removed using acetone, leaving behind an array of cylindrical pillars. This glass coverslip with an array of chromium pillars was then used to form the coverslip-sandwich arrangement shown in Fig. 2(b) using index-matching microscope oil from Nikon with n  1.515 as the liquid between the coverslips, where the space between the coverslips was controlled using a spacer. Figure 3 shows FP images of periodicallypatterned samples with square symmetry and p  300 nm that were obtained using three different microscope condensers but the same oil-immersion objective lens with NAo  1.49. Figures 3(a) and 3(b) show FP images obtained with the experimental setup sketched in Fig. 2(a). The sample arrangement corresponding to Fig. 3(a) was the coverslipsandwich arrangement sketched in Fig. 2(b), and the sample arrangement corresponding to Fig. 3(b) was formed just by the bottom coverslip with the square array of Cr pillars on top of it, which constitutes the simplest microscope condenser ever demonstrated [15]. The FP image shown in Fig. 3(c) was obtained using a previously described plasmonic UTC [17]. FP images serve as maps of the angle of inclination of the light entering the objective lens, and hence, the corresponding numerical aperture of the condenser can be extracted from the FP image [18]. Thus, when using a traditional microscope condenser that concentrates the illuminating light coming from the light source into a cone of light, the FP image of a periodically-patterned object would consist of a centered bright circular spot of radius equivalent to the numerical aperture of the condenser used, and corresponding shifted spots (diffraction spots) that are produced due to the light that is diffracted due to the periodicity of the object under observation [15]. This is the origin of the bright spot observed in the center of the FP images shown in Figs. 3(a) and 3(b). The first-order diffraction spots were not collected by the microscope objective lens and, therefore, they are not present in the FP images. More importantly, when a microscope condenser illuminates the sample with a hollow cone of light, the corresponding FP image of a periodically-patterned object 3582

APPLIED OPTICS / Vol. 54, No. 12 / 20 April 2015

Fig. 3. FP images of periodically-patterned objects with period p  300 nm obtained using (a,b) the microscope’s built-in whitelight illumination source as sketched in Fig. 2(a) with the sample formed by (a) a coverslip-sandwich arrangement sketched in Fig 2(b), and (b) a single coverslip with a square array of Cr pillars fabricated on top of it. (c) FP image obtained using a plasmonic ultra-thin condenser. Clearly, the coverslip-sandwich condenser discussed in this paper provides the best contrast between the diffracted rings and background, which is necessary for good contrast in the RP images.

consists of rings [7–10,15,16]. Therefore, the observation of rings in all of the FP images shown in Fig. 3 demonstrates that in each case the sample arrangement was illuminated using a microscope condenser. The central brightest ring observed in all of the FP images shown in Fig. 3 is the zero-order diffraction ring and the four fractions of rings observed in each of these FP images corresponds to the first-order diffraction rings that are distributed with the square symmetry corresponding to the periodicity of the object under observation [7–10,15,16]. It is worth noting that diffraction rings corresponding to two consecutive diffraction orders need to be present in a FP image in order to be able to observe the structure of a periodic-patterned sample in the corresponding RP image [7–11] and, consequently, the periodicity of the sample is visible in the RP images (not shown) corresponding to the FP images shown in Fig. 3. Bright rings with high contrast with respect to the FP image background are needed for obtaining good quality RP images; therefore, by comparing the FP images shown in Fig. 3, it can be concluded that the best microscope condenser corresponds to the coverslip-sandwich arrangement sketched in Fig. 2(b). Plasmonic UTCs are based on the excitation of surface plasmon polaritons (SPPs) by fluorescence [7–10]

and, hence, the central zero-order diffraction spot is not observed in the FP image shown in Fig. 3(c) [7– 10]. As seen in the FP image shown in Fig. 3(c), an issue of concern with these microscope condensers is the low contrast between the light that couples to SPPs and produces the condenser’s characteristic rings, and the light that does not couple to SPPs and contributes to the FP image background [7–10]. A similar problem can be observed in the FP image shown in Fig. 3(b), where the intensity of the first-order diffraction rings is much smaller than the intensity of the central zero-order diffraction spot. A notable difference between the FP images shown in Figs. 3(b) and 3(c) and the FP image shown in Fig. 3(a) is that in the FP image shown in Fig. 3(a) the intensity of the central zero-order diffraction ring is similar to the intensity of the central zero-order diffraction spot. In addition, the contrast between the first-order diffraction rings and the FP image background is larger in the FP image shown in Fig. 3(a) than in the FP images shown in Figs. 3(b) and 3(c). These differences indicate that the simple and practical coverslip-sandwich arrangement sketched in Fig. 2(b) results in a high quality microscope condenser. Nonetheless, as seen in the FP image shown in Fig. 3(c), plasmonic UTCs have the advantage of producing FP images without the central zero-order diffraction spot and, as seen in the FP image shown in Fig. 3(b), the simplest microscope condenser has the advantage of producing very large rings corresponding to NAc ∼ 1.5 > NAo . A close look at the FP image shown in Fig. 3(a) reveals the presence of these large rings with NAc > NAo when the coverslip-sandwich microscope condenser was used. 3. Characterization of Microscope Condensers

The primary advantage of using periodic-patterned objects for microscopy is that it makes it possible to quantitatively evaluate the resolution of the microscope condenser arrangement in terms of the minimum period observable, pmin , in a periodicallypatterned object, as shown below [1,7–11]: pmin 

λo : NAo  NAc

(1)

From the central zero-order diffraction spot in the FP image shown in Fig. 3(a) and using Eq. (1), we obtained that the numerical aperture of the microscope built-in white-light illumination source is ∼0.35 and, hence, the minimum observable period without using a microscope condenser is ∼370 nm, which explains why the diffraction spots were not captured by the oil-immersion objective lens. However, from the brightest zero-order diffraction ring in the FP image shown in Fig. 3(a), and using Eq. (1), we obtained NAc ∼ 1.2. The brightest ring in the central double-ring observed in the FP image shown in Fig. 3(a) has the largest radius and, therefore, the minimum observable period that can be observed using the experimental and sample setups

Fig. 4. Dependence of NAc on the spacer thickness (t). The triangles represent experimental data points, and the discontinuous line is the best linear fitting for the same. The inset shows a typical RP image of the sample with a square array of Cr pillars with p  300 nm that was used for this work. The scale bar shown in the RP image corresponds to 2 μm.

sketched in Figs. 2(a) and 2(b), respectively, is found to be pmin ∼ 211 nm. This is in excellent correspondence with the RP image shown in the inset of Fig. 4, which corresponds to the FP image shown in Fig. 3 (a). By changing the thickness, t, of the spacer in the coverslip-sandwich arrangement sketched in Fig. 2 (b), we studied the dependence of NAc on the thickness of the oil layer in between the coverslips. The rings have a finite width and, therefore, we calculated the NAc value corresponding to the center of the transversal intensity profile of the ring. As shown in Fig. 4, we found that NAc varies linearly with t in the following way: NAc  −At  B:

(2)

Here, A  0.0006 μm−1 and B  1.5045. For thickness values larger than those reported in Fig. 4, the rings have poorly defined shape, and get smaller in radius with increasing thickness. On the other hand, for thickness values less than those reported in Fig. 4, we have observed that the size of the rings associated with the coverslip-sandwich arrangement tend to be the size of the rings that were previously observed using the simplest microscope condenser [15], and the intensity of the corresponding rings also decreases consistently [Fig. 2(b)]. We also found that the coverslip-sandwich condenser sketched in Fig. 2(b) can be modified to a reconfigurable microscope condenser with variable NAc by putting a drop of liquid with a low boiling point over the top coverslip. We deposited ∼0.5 ml of isopropyl alcohol on the top coverslip using a regular plastic dropper that results in a relatively large drop of ∼1.5 cm diameter, whose shape depends on several parameters such as its interaction with the surface of the coverslip [19– 21]. As shown in Fig. 5, as the drop evaporates and its shape changes with time, brighter additional rings appear in the FP images shown in the Figs. 5(d), 20 April 2015 / Vol. 54, No. 12 / APPLIED OPTICS

3583

Fig. 6. (a,c) FP and (b,d) RP images of a square array of Cr pillars with p  300 nm obtained using the arrangement sketched in Fig. 2(b) with (a,b) water, and (c,d) with index-matching oil filling the space between the coverslips. The scale bar shown in RP images corresponds to 2 μm.

4. Imaging of Biological Samples

Fig. 5. Temporal evolution of the (b,d,f,h,j) FP, and corresponding (a,c,e,g,i) RP images of the square array of Cr pillars. The images were obtained using the microscope’s built-in illumination source with the coverslip-sandwich arrangement sketched in Fig. 1(b) with a drop of isopropyl alcohol deposited over the top coverslip. The five pairs of FP and RP images were obtained at intervals of ∼1 min . The scale bar shown in all RP images corresponds to 2 μm.

5(f), and 5(h) that were obtained at intervals of ∼1 min . The numerical aperture of the additional rings related to the drop varies progressively with time from 1.00 in the FP image shown in the Fig. 5(d) to 1.17 in the FP image shown in Fig. 5(h). As a consequence, a progressive increase in the contrast can be observed in the corresponding RP images shown in Figs. 5(c)–5(g). However, the additional ring does not improve the image resolution because the resolution is determined by the ring with the largest and constant radius corresponding to NAc ∼ 1.47, which is present in all the FP images shown in Fig. 5. The variation of the additional ring when the drop changes its shape suggests that the geometry of the top surface of the sample’s arrangement has a strong influence on the performance of the resulting condenser. The primary advantage of reconfiguring the microscope condenser in this way is that the ambiance of the object under observation is not altered at all and, hence, this technique is useful for samples that need to be kept immersed in a specific medium, or have to be sealed from surroundings. 3584

APPLIED OPTICS / Vol. 54, No. 12 / 20 April 2015

Using oil as a buffer medium is not practical for biomedical imaging applications, and therefore, we have explored the use of water as the liquid mounting medium in between the coverslips. The images shown in Figs. 6(a) and 6(b) were obtained using the microscope setup sketched in Fig. 2(a), and sample arrangement sketched in Fig. 2(b), with no drop on top of the sample arrangement and water as the liquid mounting medium between the coverslips. The bright rings observed in the FP images shown in Fig. 6(a) suggest that this sample arrangement, which is well known in biomedical imaging applications, is a microscope condenser with NAc ∼ 1.17 when the object under observation was immersed in water. On the other hand, Figs. 6(c) and 6(d) were obtained using the same sample arrangement, but with index-matching microscope oil as the liquid mounting medium between the coverslips. The bright rings observed in Fig. 6(c) suggest that such a sample arrangement behaves like a microscope condenser with NAc ∼ 1.40. In both cases, a clearly resolved array of Cr pillars with p  300 nm in the RP images shown in Figs. 6(b) and 6(d) demonstrates that the improvement in resolution is due to the resulting microscope condenser. It should be noted that the improvement in resolution is not a consequence of immersing the array of Cr pillars in the liquid-filled cavity formed by the two coverslips. The FP image (diffraction pattern) corresponding to a two-dimensional periodic lattice structure illuminated by a collimated light beam impinging perpendicularly should be an array of spots [4–6], but in the FP images shown in Figs. 6(a) and 6(c) arrays of (fractions of) rings can be observed, which is the

signature that the array of Cr pillars has been illuminated by the light passing through a microscope condenser [7,15]. The higher-order diffraction spots, on the other hand, are not visible in FP images here because, as per Eq. (1), a lattice with a period smaller than ∼370 nm diffracts the perpendicularly incident light at such high angles that cannot be captured by the oil-immersion objective lens with NAo ∼ 1.49. Therefore, the illumination in the form of a hollow cone resulting from the coverslipsandwich microscope condenser presented here turns out to be more useful. In order to explore the imaging of biological samples, we deposited a small amount of a water solution with living E. coli on a coverslip that had an array of Cr pillars with p  300 nm fabricated on it, and covered it with a second coverslip, thereby creating the arrangement sketched in Fig. 2(b). Figures 7(a) and 7(b) show the FP and RP images obtained with this arrangement without a drop of oil on top of the sample arrangement. In the RP image shown in Fig. 7(b), several bacteria and the periodicity of the array of Cr pillars are clearly visible. From the RP image in Fig. 7(b), the cylindrical shape of the E. coli bacteria was directly found to have an average diameter of ∼650 nm. This confirms both the utility for biomedical imaging applications and the improvement of resolution provided by the arrangement sketched in Fig. 2(b), which is commonly used in biomedical labs to prepare samples for optical microscopy. The appearance of bright rings in the FP image shown in Fig. 7(b) confirms that the improvement in image resolution was produced by the

illumination of the sample with hollow cone of inclined light from microscope condenser with NAc ∼ 1.34. The FP and RP images shown in Figs. 7(c) and 7(d), respectively, were obtained by adding a drop of index-matching microscope oil on top of the sample arrangement. The resulting microscope condenser with NAc ∼ 1.19 provides the required resolution for imaging both the array of Cr pillars and individual bacteria.

Fig. 7. (a,c) FP and (b,d) RP images of E. coli bacteria in a water solution contained between two coverslips. A square array of Cr pillars with p  300 nm was fabricated over the bottom coverslip. The images were obtained using the arrangement sketched in (a, b) Fig. 1(a), and (c,d) Fig. 1(b) with a drop of index-matching oil over the coverslip-sandwich arrangement. The scale bar shown in RP images corresponds to 2 μm.

Fig. 8. (a) Photograph of the part of the experimental setup close to the sample, showing the spatial filter used to block the light from the microscope’s built-in illumination source falling directly on the top aperture of the metal cage of the oil-immersion objective lens. (b,d) FP and corresponding (c,e) RP images obtained (d,e) with and (b,c) without introducing the spatial filter. The scale bar shown in all RP images corresponds to 2 μm.

5. Discussion on Origin of Observed Rings

It should be noted that the planar geometry of the coverslip-sandwich arrangement sketched in Fig. 2 (b) cannot change the direction of the light coming perpendicularly to the sample from the microscope built-in illumination source. This is in correspondence with the small diameter of the zero-order diffraction spot observed at the center of the FP images shown in Figs. 3(a) and 5. Therefore, just like for the simplest microscope condenser [15], our working hypothesis is that the observed rings are related to the light that is bounced off, or is scattered by, the inner edge of the top aperture of the metal cage of the microscope objective lens toward the top surface of the sample arrangement, where it is either reflected, or totally-internally reflected back at a highly inclined angle toward the object under observation. In what follows, we describe experimental evidence supporting this hypothesis. As shown in Fig. 8(a), we placed a small (opaque cylinder) spatial filter over

20 April 2015 / Vol. 54, No. 12 / APPLIED OPTICS

3585

the microscope condenser arrangement sketched in Fig. 2(b) to block the direct entrance of the light coming from the microscope’s built-in illumination source to the top aperture of the metal cage of the microscope objective lens. FP images shown in Figs. 8(b) and 8(d) were obtained without and with the spatial filter, respectively. As confirmed by the absence of the zero-order diffraction spot at the center of the FP image shown in Fig. 8(d), the spatial filter effectively blocked the light passing through the coverslip-sandwich arrangement without deviation. However, the spatial filter did not block the light producing the condenser-like rings, which are still present in the FP image shown in Fig. 8(d). Moreover, the observation of a clearly resolved periodically-patterned object in the corresponding RP images shown in Figs. 8(c) and 8(e) demonstrates that the increment in resolution is related to the observation of rings; i.e., with the use of a microscope condenser for illuminating the object under observation with inclined light. To investigate this further without blocking the incident light with the spatial filter, we used an iris to control the diameter of the incident light from the microscope’s built-in white light illumination source, as sketched in Fig. 2(a). In Figs. 9(a), 9(c), and 9(e) we show photographs of the flat top surface of the metal cage of the oil-immersion objective lens with the coverslip-sandwich arrangement sketched in Fig. 2(b), corresponding to different iris apertures. FP images obtained with the corresponding iris apertures are shown in Figs. 9(b), 9(d), and 9(f). When the diameter of the incident beam was reduced to slightly less than the diameter of the top aperture of the metal cage of the oil-immersion objective lens, and almost none of the incident light fell on the edge of the top aperture of the metal cage [Fig. 9(a)], there was a single centered bright spot due to the transmitted light and the diffracted rings were absent in the corresponding FP image shown in Fig. 9(b), and hence, the image of the periodic-patterned object was not present in the RP image (not shown). When the diameter of the iris was increased to an extent that the incident light fell only up to the edge of the top aperture of the metal cage of the microscope objective lens [Fig. 9(c)], diffracted rings could be observed in the corresponding FP image shown in Fig. 9(d). Consequently, an image of the periodically-patterned object was clearly seen in the corresponding RP image (not shown). When the diameter of the iris was further increased so that light fell on the entire flat surface of the metal cage of the oil-immersion objective lens [Fig. 9(c)], there was no noticeable change in the intensity of the diffracted rings in the corresponding FP image shown in Fig. 9(f). These experiments seem to confirm the hypothesis that the rings observed in the FP images are related to light that is backscattered at the edge of the top aperture of the metal cage of the oilimmersion objective lens. This is reinforced by the fact that, when using the coverslip-sandwich arrangement sketched in Fig. 2(b), we have been able 3586

APPLIED OPTICS / Vol. 54, No. 12 / 20 April 2015

Fig. 9. (a,c,e) Photographs of the part of the experimental setup close to the sample, showing the light from the microscope’s builtin illumination source falling (a) only on the top aperture, (c) only up to the edge of the top aperture, and (e) on the entire flat top surface of the metal cage of the oil-immersion objective lens. (b,d,f) Corresponding FP images. The diameter of the incident light beam from microscope’s built-in illumination source was controlled using an iris, as sketched in Fig. 2(a). The circular top aperture of the metal cage of the oil-immersion objective lens has a diameter of 4 mm.

to observe rings in the FP images only by using an oil-immersion objective lens with NAo > 1. We have been able to observe first-order diffraction rings using plasmonic UTCs and a microscope objective lens with NAo < 1 that is well-separated from the sample [22,23]. However, we have not been able to do the same using a coverslip-sandwich arrangement. This suggests that the edge of the top aperture of the metal cage of the oil-immersion objective lens is a necessary part in completing the coverslip-sandwich condenser. In order to study the influence of the direction of the illumination light on the appearance of rings in the FP images, we used a HDC formed by 64 light emitting diodes (LEDs) uniformly distributed in the interior of a hemisphere [11,16] as a source using the illumination arrangement sketched in Fig. 2(c). The hemisphere has a circular opening on the top, which

Fig. 10. FP images of a square array of Cr pillars obtained using (a) the microscope’s built-in white-light illumination source, and (b) the hemispherical digital condenser sketched in Fig. 2(c), with the sample arrangement sketched in Fig. 2(b).

permits nearly perpendicular illumination from the microscope’s built-in illumination source to illuminate the sample, when required. Figure 10(a) shows a FP image corresponding to the sample arrangement sketched in Fig. 2(b) obtained using the illumination arrangement sketched in Fig. 2(c) with all the LEDs in the HDC in the “OFF” state, and the microscope built-in illumination source “ON”, which produces the central zero-order bright spot and corresponding condenser-like rings. The spots observed in the background are due to negligible reflections from the LEDs. Figure 10(b) shows a FP image obtained with a circular group of LEDs in the “ON” state, and the microscope built-in illumination source switched “OFF”, which corresponds to the absence of a central zero-order bright spot in the FP image shown in Fig. 10(b). The zero-order diffraction spots corresponding to the light from LEDs form the discrete ring observed in the FP image shown in Fig. 10(b) [11,16], but the condenser-like zero-order diffraction ring observed in the FP image shown in Fig. 10(a) is hardly observed in Fig 10(b), and the corresponding first-order diffraction rings are completely absent in Fig. 10(b). This suggests that inclined illumination cannot produce the condenserlike rings and, therefore, the coverslip-sandwich arrangement sketched in Fig. 2(b) needs perpendicular illumination to function as an efficient microscope condenser. 6. Conclusion

We have demonstrated that the advantages in image resolution and contrast provided by a practical and simple sample arrangement commonly used in biomedical imaging laboratoriesis due to the fact that such a coverslip-sandwich arrangement in combination with the use of an oil-immersion objective lens results in a highly efficient microscope condenser. We characterized this overlooked microscope condenser, which is formed when the object under observation is immersed in a liquid, is sandwiched between two glass coverslips, and is observed using an oil-immersion objective lens. We presented a technique to make these condensers reconfigurable by depositing a drop of liquid with a low boiling point on top of it. We have also provided several experimental facts that suggest that the illumination from these condensers in the form of hollow cones of light is formed when the incoming light bounces off, or gets scattered by, the inner edge of the top aperture of the metal cage of the oil-immersion objective lens toward the top surface of the sample arrangement, and is either reflected or totally-internally reflected back at a highly inclined angle toward the object under observation. This work was partially supported by the NSF CAREER Award (ECCS-0954490). References 1. H. H. Hopkins and P. M. Barham, “The influence of the condenser on microscopic resolution,” Proc. Phys. Math. Soc. Jpn. 63, 737–744 (1950).

2. H. Köheler, “On Abbe’s theory of image formation in the microscope,” Opt. Acta 28, 1691–1701 (1981). 3. A. Vainrub, O. Pustovyy, and V. Vodyanoy, “Resolution of 90 nm (λ/5) in an optical transmission microscope with annular condenser,” Opt. Lett. 31, 2855–2857 (2006). 4. E. Hetcht, Optics, 3rd ed. (Addison Wesley, 1998). 5. M. Born and E. Wolf, Principles of Optics, 5th ed. (Pergamon, 1975). 6. J. W. Goodman, Introduction to Fourier Optics (McGraw-Hill, 1968). 7. D. B. Desai, D. Dominguez, A. A. Bernussi, and L. G. de Peralta, “Ultra-thin condensers for optical subwavelength resolution microscopy,” J. Appl. Phys. 115, 093103 (2014). 8. C. Regan, R. Rodriguez, S. Gourshetty, L. G. de Peralta, and A. Bernussi, “Imaging nanoscale features with plasmon-coupled leakage radiation far-field superlenses,” Opt. Express 20, 20827–20834 (2012). 9. C. J. Regan, D. Dominguez, A. A. Bernussi, and L. G. de Peralta, “Far-field optical superlenses without metal,” J. Appl. Phys. 113, 183105 (2013). 10. R. Lopez-Boada, C. J. Regan, D. Dominguez, A. A. Bernussi, and L. G. de Peralta, “Fundamentals of optical far-field subwavelength resolution based on illumination with surface waves,” Opt. Express 21, 11928–11942 (2013). 11. L. Molina, D. Dominguez, D. B. Desai, T. O’Loughlin, A. A. Bernussi, and L. G. de Peralta, “Hemispherical digital optical condensers with no lenses, mirrors, or moving parts,” Opt. Express 22, 6948–6957 (2014). 12. G. Zheng, C. Kolner, and C. Yang, “Microscopy refocusing and dark-field imaging by using a simple LED array,” Opt. Lett. 36, 3987–3989 (2011). 13. G. Zheng, R. Horstmeyer, and C. Yang, “Wide-field, highresolution Fourier ptychographic microscopy,” Nat. Photonics 7, 739–745 (2013). 14. L. Tian, X. Li, K. Ramchandran, and L. Waller, “Multiplexed coded illumination for Fourier ptychography with LED array microscope,” Biomed. Opt. Express 5, 2376–2389 (2014). 15. D. Dominguez, M. Alhusain, N. Alharbi, A. Bernussi, and L. G. de Peralta, “Fourier plane imaging microscopy,” J. Appl. Phys. 116, 103102 (2014). 16. S. Sen, L. Molina, D. Cao, D. B. Desai, A. A. Bernussi, and L. G. de Peralta, “Versatile optical microscopy using a reconfigurable hemispherical digital condenser,” Biomed. Opt. Express 6, 658–667 (2015). 17. L. G. de Peralta, C. J. Regan, and A. A. Bernussi, “SPP tomography: a simple wide-field nanoscope,” Scanning 35, 246–252 (2013). 18. R. Rodriguez, C. J. Regan, A. Ruiz-Columbie, W. Agutu, A. A. Bernussi, and L. G. de Peralta, “Study of plasmonic crystals using Fourier-plane images obtained with plasmon tomography far-field superlenses,” J. Appl. Phys. 110, 083109 (2011). 19. N. Verplanck, Y. Coffinier, V. Thomy, and R. Boukherroub, “Wettability switching techniques on superhydrophobic surfaces,” Nanoscale Res. Lett. 2, 577–596 (2007). 20. R. P. Shaik, W. Monach, H. Krause, and H. Zappe, “Reconfigurable liquid micro-lens system for variable fiber coupling,” in IEEE/LEOS International Conference on Optical MEMs and Nanophotonics 2008, 11–14 August 2008, pp. 15–16. 21. Y. S. Tsou, K. H. Chang, and Y. H. Lin, “A droplet manipulation on a liquid crystal and polymer composite film as a concentrator and a Sun tracker for a concentrating photovoltaic system,” J. Appl. Phys. 113, 244504 (2013). 22. D. Dominguez, M. Alhusain, N. Alharbi, A. A. Bernussi, and L. G. de Peralta, “Fourier plane imaging microscopy for detection of plasmonic crystals with periods beyond the optical diffraction limit,” Plasmonics (2015), to be published. 23. L. Grave de Peralta, “Beating the diffraction limit with Fourier plane imaging microscopy based on plasmonic ultrathin condensers,” in Proceedings of the 13th International Conference of Near-Field Optics Nanophotonics, and Related Techniques, Salt Lake City, Utah, USA, 31 August–4 September 2014, p. 382. 20 April 2015 / Vol. 54, No. 12 / APPLIED OPTICS

3587