1. Introduction

X-ray microscopy is a non-destructive technique that can provide information about internal structure of optically opaque objects. The large depth of field and high spatial resolution of X-ray images is a significant advantage of the X-ray microscopy. The point projection method for X-ray microscopy is a very simple one in principle and is capable of providing very high magnification (of order several thousand) over a wide range of X-ray energies without requiring any focusing X-ray optics. Point projection X-ray microscopy was first suggested in 1939 by Marton [1] and von Ardenne [2] and developed in the 1950s by Cosslett and Nixon [3, 4] who made significant advances by using magnetic electron lenses to focus the electron beam allowing sub-micrometer size X-ray sources. A scanning electron microscope (SEM) can be utilized for the purpose of producing a fine X-ray source suitable for X-ray microscopy which was suggested by Horn and Waltinger [5] in 1978.

A new approach to X-ray projection microscopy using a scanning electron microscope as a host has recently been developed by Mayo et al. [6, 7]. The approach, which we refer to as an X-ray ultra microscope (XuM), exploits X-ray phase contrast to boost the quality and information content of images. The XuM has proven to be a versatile and useful instrument, with greatly enhanced visibility of weakly absorbing and fine scale features. An additional technique relevant to XuM imaging of very small objects has been proposed by Wilkins [8]. The technique involves the use of integrated sample cells (ISC) which combine a target, a spacer, a sample chamber and an exit window. The technique may be applied to ultra-high spatial resolution imaging of microscopic objects and features, including small biological objects such as bacteria and cells and possibly including large biological molecules and assemblies. Compared with other methods, the ISC does not require special sample preparation such as coating or staining, and the sample may be imaged in its native state. It has the advantage over the conventional XuM approach [6, 7] that few mechanical components are required and much greater mechanical stability is achievable.

Our current XuM is based around a FEI XL-30 SFEG SEM [6, 7]. It utilises a ‘right-angle’ geometry, that is, the X-ray source-detector axis is at right-angle to the electron beam. In order to demonstrate the ISC technique proposed in [8] and to be compatible with the current ‘right-angle’ geometry of the XuM, right-angle ISCs for high-resolution XuM imaging of very small objects including dry, wet and liquid samples have been developed. Two types of ISCs are described and XuM images collected using these ISCs are presented in this paper.

2. Integrated sample cells

The right-angle ISC used for high-resolution XuM imaging represents a combination of a target, a spacer, a sample chamber and an exit window. Two types of ISC have been developed, the type I ISC is designed for small dry objects and the type II ISC is especially designed for liquid or small wet objects.

The structure of the type I ISC is shown in Fig. 1(a). Here a hollow brass tube ~0.5 mm diameter was cut at 45°. Mylar film of 3.5μm thickness was glued to the 45° end of the tube and this film was covered with a thin tantalum (Ta) foil target of 0.5μm thickness. The Mylar film acts as a spacer for the sample and as a support for the Ta foil. The sample was placed inside the tube and a sealed chamber was formed by gluing a second Mylar film to the remaining end. This Mylar film acts as an exit window for the ISC. The ISC was mounted on the SEM sample stage with the axis of the tube parallel to the X-ray source-detector direction (perpendicular to the electron beam direction). The SEM stage allows X, Y and Z translation and rotation about the vertical (Z) axis. The ISC was positioned a few millimeters below the SEM pole piece and the electron beam was focused onto the Ta foil target to generate a submicron X-ray source. The angle between the electron beam and the Ta foil was 45°. Images were recorded using a direct detection charge-coupled device (CCD) X-ray camera.

 

Fig. 1. Diagram and the point-projection microscope geometry of the right-angle integrated samplecells of (a) type I used for small dry objects; (b) type II used for wet or liquid objects.

Download Full Size | PDF

In the type II ISC (see Fig. 1(b)) an enclosure was used to support the 0.5μm thick Ta foil target at an angle of 45° to the beam direction. A wet or a liquid sample was placed inside a Teflon tube with inner diameter of 200 μm and wall thickness of 100μm and both ends of the tube were tightly sealed. This tube was mounted horizontally within the enclosure with the axis of the tube aligned perpendicular to the X-ray source-detector direction. The outer wall of the tube was in close proximity to the Ta foil. In the type II ISC, the wall of the Teflon tube was used as a spacer, as a support for the Ta foil target and as an exit window.

Figure 1 also shows how the point projection geometry results in the geometric magnification of the image in the detector plane. The magnification of the image is as measured at the CCD camera. For a target-sample distance R ₁ and a sample-detector distance R ₂, the magnification, M, is M = (R ₁+R ₂)/R ₁.

It is known [6, 7] that in imaging with a XuM the spatial resolution is determined primarily by the X-ray source size (s), the X-ray wavelength (λ) and the source-sample distance (R ₁). Indeed, as the result of Fresnel diffraction and the convolution with the point-spread function of the system, the width of the first Fresnel fringe referred to the object plane is approximately equal to (at high magnification values M>>1) $\sqrt{R_1\lambda +s^2}.$ Therefore, one needs to optimise R ₁, λ and s to achieve the required resolution. The largest source size compatible with the required resolution is used in order to maximise source intensity (minimise acquisition time) and conditions are adjusted so that $\sqrt{R_1\lambda }$ is approximately equal to the source size. To improve resolution, R ₁ is decreased in order to decrease the width of the Fresnel fringe, the latter width determining the lateral spread (blurring) of the image of any edge-like feature in the sample. Another reason for reducing the source-sample distance is to reduce the exposure time when imaging very small features. Therefore, the opportunity for having the source-sample distance R ₁ as small as a few microns is a significant potential benefit of the ISC configuration. Another important advantage of the ISC compared to conventional XuM is the ability to image wet and liquid samples in the high-vacuum environment of the host SEM.

3. Experimental results and discussion

X-ray images from four typical samples were recorded using the type I and type II right-angle ISCs fitted with a 0.5μm thick Ta (Ta Lα characteristic line is 8.14 keV) foil target inclined at 45° to the electron beam direction. The X-ray source size was approximately 100–200nm. The SEM objective aperture size and spot size were kept fixed throughout but the accelerating voltage was varied from 15 kV to 30 kV. The X-ray camera uses a direct detection, deep depletion CCD with 1340 ×1300 pixels each 20μm by 20μm in size and is fitted with a 250μm thick Be window. The target-detector distance, R ₁+R ₂, was fixed at 259mm.The experimental results and discussions are presented below. Each image is the average of a series of individual frames. Each frame in an image series has been aligned to the first frame to correct for any drift during the experiment and dark current and flat field corrections have been applied. The alignment is performed by finding the maximum of 2D correlation between pairs of frames [9].

3.1. Images of the foreleg joint of an ant and a Bryozoan shell (wet samples)

Figure 2(a) shows an image of a joint in the foreleg of an ant recorded using the type I ISC. The sample was fresh and untreated, and was directly sealed into the ISC chamber. The width of the foreleg was measured using an optical microscope prior to mounting in the ISC. Using this measurement the magnification of the X-ray image was found to be about 650 which corresponds to a target-sample distance, R₁, of about 400μm. The image was acquired at 25kV excitation voltage and a 20 minutes exposure. The spatial resolution estimated from the size of the smallest features visible in the image is about 0.2μm.

Figure 2(a) shows strong phase-contrast effects in the image in which edges and boundaries have been enhanced by Fresnel diffraction. The image clearly reveals the fine structure of the joint of the ant’s foreleg, including the skeleton of the foreleg with a lot of fine weakly absorbing details, even very fine hairs on the foreleg, visible in the image due to the presence of phase contrast. Other dark-spot-like features visible are dusts on the ant’s foreleg.

Figure 2(b) is an image of a Bryozoan shell sample collected using the type I ISC and an SEM accelerating voltage of 30kV. The sample was dry and untreated. Here magnification was calibrated using the position of a feature in two images recorded with known displacement achieved using the SEM X-stage movement. Magnification of the X-ray image was found to be about 300 which corresponds to a target-sample distance, R₁, of about 870μm. The image was formed from 100 individual image frames each with 1 minute exposure.

 

Fig. 2. XuM images of (a) an ant’s foreleg-joint and (b) a Bryozoan shell recorded using a type I ISC.

Download Full Size | PDF

Figure 2(b) shows an image with both phase contrast and absorption contrast where many ‘tube-like’ features overlay one another. Many fine resolution features of the Bryozoan shell sample were clearly revealed from the image. The spatial resolution estimated from the size of the finest features in the image is about 0.2μm.

3.2. Infected Blood cells

Figure 3 shows an image of a sample containing purified malaria infected red blood cells. The sample was fixed in 1% glutaraldehyde, 0.5% paraformaldehyde in 0.1M sodium cacodylate pH 7.4 and postfixed with 1% osmium tetroxide (OsO₄) and 3% aqueous uranyl acetate. The blood cells were spread on a silicon nitride membrane of 150 nm thickness that was held by a thick silicon frame of 200μm after dehydration. Then the silicon frame was fixed in the chamber of type I ISC. The image was collected using a type I ISC and recorded using 15kV excitation voltage. The exposure time was 1 minute for each frame; 20 frames were collected. Magnification was calibrated as for the Bryozoan shell sample and was found to be 2230 corresponding to a target-sample distance, R₁, of approximately 116 μm.

Figure 3 shows images with significant absorption and a black-white fringe at the cell boundaries. The parasite is visible in some cells in Fig. 3 as a darkened region. The spatial resolution estimated from the finest details visible in the image is about 0.1 μm.

 

Fig. 3. XuM image of purified malaria infected red blood cells collected using a type I ISC.

Download Full Size | PDF

3.3. Polystyrene particle suspensoid (liquid sample)

Figure 4 is an image of a commercial polystyrene particle suspensoid (product of SpheroTech Inc. in Illinois, USA) collected using a type II ISC. The polystyrene particle suspensoid is of 1μm diameter particles suspended in distilled water. The liquid sample, without any treatment, was directly siphoned into a squashed fine Teflon tube using a modified medical syringe. The image was recorded using a Ta target at 15kV and the exposure time was 20 minutes. R₁ was approximately 170 μm and the magnification, M, of the image was 1500.

 

Fig. 4. XuM image of polystyrene particles suspended in the distilled water collected using a type II ISC.

Download Full Size | PDF

Figure 4 shows the image from a distribution of the polystyrene particles. Interpretation of the image is rendered complicated by possible superposition of information from multiple layers of spheres and possibly clustering of spheres in three dimensions, both made feasible by the fact that the tube has an inner diameter of 200 μm. The size of the polystyrene particles revealed here would appear to be of order 1 micron, which is consistent with the description of the product. The spatial resolution in this image is better than 1μm, as individual polystyrene particles are clearly discernable, however, a more accurate estimation is difficult due to poor contrast and the lack of suitable fine features in the sample.

4. Conclusions and future prospects

Compared with other techniques for X-ray microscopy, the essential difference, apart from resolution, is that ISCs can be used in a conventional laboratory-based context and may be useful for routine screening of samples (such as blood and plant cell samples) compared with using a synchrotron, where access is much more restricted and often inconvenient. The ISC approach has some other advantages, e.g., it is based on a widely available source (SEM) and comparatively inexpensive components (integrated sample cell plus CCD) and is relatively easy to align.

Right-angle ISCs, including both type I and type II ISCs described in this paper, are compatible with the right-angle geometry of the XuM and expand the capability of the XuM technique. Comparative advantages of ISCs over the standard XuM include the following:

One restriction with the ISC is that it provides a fixed R₁ magnification determined by the spacer thickness. It is envisaged that a range of ISCs could be developed with target structure, spacer thickness and sample enclosure optimized for particular applications, such as routine inspection of blood cells.

The volume of the ISC can be made quite small. This might even be made adjustable in situ by use of an appropriate gasket and applied pressure, with possibility of adjustment to improve the visibility of certain features of interest in the sample.

The XuM images presented here, employing the use of the right-angle ISCs, clearly demonstrate the ability to characterize non-destructively very small objects including dry, wet and even samples in liquid, such as blood cells, in their natural state. The technique may be applied to ultra-high resolution imaging of microscopic objects and features, including small biological objects such as bacteria and cells, and possibly large biological molecules.

It is well-known that radiation damage is a sensitive function of resolution. We are well outside the resolution level where radiation damage would be expected to be a problem for imaging of, say, frozen hydrated samples of biological cells [10].