FIELD

[0001] The present disclosure relates to the layer-by layer production of three-dimensional objects from a curable composition. The methods of the present disclosure include selectively-curing at least some of the layers such that portions of these layers remain uncured prior to the formation of one or more subsequent layers. Further steps to control the amount of uncured resin in the final object are also disclosed.

SUMMARY

[0002] Briefly, in one aspect, the present disclosure provides layer-by-layer methods of forming a three-dimensional object from a composition comprising a polymerizable material. The three- dimensional object comprises interior layers comprising an array of voxels comprising border voxels and enclosed voxels corresponding to a cross-sectional area of the three-dimensional object bounded by the border voxels. Each voxel in a layer has a width, W, and length, L, measured along orthogonal axes in the plane of the layer, and each layer has a thickness, T, measured perpendicular to the layer.

[0003] The methods of the present disclosure comprise a selective curing step comprising: providing an uncured layer of the composition in a build region; identifying enclosed voxels selected for curing and enclosed voxels selected to remain uncured; creating a selectively-cured layer by curing the

polymerizable material in the border voxels; and curing the polymerizable material in the enclosed voxels selected for curing, wherein the enclosed voxels selected for curing comprise less than all the enclosed voxels; and translating the selectively -cured layer to recreate the build region; wherein curing the polymerizable material comprises irradiating the composition in the build region to a depth of cure, D, wherein D is greater than or equal to T.

[0004] In some embodiments, the methods further comprise repeating the selective-curing step N times to form N consecutive selectively -cured layers, wherein D is great enough to cure M adjacent layers, and wherein N is at least 2 and no greater than M. In some embodiments, the enclosed voxels selected for curing of each of the N selectively-cured layers are identified such that, collectively, the subsets of enclosed voxels enclosed voxels selected for curing provide a nesting pattern.

[0005] In some embodiments, the methods further comprise performing a full-curing step after each set of N selective-curing steps.

[0006] In some embodiments, a ratio of the enclosed voxels selected to remain uncured divided by the total number of enclosed voxels in the layer, RUC, is greater than or equal to 0.1 and no greater than 0.9. In some embodiments, M and RUC are selected such that the percent, P%, of uncured enclosed cells in the three-dimensional object, based on the total number of enclosed cells, is no greater than 10%, wherein P% = l00*RUC ^M . [0007] In some embodiments, the enclosed voxels are grouped in to cells comprising a plurality of enclosed voxels, and wherein each of the enclosed voxels in a cell are either selected for curing or selected to remain uncured.

[0008] The above summary of the present disclosure is not intended to describe each embodiment of the present invention. The details of one or more embodiments of the invention are also set forth in the description below. Other features, objects, and advantages of the invention will be apparent from the description and from the claims.

BRIEF DESCRIPTION OF THE DRAWINGS

[0009] FIG. 1 is a diagram of an exemplary stereolithography apparatus (SLA).

[0010] FIG. 2A is a cross-section of a layer shown as an array of voxels.

[0011] FIG. 2B is layer 105 of FIG. 1, fully-cured.

[0012] FIG. 2C is layer 105 of FIG. 2B, with border (B), enclosed (I) and open (O) voxels identified according to some embodiments of the present disclosure.

[0013] FIG. 2D is layer 105 of FIG. 2C, with all border voxels cured and enclosed voxels selected for curing (C) and enclosed voxels selected to remain uncured (U) identified.

[0014] FIG. 2E is layer 105 of FIG. 2D at the completion of a selective curing step according to some embodiments of the present disclosure.

[0015] FIG. 3 illustrates a prior art layer-by-layer process.

[0016] FIG. 4 illustrates a layer-by-layer process using selective curing according to some embodiments of the present disclosure.

[0017] FIG. 5A is a diagram of an individual voxel.

[0018] FIG. 5B is a diagram of the voxel of FIG. 5A as a face-connected enclosed voxel.

[0019] FIG. 6 is a diagram a portion of an exemplary stereolithography apparatus (SLA).

[0020] FIG. 7A illustrates two-layer nested pattern according to some embodiments of the present disclosure.

[0021] FIG. 7B illustrates three-layer nested patterns according to some embodiments of the present disclosure.

[0022] FIG. 8 illustrates fully-cured layers alternated with selectively-cured layers according to some embodiments of the present disclosure.

[0023] FIG. 9 illustrates the 5-voxel x 5-voxel cell used to selectively cure Examples EX-15 and EX- 16.

BACKGROUND

[0024] In many additive or three-dimensional (“3D”) fabrication processes, each individual layer is formed by irradiating and curing a photocurable resin located in a region sometimes referred to as the “build plane.” Generally, the build plane is a layer extending in the X- and Y-dimensions, perpendicular to the direction of irradiation. The radiation penetrates to some depth referred to as the Z-direction, defining the thickness of the layer. The depth of penetration and therefore the thickness of the layer is orders of magnitude less than the X and Y -dimensions, such that the layer is a substantially 2-dimensonal plane.

[0025] There are a variety of layer-by-layer, three-dimensional fabrication processes, and the methods of the present disclosure may be used in any such processes. Thus, details of specific manufacturing techniques are provided to provide context and illustrate elements of the methods, but one of ordinary skill in the art could readily apply them to any layer-by-layer manufacturing technique, including step-by- step and continuous liquid interface production (“CLIP”) processes.

[0026] For example, in the traditional vat polymerization technique, objects are created in a“top down” process where each new layer is formed on top of the previously cured layer. In the inverse vat polymerization technique, objects are created in a“bottom up” process wherein each new layer is formed at the bottom of the previously cured layer.

[0027] Typically, the compositions used for 3D fabrication processes have consisted of the photocurable resin and optionally, non-volatile additives such as fillers. In such applications, the intent is to retain the cured resin and fillers in the completed object. However, the range of materials being 3D- fabricated has expanded and now includes compositions where the photocured resin is removed from the object following the 3D-build process. For example, the photocurable resin may serve as a carrier for other components. When cured in the 3D manufacturing process, the resin serves as a matrix to hold other components in the desired shape. In subsequent steps, the cured resin may be removed, e.g., by dissolving or burning-off the resin. To retain the desired shape of the object, one or more of the other components are hardened (e.g., cured or sintered) prior to or during the matrix resin removal process.

[0028] For example, WO 2017/127561 Al (“Additive Process of Fluoropolymers”) describes 3D printable compositions for making shaped fluoropolymer articles. Such compositions may include a fluoropolymer dispersed in a polymerizable binder material. The compositions may also include solvents (e.g., water) resulting in gels (e.g., hydrogels). In subsequent processing steps the solvent is removed (e.g., evaporated), the binder is burned off, and the remaining fluoropolymer particles are sintered to form the finished product.

DETAILED DESCRIPTION

[0029] Generally, the devices used to practice the methods of the present disclosure are not particularly limited, and any known layer-by-layer 3D-printing apparatus may be used. As one example, a diagram of an inverse stereolithographic apparatus (SLA) 100 is shown in FIG. 1. Apparatus 100 includes base 120, which contains light source 122, light controller 124, and vat 126 containing solution 170. Generally, the composition of solution 170 is not particularly limited provided it contains a curable resin. Generally, any known light source may be used, e.g., lasers and light emitting diodes (LEDs). The wavelength of light may be selected to match the photocuring parameters of the resin, including, for example, visible and ultraviolet wavelengths. [0030] Light 127 is directed through window 130 into layer 101 of solution 170. Initially, the photocurable resin in layer 101 is uncured. The light controller responds to data defining the features of an individual layer of the 3D object and directs light to only those regions of layer 101 where the photocurable resin in solution 170 is to be cured. Upon exposure to light 127, the resin in layer 101 is photocured in the desired regions. Layers 102 through 110 have already been built according to this process and comprise the cured resin.

[0031] In view of the range of materials, applications, and potential post-processing steps, as used herein, the terms“cured” and“uncured” are relative. Generally, the resin is“cured” if it has been polymerized or crosslinked to an extent sufficient to achieve the desired structural integrity. For example, in a traditional process using photocurable resins and optional additives, a higher degree of

polymerization/crosslinking may be desired as the resin provides the mechanical properties of the final object. In systems forming gels, a lower degree of polymerization/crosslinking may be sufficient to hold the object in its desired shape throughout the subsequent drying and sintering steps.

[0032] During the build process, build platform (also referred to as a build plate) 160 is translated away from solution 170 in the direction indicated by arrow Z. In step-wise methods, the build platform is incrementally translated following each exposure step. In some applications, the position of the build platform is cycled away from and back toward the solution to aide in providing a layer of uncured resin between the lower surface of the previously cured layer and light source 122.

[0033] In many applications, the build process includes the creation of hundreds or even thousands of layers. For illustration purposes, object 140 is shown as having only ten layers, which are identified as layers 101 to 110. Layer 110 was the first layer formed and is adhered to build platform 160. In some embodiments, layer 110 is directly bonded to the platform. In some embodiments, additional substances may be interposed between layer 110 and the build platform. For example, in some embodiments, a release layer or sacrificial layer may be used to simplify removal of the finished object from the build platform while minimizing damage.

[0034] When creating an object using a 3D-printing technique, the object is defined by a plurality of cross-sectional layers overlaying each other in the Z-direction corresponding to the build axis, i.e., the axis along which the build platform is translated (See FIG. 1).

[0035] Referring to FIG. 2A, the cross-section of potential irradiation of a layer is divided into array 180 of voxels 190 extending in the X-axis and Y-axis forming the build plane. As used herein, the term “voxel” refers to the three-dimensional structure defined by the thickness, width, and length of a pixel. That is, the term pixel refers to the two-dimensional cross-sectional area defined by the width and length, while the term voxel refers to the volume defined by this two-dimensional area and its thickness.

[0036] For simplicity, the layer is shown as a 10 by 10 array. The size and shape of voxels 190 is not particularly limited. Generally, smaller sized voxels are desired to achieve greater precision in the dimensions of the final object. In some embodiments, the width (W) and length (L) of the voxels independently range from 40 to 70 microns, e.g., 40 to 60 microns, or even about 50 microns (i.e., 50 +/- 5 microns). Although square pixels are shown, other shapes may be used including other polygons such as rectangles.

[0037] The thickness (T) of the layer is aligned with the Z-axis and is not particularly limited.

Generally, the thickness is selected based on the depth of penetration of the light into the curable resin, and the desired geometric precision of the final object. In some embodiments, the thickness of a layer ranges from 25 to 100 microns, e.g., 25 to 75, 40 to 60, or even about 50 microns (i.e., 50 +/- 5 microns). In some embodiments, the thickness of the layers may be the same for all layers. In some embodiments, the thickness may vary between layers.

[0038] The light controller is designed to selectively irradiate the composition on a pixel-by-pixel basis. The light penetrates to provide a depth of cure (D) extending in the Z-direction. Generally, the depth of cure depends on the light intensity, the time of exposure, and the properties of the composition. Typically, the depth of cure increases with increasing intensity and time of exposure. Compositions exhibiting greater absorption or scattering of the light, will generally result in a lower depth of cure. The light controller may be used to tune the light source to provide the desired depth of cure for a given composition. As shown in FIG. 2A, in some embodiments, the depth of cure, D, corresponds to the thickness T, corresponding to a single layer of voxels.

[0039] Referring to FIG. 3, a layer-by-layer method of forming an object according to the prior art is shown. Individual layers are formed by, e.g., photopolymerization of low molecular weight species (usually‘monomers’) wherein a polymerization reaction is initiated in specific regions by radiant activation.

[0040] In step 212, a composition comprising a polymerizable material is introduced to form a build layer. In step 214, the composition in each of the voxels corresponding to the desired object is irradiated and the polymerizable material is cured. In step 218, the build platform is translated to move the now fully -cured layer away from the light source and to create a new build region. Steps 212 to 218 are repeated to form the finished object. Thus, in this prior art process, the complete cross-sectional area of the object corresponding to each individual layer is cured prior to translating that layer away from the light source.

[0041] Referring to FIG. 2B, one such exemplary layer corresponding to layer 105 of FIG. 1 is shown. Filled voxels 192 are shown with black fill to indicate that the composition in these voxels is cured and will form part of the structure of the final 3D-printed object. Open voxels 199 are unfilled and correspond to open areas in the final part (i.e., areas that do not contain cured resin). These open areas correspond to regions exterior to the finished object, and internal voids (e.g., channels and cavities) within the object. Therefore, if layer 105 of FIG. 2B was to be formed using prior art processes, each of filled voxels 192 would be irradiated and the composition in those voxels would be cured in step 214, prior to translating the build platform in step 218.

[0042] As this process creates discrete, fully cured layers in succession with Z-directional displacement between layers, the predominant polymer network is believed to be in the X-Y plane. This can lead to inferior mechanical properties in the Z-direction compared to the properties within the X-Y plane as loads must be transferred across interfaces between cured layers.

[0043] In some processes, Z-direction properties are further limited because of polymerization gradients. Generally, the portion of a layer closest to the radiation source will receive the highest intensity of radiation. Thus, this portion can have the highest degree of polymerization and the fewest remaining chemically active sites available for further polymer propagation. By contrast, the portion of a layer furthest from the radiant energy source may receive the least energy due to energy absorption or scattering from material within the layer. When forming a second layer adjacent a previously cured layer, the most reacted surface from the previous layer must react with the region of the new layer that is receiving the lowest energy dose. This can lead to unfavorable polymerization conditions, limiting polymerization across the boundary of adjacent layers and reducing interlayer adhesion.

[0044] As a result, the prior art layer-by-layer production methods can lead to the formation of objects with significant anisotropy. The mechanical properties, such as modulus and strength, can be much lower between layers in the Z-direction as compared to the properties in the build plane (i.e., the X-Y plane).

For example, poor interlayer adhesion can lead to separation between layers in the Z-direction at low loads compared to the loads required to deform or fracture the object in the X- or Y-directions.

[0045] These problems may be particularly important when producing parts using a relatively small amount of reactive binder to hold relatively inert particles in a shape or where the overall strength is further reduced by the presence of a solvent such as when manufacturing a gel. For example, where the binder material needs to maintain a shape and interlayer continuity while volatile components including solvents are removed, significant weakness in any dimension can be unacceptable.

[0046] Thus, although three-dimensional, layer-by-layer manufacturing processes have many advantages, such processes frequently result in anisotropic material properties, and challenges in improving interlayer adhesion remain.

[0047] An exemplary method of the present disclosure is shown in FIG. 4 with reference to FIGS. 2B through 2E. Again, FIG. 2B illustrates the desired final cross section of the object corresponding to layer 105. In the final object, the composition in filled voxels 192 will be cured, while open voxels 199 will be empty.

[0048] Referring to FIG. 2C, the cross-sectional solid area of layer 105 is defined by first identifying the border voxels 193 that form the boundaries of the object. These border voxels are identified by the letter“B”. Next, the enclosed voxels 194 that form the interior solid portion of the object are identified by the letter“I”. Finally, the open voxels 199 are identified by the letter“O” as they form the open or unfilled exterior and interior portions of the object.

[0049] There are several options for distinguishing a border voxel from an enclosed voxel. Referring to FIG. 5A, voxel 194a, has six faces F, twelve edges E, and eight comers, C. Referring to FIG. 5B, at a minimum, enclosed voxel 194a shares all its faces with adjacent filled voxels 192 (either border voxels or other enclosed voxels). As a voxel has six faces, such an enclosed voxel is said to be“six-connected.” Such an enclosed voxel may also be referred to as a“face-connected” voxel. In such embodiments, a border voxel is less than six-connected, that is, a border voxel shares at least one if its faces with an open voxel.

[0050] While this definition may be suitable for some applications, in some embodiments, it may be desirable to ensure that all edges of an enclosed voxel are also adjacent filled voxels. In such

embodiments, an enclosed voxel must share all its faces and edges with other filled voxels. As a voxel has six faces and twelve edges, such an enclosed voxel is said to be“eighteen-connected.” Such an enclosed voxel may also be referred to as a“face and edge-connected” voxel. In such embodiments, a border voxel is less than eighteen-connected.

[0051] In some embodiments, it may be desirable to ensure that even all comers of an enclosed voxel are adjacent filled voxels. In such embodiments, an enclosed voxel must share all its faces, edges, and comers with other filled voxels. As a voxel has six faces, twelve edges, and eight comers, such an enclosed voxel is said to be“twenty-six-connected.” Such an enclosed voxel may also be referred to as a “face, edge, and comer-connected” voxel. In such embodiments, a border voxel is less than twenty-six- connected.

[0052] As used herein, unless the context requires otherwise, the term enclosed voxel includes face- connected, face and edge-connected, and face, edge, and comer-connected voxels. The terms six- connected or face-connected, eighteen-connected or face and edge connected, and twenty-six-connected, or face, edge, and comer connected are used when referring to specific subsets of enclosed voxels. The term border voxel then depends on how the enclosed voxels are defined in the specific embodiments.

[0053] To determine whether a filled voxel is a border voxel or an enclosed voxel, it is not sufficient to consider only the layer in the build region. Rather, one must also consider both the previous layer and the subsequent layer to determine if the desired face-, face and edge-, or face, edge, and comer- connections are present. However, for the sake of simplicity, the following description only refers to a single layer in the build region. For this purpose, it is assumed that the filled and open voxels in the preceding and subsequent layers will be such that the desired conditions are met in the definitions of the border and enclosed voxels, for example, if the layer being built is identical to the preceding and subsequent layers.

[0054] Referring to FIG. 2C, face-connected voxels are considered enclosed voxels. Again, assuming every filled voxel in the build plane is adjacent a filled voxel in both the preceding and subsequent layers, the upper and lower faces of every filled voxel will be shared with an adjacent filled voxel. Thus, any voxel sharing at least one face in the X-Y plane with an open voxel 199 is a border voxel 193. As a result, all surfaces of the object, including both exterior surfaces and interior surfaces are bounded by border voxels 193. Enclosed voxels 194 do not share any faces with an open voxel 199. Thus, each face of an enclosed voxel 194, including its upper and lower faces, is shared with a face of either a border voxel 193 or another enclosed voxel 194. If enclosed voxels had been defined as face and edge- connected voxels, then voxel 194a would be a border voxel as it shares an edge with open voxel 199a. [0055] As shown in FIG. 4, in step 312, a composition comprising a polymerizable material is introduced to form a build layer.

[0056] In step 313, the composition in each border voxel is irradiated and the polymerizable material is cured. Referring to FIG. 2D, this results in cured material in each of the border voxels 193. In the methods of the present disclosure, when each layer is formed, all the border voxels are cured, while only some of the enclosed voxels are selectively cured. By curing all the border voxels, surface defects can be substantially reduced or eliminated. In addition, the resin in the uncured enclosed voxels is enclosed and flow of the uncured resin into areas intended to remain open is inhibited or prevented.

[0057] Referring to FIGS. 2C and 2D, in the methods of the present disclosure not all the enclosed voxels of a layer are cured in the same step. Instead, enclosed voxels 194 are separated into two subsets: the enclosed voxels selected for curing 195 (identified with the letter“C”), and the enclosed voxels selected to remain uncured 197 (identified with the letter“U”).

[0058] For any given layer, the percentage of enclosed voxels selected for curing may range from 0 to 100%, and, correspondingly, the percentage of enclosed voxels selected to remain uncured may range from 100% to 0%. However, in the methods of the present disclosure, for at least some layers, the percentage of enclosed voxels selected to remain uncured is greater than 0%.

[0059] In step 315, the composition in each enclosed voxel selected for curing 195 is irradiated and the polymerizable material is cured. Steps 313 and 315 may be performed simultaneous or in either order. In some embodiments, the border voxels and the enclosed voxels selected for curing are cured in the same step.

[0060] In step 318, the build platform is translated to move the partially-cured layer, shown in FIG. 3E, away from the light source to create a new build region. That is, step 318 occurs prior to irradiating the composition in the enclosed voxels selected to remain uncured 197. In FIG. 2E, cured border voxels 193 are shown with black fill to distinguish from cured enclosed voxels 195, which are shown with a hashed line fill. Uncured enclosed voxels 197 are identified with the letter“U” and open voxels 199 are unfilled. Steps 312 to 318 are repeated to form the finished product.

[0061] If no further curing steps are performed, the final object would comprise a cured border defined by the border voxels, and an enclosed interior volume bounded by the border voxels. The composition of some or all the enclosed voxels would contain uncured composition. In some

embodiments, such an object may be suitable for its intended application.

[0062] However, in many applications, a fully-cured object is preferred. In some embodiments, the object may be post-cured after the 3D-printing process is complete. For example, in some embodiments, the object may be subjected to irradiation (e.g., visible light, infrared light, ultraviolet light, microwave, e- beam or other forms of irradiation to cure the composition in the enclosed voxels that were selected to remain uncured in the 3D-printing operation. In some embodiments, the composition in the enclosed voxels that were selected to remain uncured in the 3D-printing operation may be thermally cured. [0063] In some embodiments, the composition may be selected such that it post-cures without the need for external energy. For example, in some embodiments, the composition may be selected so that it self-cures. In such embodiments, the 3D-manufactuimg process may be used to rapidly form the borders and, optionally selected interior portions of the object, while the remaining interior portions cure more slowly, subsequent to the 3D-manufacturing process.

[0064] In many applications, however, it can be desirable to have a fully-cured object upon completion of the 3D-manufacturing process. Thus, in some embodiments, the methods of the present disclosure may be modified to produce a build region extending across more than one build layer.

[0065] A diagram of inverse stereolithographic apparatus 400 is shown in FIG. 6. Apparatus 400 includes light source 422 and light controller 424. Light 427 is directed through window 430 into one or more layers of object 440. The light controller responds to data defining the features of an individual layer of the 3D object and directs light to only those regions of layer 401 where the photocurable resin is to be cured. Upon exposure to light 427, the resin in layer 401 is photocured in the desired regions.

[0066] Interior layers 402 through 405 have already been built according to the methods of the present disclosure. Therefore, these layers will include some combination of open voxels (“O”) that do not contain resin, border voxels (“B”) containing cured resin, enclosed voxels containing cured resin (“C”), and enclosed voxels containing uncured resin (“U”). To produce a finished part while minimizing or eliminating the percent of enclosed voxels containing uncured resin, in some embodiments, one or more of the light source, the light controller, and the layer thicknesses can be modified to provide a depth of cure, D, that exceed the thickness of a single layer, T.

[0067] As shown in FIG. 6, when irradiating layer 401, the depth of cure, D, is sufficient to cure layers 402 and 403 as well. Therefore, when irradiating and curing voxel 401a in layer 401, voxels 402a and 403a are also irradiated with sufficient energy to cure the composition. As formed, voxels 402a and 403a were enclosed voxels containing uncured resin. However, after forming layer 401 and irradiating voxel 401a with a depth of cure D, the resin in voxels 402a and 402b is cured converting them to enclosed voxels containing cured resin.

[0068] Because the depth of cure, D, was not sufficient to cure layer 404, the resin in voxel 404a remains under-cured or uncured. Depending on the desired end-use, some level of uncured voxels may be acceptable. However, in some embodiments, it may be desirable to minimize or eliminate the presence of uncured cells

[0069] In some embodiments, the algorithms used to select the subset of enclosed voxels containing uncured resin may be programmed to ensure the total thickness, T(total), of adjacent uncured voxels is no greater than the depth of cure D as shown in Formula 1.

T(total) < D (1)

If the thickness of each layer is T, then the maximum number of layers that can be cured simultaneous,

M, is

M < D/T (2) If the thickness of the layers varies, then

M < D/Tavg (3)

where Tavg is the average thickness of the layers extending across the depth of cure, D.

[0070] In some embodiments, the algorithms used to select the subset of enclosed voxels containing cured and uncured resin may be programmed to provide a fully-nested pattern of enclosed cells containing cured resin. For example, if the depth of cure is sufficient to cure M layers, the number of repeating patterns needed to achieve a fully-nested pattern should be less than or equal to M. For example, if the depth of cure is sufficient to cure three layers, an algorithm delivering a pattern of cured (“C”) and uncured (“U”) enclosed cells requiring two nested patterns (FIG. 7A) or three nested patterns (FIG. 7B) may be used.

[0071] As shown in FIG. 7A, nested pattern 701 consists of two complimentary patterns used in adjacent layers 702 and 704. To achieve a nested pattern, each uncured enclosed cell of layer 702 is adjacent a cured enclosed cell in layer 704. In some embodiments, each cured enclosed cell (C) of layer 702 is adjacent an uncured enclosed cell (U) in layer 704. In such an alternating pattern of nested cells, when moving vertically through a stack of adjacent enclosed cells, the algorithm would alternate cured and uncured enclosed cells. In some embodiments, there may be an overlap in cured cells. In some embodiments, a cured enclosed cell in one layer may be adjacent another cured cell in in the adjacent layer. As a result, alternating patterns are a subset of fully-nested patterns. In some embodiments, the patterns can vary across the Z-direction. For example, the pattern for each even numbered layer may be randomly or independently selected and the patterns for every odd numbered layer would then be the complimentary patterns.

[0072] Referring to FIG. 7B, fully-nested pattern 711 extends across layers 712, 713, and 714. Fully- nested pattern 715 also extends across three layers (716, 717, and 718). The algorithm is programmed to ensure that, when there are three enclosed cells aligned across all three layers, at least one of the enclosed cells will be a cured enclosed cell.

[0073] In some embodiments, each fully-nested pattern is the same across the total Z-dimension of the object. In some embodiments, the fully-nested patterns may be independently selected. In some embodiments, referring to pattern 711, a three-layer alternating pattern may be used, wherein there is one and only one cured enclosed voxel in each group of three adjacent enclosed voxels across the three layers. In some embodiments, for example, in pattern 715, there may be two or even three cured enclosed voxels in each group of three adjacent enclosed voxels across the three layers. In some embodiments, the patterns in one or more layers may be random, and the pattern in one or more layers may be selected to ensure there is at least one cured enclosed voxel in each group of three adjacent enclosed voxels across the three layers.

[0074] For a depth of cure D, extending across a greater number of layers, e.g., M =4, M = 5, or even greater, fully -nested patterns may be created that require any number of layers up to M. For example, if D is such that M = 5, i.e., 5 layers can be cured simultaneously, then nested patterns requiring, 2, 3, 4, or even 5 layers may be used. The paterns selected for each group of layers may be the same or may be independently selected. For example, in some embodiments, the enclosed voxels selected to be cured and uncured may be selected randomly, while the enclosed voxels selected to be cured in at least one layer may be selected to ensure that there is at least one cured enclosed cell in each group of enclosed cells extending across the M adjacent layers.

[0075] In some embodiments, the algorithms used to select the subset of enclosed voxels may provide for the periodic placement of fully-cured layers. For example, referring to FIG. 8, if the depth of cure is sufficient to cure three layers, an algorithm delivering a patern of cured (“C”) and uncured (“U”) enclosed cells may be used for sets of two layers (801 and 803), interleaved with layers (802 and 804) containing only cured voxels, i.e., either border voxels or, as shown, cured enclosed voxels (“C”). The patern of cured and uncured enclosed voxels in sets of layers 801 and 803 is not limited and may include any desired patern, including random paterns of cured and uncured enclosed voxels. In some embodiments, two or more fully-cured layers may be interspersed between selectively-cured layers.

[0076] In some embodiments, it may be desirable to avoid fully-cured layers and fully-nested paterns. When using random paterns of cured and uncured enclosed cells, the probability of exceeding a maximum threshold of uncured enclosed cells in the finished object can be controlled by selecting the ratio of enclosed cells that are left uncured in each layer, RUC, where RUC is equal to ratio of the number of enclosed cells left uncured in a layer divided by the total number of enclosed cells in the layer.

[0077] If, based on the depth of cure, D, and the thickness (or average thickness) of the layers, T, M layers can be cured, then uncured enclosed cells will remain if there are M consecutive layers with adjacent uncured enclosed cells when the M+l layer is cured. For each voxel, the probability of this occurring, Pv, is:

Pv = RUC ^M (4)

[0078] Each layer may comprise millions of voxels and the final object may contain thousands of such layers. As a result, using completely random paterns will result in some percentage of uncured voxels in the finished product. Equation (4) may be used to estimate the final percent of uncured enclosed cells based on the total number of enclosed cells, P%.

P% = l00*RUC ^M (5)

[0079] Using Equation (5), the ratio of uncured cells (RUC) and the depth of cure expressed as the number of layers that may be cured (M) can be selected to achieve a desired percent of uncured enclosed voxels, P%. For example, in some embodiments, it may be desirable to limit the percent of uncured enclosed voxels in the final object to be no greater than 10%, e.g., no greater than 5%, no greater than 1%, no greater than 0.5% or even no greater than 0.1%.

[0080] Referring to Table 1, if the threshold value of uncured enclosed voxels in the final object, P%, was selected as 1.0%, the depth of cure could be as low as 2 layers when using an RUC of 0.10 (i.e., 10% of the enclosed cells are randomly selected to remain uncured). However, if RUC is increased to 0.30 (i.e., 30% of the enclosed cells are randomly selected to remain uncured), the depth of cure would have to be increased to 4 layers to achieve a P% value of less than or equal to the threshold value of 1.0%.

Table 1: Percent uncured voxels (P%) as a function of RUC and M.

[0081] Equation (5) may be rearranged to determine the maximum value of RUC that may be used based on the depth of cure expressed in layers, M, and the threshold value, P, where P = R%/100.

RUC = VP (6)

[0082] Referring to Table 2, if the maximum percent of uncured enclosed voxels in the final object, P%, is selected to be no greater than 1% and the depth of cure is only 2 layers, then the maximum value of RUC would be no greater than 0.10 (i.e., up to 10% of the enclosed cells could be randomly selected to remain uncured). However, if the depth of cure is increased to 5 layers, up to 40% of the enclosed cells could be randomly selected to remain uncured (i.e., RUC could be as great as 0.40) while remaining under the threshold of 1%.

Table 2: Maximum RUC based on the percent of uncured enclosed voxels, P% and the depth of cure expressed as the number of layers cured, M.

Maximum RUC expressed as percent

[0083] In some embodiments, RUC is at least 0.1, e.g., at least 0.2 or even at least 0.3. In some embodiments, RUC is no greater than 0.9, e.g., 0.8, or even no greater than 0.7. In some embodiments, RUC is between 0.2 and 0.8, e.g., between 0.3 and 0.7, or even between 0.4 and 0.6, including the end points. In some embodiments, RUC is the same for all layers. In some embodiments, RUC may be independently selected for each layer. In some embodiments, RUC may be selected for a group of adjacent layers such the sum of RUC for the adjacent layers is at least 1.0, e.g., at least 1.1.

[0084] Examples REF-l and EX-l.

[0085] Dimensionally accurate, 3D-formed polytetrafluoroethylene (PTFE) containing hydrogels are of interest because they can be further processed (dried, de-bindered, and sintered) to obtain dense PTFE structures with features more complex and detailed than traditionally processed/machined PTFE.

Generally, the methods use low viscosity aqueous polymer emulsion/binder solutions combined with forming techniques including Stereolithography (SLA) type 3D printing to produce controlled structure (3D design) hydrogels.

[0086] A common technique for SLA printing involves what is known as‘reverse vat’ wherein (unlike standard vat) the photo curing radiation source is located below a vat containing a printable liquid. In most cases, the reverse vat technique involves a series of discrete layer formations. In this case there is a layer illumination step and a translation step. The translation step involves moving the build surface and incomplete part to a net upward position corresponding to the layer thickness. Illuminations and translations alternate until the printing step is complete.

[0087] One disadvantage of the reverse vat technique for discrete layer type printing is that the method involves curing a binder against a transparent window. This means that each cured layer must release from the window surface before creating the next layer. Low surface energy materials such as fluoroplastic films frequently form a release surface in such window constructions providing a lower tendency to bond to photocurable binders. However, even if only due to the initial suction created by the upward movement of the part during printing, there is some force tending to separate printed layers in a part that occur as a part is raised away from the lower window of the vat during the translation phase.

[0088] A second disadvantage is that the lower portion of each layer has the closest proximity to the radiation source and therefore is least likely to have active sites for further crosslinking. As the top surface of the subsequent layer that is in contact with this heavily cured surface of the previous layer is furthest away from the radiation source, the formation of crosslinking bonds between layers is not favored. Also, as discrete layers are cured, bridging between layers is not likely, resulting in potential cleavage planes.

[0089] The resulting gels are then subjected to subsequent steps including drying, de-bindering (via burnout), and sintering. All these steps can involve volume changes that result in stresses likely to further weaken inter-laminar cohesion. For this reason, increased interlaminar adhesion is useful.

[0090] A 3D-printable composition containing a polytetrafluoroethylene (PTFE) dispersion was prepared as follows. First, 80 grams of a modified PTFE dispersion was weighed into a first bottle followed by agitation by a lab bottle roller. Separately, a first acrylic monomer (7 g of SR 415 from Sartomer) and a second acrylic monomer (7 g of SR 344 from Sartomer) were weighed into a second bottle and mixed. Subsequently, 0.288 g of a photoinitiator (OMNIRAD TPO-L), 0.115 g of an inhibitor (BHT from Sigma Aldrich), and 0.058 g of an optical brightener (MAYZO OB-M1) were added to the second bottle and the contents were agitated on a lab bottle roller for at least 30 minutes. Finally, 20 g of water were added with further mixing on the lab bottle roller to form the binder mixture. Upon complete mixing, the binder mixture was slowly added to the dispersion and the resulting 3D-printable composition was further agitated on a lab bottle roller for the entire time before use.

[0091] The part design was a 40 mm by 20 mm sheet. Samples were printed layer by layer to a thickness of 1. 5mm using a RAPIDSHAPE HA40 SLA type printer (Rapidshape GmbH, Heimsheim, Germany) with the conditions shown below in Table 3. The object was a solid sheet, bounded on the top and bottom by fully-cured layers, where each interior layer was identical and consisted of a border surrounding a solid rectangular interior.

Table 3: Summary of printing parameters.

[0092] Reference Example REF-l was prepared by curing the entirety of each layer in each step. That is, when an individual layer was printed, all voxels - both the border voxels and all enclosed voxels - were irradiated to cure the resin, forming the hydrogel.

[0093] Example EX-l was prepared using a method of the present disclosure. Top and bottom layers were fully cured as they consisted of only border cells forming the top and bottom surfaces of the object. When forming the interior layers, all border voxels were irradiated to cure the resin. The enclosed voxels were divided into an array of 1 mm by 1 mm cells. The enclosed voxels in these interior layers were irradiated in a checkerboard pattern. The cells selected for irradiation were alternated in alternating layers such that each cell was irradiated in alternating steps. As the exposure energy was selected to ensure a cure depth of at least two layers, i.e., at least 100 microns, all enclosed voxels were exposed to sufficient energy to cure the resin and form the hydrogel.

[0094] Following each print, the hydrogel samples were rinsed in deionized water, residual surface liquids were blown off with a lightly pressurized nitrogen gas stream, and the samples were post cured under ultraviolet light for 30 seconds (DYMAX light curing system Model 2000 Flood with a 400 Watt EC power supply). These samples were then subjected to solvent exchange with ethanol by submerging the printed hydrogels in two sequential soaks in 200 proof ethanol for a minimum of one day each.

Preparation for supercritical C02 extraction was performed by transferring solvent exchanged hydrogels from the second ethanol bath into metal carriers. To minimize ethanol evaporative drying, samples were periodically wetted with ethanol during the transfer. When carriers were loaded they were placed into an extraction chamber.

[0095] The supercritical extraction was performed using a 10-Liter laboratory-scale supercritical fluid extractor unit designed by and obtained from Thar Process, Inc., Pittsburgh, PA, USA. The PTFE-based gels were mounted in a stainless-steel rack. After the extractor vessel lid was sealed in place, liquid carbon dioxide was pumped by a chilled piston pump (set point: -8.0°C) through a heat exchanger to heat the C02 to 50°C and into the 10-L extractor vessel until an internal pressure of 13.3 MPa was reached.

At these conditions, carbon dioxide is supercritical. Once the extractor operating conditions of 13.3 MPa and 50°C were met, a heated needle valve regulated the pressure inside the extractor vessel. The C02 and dissolved ethanol flowed downstream into a 5-L cyclone separator vessel that was maintained at room temperature and a pressure of less than 5.5 MPa, where the extracted ethanol and gas-phase C02 were separated and collected throughout the extraction cycle. Supercritical carbon dioxide (scC02) was pumped continuously through the 10-L extractor vessel for three hours from the time the operating conditions were achieved. After the three-hour extraction cycle, the extractor vessel was vented into the cyclone separator over one hour from 13.3 MPa to atmospheric pressure at 50°C before the lid was opened and the stainless-steel rack containing the dried aerogels were removed.

[0096] The dried aerogels were placed on a bed of ceramic (La Zr A1 oxide) beads in aluminum pans for burnout and sintering in a Despatch Industries Model RAF 1-42-2E programmable air circulating oven programmed according to the following program:

After step 9, the oven was opened and the samples were allowed to cool to room temperature.

[0097] Tensile specimens were prepared and tested according to ASTM D1708. The test was conducted at an extension rate of 12.7 mm per minute. Tensile Pressure at break and Elongation at break were recorded for each sample and the averages and standard deviations are reported below.

Table 4: Mechanical properties of REF-l and EX-l.

[0098] Examples REF-2 and EX-2 to EX- 16.

[0099] A 3D-printable, UV-curable acrylate composition was prepared. The composition contained 45 wt.% urethane elastomer (EXOTHANE 10, available from Esstech, Inc.), 26 wt .% 2-phenoxyethyl methacrylate, 18 wt.% 2-ethylhexyl methacrylate, and 11 wt.% isobomyl methacrylate to yield 100 parts by weight resin. The composition further included 0.5 parts per 100 parts resin (phr) photoinitiator (TPO, available from IGM Resins USA Inc.), 0.025 phr inhibitor (BHT, (2,6-Di-tert-butyl-4-methylphenol), product #34750, available from Sigma-Aldrich, St Louis, MO, USA), and 0.025 phr naphthalimide acrylate.

[0100] Dog -bone samples were designed according to the dimensions of ASTM D-638 Type V. The design was modeled in a 3D computer-aided drafting program (from Dessault Systemes SOLIDWORKS Corp.) and converted to .STL format for import into software program to convert the solid model into individual layers (NETFABB software from Source Graphics). The dog -bone samples were designed such that the layers were stacked with their X-Y planes perpendicular to the lengthwise direction of the dog -bone samples.

[0101] A .TGZ file containing a pair of linked files for each layer was prepared. One file was a .PNG image defining the two-dimensional image mask for the layer. The second was an .XML file defining the corresponding parameter set for that layer. These files were customized in a scripting language (Python Software Foundation). Parameter customization was handled through regular expression matching and replacement of individual parameters. The final values of the modified parameters are listed in Table 5.

Table 5: Modified printing parameters

* 10 second exposure at 10 Watts [0102] Sixteen print profiles were prepared, including one fully-cured control, REF-2. For each sample, the layer-by-layer images were loaded into memory to produce a three-dimensional matrix of the build. The border voxels in each layer were identified based on networks of 26-connected pixels. That is, enclosed voxels were 26-connected (i.e., an enclosed voxel was face, edge, and comer connected to adjacent filled voxels), while border voxels were less than 26-connected (i.e. any voxel having an open voxel adjacent any of its faces, edges, or comers was identified as a border voxel). This analysis was conducted by ascertaining the status of the voxels in the layer being built as well as the preceding and succeeding layers.

[0103] When processing a layer, all border voxels were illuminated to cure the resin. With REF-2, all enclosed voxels were also irradiated to cure the resin. For each of Examples EX-2 to EX- 16, a patern was applied to the enclosed pixels, as summarized in Table 6.

Table 6: Irradiation paterns.

[0104] In Table 6, cell size refers to the size in voxels that are grouped in a cell and treated collectively when establishing the patern. For example, a cell size of 3 x 3, means each cell is a group of enclosed voxels forming a 3 x 3 array. (For example, EX-4.) RUC is the ratio of cells selected to remain uncured over the total number of enclosed cells. The“Patern” determines how the RUC value is used.

In the 50% checkered patern, alternating cells are selected for irradiation forming a two-dimensional checkerboard array of cured and uncured enclosed cells. EX-7 had alternating cells where each cell was an individual enclosed voxel, while EX-8 had alternating cells where each cell was a 3 x 3 array of enclosed voxels.

[0105] In the random paterns, the algorithms used the RUC value as a probability applied to each cell. Thus, for a 50% RUC random patern (EX-2 through EX-10), each cell had a 50% probability of being selected to remain uncured. Thus, it is possible to have two more cured or uncured enclosed cells next to each other, but the average number of uncured cells should remain approximately 50% across an entire layer. For a 40% RUC random patern (EX-l 1 and EX- 12), each cell had a 40% probability of being selected to remain uncured. [0106] In some embodiments, the patterns were selected to repeat. In Table 6,“alternating” is associated with the checkered patterns (EX-7 and EX-8). This means the same pattern alternated so that every even numbered layer had the same checker pattern, and every odd numbered layer had the same complimentary checker pattern. The term“2x alternating” means that every adjacent pair of layers had complimentary patterns resulting in a nesting pattern. However, these patterns changed over the Z- direction of the object. Thus, one layer had a first random pattern and its adjacent layer had the inverse of that pattern. However, the next layer had a new randomly selected pattern, with its adjacent layer having the inverse of that pattern. As a result, if the RUC for a layer was 50%, the RUC for its adjacent layer was the inverse and also had an RUC of 50% (EX-l to EX-5). If the RUC for a layer was 40%, its adjacent layer was the inverse and had an RUC of 60% (EX- 13 and EX- 14).

[0107] Examples EX- 15 and EX- 16 were based on a pattern designed to provide greater lengths of uncured enclosed cells across adjacent layers. As shown in FIG. 9, the base pattern was a 5 x 5 cell 1190 with thirteen uncured enclosed voxels 1194 and twelve cured enclosed voxels 1193, resulting in an RUC of 52%. The cells 1190 were placed in a row along one axis, and were offset by one voxel in adjacent rows within the same plane. This pattern was repeated for every layer, except that the entire pattern is shifted down one column of voxels between each layer (EX- 15) or two columns of voxels between each layer (EX- 16).

[0108] Ten samples were prepared for each pattern by printing the UV-curable acrylate composition to form the dog bone shapes. The printed samples were washed with isopropyl alcohol. The washed samples were post-cured for three hours in a UV chamber and dried for 18 hours in a vacuum oven at 100 °C.

[0109] The tensile strength at break, modulus, and elongation at break were measured according to ATSM Test Method D638 on an INSIGHTS MTS with a 5 kN load cell at a rate of 10 mm/minute. The dog-bone samples were prepared such that the layers were stacked with their X-Y plane perpendicular to the lengthwise direction of the dog-bone samples. As a result, the load applied in the tensile tests was perpendicular to the layers, i.e., parallel to the thicknesses of the layers, such that the load must be transmitted across adjacent layers. Digital image correlation was used to measure the strain and strain distributions in the samples. The tensile strength, modulus, and elongation to break were determined according to ASTM D638-10. Ten replicates were tested and the averages and standard deviations are summarized in Tables 7, 8 and 9.

Table 7: Mechanical properties.

[0110] The ratio of the tensile strength of the selectively-cured to the tensile strength of the control sample (REF-2) is shown in Table 8. The tensile strength at failure for all selectively-cured samples was higher than the control. This indicates that the interlayer strength was greater.

Table 8: Tensile strength improvement compared to the control (REF -2) as a percent.

[0111] As shown in Table 9, the modulus was also higher for all selectively cured samples as compared to the control. Generally, there is a trade-off between the modulus and the elongation at break. As a result, through selective curing, samples can be produced with higher or lower moduli and elongations at break as compared to samples prepared from standard, fully-cured layers, while retaining the benefits of the improved tensile strength at break.

Table 9: Modulus and Elongation at break compared to the control (REF-2) as a percent.

[0112] Various modifications and alterations of this invention will become apparent to those skilled in the art without departing from the scope of this invention.