DIY XES - development of an inexpensive, versatile, and easy to fabricate XES analyzer and sample delivery system

Abstract

The application of X-ray emission spectroscopy (XES) has grown substantially with the development of X-ray free electron lasers, third and fourth generation synchrotron sources and high-power benchtop sources. By providing the high X-ray flux required for XES, these sources broaden the availability and application of this method of probing electronic structure. As the number of sources increase, so does the demand for X-ray emission detection and sample delivery systems that are cost effective and customizable. Here, we present a detailed fabrication protocol for von Hamos X-ray optics and give details for a 3D-printed spectrometer design. Additionally, we outline an automated, externally triggered liquid sample delivery system that can be used to repeatedly deliver nanoliter droplets onto a plastic substrate for measurement. These systems are both low cost, efficient and easy to recreate or modify depending on the application. A low cost multiple X-ray analyzer system enables measurement of dilute samples, whereas the sample delivery limits sample loss and replaces spent sample with fresh sample in the same position. While both systems can be used in a wide range of applications, the design addresses several challenges associated specifically with time-resolved XES (TRXES). As an example application, we show results from TRXES measurements of photosystem II, a dilute, photoactive protein.

Introduction

X-ray emission spectroscopy (XES) is a valuable technique used to study electronic structure, commonly of 3d transition metals because of their important role in catalysis and biology. This element-specific probe can give bulk information about the electronic structure and, depending on the emission lines recorded, can be used to identify the oxidation state, ligand type, and covalency.^[1–3] As hard X-ray sources have improved significantly over the last decade, XES has become more widespread. XES measurements with benchtop sources are generally limited to concentrated samples, but new sources, such as metal liquid jets,^[4] are pushing those limits with even higher X-ray fluxes. Time resolved XES (TRXES) measurements have also become more common with the development of X-ray free electron lasers (XFELs),^[5–10] advanced synchrotron beamlines,^[11–14] and is starting to emerge in a few laboratory based sources^[15–17].

Though proven to be an important measurement modality, XES is not without challenges. Signals are inherently weak due to low fluorescent yields, limited analyzer reflectivity, and small solid angle capture. These effects can severely debilitate XES measurement capabilities when using low intensity X-ray sources or when samples are dilute, X-ray damage sensitive, or limited in volume. To combat these limitations, large arrays of crystal analyzers have been employed to maximize solid angle capture^[18]. Although this method is important for facilities with dedicated setups and larger budgets, smaller laboratories are often limited in physical space or funding. In some cases, energy resolution can be sacrificed by using highly annealed pyrolytic graphite analyzers with greater integral reflectivity, which allows a larger range of energies to reflect a single energy^{[19, 20]}. Although such analyzers can have many important applications, a delicate balance must be struck between the capacity to measure a spectrum and the resolution required to interpret it. In an effort to maintain a high energy resolution and still achieve high signal collection through large solid angle, a new cost effective analyzer fabrication process is described, which can be used to create larger, monocrystalline analyzers for single or multi analyzer spectrometers. Even single analyzer setups can achieve a high collection efficiency^[21] when large analyzers are used. Here, we describe a new fabrication protocol to create diced or segmented von Hamos analyzers which simulate a cylindrical curve through a series of flat segments. These have additional applications outside of TRXES including resonant inelastic scattering and wavelength dispersive absorption.

Another limiting factor of TRXES is sample replacement. Typically for photoactive samples measured at XFELs, a liquid jet is used to rapidly replenish the sample between X-ray pulses. As the sample moves toward the interaction region, the sample is exposed to a laser pump pulse and the time of flight is used to alter the delay before the X-ray probe. Even with a high pump and probe rate, the liquid jet sample delivery methodology can be extremely wasteful with large portions of the jet passing through the X-ray interaction region without being measured. This increases sample volume requirements which can become prohibitive for many systems. Fixed targets allow for minimal volumes to be used in some cases, however, significant work is required for the fabrication of substrates and sample loading which can make this methodology prohibitive and costly^[22–24]. A recently reported drop on tape system has largely overcome these deficiencies^[25]. Specifically, this system uses an acoustic droplet ejection system to deposit drops on a thin film which are translated to predetermined locations for both the pump and probe excitations. Here we present a design with a much lower price point that circumvents the need for acoustic ejection. This system is fairly compact and can be easily integrated with synchrotron or laboratory-based sources. We also note that applications of this system are also useful for traditional XES of liquid samples that are easily damaged by X-rays or when limited sample volume is available. This sample delivery system is also useful for room temperature measurements or for automated sample replenishment during longer duration measurements, which are common with laboratory sources.

Instrumentation and design

Analyzer fabrication

Previously reported fabrication processes for segmented Hamos analyzers include silicon on insulator wafers used in deep reactive etching^[26] or smaller wafers partially diced and broken during pressing^[27]. Both methods successfully produced high quality segmented von Hamos crystals, however the method presented here is compatible with any crystal type, avoids the need of specialized equipment beyond a dicing saw, and avoids possible complications with fragility or small strip widths. As such, this process is a, low cost, and accessible approach to create custom von Hamos analyzers.

This fabrication process for von Hamos analyzers expands on the prior report which outlines the development of an application-based XES spectrometer^[21]. The decision process for choosing an analyzer crystal/wafer material and radius are discussed. An overview for the analyzer fabrication process is given, Figure 1, and techniques to improve the reliability of this process are also addressed.

Figure 1.

An overview of steps involved in the fabrication of diced von Hamos analyzers including: (A) dicing wafer to shape, (B, C) binding wafer to plastic, (D) dicing wafer only into strips while removing excess plastic, and (E,F) fixing the diced wafer‐bound substrate onto an Al base. (B) Emphasizes the process of creating a layer of epoxy between the plastic and wafer without bubbles (glue is shown as red) by bending the plastic cylindrically and pressing onto a line of glue down the center of the wafer.

Central to analyzer design are the choices of crystal material, analyzer size, and radius of curvature. The material and lattice plane orientation define the reflectivity, or integral intensity of the rocking curve, and reflection angles for the spectral energy range. These factors will significantly impact the overall signal intensity and energy resolution as shown in Figure 2. Crystal parameters and reflection angles can be found in tabulated form using the online analyzer atlas^[28] or can be calculated using the programing package XOP^[29]. Bragg’s law can be used to calculate the angle, θ, at which an X-ray of a specific energy, E_p, will be reflected and can be written as

$$\theta =\text{arcsin}\left(\frac{E_0}{E_p}\right) \mathit{\text{where}} E_0=\frac{\mathit{\text{nhc}}}{2d}$$

where E₀ is lowest energy a crystal will diffract for a given reflection, n is an integer, h is Planck’s constant, c is the speed of light and d is the lattice spacing of the crystal. In addition to concerns about energy resolution, the distance between the sample and reflected emission spectra to be recorded by the X-ray detector has to be considered to allow adequate space for both the sample environment and the detector. This distance, D, is simply calculated as:

$$D=\frac{2R}{\mathit{\text{tan}}\left({\theta }_{\mathit{\text{low}}}\right)}$$

Where R is the radius of curvature of the analyzer and θ_low is the Bragg angle corresponding to the lowest recorded energy. Another consideration is the spectral length, L, or the spatial separation between the lowest desired energy and the highest at the analyzer focus, which should not be longer than the active area of the X-ray detector used. This length can be expressed as

$$L=\frac{2R}{\mathit{\text{tan}}\left({\theta }_{\mathit{\text{high}}}\right)}-\frac{2R}{\mathit{\text{tan}}\left({\theta }_{\mathit{\text{low}}}\right)}$$

where θ_high is the Bragg angle corresponding to the highest recorded energy.

Figure 2.

The energy resolution for a von Hamos analyzer given the angle of reflectance. In all graphs the rocking curve is taken to be μrad, the X‐ray spot size is 100 μm FWHM. (a) Different components that contribute to the final or total energy resolution, (b) the total energy resolution is found for two radii and two pixel sizes, corresponding to a Pilatus (172 μm) and a Mythen detector (50 μm) with the highlighted section expanded for reference.

Careful consideration must be given to balance between the Bragg angle affecting energy resolution and radius of curvature affecting solid angle coverage and spectrometer size.

Analyzer size is limited only by the dicing saw working area. Although finding large wafers commercially can be challenging, single crystal ingots can be cut at an angle or axially to give increased cross-sectional areas with the correct orientation. Here we used a Si(110) wafer, ≥200 mm in diameter, with a polished surface for X-ray diffraction. We then cut the wafer to our preferred outer dimensions of 190 mm by 45 mm, Figure 1a.

A thin flexible substrate of 1/16 inch (1.6 mm) thick polycarbonate was prepared for bonding to the cut wafer by lightly scarring the bonding surfaces with a heavy grit sand paper to improve adhesion. A single line of epoxy was laid down the center of the bonding surface of the optic. A volume of ~4.5 of Epotek 301 was used for ever cm² of wafer surface to be bonded. To avoid trapping air bubbles in the epoxy, the plastic was bent cylindrically, pressed into the glue, and then slowly released while maintaining pressure along the center of the optic/plastic surface, Figure 1b. The optic/plastic was then placed in parallel plate press with ~90 kPa pressure. The epoxy was left for a minimum of 72 hours to ensure proper curing.

The wafer is then mounted for the dicing process on dicing tape with minimum amounts of heat to prevent surpassing 65 °C, the glass transition temperature of the Epotek 301 epoxy. The plastic-bonded wafer was diced into small strips while leaving the plastic substrate uncut except to remove excess plastic around the perimeter, Figure 1d. The final focus of the analyzer in the non-dispersive direction is about twice the width of each strip, or segment, such that smaller strips lead to a tighter focus but smaller strips require additional cuts in the dicing saw reducing total reflectivity surface. Analyzers described here used strip widths of 0.75–1 mm to maintain a tight focus. A dicing blade minimum width of 100–300 μm thick was also used.

A cylindrical convex base with the chosen analyzer radius and a cylindrical concave base, with a larger radius offset for the thickness of the crystal and plastic, are then used to give the analyzer the final form, Figure 1e. For analyzers developed in our lab, wire electrical discharge machining was used to create both cylindrical pieces out of aluminum. Here, ~4.5 mg/cm² Epotek 301 was used between the plastic and the concave base. A uniform 4 μm thick Kapton film was placed over the convex pusher and sprayed with a lubricant to prevent the convex pusher being glued to the top of the analyzer. The analyzer was then pressed with no higher than 90 kPa as higher pressures increase the change of fracturing the crystal strips and 90 kPa was found to be more than sufficient for our 0.5 m radii analyzers. Although the same radius and material were chosen for all large analyzers, smaller prototypes of shorter radius (0.25 m) were also developed with this method without the need for modification.

Spectrometer

The spectrometer used with this setup follows a similar methodology to that published previously,^[21] with the distinction that the 3D-printed He chamber accommodates three analyzers, Figure 3. The crystals are held in place with additional 3D prints which captivate the crystals against the back wall of the spectrometer. The spectrometer wall is oriented so that reflections of each crystal are parallel, but not overlapping, on the detection surface. This design approach makes the system independent of internal motors, reducing cost and setup time while maintaining a higher energy resolution than systems that use overlapping spectra. The size of the entrance aperture was determined based on ray tracing from the sample source to the crystal analyzers with and additional ~3 mm clearance for thick samples or misalignment. Both entrance and exit aperture are covered with a Kapton film, whereas the top is sealed with an O-ring and a thick plastic cover. A constant flow of He inside the spectrometer is used to reduce X-ray scattering and absorption by purging other heavier element gasses from the spectrometer.

Figure 3.

A) (A) The 3D‐printed spectrometer with three von Hamos analyzers inside. Analyzers are shown in orange for visibility. (B) The Pilatus 100 k image of the emission spectra from MnO showing the reflection of each of the three analyzers.

With the three crystal spectrometer, a 2D position sensitive X-ray detector is required. For the Mn Kβ spectrometer reported here, we use a Pilatus 100k (Dectris) ^{[30, 31]}.

Sample delivery system

A sample delivery system has been developed to run in two different modes: deposit discrete nanoliter droplets of liquid sample on a thin film similar to a system used at the Linac Coherent Light Source^[21], or deposit a continuous line of liquid on the thin film. This system can be used to automate sample delivery for time-resolved studies, room temperature measurements, and rapid exchange of samples sensitive to X-ray induced damage. Additionally, it can greatly reduce the sample volume required for measurement. The cost effective design makes this system practical for X-ray experiments done at both synchrotron-based and benchtop-based sources as well as other experiments including laser based spectroscopy. The physical sample delivery system can be outlined in three parts: the syringe pump, deposition system, and sample substrate replacement. For control, an Arduino microcontroller is used to handle the motors (StepperOnline part #17HS19–0406S and 17HM19–0406S) and timing.

Syringe pump

The syringe pump system uses a mounted trapezoidal thread lead screw to convert stepper motor rotation (0.9 degree per step) into a linear motion. A syringe is mounted firmly and the plunger pressed with each incremental microstep (see Figures 4a and 5). The sample is then transported to the deposition system nozzle using a 19 gauge PTFE tubing with Luer Lock connector (Hamilton item # 90619). A smaller diameter nozzle (0.016 in. ID 0.035 in. OD) was taken from plastic syringe needles (JensenGlobal part number # JG22–0.5XPRO) and glued inside the end of the tubing to produce smaller droplets.

Figure 4.

(a) Overhead view of the syringe pump setup with a stepper motor driving a lead screw. A metal L‐beam is then used to push the plunger on the syringe. (b) The deposition system with the rotary arm is shown. The tubing holding the plastic nozzle that deposits the sample is mounted in the rotary arm and positioned using the manual three‐axis stage. (c) The sample substrate replacement assembly with: A roll of thin plastic with the plastic cut to six segments (right), a thin Teflon sheet cylindrically bent to a 25 cm radius of curvature, the stepper motor attached to a wide winding spool, or spindle, for collecting used substrate (left), the 3D‐printed frame (red) with the 25 cm curvature, and springs on top of the roll of plastic substrate which can be used to control the tension in the film by tightening screws (not show).

Figure 5.

The combined sample delivery assembly is shown. One possible orientation for incoming x‐rays is shown in blue.

Deposition system

The deposition system uses a rotating arm attached to a stepper motor, Figure 4b and 5. When depositing single nanoliter droplets, a metal arm swings in and the nozzle barely contacts the film. The sample adheres making a small droplet, and the arm returns to the original position. The nozzle diameter and the syringe pump rate determine the sample spot size which we tuned to be approximately 400 μm in diameter. If a continuous stream is desired the nozzle is left in place, lightly touching the substrate. For alignment purposes, the arm/stepper motor system is mounted on a manual three axis stage. This allows adjustments to ensure correct sample positioning on the film and to maintain a soft contact between the nozzle and thin film while leaving the film in the correct sample position.

Sample substrate replacement

As the substrate for the sample droplets, we used 3.6 μm thick Mylar or polypropylene. The film was purchased (Chemplex) with 63.5 mm wide rolls. Each roll was cut into strips (Figures 4c and 5) on a lathe with a razor blade at 1–1.5 cm intervals. Once cut, the roll was mounted to a rod with slightly smaller diameter than the internal diameter of the film roll. The film is pulled tightly against a Teflon sheet which is mounted to a 3D printed support to give it a 0.25 m radius of curvature. A 12 mm × 20 mm window in the Teflon is where the nozzle contacts the film, and the droplets are deposited and measured. The curvature ensures the film is tight across the window as thin substrate is slowly wound around a winding spool attached to a stepper motor. The large diameter of the winding spool ensured only slight variations in the spool width as the film was wound, making negligible film loss from over stepping. Due to the small step size of 1–2 mm, a single roll of film (6 × 10 mm strips) can be used with ~0.25–0.5 million drops. During measurement at a synchrotron source, the thing plastic substrate lead to no detectable background from X-ray scattering.

Software control

An Arduino Uno with an Adafruit Motorshield and 12 V power supply were used to control timing and motor stepping. When the Arduino receives a high transistor-transistor logic, or TTL, signal it initiates the following sequence: (1) simultaneously activate the syringe pump, move the thin film, and start moving the needle into position; (2) as the needle moves into position all other motors finish and the needle motor speed is reduced to slow the impact with the thin film; (3) after the needle barely touches the thin film, it returns to its original position; (4) a TTL signal is optionally sent to open any optical or X-ray shutters; and 5) wait for another signal to repeat. For the current setup the whole process can run at frequencies up to 10 Hz. Faster frequencies are possible if the distance travelled by the needle arm are reduced, or if the motorshield, which is limited by the I²C communication rate, is replaced. Microstepping throughout this sequence was found to be essential to reduce vibrations in the thin film from both the winding spool and the needle impact. The syringe pump is also set to microstepped to limit the deposition to nanoliters.

Time-resolved X-ray emission spectroscopy

Beamline setup

Using the spectrometer and sample delivery systems described herein, Mn Kβ emission spectra were collected at beamline 14-ID B^[13] at the Advance Photon Source at Argonne National Laboratory. A pink beam was used with the U27 undulator gap set to 11.7 mm, which corresponds to a central energy of about 7.85 keV. Two sets of Kirkpatrick Baez mirrors were used for focusing the beam to spot size of ~100 μm × 100 μm and collimating the X-rays. Each sample was exposed to one 22-μs pulse train containing 1.5 × 10¹¹ photons and the sample was replenished at 10.3 Hz. To incorporate the sample delivery system with the X-ray spectrometer, the sample delivery system was mounted upside-down as compared with that shown in Figure 5 and the thin film was angled at ~40° from the X-ray path.

The X-rays were made monochromatic by a Si(111) channel-cut monochromator for calibration of the emission spectrum recorded by the Pilatus. The undulator gap was set to 10.5 mm when the monochromator was used in order to maximize the photon flux at lower energies. Scattering peaks were recorded with steps of 2 eV across the energy range.^[32]

Laser advancement and timing

Samples of Photosystem II were either exposed to no laser pump or one flash, corresponding to the majority S₁ and S₂ states respectively. Both data sets were collected at 10 Hz with the latter being measured 500 μs after illumination from a 527 nm laser with 220 μm diameter spot size, a pulse length of ~200 ns and an energy density of 11 mJ/mm².

Spectroscopic results from photosystem II

To demonstrate the capability of the analyzer and sample delivery systems, we measured Mn Kβ emission from the oxygen evolving complex (OEC) of photosystem II at room temperature. The OEC, which features a unique Mn₄CaO₅ core, was chosen as an example case because it has very low Mn Content (~500 μM) and is highly X-ray damage sensitive.^[33–35] The preparation and characterization of these samples prior to beamtime followed previously reported protocols,^[27] with the exception D₂O buffer for the final resuspension and centrifugation steps before freezing. At the beamline a solution of 0.4 M sucrose, 5 mM CaCl₂, 5 mM MgCl₂, and 15 mM NaCl, 50 mM MES, pD 6.0 in D₂O was used for resuspension of PS II membrane fragments. A solution of 50 mM PPBQ, an artificial quinone that acts as an electron acceptor, in dimethyl sulfoxide was added to obtain a PPBQ concentration of 500 μM. The sample was then loaded into a syringe connected to the delivery system. All preparation was done in the dark with a dim green light.

Using TRXES, we were able to capture the shift between two different oxidative states, or S-states, of the oxygen-evolving complex (OEC) at room temperature. The data after smoothing using a 9-point boxcar average and a linear background subtraction are shown in Figure 6. From the spectra, we detected a shift in the Kβ_1,3 peak to lower energy. This shift indicates an average higher oxidation state (reduced localized spin density) for the average Mn atom in the OEC which is in agreement with data achieved previously,^[14] only the detection system reduced the data collection requirement according to the 6-fold increase in solid angle because of the new spectrometer design, and the sample delivery system conserved sample and significantly reduced the labor required.

Figure 6.

The Mn Kβ X‐ray emission mainlines for the S1 and S2 majority states. The spectra are normalized by the integral emission intensity between 6470 and 6495 eV. The difference spectrum shows the small shift in spectral position between the two data sets.

Conclusions

The development of more advanced X-ray sources provides more accessibility and capability for XES. Here, we present a way to improve detection of the emission signal by capturing more fluorescence using large area, small radii of curvature, von Hamos X-ray analyzers fabricated a new protocol. The outlined method allows for advanced customization while maintaining a low price point for analyzer development.

An application specific spectrometer design is outlined for Mn Kβ emission spectroscopy that can be adapted for any energy range. The system overcomes the need for motors by using a fixed geometry inside a He chamber that was 3D printed. This system allows for easy integration into beamlines and laboratory sources for XES and TRXES.

A sample delivery system was also developed that can deposit nanoliter droplets or a continuous stream onto a thin film for measurement. This system can be easily recreated with minimal effort and can be used for X-ray and laser spectroscopies conducted at synchrotrons or in the lab. This enables room temperature X-ray spectroscopy of liquid samples by automatically and reproducibly positioning nanoliters of liquid in a small interaction region.

We emphasize that with the spectrometer and sample delivery system mentioned herein, we were able to obtain time-resolved spectra at room temperature using synchrotron source to study a protein that is radiation damage sensitive, low concentration of Mn ions, and limited volume. TRXES results measured from PSII show the effective integration of these systems at a synchrotron beamline and infer their potential for widespread application in XES.