High‐Resolution Patterning of Organic–Inorganic Photoresins for Tungsten and Tungsten Carbide Microstructures

Tungsten is an important material for high‐temperature applications due to its high chemical and thermal stability. Its carbide, that is, tungsten carbide, is used in tool manufacturing because of its outstanding hardness and as a catalyst scaffold due to its morphology and large surface area. However, microstructuring, especially high‐resolution 3D microstructuring of both materials, is a complex and challenging process which suffers from slow speeds and requires expensive specialized equipment. Traditional subtractive machining methods, for example, milling, are often not feasible because of the hardness and brittleness of the materials. Commonly, tungsten and tungsten carbide are manufactured by powder metallurgy. However, these methods are very limited in the complexity and resolution of the produced components. Herein, tungsten ion‐containing organic–inorganic photoresins, which are patterned by two‐photon lithography (TPL) at micrometer resolution, are introduced. The printed structures are converted to tungsten or tungsten carbide by thermal debinding and reduction of the precursor or carbothermal reduction reaction, respectively. Using TPL, complex 3D tungsten and tungsten carbide structures are prepared with a resolution down to 2 and 7 μm, respectively. This new pathway of structuring tungsten and its carbide facilitates a broad range of applications from micromachining to metamaterials and catalysis.


Introduction
Tungsten is an important high-performance material owing to its high electrical as well as thermal conductivity, high thermal and chemical stability, and excellent X-Ray absorption ability. [1] From all elements, tungsten possesses the highest melting and boiling point of 3422 and 5555°C, respectively, and exhibits a very high hardness of 9.0 on the Mohs scale. [2,3] Due to these prominent material properties, tungsten is utilized in micromachining applications such as probes in high-precision coordinate measuring systems [4,5] and electrodes. [6] It serves as collimators and detectors for computed tomography (CT) or gamma ray detectors, [7,8] and it is a popular material for metamaterial absorbers [9][10][11][12] as well as emitters. [13,14] In addition, tungsten is considered to be one of the best materials for the harshest conceivable environments such as fusion reactors as a plasma-facing material (PFM). [15][16][17] Tungsten carbide, on Tungsten is an important material for high-temperature applications due to its high chemical and thermal stability. Its carbide, that is, tungsten carbide, is used in tool manufacturing because of its outstanding hardness and as a catalyst scaffold due to its morphology and large surface area. However, microstructuring, especially high-resolution 3D microstructuring of both materials, is a complex and challenging process which suffers from slow speeds and requires expensive specialized equipment. Traditional subtractive machining methods, for example, milling, are often not feasible because of the hardness and brittleness of the materials. Commonly, tungsten and tungsten carbide are manufactured by powder metallurgy. However, these methods are very limited in the complexity and resolution of the produced components. Herein, tungsten ion-containing organic-inorganic photoresins, which are patterned by two-photon lithography (TPL) at micrometer resolution, are introduced. The printed structures are converted to tungsten or tungsten carbide by thermal debinding and reduction of the precursor or carbothermal reduction reaction, respectively. Using TPL, complex 3D tungsten and tungsten carbide structures are prepared with a resolution down to 2 and 7 μm, respectively. This new pathway of structuring tungsten and its carbide facilitates a broad range of applications from micromachining to metamaterials and catalysis. the other hand, is an important material for cutting tool manufacturing, wear resistance, mining, and other heavy industries, [18] due to its exceptional hardness of 9-9.5 on the Mohs scale. [19] Additionally, porous tungsten carbide has been reported as an excellent replacement or scaffold material for electrochemical reactions such as hydrogen evolution reaction [20,21] and methanol electro-oxidation, [22,23] which typically rely on very expensive and scarce platinum group metals (PGMs). The most commonly used method to produce tungsten carbide is by carburization, which involves blending both elements together using ball mills or double-cone blenders. Subsequently, the powder is reacted at 1300-1700°C in hydrogen atmosphere. [1] Unfortunately, many of their advantageous physical properties, for example their high melting temperature as well as their brittleness, [1] make it notoriously difficult to structure tungsten and tungsten carbide since traditional machining techniques such as forging, casting, cutting, and milling cannot be used. [24][25][26] The conventional method to shape tungsten and tungsten carbide is by powder metallurgy, wherein a high-purity tungsten powder (and potentially a carbon source) is compressed into the desired shape to form the green compact. Subsequently, the sintering step is carried out at 2000-3050°C (tungsten) and 1350-1600°C (tungsten carbide) in high-purity dry hydrogen atmosphere. [1] The main drawbacks of powder metallurgy techniques are the low geometrical freedom of the shaped tungsten parts and the necessity of high-temperature post treatment. [27,28] Although powder injection molding (PIM) surpasses classical powder metallurgy processes in terms of the achievable complexity, injection molding methods are only suitable for large-scale productions due to the necessity of the expensive tools. Additionally, the resolution of the sintered tungsten parts is limited to hundreds of micrometers. [24,29] Common machining techniques for high-melting and brittle materials such as tungsten and tungsten carbide are noncontact subtractive machining methods. These techniques are generally used for the manufacturing of, for example, micropillars and microgrooves. [30][31][32][33] Electric discharge machining (EDM) was utilized to manufacture tungsten micropillars with a minimum diameter of 7 μm and an aspect ratio of 14 as well as microgrooves with a width of 142 μm in a 500 μm-thick tungsten plate. [34,35] Tungsten carbide was machined using EDM to manufacture micropillars with a minimum diameter of 3 μm and microholes with diameters of 60 μm in a 300 μm-thick tungsten carbide strip. [33,36] However, the achievable geometries utilizing ECM are restricted to cylindrical objects and true 3D structuring is not possible. Using a combination of EDM and reverse EDM (rEDM), tungsten micropillar arrays with a pillar diameter of 35 μm and an aspect ratio of 42 were manufactured. [6] With this technique it is possible to structure more complex geometries compared to EDM, including disc-type or U-shaped electrodes. However, complex 3D objects are not accessible. [6] For achieving more complex geometrical features and greater freedom of design, 3D printing techniques such as selective laser melting (SLM) and electron beam melting (EBM) have been reported for the manufacturing of tungsten and tungsten carbide components. [37][38][39][40] SLM was, for example, utilized to manufacture pure tungsten 3D multicollimators with a resolution of 460 μm. [7] However, the surface quality of SLM-manufactured parts is poor with surface roughness R a values ranging from 9 to 14 μm. [41] This is mainly due to the intrinsic material properties of tungsten, for example, the high melting temperature, the high surface tension of the melt, and high thermal conductivity, which lead to balling effects and surface defects as well as the size of the tungsten powders with particle sizes of tens of micrometers. [38,42,43] In addition, the final tungsten parts manufactured by SLM and EBM suffer from microcrack formation due to high thermal stresses during solidification or recrystallization. [40,44,45] In a different approach, tungsten microlattices were produced by room-temperature extrusion-based 3D printing of a tungsten(VI) oxide containing ink, subsequent thermal treatment, reduction, and sintering under hydrogen atmosphere to metallic tungsten. After the thermal treatment, the parts underwent a linear isotropic shrinkage of 50% to strut diameters of 80 μm. [46] This method, in general, allows manufacturing of 3D objects. However, the resolution is limited to the nozzle diameter through which the ink is extruded. Recently, it was reported that digital light processing (DLP) was utilized to structure tungsten oxidebased catalysts, tungsten, and a tungsten nickel alloy. [47][48][49] In the former two applications, an ammonium metatungstate containing aqueous ink was structured, then dried for 3-4 days at room temperature, and subsequently thermally reduced to tungsten oxide and porous tungsten, respectively. [47,48] The resulting tungsten oxide was successfully utilized as a catalyst for the hydrogenation of alkynes and nitrobenzenes, whereas the resulting porous tungsten reached feature sizes of around 700 μm. [47,48] In the latter, 3D-architected hydrogels were manufactured as scaffolds by DLP. The scaffolds were then infused with the metal precursor solutions (tungsten and nickel), calcinated, and reduced to metal replicas of the initial scaffold with feature sizes of 100-150 μm. [49] These methods are capable of shaping 3D tungsten objects; however, the resolution is limited to minimum feature sizes of hundreds of micrometers after exploiting the shrinkage of the thermal reduction processes. For shaping high-resolution 3D tungsten microstructures, techniques such as focused electron/ion beam-induced deposition (FEBID/FIBID) can be utilized. [50][51][52] Using FIBID, horizontally suspended tungsten nanowires with a diameter of 98 nm and a length of 3.5 μm were manufactured. [52] The resulting nanowire, however, displays up to 25 at% impurities of gallium in the core and up to 30 at% of carbon in the shell. The low purity of the resulting metal parts is a well-known issue in FEBID/FIBID due to the use of organic precursor materials as well as the ion beam source in FIBID, especially. [53] The produced tungsten nanowires can be purified of the gallium contamination by applying high voltage, which in turn leads to a segregation of tungsten and gallium and evaporation. [54] Recently, it was reported that organic-inorganic photoresins can be structured using two-photon lithography (TPL) to form complex 3D nanoarchitectured metal and metal oxide structures. [55][56][57] The printed organic-inorganic polymer parts were subsequently thermally reduced to their metal or metal oxide counterparts, respectively. By exploiting the shrinkage during thermal reduction, nickel, zinc oxide, and platinum objects with a resolution of 25-100 nm (nickel), [55] 250 nm (zinc oxide), [56] and 300 nm (platinum) [57] were fabricated. However, so far, the choice of materials for this process is limited and, until now, inaccessible for microstructuring of tungsten and tungsten carbide.
In this article, we demonstrate, for the first time, that tungsten as well as tungsten carbide microstructures can be shaped using TPL of a tungsten ion-containing organic-inorganic photoresin. The printed organic-inorganic polymer parts are subsequently thermally debinded and thermally reduced in a 5 vol% hydrogen and 95 vol% argon atmosphere to their respective tungsten parts with a final resolution of 2 μm after thermal reduction. To obtain tungsten carbide, the polymerized parts were converted by carbothermal reduction (CTR) with a final resolution of 7 μm. The organic binder of the organic-inorganic photoresin hereby supplied the carbon source for CTR. Elemental analysis of the reduced tungsten samples showed 93 at% tungsten and 7 at% oxygen. The tungsten carbide samples revealed an atomic ratio of tungsten to carbon of 45:55 and X-ray diffraction (XRD) shows that this manufacturing technique results in tungsten carbide. Surface and pore size analysis of the manufactured tungsten carbide show a surface area of 49.87 m 2 g À1 , a median mesopore radius of 9.44 nm, and a median micropore radius of 0.45 nm.

Results and Discussion
The tungsten-containing photoresin was prepared by mixing 33 wt% of an aqueous sodium metatungstate (SMT, Na 6 [H 2 W 12 O 40 ]) solution at a concentration of 3 g mL À1 with 67 wt% of an acrylic binder matrix. The composition is based on the compromise between the highest amount of tungsten precursor in the formulation and sufficient mechanical stability of the printed parts.
Before adding the SMT solution, the pH level of the binder matrix was adjusted to 3 by adding 26 wt% (w.r.t matrix mass) of glacial acetic acid. This is a necessary step to prevent precipitation of SMT, since it is only stable at pH levels between 3 and 4. [1] At lower pH levels of around 1, SMT condensates to a polymeric form, which is insoluble in aqueous solutions. At higher pH levels of 6-7, SMT dissociates to WO 3 and WO 4 À2 fragments, which are also insoluble in aqueous solutions. [1] The organicinorganic photoresin was blended with a photoinitiator (PI) with a high two-photon absorption cross section to demonstrate structuring the material at 780 nm by TPL. [58] Figure 1 displays the Figure 1. Structuring tungsten using organic-inorganic photoresins. a) Schematic of the TPL process. The laser is focused into a material vat containing the resin. The resin only polymerizes within the focal point (voxel) of the laser. For layer-wise 3D structuring, the laser is scanned along the x/y-plane and the objective together with the vat is lowered along the z-axis using a piezostage. b) UV-vis spectrograph of the tungsten containing organic-inorganic photoresin displaying a transparency of 92.7% at 780 nm at a thickness of 200 μm. c-e) Schematic of the tungsten and tungsten carbide manufacturing process. (c) The organic-inorganic photoresin consists of a homogeneous mixture of an acrylic binder matrix and the metal precursor SMT. The material is crosslinked via radical polymerization to the organic-inorganic polymer component. Ethanol and water are used to remove uncured photoresin. (d) Thermal reduction procedure to obtain tungsten. The organic-inorganic polymer part is debinded at 600°C in air, the polymeric matrix is thermally decomposed, and the precursor is transformed to sodium tungsten oxide. The debinded part is subsequently thermally reduced in a 5 vol% hydrogen atmosphere at 900°C to the respective tungsten metal part. (e) Carbothermal reduction reaction to obtain tungsten carbide. The organic-inorganic polymer part is debinded in nitrogen atmosphere at 600°C. Carbon from the polymeric matrix stays within the debinded part and is subsequently utilized as a carbon source for the carbothermal reduction reaction in a 5 vol% hydrogen atmosphere at 1000°C to obtain tungsten carbide.
www.advancedsciencenews.com www.aem-journal.com tungsten microstructuring process using TPL schematically. The organic-inorganic photoresins were structured using a commercially available TPL printing system (NanoOne, UpNano) in Vat mode, which is displayed in Figure 1a. The resin is poured into the material vat and the latter is placed above the objective with water as an immersion solvent. A sample holder containing the substrate is immersed into the vat. The laser is focused through a high-precision glass cover into the resin and the focal point is maintained at a constant height of 200 μm above the cover glass. The laser is scanned along the xy-plane and the vat together with the objective are lowered along the z-axis. For high-resolution structuring of the organic-inorganic photoresin, high optical transparency of the material at the wavelength of polymerization is required. Therefore, UV-vis spectrum of the tungsten ion containing organic-inorganic photoresin without PI was recorded (see Figure 1b). The organic-inorganic photoresin demonstrated a high transmission of 92.7% at 780 nm at a thickness of 200 μm, which makes the material suitable for the TPL process. Figure 1c-e) shows the tungsten and tungsten carbide manufacturing processes schematically. The tungsten ions are dissolved and homogeneously distributed within the liquid organic-inorganic photoresin (see Figure 1c). After the resin is crosslinked by means of radical polymerization, the tungsten ions are entrapped and homogeneously distributed inside the organic-inorganic polymer. Subsequent to the polymerization step, noncured material was developed in a 1:1 (v:v) mixture of ethanol and water for 10 min. To convert the organic-inorganic polymer to tungsten, the polymerized parts were thermally debinded at 600°C in air to remove the polymeric binder matrix (see Figure 1d). During this step, the mixed tungsten oxides from SMT are transformed to sodium tungsten oxide. The debinding protocol (see Table S1, Supporting Information for the debinding protocol) was developed based on thermogravimetric analysis of the organic-inorganic polymer (see Figure S1 Supporting Information for the TGA analysis). As can be seen at 100°C residual water evaporates and at 320°C the polymeric matrix decomposes. Dwell times were introduced into the debinding protocol during the phases of maximum mass losses. The final temperature was set to 600°C at which point no additional mass loss was detectable. Finally, sodium tungsten oxide was thermally reduced to tungsten metal at 900°C in a 5 vol% hydrogen atmosphere (see Table S2 Supporting Information for the reduction protocol).
To convert the printed organic-inorganic polymer part to tungsten carbide, the samples were debinded in a nitrogen atmosphere at 600°C (see Figure 1e). Carbon from the polymeric matrix stays within the debinded part together with sodium tungsten oxide. This allows to use the polymeric binder matrix as the carbon source for the subsequent CTR where it crystallizes with tungsten to tungsten carbide. To perform the CTR, the atmosphere was changed to 5 vol% hydrogen and 95 vol% argon and the temperature was elevated to 1000°C (see Table S3 Supporting Information for the CTR protocol).
The microstructures of the debinded and reduced tungsten as well as the tungsten carbide samples were analyzed by scanning electron microscopy (SEM). The elemental composition and the atomic structure was investigated by energy-dispersive X-ray spectroscopy (EDX) and XRD, respectively (see Figure 2). After the debinding step at 600°C in air, the atomic ratio of tungsten and oxygen was 19.12% and 65.3%, respectively, showing that the precursor SMT was transformed to tungsten oxide (see Figure 2a and Table S4 Supporting Information for the associated SEM-EDX data). After further examination using XRD, the resulting tungsten oxide was identified as a mixture of various sodium polyoxotungstates with Na 2 W 4 O 13 as the main phase as well as traces of Na 2 W 2 O 7 and Na 0.05 W 2 O 6 (see Figure 2d and S2 Supporting Information for a more detailed XRD pattern). Apart from the atomic ratio, the analysis of the microstructure using SEM reveals a typical platelet-like morphology characteristic for Na 2 W 4 O 13 (see Figure 2f ). [59] After the reduction step in 5 vol% hydrogen atmosphere at 900°C, the atomic ratio of tungsten and oxygen changed to 67.68% and 32.32% (see Figure 2b, Table S5 Supporting Information for the associated SEM-EDX data). However, using a 200 keV scanning transmission electron microscope-EDX (STEM-EDX), an atomic ratio of tungsten to oxygen of 93.12% to 6.88% was detected (see Table S6 Supporting Information for the STEM-EDX data at 200 keV). The different results from SEM-and STEM-EDX can be explained by the penetration depth of the electrons in EDX experiments, which depend on the acceleration voltage, the density of the sample, as well as the incident angle of the electron beam. [60] Using Monte Carlo simulations, the penetration depth of electrons at an acceleration voltage of 20 keV into a tungsten sample is estimated to 200 nm. [60] In addition, tungsten slowly oxidizes even at room temperature. [1,61] The combination of oxide skin formation on the surface of the tungsten samples as well as the penetration depth of the electron beam leads to different representations of the atomic ratio of tungsten to oxygen. EDX mapping confirmed that the reduced sample has an oxide skin and the bulk part consists of tungsten (see Figure S3 and S4, Supporting Information, for the EDX mapping at 20 and 200 keV, respectively). Since the 200 keV STEM was performed on fragments taken from the octet lattices, it was ensured that the electrons penetrated through the entire sample, leading to a more accurate representation of the atomic composition of the analyzed samples. The results of the STEM-EDX measurements are in agreement with the XRD measurements. Evaluation of the tungsten sample using XRD shows that tungsten is the main phase and there are traces of sodium tungstate present (see Figure 2e and S5 Supporting Information for a more detailed XRD pattern). The analysis of the microstructure of the reduced samples using SEM reveals a typical icositetrahedron morphology of tungsten particles (see Figure 2g). [1] For the fabrication of tungsten carbide structures, the composition of the organic-inorganic photoresin was adapted to approach a 1:1 atomic ratio of tungsten to carbon after CTR. Using the initial composition, the atomic tungsten to carbon ratio after CTR was 39:61. To compensate for the lower amount of tungsten compared to carbon, the amount of the SMT solution was raised to 47 wt% and the amount of acrylic binder matrix was decreased to 53 wt%. Using this adapted organic-inorganic tungsten carbide resin, the atomic tungsten to carbon ratio after CTR was 45:55 (see Figure 2c, Table S7, and Figure S6 Supporting Information for the associated SEM-EDX data). An investigation of the debinded sample using XRD reveals that the main phases consist of a mixture of cubic and tetragonal sodium tungsten bronzes (see Figure 2d and S7 Supporting Information for a more detailed XRD pattern). Sodium tungsten bronzes (Na x WO 3 ) are nonstochiometric metal oxides with various ratios of sodium versus tungsten oxide. [62] In general, a higher amount of sodium (x > 0.4) results in a cubic structure, whereas a lower amount (x < 0.4) in a tetragonal structure. [63] Amorphous carbon cannot be detected using powder XRD in the sense of "Bragg reflections." However, the amorphous carbon can be seen in form of the raised background in the range between about 20°a nd 45°2θ. XRD analysis of the tungsten carbide sample shows that carbon and tungsten successfully crystallized to tungsten carbide rather than having both elements separately (see Figure 2e and S8 Supporting Information for a more detailed XRD pattern). Further, the XRD data shows that there are amounts of tungsten in the sample. The analysis of the microstructure shown in Figure 2h) shows a significant decrease in Figure 2. Characterization of the structured tungsten and tungsten carbide. SEM and EDX was performed on printed octet lattice samples and XRD on cast samples. a) EDX spectrum of a debinded sample showing an atomic ratio of tungsten and oxygen of 19.12% to 65.3%, respectively. b) EDX spectrograph of a thermally reduced sample showing high-purity tungsten with 96.01 wt% tungsten and 3.99 wt% oxygen. c) EDX spectrum of a tungsten carbide sample. By performing the debinding step in nitrogen atmosphere, carbon remains in the debinded sample and alloys with tungsten in the reduction step to tungsten carbide. The elemental analysis reveals an atomic ratio of tungsten to carbon of 45:55. d) XRD pattern of thermally debinded samples for the tungsten manufacturing (top) and tungsten carbide manufacturing (bottom) process. The former (tungsten manufacturing path) shows that debinding in air atmosphere leads to a mixture of sodium tungstates with the main phase being Na 2 W 4 O 13 . The latter (tungsten carbide manufacturing path) shows that debinding in nitrogen atmosphere leads to a mixture of cubic and tetragonal sodium tungsten bronzes. e) The analysis of the XRD pattern of reduced tungsten (top) and tungsten carbide (bottom) samples shows that both materials were manufactured successfully with traces of Na 2 WO 4 in the case of tungsten and tungsten in the case of tungsten carbide. f ) SEM image of the microstructure of a debinded sample displaying the typical platelet-like morphology of Na 2 W 4 O 13 (scale bar: 3 μm). g) SEM image of the microstructure of a reduced tungsten sample showing an icositetrahedron morphology, which is characteristic for tungsten (scale bar 1.5 μm). h) SEM image of the microstructure of a tungsten carbide sample obtained via CTR showing a ultrafine porous tungsten carbide structure (scale bar: 1.5 μm).
www.advancedsciencenews.com www.aem-journal.com particle size compared to the sodium polyoxotungstates and tungsten samples. The morphology correlates to ultrafine porous tungsten carbide. [64] The surface area as well as the pore size of the manufactured tungsten carbide were analyzed using Brunauer, Emmett, and Teller (BET) measurements based on nitrogen adsorption (see Figure S9-S11 Supporting Information for the BET analysis). The resulting surface area was 49.87 m 2 g À1 , the median mesopore radius was 9.44 nm, and the median micropore radius was 0.45 nm. The degree of porosity of the prepared tungsten and tungsten carbide octet lattices was determined by measuring the volume of the respective reduced samples and comparing it to estimated fully dense tungsten and tungsten carbide, respectively. The tungsten samples showed a porosity of 45.9 AE 1.7% and the tungsten carbide samples of 39.5 AE 3.7%. Such a nanoporous tungsten carbide could be utilized in catalysis applications, where it is used to replace PGMs or serves as a scaffold to minimize the necessary amount of PGM. [20,65] Using this approach, we manufactured tungsten micropillar arrays as well as complex 3D tungsten and tungsten carbide microstructures (see Figure 3). Figure 3a-c) shows micropillar arrays with an aspect ratio of 2, where the polymerized state is shown in Figure 3a, the debinded state in Figure 3b, and the reduced state in Figure 3c. The final diameters of the reduced micropillars were 2.1 AE 0.3 μm. Figure 3d-f ) shows complex 3D www.advancedsciencenews.com www.aem-journal.com objects manufactured by TPL from the organic-inorganic polymer state (Figure 3d) to the debinded (Figure 3e) and reduced specimen (Figure 3f ). The reduced high-purity tungsten samples had beam diameters of 8.9 AE 0.1 μm. Manufacturing tungsten at that high resolution and 3D complexity is a useful tool for applications in the field of catalysis as well as metamaterial emitters and absorbers. [12,66] Figure 3g-i) shows an organicinorganic polymer part using the adapted composition for tungsten carbide manufacturing, the debinded specimen, and the resulting tungsten carbide sample. The beam diameters of the tungsten carbide samples after CTR were 6.9 AE 0.2 μm. We further analyzed the shrinkage behavior of the organicinorganic materials during the debinding and reduction processes, as well as the material properties of the reduced specimen. The shrinking analysis using SEM images of the tungsten and tungsten carbide manufacturing process can be seen in Figure 4. The samples manufactured by TPL are freestanding, meaning that they can shrink isotropically in all directions of space. The shrinkage during the tungsten manufacturing was analyzed by measuring ten beam diameters of three octet lattices after polymerization, debinding, and reduction (see Figure 4a and Table S8 Supporting Information for the shrinking analysis of tungsten). From the polymeric state, the beam diameters shrank from 19.1 AE 0.3 to 9.8 AE 0.1 μm in the debinded state by a shrinking factor of 1.93 AE 0.05. From the debinded state to the reduced state, the beam diameters shrank to 8.9 AE 0.1 μm by a shrinking factor of 1.10 AE 0.03. The total shrinking factor from the polymer state to the reduced state was 2.13 AE 0.08. The tungsten carbide samples were analyzed accordingly (see Figure 4b and Table S9 Supporting Information for the shrinking analysis of tungsten carbide). The beam diameters of the polymeric octet lattices shrank from 14.6 AE 0.8 to 11.1 AE 0.3 μm in the debinded state by a factor of 1.31 AE 0.07. From the debinded state to tungsten carbide, the beam diameters shrank to 6.9 AE 0.2 μm by a factor of 1.65 AE 0.25. The total shrinkage factor from the polymer state to the reduced state was 2.17 AE 0.38.
The material properties such as the surface roughness of tungsten carbide and tungsten samples, respectively, the hardness and modulus of tungsten carbide, as well as stress-strain measurements of tungsten and tungsten carbide are displayed in Figure 5. For the surface roughness, plates with an edge length of 3 mm and a thickness of 500 μm were printed using TPL. After reduction to tungsten and tungsten carbide, the plates shrank to an edge length of 1.38 mm and a thickness of 230 μm. The corresponding surface roughness R a of each state (polymer, debinded, and reduced) are shown in Figure 5a. For the tungsten manufacturing path, the R a in the polymeric state was 138 nm. After the debinding step, the R a increased to 758 nm and after reduction decreased again to 317 nm. For the tungsten carbide manufacturing path, the R a in the polymeric state was 128 nm and after thermal debinding in nitrogen atmosphere the R a increased to 414 nm. After CTR the tungsten carbide sample showed a R a of 147 nm. The R a of printed octet lattices in the polymeric, debinded, and reduced states were measured to compare the values to the R a of the printed larger plates. The measurements were taken on individual beams going diagonally across the octet lattice. For the tungsten manufacturing path, the R a of the polymeric, debinded, and reduced states were 12, 346, www.advancedsciencenews.com www.aem-journal.com and 214 nm, respectively. For the tungsten carbide manufacturing path, the R a were 11, 239, and 162 nm, respectively. The deviations in R a from the printed plates to the printed octet lattices (see Table 1) were highest in the polymeric state and are approximating in the reduced state. For the hardness experiments, circular plates with a diameter of 1 cm and a thickness of 500 μm were prepared by UV casting. After reduction to tungsten carbide, the plates shrank to a diameter of 0.47 cm and a thickness of 230 μm. The indentation hardness (HIT) and modulus (EIT) of the manufactured tungsten carbide was measured and compared to literature values (see Figure 5d, S12 and Table S11 Supporting Information for the indentation experiments). After the thermal reduction at 1000°C, the tungsten carbide samples displayed a HIT of 64.9 AE 11.1 MPa and an EIT of 1.314 AE 0.295 GPa, which is 0.36% and 0.20%, respectively, of commercial tungsten carbide. [67] The hardness and modulus of the manufactured tungsten carbide seem low in comparison to the literature values, which can be explained by the porous appearance of the samples. [68] To assess, if the prepared tungsten and tungsten carbide octet lattice structures show a higher modulus, uniaxial compression experiments were carried out using flat punch tip and a speed of 0.002 mm s À1 . The load-displacement data were converted to engineering stress-strain curves by normalizing by sample height and cross-sectional area. The curves for the tungsten (see Figure 5d) and tungsten carbide sample (see Figure 5e) show drops in stress after an initial rise, which can be attributed to the breaking of individual struts of the lattice structure. The tungsten and tungsten carbide samples reached a maximum stress of 0.9 MPa at a strain of 17% and of 5.1 MPa at a strain of 38%, respectively. The octet lattice samples showed a lower mechanical stability compared to the cast plate. Due to the high porosity, the beams break more easily because the pores serve as breaking points. In summary, we demonstrated a novel approach for the microstructuring of tungsten and tungsten carbide using highresolution microstructuring via TPL. We manufactured complex  3D tungsten microstructures with a resolution of 9 μm and pillar arrays with a resolution down to 2 μm. We further showed a novel method to manufacture ultrafine tungsten carbide structures utilizing the polymeric binder matrix as a carbon source for the CTR with a resolution down to 7 μm. Such small objects are highly sought-after for applications like emitter tips, probes, microtools as well as metamaterials or catalysis, where high surface areas as well as the physicochemical properties of tungsten and tungsten carbides are highly desirable. Additionally, ultrafine, nanoporous tungsten carbide structures with a high surface area have many potential applications in catalysis, since the material may replace, or at least reduce, PGM for electrochemical reactions like hydrogen evolution reaction or methanol electrooxidation. [20][21][22][23]
Resin Preparation: The binder matrix for the tungsten resin consisted of 67 wt% PEGDA 575 and 33 wt.% Genomer 7311. The binder matrix was mixed with 25 wt% (w.r.t binder matrix mass) glacial acetic acid. The tungsten precursor SMT was dissolved in DI water at a concentration of 3 g mL À1 . The binder matrix and the SMT solution were mixed using a magnetic stirrer. For tungsten manufacturing, 33 wt% of SMT solution were mixed with 67 wt% of binder matrix and for tungsten carbide manufacturing, 47 wt% of SMT solution were mixed with 53 wt% of binder matrix. Then, 0.4 wt.% of MEK was added to the organic-inorganic photoresins and stirred until completely dissolved. The samples for the hardness measurements were manufactured by UV casting. Here, the PI was changed to 0.1 wt% TPO and the samples were cured using a highpressure mercury lamp of type Superlite 400 (Lumatec, Germany) at a wavelength of 415 nm and an exposure dose of 690 mJ cm À2 .
Surface Functionalization: The fused quartz substrates were treated for 30 min in acidic methanol (methanol:HCl, 1:1 (v:v)). Subsequently, the substrates were washed with IPA and deionized (DI) water and dried with nitrogen. The substrates were immersed in a 100 mM solution of MACS in dry toluene for 60 min. The substrates were again washed with IPA and DI water and subsequently dried using nitrogen.
Two-Photon lithography: TPL was performed using a NanoOne highresolution printing system (UpNano GmbH, Austria) equipped with a 20Â water immersion objective (NA 0.7, UAPON20XW340, Olympus, Austria) in Vat mode. The laser (80 MHz repetition rate, 90 fs pulse length, 780 nm center wavelength, 1391.6 GW cm À2 intensity at focus, Equation S(7), Supporting Information, for the calculation of intensity at focus) was focused through a high-precision cover glass into a material vat containing the resin. The focal point was maintained at a constant height above the glass. For layer-wise 3D structuring, the laser was scanned along the XY-plane by a galvanometer scanner and the objective together with the vat were lowered along the z-axis using a piezostage. The tungsten microstructures were written with a laser power of 25 mW at a scanning speed of 200 mm s À1 , at a hatching distance in the XY-plane (Δxy) of 0.3 μm, and at a slicing distance in the Z-plane (Δz) of 0.5 μm.
Heat treatment: Thermal debinding of the binder matrices and reduction of the tungsten precursor were performed using an ashing furnace of the type AAF (Carbolite/Gero, Germany). For the reduction of the sodium tungsten oxide, as well as for the carbothermal reduction reaction to tungsten carbide, the samples were additionally treated in forming gas atmosphere (5 vol% hydrogen, 95 vol% argon) using a tube furnace of type OTF-1200X (MTI, UK). To ensure an isotropic shrinkage of the octet lattice structures manufactured by TPL, the samples were removed from the substrate using a razor blade and subsequently placed on a separate fused quartz glass slide. The debinding and reduction protocols are listed in Table S1-S3 in the Supporting Information.
Material characterization: The surface roughness was measured using a white light interferometer of type NewView 9000 (Zygo, USA). Scanning electron micrographs were taken using an ultrahigh resolution focused ion beam SEM of type Scios 2 HiVac DualBeam (Thermo Fisher Scientific, Germany). Transmission electron micrographs were taken using a highresolution transmission and scanning electron microscope of type Talos F200X G2 (Thermo Fisher Scientific, Germany) in STEM mode. SEM-EDX spectra were taken using an Octane Elite EDS System (EDAX, Germany) and STEM-EDX using a FEI Super-X EDX system equipped with 4 SDDs (Silicon Drift Detectors) (Thermo Fisher, Germany). For the SEM and EDX measurements, the samples were sputtered with 30 nm of an 80:20 mixture of gold and palladium. To avoid contamination by the sputtered material into the tungsten and tungsten carbide parts during the reduction process and carbothermal reduction reaction, respectively, the samples from the organic-inorganic polymeric state to the reduced state were not identical but exemplary. However, the respective samples were prepared identically. The STEM-EDX measurements were taken without prior sputtering. The optical transparency spectra of the organicinorganic photoresin was measured using a UV/Vis spectrophotometer of type Evolution 201 (Thermo Scientific, Germany) at a sample thickness of 1 cm. The transmission spectra were calculated to the respective thickness of the organic-inorganic photoresin (see Equation S(2) Supporting Information for the calculation of the transmission spectra). TGA was performed using a simultaneous thermal analyzer of type STA 449 F5 Jupiter (Netzsch, Germany). The samples were heated at 3°C min À1 from room temperature to 600°C and dwelled for 1 h (see Figure S1 Supporting Information for the TGA of the organic-inorganic polymer). The hardness and modulus of the tungsten carbide samples were measured using a nanoindenter of type Step 700, a Berkovich tip, and a tester of type NHT3 (Anton Paar, Germany). The samples were analyzed at a load of 100 mN and the data of 26 measurements were evaluated by the method of Oliver & Pharr (see Figure S12 and Table S11 Supporting Information for the nanoindentation experiments). [69] The compression experiments were performed using a force sensor of type XFTC300 (TE connectivity, Germany) with a range of 5 N. The samples were compressed using a flat punch indenter driven by a piezostage of type P-843.4 (Physik Instrumente, Germany) with a range of 60 μm. XRD) patterns were recorded in Bragg-Brentano geometry using a D8 DISCOVER diffractometer (Bruker, Karlsruhe, Germany) equipped with Cu Kα radiation, a variable divergence slit, and a LYNXEYE XE-T detector. Experiments were carried out in a 2θ range of 5/10 to 90°with a step size of 0.025°. The surface area and pore sizes of tungsten carbide samples were measured using a surface analyzer of type Sorptomatic 1990 (Porotec, Germany) using nitrogen as an adsorbate.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.

Forschungsgemeinschaft (DFG, German Research Foundation) under
Germany's Excellence Strategy, EXC-2193/1, 390951807. This work is part of the Research Cluster "Interactive and Programmable Materials (IPROM)" funded by the Carl Zeiss Foundation. Open access funding was enabled and organized by Projekt DEAL.
Open Access funding enabled and organized by Projekt DEAL.