A transmission X-ray microscope has been designed and commissioned at the 18-ID Full-field X-ray Imaging beamline at the National Synchrotron Light Source II. This instrument operates in the 5–11 keV range, and, with the current set of optics, is capable of 30 nm spatial resolution imaging, with a field of view of about 40 μm. For absorption contrast, the minimum exposure time for a single projection image is about 20 ms and an entire 3D tomography data set can be acquired in under 1 min. The system enables tomographic reconstructions with sub-50 nm spatial resolution without the use of markers on the sample or corrections for rotation run-outs.

Transmission X-ray microscopes (TXMs) are X-ray analogs of the optical microscope (Fig. 1) which have been widely used in various fields of science, engineering, and technology.1–3 There is a one-to-one correspondence in its optical scheme: a condenser that illuminates the sample and a magnification lens (objective) downstream of the sample that images the sample onto a detection system. The pinhole blocks any stray unfocused beam to ensure annular illumination, and the phase ring is only used in the Zernike phase contrast mode. Due to the much smaller X-ray wavelength, the spatial resolution of an X-ray microscope is much better than an optical microscope. In addition, the penetrating power of the X-rays enables visualization of a sample’s internal morphology. Hard X-ray TXMs are in operations at several synchrotron facilities world-wide.4–7 These instruments are either based on an in-house design or are from Xradia/Zeiss,8 who also sells laboratory based TXMs. In general, due to photon flux, laboratory-based TXMs generally operate at the 100 nm spatial resolution range while synchrotron based TXMs operate in the 20–60 nm spatial resolution range. Typical absorption based TXM acquisition times for a full 3D tomography data set are in the few-to-tens of hours time scale for laboratory-based instruments and tens of minutes time scale for synchrotron-based systems.

FIG. 1.

Schematic of a TXM. The condenser focuses the X-rays onto the sample. The zone plate (ZP) magnifies and images the sample onto a detector. A phase ring can be inserted at the back focal plane of the zone plate for Zernike phase-contrast. A Bertrand lens, located downstream of the phase ring, that is used to visualize the back focal plane of the objective zone plate during alignment of the phase ring is not shown.

FIG. 1.

Schematic of a TXM. The condenser focuses the X-rays onto the sample. The zone plate (ZP) magnifies and images the sample onto a detector. A phase ring can be inserted at the back focal plane of the zone plate for Zernike phase-contrast. A Bertrand lens, located downstream of the phase ring, that is used to visualize the back focal plane of the objective zone plate during alignment of the phase ring is not shown.

Close modal

Although the Xradia/Zeiss TXMs have been extremely productive instruments at synchrotrons, they are closed, turn-key instruments that use proprietary data acquisition and data analysis software. As such, they are not well-suited in an environment where there is a desire to (1) constantly improve the capabilities (e.g., spatial or temporal resolution) of the instrument, (2) integrate the TXM controls with beamline controls, (3) implement new data acquisition strategies (e.g., fly scans), and (4) integrate workflows and take advantage of new developments in analysis and reconstruction. Because of these limitations, the National Synchrotron Light Source II (NSLS-II) decided to design and build its own TXM. The first imaging results of this new instrument have been reported recently.9 Here, we present the mechanical details of the instrument design, characterization, and performance.

The overall layout of the NSLS-II 18-ID beamline is shown in Fig. 2. The beamline utilizes a damping wiggler as its source. The damping wiggler is a 7 m long device with 100 mm period, ∼1.85 T magnetic field, and the electron deflection parameter K = 16.5. The beamline’s fixed masks (not shown) limit the angular acceptance of the beamline to, on-axis, 0.21 mrad × 0.21 mrad of the wiggler radiation fan. A top-side-contact water-cooled bendable mirror collimates the white beam. This collimation is necessary to achieve the energy bandwidth for X-ray Absorption Near Edge Spectroscopy (XANES) measurements. This is the most challenging optical component on the beamline due to its high power load (∼1300 W at 500 mA ring current). The collimating mirror (CM) has 3 stripes Rh, Cr, and Si and operates at a fixed incidence angle of 4.4 mrad, leading to cut-off energies of about 15, 12, and 7 keV, respectively. Downstream of the CM is the cryogenically cooled Si(111) double crystal monochromator (DCM). At 500 mA ring current, the power incident on the DCM is 956 W, 745 W, 415 W, respectively, for the Rh, Cr, and Si stripes on the CM. Downstream of the DCM, a Rh-coated cylindrical mirror is bent in the tangential direction to focus the beam onto the secondary source aperture (SSA), which is in the 18-ID-B hutch, and is at 61 m from the source. The CM, DCM, and TM are all located inside the 18-ID-A hutch, also known as the first optical enclosure (FOE). Further details of the beamline optical design, thermal calculations, and ray trace simulations are described elsewhere.10 

FIG. 2.

Schematic of the beamline layout. There are two major hutches. The upstream hutch (18-ID-A, first optical enclosure) is where all the beamline optics reside. The first optic is the water-cooled bendable collimating mirror (CM), followed by a cryogenically cooled Si(111) double crystal monochromator (DCM), and followed by a toroidal mirror (TM), which is realized by bending a cylindrical mirror. Both mirrors are 1450 mm long and operate at a fixed 4.4 mrad incidence angle.

FIG. 2.

Schematic of the beamline layout. There are two major hutches. The upstream hutch (18-ID-A, first optical enclosure) is where all the beamline optics reside. The first optic is the water-cooled bendable collimating mirror (CM), followed by a cryogenically cooled Si(111) double crystal monochromator (DCM), and followed by a toroidal mirror (TM), which is realized by bending a cylindrical mirror. Both mirrors are 1450 mm long and operate at a fixed 4.4 mrad incidence angle.

Close modal

The beamline layout and optics are optimized for maximizing the photon flux at the sample position consistent with the required phase space of the microscope—estimated at 20 µm field of view (FOV) × 5 mrad = 100 μm mrad. The sample is fixed at 2 m downstream from the SSA, i.e., 63 m from the source. These distances set the overall scale of the beamline and the instrument. Although we designed the TXM to easily accommodate future upgrades of X-ray optics, our plan for initial operations was to reuse the X-ray optics from an Xradia/Zeiss TXM.11 The parameters of these X-ray optics are given in Table I. The experimental hutch (18-ID-B) houses four major sections of the instrument: Secondary Source Aperture (SSA), sample station, detector station A, and detector station B (Fig. 3). Although the SSA is not a part of the TXM itself, we include the description here because it resides in the same experimental hutch and it houses components that are pertinent to the TXM operations.

TABLE I.

Current X-ray optics parameters.

Elliptical capillary condenser (5–11 keV)
Major axis (mm) 1000 
Minor axis (mm) 0.37 
Length (mm) 100 
Entrance radius (mm) 0.195 
Exit radius (mm) 0.115 
Secondary source to center distance (mm) 1900 
Exit to sample distance (mm) 50 
Objective zone plate (high resolution) 
Diameter (µm) 100 
Outermost zone (nm) 30 
8 keV focal length (mm) 19.4 
Objective zone plate (large field of view) 
Diameter (µm) 200 
Outermost zone (nm) 30 
8 keV focal length (mm) 38.8 
Elliptical capillary condenser (5–11 keV)
Major axis (mm) 1000 
Minor axis (mm) 0.37 
Length (mm) 100 
Entrance radius (mm) 0.195 
Exit radius (mm) 0.115 
Secondary source to center distance (mm) 1900 
Exit to sample distance (mm) 50 
Objective zone plate (high resolution) 
Diameter (µm) 100 
Outermost zone (nm) 30 
8 keV focal length (mm) 19.4 
Objective zone plate (large field of view) 
Diameter (µm) 200 
Outermost zone (nm) 30 
8 keV focal length (mm) 38.8 
FIG. 3.

Sketch of the TXM instrument. The outline of the experimental hutch footprint is also shown.

FIG. 3.

Sketch of the TXM instrument. The outline of the experimental hutch footprint is also shown.

Close modal

Based on simulations, the expected photon flux at the entrance to the condenser is 5.4 × 1013 photons/s at 500 mA ring current. The measured flux is 3.9 × 1013 photons/s at 400 mA, which, when scaled to the current, is ∼90% of the expected value. Based on the experience using the Xradia/Zeiss TXM at NSLS11 and scaling by the photon flux, it is expected that the exposure times of 20 ms per image are feasible for the TXM instrument at Full-field X-ray Imaging beamline (FXI).

As mentioned in Sec. I, one general design goal for the TXM is to have a flexible platform that can accommodate future improvements in performance—either in terms of speed or spatial resolution. Since the primary strength of the instrument will be its speed, the design must also be as accommodating as possible to sample environment cells so that in situ/in-operando studies that take advantage of the instrument speed can be performed.

The baseline goal for the instrument stability is to achieve 30 nm spatial resolution measurements with the TXM, with the hope that the instrument may be able to go down to 20 nm in resolution. Based on the experience of the Hard X-ray Nanoprobe (HXN) at NSLS-II, it was clear that trying to achieve better than 20 nm spatial resolution would require significant sacrifices in terms of instrument flexibility and ease of use. Speedwise, based on the expected photon flux, the TXM instrument is designed to be able to collect images at 50 frames per second, corresponding to the expected minimum exposure time of 20 ms per image. In summary, the main design goal is not driven by trying to attain the highest spatial resolution possible but rather to build an instrument that is as fast, flexible, and easy to use as possible at the 30 nm resolution range.

The design philosophy for the instrument is comprised of several elements with the intent of maximizing the stability of the optical elements and the sample. These elements are simplicity, modularity, uniformity, and spatial minimization of the support and motion stages of the optics and the sample. Simplicity was achieved by implementing only the needed degrees of freedom for each optical element, thereby eliminating the introduction of possible sources of instability from unneeded motion stages. Each of the degrees of freedom was then implemented by a common, modular motion stage design. This ensures that each optical element shares common motion, and thermal and mechanical support characteristics. Minimizing the spatial extent of the support and motion stages was achieved by providing a solid granite mounting surface as close as possible to the optical elements and the sample. Large granite blocks are used, with mounting surfaces located above and below the X-ray beam. This maximizes the stiffness of the support and motion stages. The modularity and uniform design of the support and motion stages ensure a common response to vibration and thermal drift issues.

The design decisions of the instrument and the selection of specific commercial components are based primarily on the experience of two of the team members (DSC and EN). All the major commercial components used here have been characterized and used before at other NSLS-II beamlines, and thus, their level of performance is known. No finite-element-analysis of the instrument or subsections was done.

The location of the SSA, 61 m from the center of the DW source, is determined by the beamline optics’ design.10 The SSA station provides beam intensity monitoring, secondary source aperture definition, phase-diffusion, and beam intensity attenuation. The devices providing these capabilities are contained within a clear polycarbonate chamber. The chamber can be filled with helium (He) to reduce X-ray absorption. The chamber is fitted with an Advanced Micro Instruments Model 60-P2 Oxygen (O2) sensor with a single channel readout. This sensor provides an indirect indication of the helium concentration in the chamber. When the O2 sensor indicates near-zero-% O2, it is assumed that the chamber is mostly filled with He. Beam intensity monitoring is provided by two Advanced Design Consulting Model MIC-205 ion chambers, in series. These ion chambers’ signals are read by an FMB Oxford I-404 Quad Current Integrator module. The module contains a high voltage supply that is used by both ion chambers. The signal currents of the ion chambers are connected to two of the four current inputs. However, it should be noted that since the beam intensities at the SSA location are ∼1013 photons/s, the ion chambers do not provide a reliable quantitative measure of the full beam intensity. A silicon diode is temporarily inserted into the beam path when a quantitative measure of full beam intensity is needed. Nonetheless, the ion chambers are useful for energy calibration with appropriate foils and also for easy nonperturbative monitoring of relative intensity.

The secondary source aperture is defined by a set of JJ X-ray Model AT-F7-HV slits. These slits are motorized and provide for the definition of a slit opening and slit position for each of the horizontal and vertical directions. The aperture size can be set with 1 μm resolution, and the position can be set to 1/3 μm resolution. This aperture is located approximately 2 m upstream of the TXM sample location.

Downstream of the SSA, a spinning disk of paper can be used as an X-ray phase diffuser to make the field of view more uniform. A 12 V direct-current (DC) motor (Ametek Pittman Model 6312S001-R1) is used to spin the disk. The motor is powered by a simple bench DC power supply. The rotational speed of the motor is controlled by manually setting the output voltage of the supply. Typically, the voltage is set between 4 and 5 V, corresponding to about 50–60 revolutions per second. At this rotation speed, the disk rotates at least one full turn for all practical exposure times. The disk is constructed of variable layers of common white filter paper. This phase diffuser was implemented as a complement to the condenser shaker (see below). Beam intensity attenuation is provided by using an XIA Model PF4 pneumatic-actuated attenuator. Attenuation of the beam is achieved by inserting up to four thicknesses (1, 2, 4, and 8) of 38 μm thick UHV aluminum foil. At 9 keV, this provides up to 244× attenuation when all four attenuators are inserted.

The sample station holds all the critical components of the TXM: the sample stage and all the X-ray optical elements. The sample station is supported by a large granite base (Fig. 4, A) with a cut-out in the middle. This granite base is supported by three Unisorb Model RK2 leveling-wedge units, which provide the height and leveling adjustment for the base. The remaining three units provide stabilization through spring washers between the units and the base. The three main leveling units support approximately 80% of the weight of the sample station. Vibration isolation pads, made from alternating layers of stainless steel and 3M 468 tape (a total of 8 stainless steel and 7 tape layers per pad; total thickness per pad is about 3 mm), are inserted between the leveling wedge supports and the granite block. Figure 5(a) shows the vibration spectrum measured with a Data Physics accelerometer system on the floor and the top of the granite base (Fig. 4, A) prior to installation of the vibration isolation pads. The black curve represents floor vibrations, and the red curve represents the spectrum measured on the top surface of the microscope granite base. As seen from the figure, without the vibration isolation pads, cultural noise (typically below 50 Hz) gets significantly amplified on the granite base compared to the floor. After installation of the vibration damping pads, the vibration spectrum was measured again; see Fig. 5(b). The figure indicates that the granite base vibrations above 20 Hz (the red curve) are efficiently reduced when compared to the floor vibrations (the black curve). The height of the granite base is driven by the desire, for stability, to minimize the distance between the x-ray beam and the granite surface. Thus, the final height was determined after all the components on the granite have been selected and details of the stages completed. The width of the granite base was mostly driven by the desire to have space lateral to the beam for mounting auxiliary components such as the visible light microscope, the telescope, and electrical connection interfaces. Because the cost of bulk granite is not prohibitive, it is simpler to make the entire base out of granite instead of a “granite table top and a steel base” combination.

FIG. 4.

A cut through the sample station showing the details. A: granite base, B: granite support for the sample stages, C1-C4: gantry assembly, D: linear sample stage, E: rotary sample stage with a slip ring, F: SmarAct rails (two on C1 and one on C4) for the X-ray optical elements, G: manual telescope on a rail, H: visible light microscope, and I: pinhole. The zone plate, phase ring, and Bertrand lens are not shown in this view. A few breadboard pieces are attached to A and C1 for convenience and general use. The coordinates used are shown, with the beam propagating in the z-direction.

FIG. 4.

A cut through the sample station showing the details. A: granite base, B: granite support for the sample stages, C1-C4: gantry assembly, D: linear sample stage, E: rotary sample stage with a slip ring, F: SmarAct rails (two on C1 and one on C4) for the X-ray optical elements, G: manual telescope on a rail, H: visible light microscope, and I: pinhole. The zone plate, phase ring, and Bertrand lens are not shown in this view. A few breadboard pieces are attached to A and C1 for convenience and general use. The coordinates used are shown, with the beam propagating in the z-direction.

Close modal
FIG. 5.

(a) Vibration spectrum measured on the floor (the black curve) and the top of a granite (the red curve) prior to installation of the isolation pads (b) Vibration spectrum measured after adding vibrational damping (black—floor, red—top surface of the microscope granite base).

FIG. 5.

(a) Vibration spectrum measured on the floor (the black curve) and the top of a granite (the red curve) prior to installation of the isolation pads (b) Vibration spectrum measured after adding vibrational damping (black—floor, red—top surface of the microscope granite base).

Close modal

The sample motion stages sit on a separate granite block (Fig. 4, B) supported by a manually adjusted three-point kinematic mount which provides leveling and height adjustment, relative to the granite base. This granite block sits inside a well that is carved into the granite base (Fig. 4, A). The X-ray optical element motion stages are mounted on an independent granite gantry assembly. This gantry assembly consists of separate pieces of granite (Fig. 4, C1-C4) that are rigidly held together. C1 has a cut-out to accommodate the sample stages. The gantry assembly can be aligned to the beam independently from the sample block. On this gantry assembly, three SmarAct rails are mounted (Fig. 4, F) to accommodate the stages for the X-ray optical elements. One rail sits on the “overhead” piece of granite (Fig. 4, C4) directly over the X-ray path while two rails sit on the lower granite (Fig. 4, C1), laterally displaced from the X-ray path. These rails provide a 400 mm travel range for the SmarAct carriages mounted on them (see below) upstream of the sample position and 520 mm downstream. A close-up of the region around the sample is shown in Fig. 6.

FIG. 6.

Detailed view of the sample region, including details of the sample stage stack (the inset). The distance between the pinhole and the sample is ∼10 mm. The distance between the sample and the zone plate is determined by the zone plate focal length; for our current set of optics, this is ∼19 mm and 38 mm (Table I). For simplicity, the visible light microscope is not shown.

FIG. 6.

Detailed view of the sample region, including details of the sample stage stack (the inset). The distance between the pinhole and the sample is ∼10 mm. The distance between the sample and the zone plate is determined by the zone plate focal length; for our current set of optics, this is ∼19 mm and 38 mm (Table I). For simplicity, the visible light microscope is not shown.

Close modal

The X-ray optical elements are the condenser lens, with a beam stop attached to its upstream end, a pinhole, a zone plate, a phase ring, and a Bertrand lens (Fig. 7). Each of these optical elements is mounted on a modular stack of motion stages of a common design.

FIG. 7.

Picture of the TXM, showing the area around the sample.

FIG. 7.

Picture of the TXM, showing the area around the sample.

Close modal

Each motion stack consists of a SmarAct Model SLLA-42 linear rail positioner for motion along the beam direction, an Attocube Model ECSx5050 linear positioner for motion in the horizontal direction perpendicular to the beam, and an Attocube Model ECSz5050 linear positioner for motion in the vertical direction perpendicular to the beam. Piezo based actuator systems were chosen due to their compact size, minimal heat dissipation, and relatively high (above 300 Hz) resonance frequencies. Similar piezodriven actuators were successfully implemented in various X-ray microscopy systems built and commissioned earlier.12 The pinhole and the zone plate motion stacks hang vertically down from the rail located above the beam on the “overhead” piece of the gantry assembly (see Fig. 4, C4). This scheme provides a short distance from the granite support to the optical element, while allowing the element to be positioned near the sample with minimal intrusion into the space around the sample. The phase ring and Bertrand lens motion stacks are mounted on the lower sets of rails. These optical elements are cantilevered out from the tops of their motion stacks into the beam path on low-mass, high-stiffness, additive-manufactured aluminum arms. The SmarAct motion stages are controlled from a SmarAct Model MCS controller. The Attocube motion stages are controlled from an Attocube Model ECC100 controller.

The condenser lens, with the attached beam stop, is positioned by a combined-motion mechanism, using the same modular motion stages as the other optical elements (Fig. 8). The mechanism provides all six degrees-of-freedom (x, y, z, pitch, roll, and yaw), one of which is not used in practice (the roll). The mechanism is supported on a plate spanning the two lower inboard/outboard SmarAct rails, rigidly connecting two of the passive carriages (one on each rail). This platform provides a wide base which can be manually positioned along the beam direction and then locked into the position. The mechanism consists of two identical stacks of two Attocube Model ECSx5050 and one ECSz5050 linear positioners, located above the inboard/outboard SmarAct carriages. Spanning the tops of these stacks is a rigid bridge supported on a ball and groove (the inset of Fig. 8) three-point kinematic mount with magnetic preloading. This bridge supports an Attocube Model ECGt5050 goniometer, a Piezosystem Jena Tritor 38 condenser shaker, and the condenser lens module. The two stacks of three linear motion stages, when used in combination, provide the three x, y, and z linear degrees-of-freedom and the roll (not used) and yaw degrees-of-freedom. The pitch for the condenser module is provided by the goniometer stage. The vertical distance between the center of the condenser lens to the top mounting surface of the SmarAct carriages is 65 mm. This height is constrained by the need to incorporate the pitch and shaker components underneath the condenser box. Typically, we shake the condenser only in the vertical direction at 100 Hz, which is sufficient for exposure times >20 ms.

FIG. 8.

Close-up of the condenser stage. A: manual SmarAct linear stages, B1 and B2: Attocube ECSx5050 and ECSz5050, C: Attocube pitch, and D: condenser box. The Piezo-Jena shaker is between C and D and is hidden in this view. The inset shows one of the ball-groove kinematic mounts that enable the relative motions between the two motion stacks.

FIG. 8.

Close-up of the condenser stage. A: manual SmarAct linear stages, B1 and B2: Attocube ECSx5050 and ECSz5050, C: Attocube pitch, and D: condenser box. The Piezo-Jena shaker is between C and D and is hidden in this view. The inset shows one of the ball-groove kinematic mounts that enable the relative motions between the two motion stacks.

Close modal

The sample stage stack (see the inset of Fig. 6) starts with a horizontal linear translation stage that provides motion perpendicular to the beam in the x-direction. This linear stage is a Kohzu Model XA16A-L21, with recirculating ball rail bearings and a ball screw, and it is mounted directly onto the sample stage granite table (Fig. 4, B) via an adapter plate. The rotary stage, a Physik Instrumente (PI) Model A-623 with air-bearings and rotary servo motor, sits on top of this horizontal translation stage. The horizontal translation allows the rotation axis of the rotary stage to be brought into the beam. Located on top of the rotary stage are an Attocube Model ECSxy5050 linear positioner and an Attocube Model ECSz5050 linear positioner. These allow the sample to be aligned to the center of rotation of the rotary stage and positioned vertically into the beam. The PI rotary stage is equipped with a Moog Model AC6355-56A slip ring module. This slip ring module allows continuous rotation of the sample without interference from the cables. Opto-electronic (digital) encoders are used in all motion stages above the slip ring to ensure that sliding contacts do not introduce any errors in the position readouts. The slip ring module provides 56 channels, 36 of which are used by the Attocube stages, 4 are frame electrical grounds, and the remaining 14 are available to the user through a set of DB15 connectors. The user may use these as needed to introduce electrical signals into and out of their sample environment. A Newport Model BK-1A-T Kinematic Base Top Plate, attached to the top of the Attocube, provides a flexible, convenient, and position-reproducible sample holder. The system is built to accommodate a sample load as high as 500 g.

Prior to installation at the beamline, the rotary stage was evaluated in the Nanopositioning Laboratory of NSLS-II. Rotational errors were inferred using the setup and method described elsewhere.13 A diamond turned reference cylinder was mounted on top of the rotary stage, and capacitive displacement sensors measured its displacement during rotational motion. From the measurements, the rotation run-out is estimated to be about ±35 nm peak to peak. These errors are consistent with the sub-50 nm resolution in the 3D reconstructed data. Even with a slip-ring installed, the rotational performance of the air-bearing stage exceeds the performance of any conventional ball-bearing rotary stage.

The X-ray sample and optical elements are in the room ambient environment. To reduce air attenuation, telescoping airtight tubes are provided upstream of the condenser box and downstream of the zone plates. The upstream tube is connected to the SSA box, while the downstream tube is connected to the detector box (see below), both of which are filled with He during operations. The tubes are provided with Kapton windows for X-ray transmission. During operations, the X-rays typically traverse ∼0.25 m of air. At 8 keV, this corresponds to an attenuation loss of about 26%.

An enclosure for the sample station has been constructed but not yet installed. This enclosure should reduce the experimental hutch temperature variations seen by the TXM instrument and attenuate acoustic vibrations. However, to date, our data have been successfully reconstructed and, as seen below, the image stability is adequate even without this enclosure. Note that due to the use of the air bearing stage, this enclosure is not airtight and cannot be used to purge the air within with He. Figure 9 shows the air temperature variation at the sample position over 1 h, including the effect of opening and closing the hutch doors, without the sample station enclosure. The standard deviation of the temperature over 1 h is 0.06 °C. Peak-to-peak, the temperature varies by ∼0.25 °C. Opening and closing the hutch door does not have much effect on the temperature.

FIG. 9.

Temperature variation at the sample position over 1 h. The beige shaded regions indicate the times when the hutch door was open.

FIG. 9.

Temperature variation at the sample position over 1 h. The beige shaded regions indicate the times when the hutch door was open.

Close modal

A visible-light microscope is located on the inboard side at the sample location, looking at the sample from a direction perpendicular to the beam. The microscope is motorized for moving the image left-right and for focus. The motion stages are Newport Model MFA-PPD linear stepper-motor stages. The microscope is constructed from standard commercial optical structural elements (cage frames, tubes, thread adapters, etc.). The microscope uses a Mitutoyo 10× Plan Apochromat objective with a 0.28 NA and 34 mm working distance. The camera is an Allied Vision Mako G507B. On the outboard side, a telescope (Navitar 7000 Zoom lens on a WATEC 902B camera) on a manual slide with encoder that runs along the entire length of the granite base is used for rough alignment of the optical components in the beam direction (z). A small display monitor with cross-hairs is located next to the telescope for easy viewing of the telescope camera. Another display is mounted nearby for the manual slide encoder readout. This telescope is very useful for trouble-shooting and initial setup.

There are two identical detector stations. One station is located immediately downstream of the sample station. The other is located 6 m downstream from the sample location. The reason we chose to have detector stations separate from the sample station is to provide more flexibility for the objective zone-plate to detector distance so that future advances in zone plates and detector pixel sizes can be accommodated. Since the required stability at the detectors is in the micron-level (after magnification from the zone plate objective), the separation of the sample and detector stations is not a problem. Each station is provided with a large clear polycarbonate chamber. The chamber can be filled with helium (He) to reduce X-ray absorption. Each of these chambers is fitted with an O2 sensor. The detector modules and their motions are completely contained within the chambers.

Each detector station provides three degrees-of-freedom for a detector-module mounting platform. The platform is a Thor Labs Model MB2530/M Breadboard, 250 mm × 300 mm solid aluminum (Fig. 10, D). This platform is positioned using three motion stages: a Parker Daedal Model 406XR with 1000 mm travel to provide motion along the beam direction (Fig. 10, A), a Kohzu Model XA16F-L22 with 200 mm travel to provide horizontal motion perpendicular to the beam (Fig. 10, B), and a Kohzu Model ZA16A-W2C with 16 mm travel to provide vertical motion (Fig. 10, C). The motorized motion along the beam direction enables automation to maintain constant magnification over the 5–11 keV range of the instrument. This is extremely useful for automated XANES measurements across several elements.

FIG. 10.

Detailed view of the detector station. A: Long linear stage along the beam direction, B: linear stage perpendicular to the beam direction, C: vertical stage, D: detector module mounting plate, E and F: rolls for the two cameras, Inset: close-up of the scintillator and focus stage, G: tip-tilt stage for scintillator, and H: linear stage for focus.

FIG. 10.

Detailed view of the detector station. A: Long linear stage along the beam direction, B: linear stage perpendicular to the beam direction, C: vertical stage, D: detector module mounting plate, E and F: rolls for the two cameras, Inset: close-up of the scintillator and focus stage, G: tip-tilt stage for scintillator, and H: linear stage for focus.

Close modal

There are currently two detector modules in use: one low-resolution and one high-resolution. The detector modules share many design and construction features. Standard commercial optical structural elements (breadboards, cage frames, tubes, thread adapters, etc.) are used where possible. The low-resolution camera is used during the TXM and beamline alignment process. A 100 µm thick YAG scintillator coupled to a 2× Mitutoyo objective lens with a tube lens is attached to a Manta MG 235B camera from Allied Vision Technology. This camera has 5.86 µm × 5.86 µm pixel sizes, with 1936 (H) and 1216 (V) pixels. The scintillator is mounted on a SmarAct Model STT-25.4 tip-tilt stage for angular alignment (Fig. 10, G in the inset), which is mounted on a SmarAct Model SLC-2430-S translation stage (Fig. 10, H in the inset) to allow for the scintillator to be placed at the correct distance from the objective lens for a proper image focus. Camera roll, for alignment of the pixel columns parallel to the sample rotation axis, is enabled by a SmarAct Model SR-7012-S rotation stage (Fig. 10, E). For the high-resolution camera, we reused the Xradia/Zeiss 20 µm thick CsI scintillator coupled to a Nikon 10× objective with a tube lens. The scintillator unit is fixed to the objective; thus, there are no angular adjustments between the scintillator and the objective. Focusing is achieved by a linear translation of the scintillator/objective unit relative to the tube lens, which is rigidly attached to the camera. The detector is a Neo Andor camera. The pixel size is 6.5 µm × 6.5 µm, and there are 2560 (H) × 2160 (V) pixels. With our network and data acquisition system, full frame images can be captured at about 20 frames per second. For this much bigger camera, the camera roll is enabled by a Kohzu Model RA13A-C1-MOD56DIA rotary stage with a Thor Labs Model ZFS25B actuator (Fig. 10, F).

After testing the various components in the Nanopositioning Laboratory, the microscope was assembled and installed at the 18-ID beamline of NSLS-II. Figure 11 demonstrates mechanical and vibrational characteristics of the microscope inside the experimental hutch. Figure 11(a) shows the vibrational spectrum measured on top of the microscope granite slab using a Data Physics accelerometry system. As seen from the figure, most of the mechanical resonances are located below 100 Hz, with the peak near 60 Hz. Figure 11(b) depicts the time domain response at the sample location recorded with a Fabry-Perot Attocube FPS3010 interferometry system. The interferometer head was mounted on the breadboard adjacent to the manual telescope (see Fig. 4, G) approximately 300 mm away from a mirror installed at the sample location. The setup thus measures relative motion in the x-direction. No enclosure was present during the measurements; therefore, air fluctuations and acoustic noise in the hutch contributed to the overall noise level. As seen from the figure, the peak-to-peak noise is typically below 20 nm (4.1 nm rms value). We believe this can be further improved by enclosing the microscope inside of a thermal- and acoustic-attenuating enclosure. Figure 11(c) shows the FFT spectrum of the time domain signal measured at the sample location. The majority of the mechanical resonances are located around 200 Hz, well above the vibrations on the granite base, as shown in Fig. 11(a). The fact that the microscope resonance frequencies (∼200 Hz) are >3 times higher than the granite base (∼60 Hz) makes the microscope less sensitive to the floor vibrations.

FIG. 11.

(a) Mechanical resonance spectrum measured on top of the microscope’s granite slab. (b) Time domain response measured at the sample location utilizing FPS3010 fiber-optic interferometer. Measurement time constant equals 600 µs. (c) FFT spectrum at the sample location showing that the majority of the microscope’s mechanical resonances are around 200 Hz. (d) 200 nm steps performed by the sample centering stage.

FIG. 11.

(a) Mechanical resonance spectrum measured on top of the microscope’s granite slab. (b) Time domain response measured at the sample location utilizing FPS3010 fiber-optic interferometer. Measurement time constant equals 600 µs. (c) FFT spectrum at the sample location showing that the majority of the microscope’s mechanical resonances are around 200 Hz. (d) 200 nm steps performed by the sample centering stage.

Close modal

Figure 11(d) shows interferometrically measured steps performed by the sample centering stage mounted on top of the rotary stage. Sub-200 nm steps can be reliably performed using stick-slip Attocube actuators wired through the slip ring.

As a demonstration of the capabilities of the instrument, a small piece of copper was placed in a Kapton tube and silver nitrate solution was added to it, causing the growth of silver dendrites on the copper surface. The sample was rotated at 3°/s, and the camera acquired the projection images in a “fly-scan” mode. A complete tomography data set (800 projections, 0.05 s exposure time per projection) was collected in 60 s as the sample rotated through 180°. Thus, the 3D silver dendritic growth can be tracked with 1 min time resolution.9 A 3D tomography reconstruction using the “Gridrec” implementation in TomoPy is shown in Fig. 12. Aside from finding the center of rotation in the projection images, no image alignment was performed. A slice through the reconstruction demonstrates the 50 nm spatial resolution.

FIG. 12.

(a) Reconstructed 3D image of silver dendrites growing on a copper substrate [(b) and (c)] slice through the 3D reconstruction at positions indicated in (a), showing 50 nm, 100 nm, and 200 nm features. The X-ray energy was 8.95 keV. 800 projections were taken, with 0.05 s exposure time. The sample was rotated at 3°/s, and the total data acquisition time was 60 s.

FIG. 12.

(a) Reconstructed 3D image of silver dendrites growing on a copper substrate [(b) and (c)] slice through the 3D reconstruction at positions indicated in (a), showing 50 nm, 100 nm, and 200 nm features. The X-ray energy was 8.95 keV. 800 projections were taken, with 0.05 s exposure time. The sample was rotated at 3°/s, and the total data acquisition time was 60 s.

Close modal

In order to characterize the stability of the TXM, a calibration star pattern was imaged repeatedly over 12 h. The results of these scans are shown in Fig. 13. The image shifts are measured relative to the image at “time zero.” For short term stability, the star pattern was imaged at 20 frames per second over a period of 75 s. Here, the image shifts are seen to be <1 pixel (a demagnified size of 17.2 nm) in each direction over the 75 s of the scan. For the long term stability, the star pattern was imaged at 1 frame per 5 min over 12 h. Over this long time scale, the image shifts are <6 pixels (103 nm) in each direction. Currently, we cannot determine if these shifts are due to sample or zone plate motion. Due to the large zone plate magnification, these shifts are unlikely to be due to the detector motion. These small drifts are not a big issue for the current 2D-30 nm or 3D-50 nm spatial resolution imaging due to the short acquisition times.

FIG. 13.

Stability of the TXM as measured by image motion relative to a “time-zero” image; (a) short term stability with images acquired at 20 frames per second for 75 s and (b) long term stability with images acquired at 1 frame every 5 min for >12 h.

FIG. 13.

Stability of the TXM as measured by image motion relative to a “time-zero” image; (a) short term stability with images acquired at 20 frames per second for 75 s and (b) long term stability with images acquired at 1 frame every 5 min for >12 h.

Close modal

Stability of the microscope is determined by two important parameters: stiffness of its critical components (which affect amplitude of vibrations and susceptibility to the cultural noise) and long-term stability determined by thermal drifts induced by self-heating and ambient temperature variations. In a typical synchrotron environment, cultural noise (pumps, controllers, fans, cooling equipment, etc.) causes vibrations in the frequency range well below 100 Hz with a peak around 60 Hz, as seen in the panel of Fig. 11(a). Introduction of isolation pads under the granite block (Fig. 5) helps to reduce cultural noise measured on the top surface of the microscope granite base. During the design phase of the microscope, resonance frequencies of the motion stages were taken into account when motion components were selected. Small form factor, long travel range piezo driven stick-slip actuators are known to have resonance frequencies well above 200 Hz.14 The central resonance peak at around 200 Hz measured at the sample location [Fig. 11(c)] is mainly determined by the air-bearing rotary stage used for the fast sample rotation. The fact that the sample stage fundamental resonance frequencies are at least 3 times higher than the background resonance spectrum makes it less susceptible to the floor vibrations and ultimately improves stability at the sample location during the measurements.

Thermal stability (drifts) also plays an important role during the data acquisition process. Due to the nature of friction-based stick-slip actuators, the amount of heat dissipated during short motion/holding position (not fast scanning) is in the tens of μW range. The main sources of heat pertain to the optoelectronic encoders used to measure the actual position. Depending on the manufacturer, the amount of dissipated heat varies between 200 mW and 500 mW per moving axis. Duty cycling modes are implemented in the controllers to reduce the amount of heat dissipated by an encoder.15 The compact form factor of the motion stages helps to further reduce thermal drifts and make them less susceptible to the temperature oscillations. The microscope components are mounted on a large granite block which serves as a large thermal reservoir and responses slowly to the temperature fluctuations in the hutch. As a result, the TXM components experience smaller temperature variations, thereby reducing the thermal drifts. The advantage of the developed microscope is also related to the data acquisition time, and fast measurements minimize the drifts yielding better tomographic reconstructions.

Thus far, the design of the developed TXM microscope has proven to be adequate and enables robust and fast operation. Mechanical components were chosen appropriately and ensure mechanical and thermal stability. Probably more user operation and various user experiments have to be carried out to determine if any further optimization is required.

As seen above, the performance of the TXM met the design goals. Nonetheless, it is useful to note a few lessons learned from the operations of the instrument thus far (∼1 year).

So far, we find that the visible light microscope has been of little use because samples are easily aligned using the x-ray beam together with the telescope. On the other hand, the telescope is really indispensable.

Operationwise, we have had users bumping into the pinhole during sample placement or during large-sample rotations. The former is difficult to prevent, but perhaps, an engineering solution could be found for the latter—such as a “light curtain.”

The zone plate stage as shown currently does not have any angular adjustments. For better zone plate alignment, especially for stacked zone plates, angular adjustments will be extremely helpful. We have used a SmarAct Model STT-25.4 tip tilt stage (same as the one used for the scintillator) for zone plate tests, but this stage is not optimal because the motions are not reproducible and not encoded.

The condenser shaker can be improved. As described above, the shaker is independent of the other condenser motions. Currently, we normally only shake in the vertical direction at 100 Hz and 15% amplitude (∼5 µm). This is sufficient to homogenize the beam, but it does not significantly affect the beam intensity away from the center, i.e., the overall width of the field of view does not increase very much. We find that if we try to shake with higher amplitudes, the image quality deteriorates, most likely due to vibrations being transmitted to the zone plate. Also, the shaker motion can affect the condenser y-translation—the y-encoders detect the shaker motion, which can cause the y-motion controller to move to “correct” for the motion-encoder discrepancy.

One of the major challenges encountered so far is the problem of sample interactions with the X-ray beam. Bubble formation in liquid samples and thermal-induced motion due to the X-ray beam are two common challenges. The TXM currently does not utilize a shutter to reduce the dose on the sample. A fast shutter capable of ∼10 ms opening times with adjustable duty cycles will be very useful. We plan to implement this capability soon.

In conclusion, a new transmission X-ray microscope (TXM) that is at least an order of magnitude faster than other instruments with comparable spatial resolutions has been designed and tested. This instrument enables collection of a complete 3D tomography data set in under 1 min, with sub-50 nm spatial resolution on 3D reconstructed data, without the need for projection image pre-alignment or sample markers. The instrument currently operates in an ambient environment, without any temperature/air movement controls. At the 30 nm level, the TXM is stable on the few minutes scale, which is sufficient for the data collection.

This research used the 18-ID (FXI) beamline of the National Synchrotron Light Source II, a U.S. Department of Energy (DOE) Office of Science User Facility operated for the DOE Office of Science by the Brookhaven National Laboratory under Contract No. DE-SC0012704. The authors also acknowledge useful discussions with Vincent de Andrade and Francesco De Carlo of the Advanced Photon Source at the Argonne National Laboratory.

1.
Y.
Jiang
,
J. L.
Carvalho-de-Souza
,
R. C. S.
Wong
,
Z.
Luo
,
D.
Isheim
,
X.
Zuo
,
A. W.
Nicholls
,
I. W.
Jung
,
J.
Yue
,
D.-J.
Liu
,
Y.
Wang
,
V.
De Andrade
,
X.
Xiao
,
L.
Navrazhnykh
, and
D.
Weiss
, “
Heterogeneous silicon mesostructures for lipid-supported bioelectric interfaces
,”
Nat. Mater.
15
,
1023
1030
(
2016
).
2.
L.
Mu
,
R.
Lin
,
R.
Xu
,
L.
Han
,
S.
Xia
,
D.
Sokaras
,
J. D.
Steiner
,
T.-C.
Weng
,
D.
Nordlund
,
M. M.
Doeff
,
Y.
Liu
,
K.
Zhao
,
H. L.
Xin
, and
F.
Lin
, “
Oxygen release induced chemomechanical breakdown of layered cathode materials
,”
Nano Lett.
18
,
3241
3249
(
2018
).
3.
H.
Wang
,
B. P.
Weiss
,
X.-N.
Bai
,
B. G.
Downey
,
J.
Wang
,
J.
Wang
,
C.
Suavet
,
R. R.
Fu
, and
M. E.
Zucolotto
, “
Lifetime of the solar nebula constrained by meteorite paleomagnetism
,”
Science
355
(
6325
),
623
627
(
2017
).
4.
J. C.
Andrews
,
S.
Brennan
,
P.
Pianetta
,
H.
Ishii
,
J.
Gelb
,
M.
Feser
,
J.
Rudata
,
A.
Tkachuk
, and
W.
Yun
, “
Full-field transmission x-ray microscopy at SSRL
,”
J. Phys.: Conf. Ser.
186
,
012002
(
2009
).
5.
M.-T.
Tang
,
Y.-F.
Song
,
G.-C.
Yin
,
F.-R.
Chen
,
J.-H.
Chen
,
Y.-M.
Chen
,
K. S.
Liang
,
F.
Duewer
, and
W.
Yun
, “
Hard x-ray microscopy with sub 30 nm spatial resolution
,”
AIP Conf. Proc.
879
,
1274
(
2007
).
6.
V.
De Andrade
,
A.
Deriy
,
M. J.
Wojcik
,
D.
Gursoy
,
D.
Shu
,
K.
Fezzaa
, and
F.
De Carlo
, “
Nanoscale 3D imaging at the Advanced Photon Source
,” (
2016
).
7.
J.
Lim
,
H.
Kim
, and
S. Y.
Park
, “
Hard x-ray nanotomography beamline 7C XNI at PLS-II
,”
J. Synchrotron Radiat.
21
,
827
831
(
2014
).
8.
See www.zeiss.com/microscopy/us/x-ray.html for vendor instruments.
9.
M.
Ge
,
D. S.
Coburn
,
E.
Nazaretski
,
W.
Xu
,
K.
Gofron
,
H.
Xu
,
Z.
Yin
, and
W.-K.
Lee
, “
One-minute nano-tomography using hard x-ray full-field transmission microscope
,”
Appl. Phys. Lett.
113
,
083109
(
2018
).
10.
W.-K.
Lee
,
R.
Reininger
,
W.
Loo
,
R.
Gambella
,
S.
O’Hara
,
Y. S.
Chu
,
Z.
Zhong
, and
J.
Wang
, “
FXI: A full-field imaging beamline at NSLS-II
,”
Proc. SPIE
9592
,
959209
(
2015
).
11.
J.
Wang
,
K. Y.-C.
Chen
,
Q.
Yuan
,
A.
Tkachuk
,
C.
Erdonmez
,
B.
Hornberger
, and
M.
Feser
, “
Automated markerless full field hard x-ray microscopic tomography at sub-50 nm 3 dimensional spatial resolution
,”
Appl. Phys. Lett.
100
,
143107
(
2012
).
12.
E.
Nazaretski
,
H.
Yan
,
K.
Lauer
,
N.
Bouet
,
X.
Huang
,
W.
Xu
,
J.
Zhou
,
D.
Shu
, and
Y. S.
Chu
, “
Design and performance of an x-ray scannig microscope at hard x-ray nanoprobe beamline of the NSLS-II
,”
J. Synchrotron Radiat.
24
,
1113
(
2017
).
13.
W.
Xu
,
K.
Lauer
,
Y. S.
Chu
, and
E.
Nazaretski
, “
A high precision instrument for mapping of rotational errors in rotary stages
,”
J. Synchrotron Radiat.
21
,
1367
1369
(
2014
).
14.
E.
Nazaretski
,
X.
Huang
,
H.
Yan
,
K.
Lauer
,
R.
Conley
,
N.
Bouet
,
J.
Zhou
,
W.
Xu
,
D.
Eom
,
D.
Legnini
,
R.
Harder
,
C.-H.
Lin
,
Y.-S.
Chen
,
Y.
Hwu
, and
Y. S.
Chu
, “
Design and performance of a scanning ptychography microscope
,”
Rev. Sci. Instrum.
85
,
033707
(
2014
).
15.
E.
Nazaretski
,
J.
Kim
,
H.
Yan
,
K.
Lauer
,
D.
Eom
,
D.
Shu
,
J.
Maser
,
Z.
Pesic
,
U.
Wagner
,
C.
Rau
, and
Y. S.
Chu
, “
Performance and characterization of the prototype nm-scale spatial resolution scanning multilayer Laue lenses microscope
,”
Rev. Sci. Instrum.
84
,
033701
(
2013
).