The design and fabrication of Fresnel zone plates for hard x-ray focusing up to 25 keV photon energies with better than 50 nm imaging half-pitch resolution is reported as performed by forming an ultrananocrystalline diamond (UNCD) scaffold, subsequently coating it with atomic layer deposition (ALD) with an absorber/phase shifting material, followed by back side etching of Si to form a diamond membrane device. The scaffold is formed by chemical vapor-deposited UNCD, electron beam lithography, and deep-reactive ion etching of diamond to desired specifications. The benefits of using diamond are as follows: improved mechanical robustness to prevent collapse of high-aspect-ratio ring structures, a known high-aspect-ratio etch method, excellent radiation hardness, extremely low x-ray absorption, and significantly improved thermal/dimensional stability as compared to alternative materials. Central to the technology is the high-resolution patterning of diamond membranes at wafer scale, which was pushed to 60 nm lines and spaces etched 2.2-μm-deep, to an aspect ratio of 36:1. The absorber growth was achieved by ALD of Ir, Pt, or W, while wafer-level processing allowed to obtain up to 121 device chips per 4 in. wafer with yields better than 60%. X-ray tests with such zone plates allowed resolving 50 nm lines and spaces, at the limit of the available resolution test structures.

X-ray focusing optics for hard x-ray energies and resolution below 25 nm becomes necessary to analyze in volume, centimeter-size objects with critical details such as defects in encapsulated electronic circuitry1 cracks in aviation components,2 pore sizes and composition in rocks,3 compositional details in biological samples,4 or in situ functioning investigations of battery electrodes.5,6 With the introduction of 3D printing of mechanical components in various industries, it is envisioned that the x-ray quality control and diagnostic with zoom-in capabilities to submicrometer detail will become customary for high-end components.7–9 Fresnel zone plates (FZPs) constitute handy solutions for focusing hard x-rays to tens of nanometers10–13 comparable in resolution to large adaptive mirrors,14 while also offering better resolution than capillary optics15 or compound refractive lenses, which are all much bulkier solutions.16,17 However, for good penetration into millimeter to centimeter-large objects, photon energies have to be increased to the 10-keV range or beyond. This makes necessary thicknesses of Fresnel zone rings be in the few micrometer range, to provide sufficient absorption or phase shifting. For instance, for a phase shifting of π at 25 keV, one would need >4 μm of Ir, Pt, Au, or W. Since the resolution of FZPs (δ) is given by the smallest zone width (δ1.22Δr d, where Δr is the outermost zone width),18,19 resolutions of <25 nm can be obtained only with FZPs having the finest zones of <25 nm. This pushes the FZPs aspect ratios to tens-to-one to hundreds-to-one values. Such aspect ratios cannot be obtained directly with standard electron beam techniques and require subsequent development of additional processes, such as etching the pattern into an underlying material,11,12 or sequentially repeating lithography and electroforming steps.13,20 One of the hopeful variations of this technique uses electron beam lithography (EBL) to pattern low aspect ratio metal zone plate patterns, then use them to deep-etch silicon by metal assisted etching.21 Such zone plates claim achieving 66:1 (Ref. 22) and even 500:1 (Ref. 23) aspect ratios after zone doubling. The technique is hard to control and uses silicon as scaffold, resulting in higher absorption of x-rays than diamond. Yet another way to obtain ultrahigh aspect ratio FZPs is based on sequential coating of cylindrical wires24 or capillary lumens25 with sputtering or atomic layer deposition (ALD) of high-Z/low-Z films of thicknesses following the zone widths rule, then slicing with focus ion beam disk-shape FZP structures. Such techniques, although not limited in aspect ratios, are highly demanding for the thickness control in the deposition processes achievable and result in FZPs of low aperture sizes. Stacking of zone plates is also a way to increase aspect ratios26 or reduce effective zone widths by interlacing27,28 but high aspect ratio individual FZPs are still desirable to start with, to reduce the number in the stack and reduce that way the mounting burden.

One lately developed technique to decrease zone widths and increase aspect ratios is to form low density material scaffolds by lithography and etching, followed by ALD coating of the scaffold with a high-density material.29 While long-ago known as “edge-defined technique,”30 spatial frequency doubling is popular with micro-optics and manufacturing of gratings or arrays of electrodes, where exact frequency multiplications can be obtained. However, with FZPs, the technique leads to violation of one of the zone plate equations, namely, that all zones should be equal in area. With frequency multiplication, only one zone can have the right width of the absorber/phase shifter ring—the zone equal in width to the ALD layer thickness, while all others are only partially to marginally fulfilling the requirement, leading to a drastic decrease of diffraction efficiency (practically from tens to few percent),31,32 but with important gain in imaging or focusing resolution due to the decrease of the smallest zone size (which is given by the ALD layer thickness). We present a study of ALD-based frequency multiplication with ultra-nanocrystalline diamond (UNCD) scaffolds, the formation of functional FZPs by subsequent processing to form membranes holding the optical structures, and test results of such FZPs on beam lines 1BM and 32ID of the Advanced Photon Source (APS).

The configuration of the FZP was chosen to be of the so-called composite design.18 This consists of a central FZP, varying in zone widths from a central zone (diameter of 6.32 μm) to an outermost zone width of 60 nm, then an outer FZP continues with zones from 180 to 100.5 nm. The 3× scaled-up outer FZP has its third order focus, where the central FZP has its first order focus, thus, adding to the light diffracted toward the first order focus of the central FZP an additional energy contribution. The area ratios Aouter/Ainner of the two zone plates is 227% showing that 227% more incident energy is harvested from the beam, out of which a fraction (4%–9%, depending on the diffraction efficiency of the outer zone plate in its third order focus) is pumped in the common focus. While, theoretically, the outer FZP can be continued to reach again 60 nm zones, its diameter would have increased too much for practical x-ray applications in the targeted beam line (32ID of APS), which used to work with ∼300 μm diameter beams at the zone plate objective position.33 Designs of composite FZPs with equal outermost zone widths in the inner and outer zone plates are possible. The features of the fabricated FZPs are described in Table I.

Table I.

Targeted characteristics of the fabricated FZPs.

ParameterValue
Finest zone width on scaffold 60 nm 
Finest zone width with frequency multiplication 20 nm 
Equivalent period of the frequency multiplied FZP 40 nm 
First order focus distance with frequency doublinga 10.3 cm 
First order focus distance with frequency triplinga 6.7 cm 
Inner FZP diameter 166 μ
Outer FZP diameter 300 μ
Structure height (thickness of FZP) 2.1 μ
ParameterValue
Finest zone width on scaffold 60 nm 
Finest zone width with frequency multiplication 20 nm 
Equivalent period of the frequency multiplied FZP 40 nm 
First order focus distance with frequency doublinga 10.3 cm 
First order focus distance with frequency triplinga 6.7 cm 
Inner FZP diameter 166 μ
Outer FZP diameter 300 μ
Structure height (thickness of FZP) 2.1 μ
a

Calculated for 25 keV.

The main ideas behind the fabrication scheme and frequency multiplication are shown in Fig. 1. The choice was to have 60 nm smallest zones made by EBL in 450-nm-thick hydrogen silsesquioxane (HSQ) resist, spin-coated on UNCD. Once exposed and developed, the HSQ resist transforms into an effective hard mask (similar to SiO2) for deep etching the UNCD in oxygen plasma. Reactive ion etching (RIE) transfers the pattern into UNCD. The fabrication continues with etching of windows from the back side of the support Si wafer with front-to-back alignment, using a low stress SiNx mask and potassium hydroxide (KOH) etchant. ALD of Ir, Pt, or W was performed to a thickness of 20 nm. This led to a frequency tripling for the finest zones of 60 nm, as explained in Fig. 1(b), but with one every third zones missing in the positions coinciding with the centers of the 60 nm UNCD wedges. As can be seen, when the trench and wedge sizes increase to 80 nm toward the center of the FZP (area 2), the ALD-made zones fall in antiphase and vanish reciprocally their contribution to the focus. At 100-nm trenches and wedges (area 3), the ALD layer is equivalent to a fivefold frequency multiplication, but three of every five zones are missing. This scheme continues toward higher trench and wedge widths, leading to an overall decrease in efficiency of the frequency-multiplied FZPs compared to non-frequency-multiplied ones, and excitation of higher order foci. It was shown that frequency multiplied zone plates also have the even orders of foci excited.29 This includes the appearance of a foci corresponding to frequency doubling. Overall, a low efficiency of these FZPs is expected, but with a significant gain in resolution.

Fig. 1.

(Color online) (a) Fabrication sequence: 1—EBL with HSQ resist on top of UNCD/Si wafer; 2—reactive ion etch (RIE) of UNCD to ∼2.1 μm depth; 3—back side window opening and ALD coating. (b) Frequency multiplication with missing zones approach for interpreting physical frequency doubling for a diamond-scaffold ALD-coated FZP: In the 60-nm trench region (1), one can view a 20-nm ALD coating as realizing a frequency tripling of the scaffold spatial frequency, but one of every three zones are missing. In the region (3) with 100-nm trenches, the same 20 nm of the ALD material would multiply the scaffold frequency by 5, but 3 of every 5 opaque zones are missing within a period of the scaffold. In the region (2) with 80 nm zones, the ALD-zones fall in antiphase and their contribution vanishes. In transition regions, such as, e.g., between (1) and (2), a frequency multiplication by 3 works, but the contributions to the focus gradually decrease due to the mismatch between the ideal and resulted zone positions for an integer frequency multiplication.

Fig. 1.

(Color online) (a) Fabrication sequence: 1—EBL with HSQ resist on top of UNCD/Si wafer; 2—reactive ion etch (RIE) of UNCD to ∼2.1 μm depth; 3—back side window opening and ALD coating. (b) Frequency multiplication with missing zones approach for interpreting physical frequency doubling for a diamond-scaffold ALD-coated FZP: In the 60-nm trench region (1), one can view a 20-nm ALD coating as realizing a frequency tripling of the scaffold spatial frequency, but one of every three zones are missing. In the region (3) with 100-nm trenches, the same 20 nm of the ALD material would multiply the scaffold frequency by 5, but 3 of every 5 opaque zones are missing within a period of the scaffold. In the region (2) with 80 nm zones, the ALD-zones fall in antiphase and their contribution vanishes. In transition regions, such as, e.g., between (1) and (2), a frequency multiplication by 3 works, but the contributions to the focus gradually decrease due to the mismatch between the ideal and resulted zone positions for an integer frequency multiplication.

Close modal

The zone plate structure itself had to be designed to withstand capillary forces in diverse steps of the fabrication, especially in the high aspect ratio stage. A choice was made to have buttresses connecting the zones, for increased mechanical stability. The analysis of buttress geometries reported in the literature and found in practice in APS showed that while the zone width decreases, more buttresses have to be added,3 but adding them by simply starting a new buttress line on a zone ended up pulling the zone in one direction, with no compensation from the other side. This led in many cases to deforming the respective zones and making them less effective in the zone plate function.19,34 To alleviate this, a new strategy was implemented, by dividing symmetrically existent buttress lines, rather than starting new ones. Since the buttresses consume zone plates' real estate and reduce efficiency, the number and size of buttresses was kept at minimum. This also meant that once the zones get tripled in size (at the start of the outermost part of the composite zone plate), the number of buttresses reached at the 60 nm zones was no longer necessary and therefore the buttress lines were made to merge symmetrically, then kept dividing again toward the outer side of the outer zone plate. The resulting composite zone plate design can be seen in Fig. 2.

Fig. 2.

Composite zone plate structure with novel buttresses configuration, as exposed in 350-nm-thick HSQ resist with the optimized lithographic process (SEM image).

Fig. 2.

Composite zone plate structure with novel buttresses configuration, as exposed in 350-nm-thick HSQ resist with the optimized lithographic process (SEM image).

Close modal

For the development of UNCD-based zone plates, first an optimization of the diamond films and their processing was necessary. This process had as goals:

  1. optimizing the deposition of a stack of nonconductive UNCD (300 nm)/conductive (Boron-doped35) UNCD (3000 nm)/nonconductive UNCD (700 nm) within a single reactor run;

  2. fabrication of flat UNCD membranes. This process required optimizing the deposition recipe to control stress toward a slightly tensile-stress UNCD (Fig. 3);

  3. evaluation of the material loss in polishing the UNCD films to <1 nm (rms) roughness, as necessary for the high-resolution electron-beam lithography step;

  4. optimizing the diamond structure toward low grain size, to better fit the goal of smooth side wall etching in the zone plates fabrication process (Fig. 4).

Fig. 3.

(Color online) UNCD optimization for slightly tensile stress and flat membranes. Buckled UNCD Before optimization (a) and flat UNCD membrane after optimization (b). The frame is made of Si. The size of these membranes is 5 × 5 mm. (Low resolution reflection optical microscopy images.)

Fig. 3.

(Color online) UNCD optimization for slightly tensile stress and flat membranes. Buckled UNCD Before optimization (a) and flat UNCD membrane after optimization (b). The frame is made of Si. The size of these membranes is 5 × 5 mm. (Low resolution reflection optical microscopy images.)

Close modal
Fig. 4.

(Color online) AFM images of a UNCD surface: (a) as deposited (roughness Rq = 24.4 nm rms) and (b) after polishing (Rq = 0.33 nm rms).

Fig. 4.

(Color online) AFM images of a UNCD surface: (a) as deposited (roughness Rq = 24.4 nm rms) and (b) after polishing (Rq = 0.33 nm rms).

Close modal

The goals (1) and (2) were performed simultaneously. Incorporation of conductive UNCD in the stack was done in order to allow dispersing of electrostatic charges during electron beam lithography, but also to allow processing of FZPs by electroforming in a separate fabrication process, not reported here. The goal was reached by optimizing the temperature of the substrates, the number of filaments and the CH4/H2 flow ratio in the hot filament chemical vapor deposition reactor, as well as spacing and arrangement of wafers inside the chamber. A rotating stage was used for uniformization of film thickness (below 2%). Polishing of the UNCD surface from ∼25 nm (rms) roughness as deposited, to sub-1 nm (necessary for the high-resolution EBL step) was performed. The optimized UNCD made possible flat, smooth, and slightly tensile-stressed films and membranes (Fig. 3), useful for many micro-optics and sensors applications and permitted the further development of zone plates.

Fabrication of FZPs required optimization of a series of processes, including EBL, RIE of UNCD, and the ALD of a high-Z and density metal on the UNCD scaffold.

The EBL step was pursued with hydrogen silsequioxane (HSQ) negative resist (FOX 15, Dow Corning) of thickness 350 nm. HSQ transforms into glass (SiO2) upon exposure with an electron beam, becoming insoluble, while the unexposed resist can be dissolved in a basic resist developer (Microposit MF®CD26, Shipley), leading to possible high aspect ratio structures.36 Exposed HSQ (mainly SiO2) is an efficient hard mask for etching diamond with oxygen RIE. Figure 5 presents a typical example of correctly exposed and developed FZP pattern in HSQ resist.

Fig. 5.

FZP images after EBL with 350-μm-thick HSQ resist. (a) Global view with partitioning in central and exterior zone plates, within a composit FZP design. (b) SEM image of the cental FZP. (c) High resolution SEM of the minimal zone width area (60 nm lines and spaces) showing exceptional accuracy. (d) Zoom in SEM in the central zone area.

Fig. 5.

FZP images after EBL with 350-μm-thick HSQ resist. (a) Global view with partitioning in central and exterior zone plates, within a composit FZP design. (b) SEM image of the cental FZP. (c) High resolution SEM of the minimal zone width area (60 nm lines and spaces) showing exceptional accuracy. (d) Zoom in SEM in the central zone area.

Close modal

Exposure and development were continuously optimized, in conjunction with etching, since the resistance to RIE of exposed HSQ depends on the dose and hardening into glass phase. Thus, exposure dose was ultimately dictated by the necessary degree of hardening, and was established at 4250 μC/cm2. Using a 100 keV JEOL9300FS EBL tool, lithography was carried out (on a proximity effect corrected pattern) at a current of 2 nA and shot pitch of 3 nm. Yields of up to 65% per wafer were attained. While initially the EBL was performed with a dose variation around the optimal value on each device chip (containing 5 FZPs each), in the final stage the exposure was very reliably producing good structures with the nominal exposure. Thus, it became possible to reduce the number of exposures per chip to only one, resulting in 121 FZP devices per 4 in. wafer, which could be exposed in ∼64.5 h.

The next important step developed was the transfer of the zone plates structures from HSQ resist into diamond. The starting point for diamond etching was a previously reported recipe37 based on inductive-coupled plasma (ICP) RIE, using an Oxford 100 ICP-RIE system. The reported etch rate for UNCD was ∼55 nm/min. However, the known recipe proved to etch much slower the present high aspect ratio zone plates structures. After continuous improvement, the process was migrated to a STS ICP-RIE system (STS LpX Pegasus), with a special configuration for etching deep, narrow trenches (“long funnel”). Essential in the recipe was the capability to gradually decrease the pressure, modify the gas composition (from O2 + SF6 to pure O2) and slightly increase the RF power during the etching. Figure 6 presents a FZP scaffold structure with a height of 2.1 μm (including 250 nm of HSQ resist left after etching). The low RF power (50 W) versus ICP power (1200 W) in the final stage of the process provided a smooth etch of the base diamond, mostly free of spikes and grass.

Fig. 6.

Zone plates structure as transferred into the UNCD stack to a total depth of ∼2.1 μm. (a) the central zones area, (b) zoom-in into the fine zone area, (c) focused ion beam (FIB) cross section showing an etching depth of ∼2.1 μm in the smallest zone area. The white material atop of the structures is Pt, deposited with the FIB prior to sectioning, to protect the nanostructures from redeposition and offer contrast.

Fig. 6.

Zone plates structure as transferred into the UNCD stack to a total depth of ∼2.1 μm. (a) the central zones area, (b) zoom-in into the fine zone area, (c) focused ion beam (FIB) cross section showing an etching depth of ∼2.1 μm in the smallest zone area. The white material atop of the structures is Pt, deposited with the FIB prior to sectioning, to protect the nanostructures from redeposition and offer contrast.

Close modal

Back side etching for membranes formation was performed in a 30% KOH solution at 85 °C, with the wafers mounted in a holder protecting the front side of the wafer. The wafers were thoroughly rinsed in a DI water cascade, then rinsed with IPA, and gently dry-blown with nitrogen on top of an 80 °C-heated hot plate. Figure 7 presents the front side and a detail of the chips with membranes and FZP structures on them.

Fig. 7.

(Color online) Front side (a) and detail (b) of one FZP on a chip with 5 UNCD membranes and zone plates structures (optical microscopy).

Fig. 7.

(Color online) Front side (a) and detail (b) of one FZP on a chip with 5 UNCD membranes and zone plates structures (optical microscopy).

Close modal

The released zone plate structures underwent an ALD process to form a highly conformal metal layer for spatial frequency multiplication, as explained in Fig. 1. Coatings of Ir,38 Pt, and W were performed as shown in Table II.

Table II.

ALD conditions used for metal deposition on FZPs.

ALD materialPrecursor 1Precursor 2Temperature (°C)
Ir Iridium (III) acetylacetonate O2 300 
Pt MeCpPtMe3 O2 300 
WF6 SiH4 300 
ALD materialPrecursor 1Precursor 2Temperature (°C)
Ir Iridium (III) acetylacetonate O2 300 
Pt MeCpPtMe3 O2 300 
WF6 SiH4 300 

Testing the FZPs was done on two hard x-ray beamlines of the APS) of Argonne National Laboratory: 1BM—a beam line dedicated to characterization of x-ray optics, and 32ID—the host of a transmission x-ray microscope and tomography setup. At 1-BM measurements were pursued to determine the diffraction efficiency of FZPs, while in 32-ID imaging experiments were pursued, to determine the resolution.

Diffraction efficiency measurements at 1BM were done by measuring the photocurrents with a PIN diode. Measurements were done (1) with a blank window of the chip for the incident energy (Io); (2) with the FZP in and aperture close to the focus point (If); with the beam blocked for the dark current (Id). The ratio of the area of the aperture spot Aap and the area of the FZP in the shadow image (AZP) were determined with the CCD camera image, with the PIN diode out of the optical path. The efficiency was then determined by

εm=IfIo(IoId)AZPAap.
(1)

Efficiencies were determined for different FZPs, at different photon energies from 8 to 18 keV and for different focus orders (m = 1 and 3). The tested FZPs with Ir coating were from a first batch fabricated and of poorer quality. Tests with Pt and W coated devices were from different batches of better quality. The efficiencies turned out expectedly low, with a maximum of 1.7% for Pt coated FZPs. The W-coated FZPs presented also a low efficiency at 8 keV and worse at higher energies. This was expected, since the absorption and phase shifting of W is lower than for Pt. The measured values are given in Table III. The measurement uncertainties were estimated to be ∼9.8% and originate mainly in the fluctuations of determining the intensities If, Io and Id and their error propagation through formula (1).

Table III.

Efficiencies determined experimentally for FZPs at 8 keV for focus order 1 as considered for the frequency tripled FZP, or order 3 for the scaffold FZP.

FZP #FZP typeε1 (%)
4-1 Pt 1.30 
4-2 Pt 1.28 
6-1 0.75 
1-1 Ir 0.84 
1-2 Ir 0.63 
6-2 Pt 1.42 
6-1 Pt 1.75 
FZP #FZP typeε1 (%)
4-1 Pt 1.30 
4-2 Pt 1.28 
6-1 0.75 
1-1 Ir 0.84 
1-2 Ir 0.63 
6-2 Pt 1.42 
6-1 Pt 1.75 

Diffraction efficiencies of a Pt-coated FZP at different energies for the first order focus (considered with frequency tripling) are shown in Fig. 8. Diffraction efficiencies for focus order 1 and 2 considered for the scaffold FZP pattern were higher (4.5%–4.8% for order 1 and 1.9%–2.4% for order 2, respectively) for the Pt-coated FZPs.

Fig. 8.

(Color online) Efficiencies of a Pt-coated FZP at different energies. Scale bars correspond to 10% uncertainties. The curve is for eye-guiding purpose.

Fig. 8.

(Color online) Efficiencies of a Pt-coated FZP at different energies. Scale bars correspond to 10% uncertainties. The curve is for eye-guiding purpose.

Close modal

Resolution tests were carried out with an optical scheme for a transmission x-ray microscope at beam line 32ID in APS,39 using W-coated FZPs. As optical object, a resolution target (“Siemens star”) with minimum features of 50 nm was used [Fig. 9(a)]. Images were recorded by projecting real images of the object onto a scintillator screen and further onto a CCD camera with a 5× optical microscope objective. The pixel size on the camera was 3.45 μm, but projected back via the 5× objective on the scintillator screen it was determined to correspond to 1.26 μm in the x-ray image plane. Figures 9(b) and 9(c) show images of the resolution target recorded for the first order and third order foci, respectively. The 50 nm lines are clearly visible in both images, showing that the resolution is probably better than 50 nm. Dwell times for recording the x-ray images were 10 s for the first order focus and 20 s for the third order focus.

Fig. 9.

(a) SEM of the resolution target (“Siemens star”) with 50 nm minimum features, used as object for x-ray imaging; (b) x-ray image of the resolution target obtained at 9.1 keV with the first order focus of a W-coated FZP; (c) similar, obtained with the third order focus. Scale bars in the x-ray images correspond to sizes in the object plane.

Fig. 9.

(a) SEM of the resolution target (“Siemens star”) with 50 nm minimum features, used as object for x-ray imaging; (b) x-ray image of the resolution target obtained at 9.1 keV with the first order focus of a W-coated FZP; (c) similar, obtained with the third order focus. Scale bars in the x-ray images correspond to sizes in the object plane.

Close modal

Estimation of a theoretical value for the efficiency based on the frequency multiplication with missing zones approach and general FZP theory40 is given in the supplementary material.42 For the composite FZP considered (Table I), the estimations show a value of ∼8.7% for the inner zone plate, to which the outer zone plate may add an additional contribution of 4.3%, to a total of 13% considering only the central zone plate area. Taking the whole area of the composite zone plate would mean an efficiency of ∼4% in the first order focus.

While expected to be lower than the theoretical estimations, the recorded efficiency values are low (∼1%). However, imaging can be carried out even with the lowest efficiency FZPs (the W-coated ones) with dwelling times of 10–20 s, resulting in high resolution images. This proves that the lithography met the specifications of the design, and the batch processing strategy works for FZPs. The low efficiency has as consequence the increase in necessary recording time of images, which can be a limiting factor for some applications. An analysis of the causes of low efficiency revealed that the culprits likely are: (1) the spatial frequency multiplication scheme, resulting in fulfilling the equal zone area requirement only for very few zones. An improvement may be obtained by increasing the thickness of the ALD metal to slightly larger than 1/3 of the finest scaffold zone.32 (2) The fact that the outer zone plate does not have the same frequency multiplication as the inner FZP. If the outer zone plate scaffold of the composite can be made to end in 60 nm zones, the frequency tripling from those zones would result in an increase of efficiency. (3) The larger zones have a deeper profile than the fine zones leading to thickness anomalies in the corresponding zones. (4) The side walls of the FZP scaffold are not perfectly vertical all along their depth and have a portion of visible V-shape in the lower ∼700 nm end [Fig. 6(c)] and also have some roughness. (1) and (2) can be addressed also by implementing a sequential selective ALD process for metal coating, by protecting zones once coated to a nominal thickness of metal with an adequate photoresist mask, followed by a subsequent selective ALD process to increase the thickness for larger zones and so on. Given that the photoresist patterning can be done with standard contact lithography aligners, the number of processes can be increased to limits dictated by practical considerations regarding how many times the lithography/ALD sequence can be repeated. Each such sequence will bring a certain region of the FZP to its ideal width of the frequency-multiplication scheme selected. A suited resist for a selective ALD process is photosensitive polyimide,41 which resists well both the process temperature and chemistries. (3) and (4) can be addressed by considering an optimization of the FZP pattern to comply with technology constraints and limitations. Namely, instead of varying width trenches, equal widths ones can be patterned and etched, with optimized placing. In that case, optimization of EBL and RIE will target a single-width trench, with better outcome perspectives. An optimization of FZPs for frequency multiplication has to consider also inherently equal widths of some absorber/phase shifter zones, such as deposited by ALD. These fixes will be implemented in a future work.

Fresnel zone plates were designed and fabricated based on UNCD scaffolds with ALD coating of absorber/phase shifting metals using a frequency multiplication scheme. The zone plates design consists in a composite zone plate, with novel arrangement of buttresses to prevent tearing of the ring structures by capillary and electrostatic forces. The stress and composition of UNCD, and the EBL, RIE and ALD processes were optimized. Resulting FZPs with ALD coatings of Ir, Pt, and W were tested on x-ray beamlines, showing imaging half-pitch resolution better than 50 nm (limited by the available resolution test object), while the diffraction efficiencies were (expectedly) around 1%. Ways to improve diffraction efficiency by sequential lithography and selective ALD processes were discussed. A further path for improvement is to consider optimization of FZPs design with constraints regarding trenches of same widths in the scaffold and equal width absorber/phase shifter zones.

Work of the Advanced Diamond Technology was supported by the DOE Small Business Innovative Research project through Award No. DE-SC0011265. Use of the Center for Nanoscale Materials and Advanced Photon Source, Office of Science user facilities, was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-06CH11357. ICP-RIE for deep etching of diamond was performed at the Northwestern University Micro/Nano Fabrication Facility (NUFAB), which is partially supported by the Soft and Hybrid Nanotechnology Experimental (SHyNE) Resource No. (NSF ECCS-1542205), the Materials Research Science and Engineering Center (No. NSF DMR-1121262), the State of Illinois, and Northwestern University. The authors are thankful to Liliana Stan for help with depositions in CNM and to Francesco DeCarlo and Albert Macrander for accessing beam lines 32ID and 1BM, respectively, in APS.

1.
M.
Holler
,
M.
Guizar-Sicairos
,
E. H. R.
Tsail
,
R.
Dinapoli
,
E.
Müller
,
O.
Bunk
,
J.
Raabel
, and
G.
Aeppli
,
Nature
543
,
402
(
2017
).
2.
B.
Yu
,
R. S.
Bradley
,
C.
Soutis
, and
P. J.
Withers
,
Philos. Trans. R. Soc. A
374
,
20160037
(
2016
).
3.
S. Z.
Marek
,
T.
Jerzy
,
J. W.
David
,
F. S.
Yen
,
C. W.
Chun
, and
L. F.
Ray
,
Oil Gas Res.
2
,
1000109
(
2016
).
4.
J. C.
Andrews
,
F.
Meirer
,
Y.
Liu
,
Z.
Mester
, and
P.
Pianetta
,
Microsc. Res. Tech.
74
,
671
(
2011
).
5.
L.
Nowack
,
D.
Grolimund
,
V.
Samson
,
F.
Marone
, and
V.
Wood
,
Sci. Rep.
6
,
21479
(
2016
).
6.
C.
Lim
,
H.
Kang
,
V.
De Andrade
,
F.
De Carlo
, and
L.
Zhu
,
J. Synchrotron Radiat.
24
,
695
(
2017
).
7.
L.
Lavery
,
W.
Harris
,
H.
Bale
, and
A.
Merkle
,
Microsc. Microanal.
22
,
1762
(
2016
).
8.
B.
Patterson
,
N.
Cordes
,
K.
Henderson
,
R.
Pacheco
,
M.
Herman
,
J. C. E.
Mertens
,
X.
Xiao
,
J.
Williams
, and
N.
Chawla
,
Microsc. Microanal.
22
,
1760
(
2016
).
9.
How x-ray microscopy facilitates game-changing 3D printing for parts manufacturers
,” SelectScience Interview of Branden Kappes by S. Nicholas, 16 June
2017
, http://www.selectscience.net
10.
R.
Divan
,
D. C.
Mancini
,
N.
Moldovan
,
B.
Lai
,
L.
Assoufid
,
Q.
Leonard
, and
F.
Cerrina
,
Proc. SPIE
4783
,
82
(
2002
).
11.
J.
Reinspach
,
M.
Lindblom
,
M.
Bertilson
,
O.
von Hofsten
,
H. M.
Hertz
, and
A.
Holmberg
,
J. Vac. Sci. Technol., B
29
,
011012
(
2011
).
12.
J.
Vila-Comamala
,
K.
Jefimovs
,
J.
Raabe
,
T.
Pilvi
,
R. H.
Fink
,
M.
Senoner
,
A.
Maaßdorf
,
M.
Ritala
, and
C.
David
,
Ultramicroscopy
109
,
1360
1364
(
2009
).
13.
S.
Werner
,
S.
Rehbein
,
P.
Guttmann
,
S.
Heim
, and
G.
Schneider
,
Microelectron. Eng.
87
,
1557
(
2010
).
14.
G. E.
Ice
,
J. D.
Budai
, and
J. W. L.
Pang
,
Science
334
,
1234
(
2011
).
15.
W.
Yun
,
B.
Stripe
,
S.
Lewis
,
D.
Reynolds
,
A.
Lyon
, and
J.
Kirz
, “
New developments in XRD/XRF instrumentation I section
,” in 66th Annual Denver X-Ray Conference
, Big Sky Resort, Big Sky, MT, 31 July–4 August (
2017
).
16.
B.
Lengeler
,
C. G.
Schroer
,
M.
Richwin
,
J.
Tümmler
,
M.
Drakopoulos
,
A.
Snigirev
, and
I.
Snigireva
,
Appl. Phys. Lett.
74
,
3924
(
1999
).
17.
C. G.
Schroer
 et al,
Appl. Phys. Lett.
87
,
124103
(
2005
).
18.
A. G.
Michette
,
Optical Systems for Soft X Rays
(
Plenum
,
New York/London
,
1986
).
19.
S.
Ren Wu
,
Y.
Hwu
, and
G.
Margaritondo
,
Materials
5
,
1752
(
2012
).
20.
W.
Yun
,
B.
Lai
,
A. A.
Krasnoperova
,
E.
Di Fabrizio
,
Z.
Cai
,
F.
Cerrina
,
Z.
Chen
,
M.
Gentili
, and
E.
Gluskin
,
Rev. Sci. Instrum.
70
,
3537
(
1999
).
21.
R.
Divan
,
D.
Rosenthal
,
K.
Ogando
,
L. E.
Ocola
,
D.
Rosenmann
, and
N.
Moldovan
,
J. Vac. Sci. Technol., B
31
,
06FF03
(
2013
).
22.
C.
Chang
and
A.
Sakdinawat
,
Nat. Commun.
5
,
4243
(
2014
).
23.
K.
Li
,
M. J.
Wojcik
,
R.
Divan
,
L. E.
Ocola
,
B.
Shi
,
D.
Rosenmann
, and
C.
Jacobsen
,
J. Vac. Sci. Technol., B
35
,
06G901
(
2017
).
24.
K.
Keskinbora
 et al,
Opt. Express
22
,
18440
(
2014
).
25.
J. H.
Je
and
J.
Jung
, U.S. patent US20130280421 (28 October
2011
).
26.
S.-C.
Gleber
 et al,
Opt. Express
22
,
28142
(
2014
).
27.
S.
Rehbein
,
A.
Lyon
,
R.
Leung
,
M.
Feser
, and
G.
Schneider
,
Opt. Express
23
,
11063
(
2015
).
28.
I.
Mohacsi
,
I.
Vartiainen
,
B.
Rösner
,
M.
Guizar-Sicairos
,
V. A.
Guzenko
,
I.
McNulty
,
R.
Winarski
,
M. V.
Holt
, and
D.
Christian
,
Sci. Rep.
7
,
43624
(
2017
).
29.
J.
Vila-Comamala
,
S.
Gorelick
,
E.
Färm
,
C. M.
Kewish
,
A.
Diaz
,
R.
Barrett
,
V. A.
Guzenko
,
M.
Ritala
, and
C.
David
,
Opt. Express
19
,
175
(
2011
).
30.
D. C.
Flanders
and
N. N.
Efremov
,
J. Vac. Sci. Technol., B
1
,
1105
(
1983
).
31.
I.
Mohacsi
 et al,
Opt. Express
23
,
776
(
2015
).
32.
F.
Marschall
,
J.
Vila-Comamala
,
V. A.
Guzenko
, and
C.
David
,
Microelectron. Eng.
177
,
25
(
2017
).
34.
D. L.
Olynick
,
B. D.
Harteneck
,
E.
Veklerov
,
M.
Tendulkar
,
J. A.
Liddle
,
A. L.
David Kilcoyne
, and
T.
Tyliszczak
,
J. Vac. Sci. Technol., B
22
,
3186
(
2004
).
35.
H.
Zeng
 et al,
Carbon
84
,
103
(
2015
).
36.
D. P.
Mancini
,
K. A.
Gehoski
,
E.
Ainley
,
K. J.
Nordquist
,
D. J.
Resnick
,
T. C.
Bailey
,
S. V.
Sreenivasan
,
J. G.
Ekerdt
, and
C. G.
Willson
,
J. Vac. Sci. Technol., B
20
,
2896
(
2002
).
37.
N.
Moldovan
,
R.
Divan
,
H.
Zeng
, and
J. A.
Carlisle
,
J. Vac. Sci. Technol., B
27
,
3125
(
2009
).
38.
S. T.
Christensen
and
J. W.
Elam
,
Chem. Mater.
22
,
2517
(
2010
).
39.
V. D.
Andrade
,
A.
Deriy
,
M. J.
Wojcik
,
D.
Gürsoy
,
D.
Shu
,
K.
Fezzaa
, and
F. De
Carlo
,
SPIE Newsroom
, 12 May (
2016
).
40.
J.
Kirtz
,
J. Opt. Soc. Am.
64
,
301
(
1974
).
41.
R. H. J.
Vervuurt
,
A.
Sharma
,
Y.
Jiao
,
W. M. M.
Kessels
, and
A. A.
Bol
,
Nanotechnology
27
,
405302
(
2016
).
42.
See supplementary material at https://doi.org/10.1116/1.5003412 for the estimation of a theoretical value of the diffraction efficiency of the composite FZP.

Supplementary Material