Apparatus and method for layerwise formation of 3D objects from particulate build material

Abstract

Provided is an apparatus (10) for the layerwise manufacture of 3D objects from particulate build material, the apparatus comprising a distribution device (290) to distribute each layer (160); a digital light projector comprising a digital mirror device (240) and a light source (250), wherein the digital mirror device comprises an array of individually addressable micromirrors, each arranged to redirect an energy beam incident on the array from the light source towards the layer to heat corresponding voxels of the layer; and a controller arranged to: control the distribution device (290) to distribute a layer (160); control each micromirror to selectively provide non-fusing energy beams (420H,4220L) of different non-fusing energies to the layer to heat corresponding voxels to a predefined temperature below the melting temperature; to repeat the distributing and heating to form a plurality of layers; and for at least one of the plurality of layers, to control a fusing energy beam source to provide fusing energy (410) to voxels within the object cross section of said at least one layer so as form a 3D object. A method for operating the apparatus is also provided.

Description

[0001] APPARATUS AND METHOD FOR LAYERWISE FORMATION OF 3D OBJECTS FROM PARTICULATE BUILD MATERIAL

[0002] FIELD OF THE INVENTION

[0003] The present disclosure relates to the field of powder bed fusion for the layerwise manufacture of 3D objects from particulate build material, and in particular to non-fusing heat sources and their operation. Apparatus and methods of control therefor are also disclosed. The apparatus and its operation may find particular benefits in applications requiring improved part uniformity and enhanced recyclability of unfused build material.

[0004] BACKGROUND

[0005] In powder bed fusion processes like print and fuse or laser sintering, an object is built up layer by layer from powder material by selectively fusing, or melting, the powder over successive object cross sections. This results in the object being built up within a supportive bed of unfused powder that can be removed at the end of the process by simply brushing and / or blowing the supportive powder off by pressurised gas streams. Recently, improvements to powder bed fusion processes have led to high yield through fast layer times and increasing build bed size. While laser sintering processes typically are more time consuming due to having to trace the object cross section along a route, print and fuse processes employ scanning printheads and lamps that can process each layer in a single pass.

[0006] Powder bed fusion processes require careful control of the temperature across the layer. For example, if certain areas of the layer get relatively cold, this can result in insufficient fusion of build material within object areas and / or unwanted curling of fused material in those cooler areas. Alternatively, overheating the unfused build material leads to relatively hot areas, leading to a hard cake in those areas and unwanted powder ageing / oxidation. To some degree this may be addressed by controlling the temperature of each layer before the step of fusing, for example by a static and / or scanning infrared heater that preheats each newly deposited layer to a temperature within the operating window, which is a temperature range above the solidification temperature of the powder and below the melting point of the powder. Nonetheless, such processes retain limitations that may be addressed by the present invention disclosed herein. SUMMARY

[0007] The invention is set out in the appended independent claims, while particular embodiments of the invention are set out in the appended dependent claims.

[0008] BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Reference is now directed to the drawings, in which:

[0010] Fig. 1 is a schematic cross section side view through a build bed of an apparatus for the layerwise manufacture of 3D objects according to the invention;

[0011] Fig. 2 is a variant of Fig. 1;

[0012] Fig. 3A to 3C are variants of Fig. 1 illustrating different stages of selectively heating sequential layers based on slice data;

[0013] Fig. 4 is a variant of Fig. 1 in which the top most layer comprises a fused cross section;

[0014] Fig. 5 is a flow chart of a method of using the apparatus according to the invention;

[0015] Figs. 6A to 6D illustrate different stages of processing a layer comprising an object cross section in a variant of Fig. 4 comprising a scrolling digital mirror device (DMD);

[0016] Figs. 7A to 7E illustrate different stages of concurrent heating and fusing in a variant of Fig.

[0017] 4 comprising one or more scrolling DMDs;

[0018] Fig. 8A illustrates a mask to be applied to a slice for a DMD;

[0019] Fig. 8B illustrates a slice comprising non-fusing heating data;

[0020] Fig. 8C illustrates a schematic top view of a layer heated according to a combination of the slice of Fig. 8B modified by the mask of Fig 8A;

[0021] Fig. 9 illustrates a series of mask weights used to modify the pixel data of anon-fusing slice; Figs. 10A to 10D illustrate a sequence of slices modified by a mask for a scrolling DMD;

[0022] Fig. 11 is a flow chart illustrating a method for providing modified non-fusing slice data; and

[0023] Fig. 12 is a flow chart illustrating a method for dynamically correcting modified non-fusing slice data.

[0024] In the drawings, like elements are indicated by like reference numerals throughout. It should be noted that the drawings are not to scale and that certain features may be exaggerated in size to be more clearly visible. DETAILED DESCRIPTION

[0025] In the following, a “fusing energy beam source” is operable to fuse or consolidate build material by being operated at an energy output and / or wavelength capable of raising the temperature of the build material to or above the melting temperature. In this way, fusion of the build material over object cross sections may be achieved. By contrast, a “non-fusing energy beam source” is operated at a comparatively lower energy output and / or at a different wavelength, capable of raising the temperature of the build material without it reaching or exceeding the melting temperature. Herein, a heating step may be applied at any suitable instance of the process, and may comprise a “preheating” step immediately following distribution of a new layer.

[0026] A “selective” heat source herein is a heat source able to address the layer per voxel or unit control area so as to selectively heat to fuse or warm specific areas of the layer. A “non- selective” heat source herein is a heat source that applies heat over groups of voxels at a time without being configured to selectively heat or fuse specific areas of the layer in absence of other means to enhance the absorption of radiation of the non-selective heat source at specific voxels of the layer. Such means may be a printhead selectively applying radiation absorber to specific voxels that are to fuse. Examples of non-selective heat sources are scanning infrared bar heaters conventionally used in print and fuse apparatus, or the overhead heaters fixedly arranged above the layer to help maintain a uniform temperature over the layer.

[0027] In conventional print and fuse processes, non-fusing energy is provided by non-selective heat sources. The non-selective heat source is typically an elongate infrared lamp that spans the width of the layer and that is operated as it traverses the layer to provide non-fusing energy. Such a ‘preheat’ or “warming” lamp is often applied in combination with an overhead heater comprising an array of individually controllable heating elements. The overhead heater may be operated throughout the layer cycle based on thermal feedback of the layer temperature to heat zones of a plurality of voxels at a time. This is usually done with the aim of achieving a uniform temperature across the non-fusing area within the layer, outside the object cross sections. Neither heat source is capably of providing selective control over applying non-fusing energy to the layer. To overcome present shortcomings of conventional powder bed fusion apparatus, the inventors have realised that a digital light projector (DLP) may be provided to such apparatus to achieve selective heating with non-fusing energy, and controlled to achieve certain advantageous effects according to a method described herein. The apparatus and method according to the invention and their respective variants will now be described with reference to Figs. 1 to 12.

[0028] Turning first to the improved apparatus for the layerwise manufacture of three dimensional (3D) objects from particulate material according to the invention, Fig.l illustrates a schematic cross section side view of the apparatus 10 through a a sequence of powder layers 160. The apparatus comprises a distribution device (or “distributor”, not shown in Fig. 1 but illustrated in Fig. 2) arranged to distribute each layer 160; a digital light projector comprising a digital mirror device (DMD) 240 and a light source 250; and a controller 300. The DMD comprises an array of individually addressable micromirrors. Each micromirror is arranged to reflect a light beam 450 provided by the light source 250 in the form of non-fusing energy beams 420 onto the top most layer 160 to heat corresponding voxels of the layer. The controller 300 is arranged to control the distribution device to distribute each layer 160. The distributor may be a roller or a blade configured to spread an amount of build material over the preceding layer to form a new layer. The so-formed layer stack is supported on a platform that is lowered before each new layer is distributed by the distributor. The sequence of layers 160 illustrated in Fig. 1 and selectively heated by the DLP may be part of warming sequence in which layers without fused cross sections are being distributed and heated. The controller is further configured to control the digital mirror device 240 based on slice data (or “slices”) comprising non-fusing slice data (or “non-fusing slices”) to provide non-fusing energy beams 420 of different non-fusing energies to corresponding non-fusing voxels within the layer (by switching them between the operative and the inoperative state at a switching rate, or duty cycle as defined in the non-fusing slice data) so as to heat them to a predefined temperature below the melting temperature. The non-fusing energy beams 420 are indicated by arrows, where the differences in non-fusing energy are indicated by the relative thickness of the arrows, wherein the thicker the arrow, the higher the non-fusing energy provided by the corresponding mirror of the DMD. The controller is further configured to repeat the distributing and heating of each layer to form a sequence of cross sections forming the 3D object, wherein for at least one layer 160, the controller is configured control a fusing energy beam source to provide fusing energy beams to a plurality of object voxels of an object cross section 150 within the layer 160. The fusing energy beam source is not shown in Fig. 1. The fusing energy beam source may be any suitable source capable of providing selective fusing energy to the object cross sections. It may for example be an additional radiation source such as a laser tracing the cross section, or a non-selective infrared light source applied following definition of each object cross section with absorber, for example deposited by a printhead printing. Alternatively, the digital mirror device 240 may be operated to provide either nonfusing or fusing energy beams depending on the switch rate between the operable and inoperable state of each mirror. Thus the digital light projector may comprise the fusing energy beam source.

[0029] The mirrors and their operation define the resolution of voxels that may be selectively addressed on the layer. The DMD may be expected to heat not just the surface pixel within the layer, but due to the relatively small layer thickness also most or all of the layer underneath it; thus each energy beam redirected by a mirror is expected to heat at least the corresponding voxel of the top most layer. Herein, a “voxel” is referred to as the smallest area selectively addressable by the DMD extending through a layer, wherein the layer may be from 10pm to several 100pm thick, or even of the order of millimeters thick.

[0030] The apparatus 10 is thus provided with a light source providing selective non-fusing energy beams to control the heating exposure over some or all voxels of the layer. Local overheating and / or underheating of regions of the layer may be prevented or at least reduced. This is an improvement over conventional powder bed fusion apparatus that allows voxel level “greyscale” of heating, providing fine control over the non-fusing temperature of each layer. This enables improvements in thermal uniformity and quality of the finished object and reduces ageing of the powder by providing non-fusing energy beams only to those nonfusing voxels where it isa needed. Furthermore, compared to conventional powder bed fusion apparatus, the use of a digital light projector reduces energy consumption and cost of manufacturing.

[0031] Embodiments of the apparatus and method of operation according invention and their variants will now be described in more detail with reference to Fig. 1 to Fig. 12. The different non-fusing energy levels may be provided by one or more of the following and applied at one of more steps of processing each layer. Thermal feedback for fine tuning preheating

[0032] The DMD may be used to selectively heat the layer ahead of or behind the distributor and / or fusing energy source. This is for example illustrated in Fig. 2, which is a variant of Fig. 1, and shows the distributor 290 during its pass across layer 160n-l to distribute a new layer 160n. The DMD may be operated following behind the distributor to immediately preheat the new layer. It should be noted that while Fig. 2 and subsequent Figures do not show all features of Fig. 1, these features may equally be present and be operated in the same way as described herein.

[0033] The different non-fusing energy beams may be provided based on thermal feedback from measurements of temperature of the layer. As illustrated in Fig. 1, the apparatus may comprise a thermal sensor 280 arranged to measure the temperature of the layer 160. With reference to Fig. 2, the thermal sensor 280 may be controlled to measure temperatures of the freshly distributed layer 160n. The measured temperatures may be used to determine a measured non-fusing thermal profile of the non-fusing areas of the layer 160, which is compared to the predefined non-fusing thermal profile. The measured temperatures may be used to generate a non-fusing thermal profile of the freshly distributed layer. The controller 300 may be configured to compare the measured temperatures or the determined non-fusing thermal profile to a predefined non-fusing thermal profile for that layer, and based on the comparison, to determine one or more deviations from the predefined non-fusing thermal profile for some or all non-fusing voxels. The controller may be configured to determine the different non-fusing energies to be provided by the DLP based on the determined deviations, and apply the determined different non-fusing energy beams during the operation of the DLP so as to fine tune a level of heating towards the predefined non-fusing thermal profile of the freshly distributed layer. The different non-fusing energies may be defined to elevate the temperature of the newly distributed layer 160n towards, or to, a predefined, uniform target layer temperature, and within a predefined range of temperature variation. The predefined target layer temperature may be close to the melting temperature and sufficiently below the melting temperature, or the onset of melting, of the build material so as to avoid accidental fusion. Furthermore, the predefined target layer temperature may be above the solidification temperature. The target layer temperature may be chosen to fall within 2 °C to 20 °C below the melting temperature, which may be defined by the onset of melting. In addition, the range over which the actual temperature may vary may be defined, for example within less than + / - 5 °C or less than + / -1 °C from the target layer temperature.

[0034] Furthermore, the different non-fusing energies may be determined so as to reduce a range of temperature deviations, for example due to an underlying fused area such as area 150n-l shown in Fig. 2, from the predefined non-fusing profile. As will be described below, the predefined non-fusing profile may be based on slice data and corrected based on the determined non-fusing thermal profile.

[0035] Additionally, or instead, the measured temperatures by the thermal sensor 280 may comprise temperatures of the build area 160 following the provision of fusing energy by the fusing energy source to cross sections 150 of the object, for example of cross sections 150n-l in layer 160n-l of Fig. 2. The DMD may be operated to provide different energies of nonfusing energy beams based on the per-voxel temperature map of all or part of the layer 160n- 1 following the provision of fusing energy. Where a scanning fusing lamp is applied, such a post-fuse heating step may be initiated before the scanning fusing lamp has traversed the build area.

[0036] Different energies for the non-fusing energy beams 420 due to fused cross sections in previous or future layers may be determined based on slice data. For example, a region of a new layer may be heated in a different manner to the surrounding region so as to prepare for a future fused area it is to support, or to compensate for an underlying fused area it is covering. Fig. 3 schematically illustrates by different stages of the manufacturing process of an object how the DMD may be used advantageously to provide non-fusing energy selectively to the layer.

[0037] Fig. 3A illustrates the application of different non-fusing energy beams 420 to a fresh layer 160n. A central section of the non-fusing energy beams 420 indicate beams of higher nonfusing energy to a defined region so as to condition the region to support a subsequent fused cross section 150 shown for the subsequent layer 160n+l in Fig. 3B. The cross section may be fused in any convenient way that is not shown here. A new layer 160n+2 is distributed and shown in Fig. 3C. Non-fusing energy beams 420 are applied of different energies around and above the fused cross section 150. For example, a higher non-fusing energy may be provided to a region surrounding the overlap region with the fused area 150 compared to the non-fusing energy provided to the region overlapping with the underlying fused area 150. This may slow down the dissipating of energy from the fused area 150 to allow it to improve in mechanical properties.

[0038] The DMD thus allows providing non-fusing energy beams only where they are needed, such as to non-fusing regions surrounding to object cross sections, in the horizontal as well as the vertical (layering) direction. Heating may thus be fine tuned without the need of heating the entire layer to, for example, the conventional target layer temperature, but to a relatively lower preheat temperature instead. In variants, the remaining layer may not be heated, and the DMD is only operated to heat some of the non-fusing voxels, for example one or more of the non-fusing voxels within the peripheral region surrounding an object cross section, a defined overlap, and a cover region. This reduces or delays heat dissipation from cross sectional areas that received fusing energy and ensures that the cross sections remain in a sufficiently low viscous state to fully consolidate within each cross section and between overlapping cross sections without the need to apply excessive fusing intensity to the object cross section. Potential overheating of fused cross sections may be avoided and the mechanical properties of the object may be improved. In addition, unnecessary heating may be avoided of a certain region that is already hot, for example due to being heated by an underlying fused cross section, which prolongs the reusability of the unfused powder.

[0039] Additionally, or instead, it may be desirable to compensate for heat leakage through the build volume boundaries. Typically, the side walls and support supporting the build bed present heat leakage paths through which energy provided to each layer escapes. This can cause dimensional inaccuracies particularly near the side walls and edges, radially outward from the centre of the build bed. Provision of a DMD to selectively supply heating where needed, for example to non-fusing voxels closest to the layer boundaries compared to the centre of the layer, compensates for at least some of the heat leakage through the side walls.

[0040] Adjusted slice data

[0041] The location and level of non-fusing energy to be applied by the DMD may be predicted by simulation, for example based on initial slice data. Such slice data may be modified based on a thermal model that takes into account e.g. heat dissipation, object packing densities and object model proximities in the build volume so as to reduce or avoid thermal nonuniformities in the unfused and optionally fused build material. Localised heating may allow for lower non-fusing energies to be applied elsewhere in the build bed where objects are not being formed, thereby reducing the energy used for a build compared to conventional processes. This reduces energy costs as well as aging of the unfused powder.

[0042] A non-fusing heating step using the DMD may for example provide non-fusing energy beams of higher energy to the peripheral region of the object cross section 150 compared to regions further removed from the object cross section 150. The non-fusing energy may be varied gradually away from the periphery of the fused area 150. This is further illustrated in Fig. 4, which illustrates selective non-fusing energy beams 420 provided such that energy beams 420H closest to the periphery of the object cross section 150 are of higher energy compared to energy beams 420L further away from the periphery. Thus non-fusing energy beams 420 may reduce in energy outwards from the object cross section 150 over a peripheral region surrounding the object cross section. Fig. 4 shows a further variant in which the DMD may, optionally as indicated by the dashed arrows, further be operated to apply fusing energy beams 410 to the object cross section. In addition, the fusing energy may be varied, for example so as to increase from a centre of the object cross section outwards, as shown.

[0043] In summary, the DMD may be operated to provide different non-fusing energies to corresponding voxels as may be defined by one or more of the following:

[0044] (i) non-fusing energy beams of higher non-fusing energies are provided to voxels within an underlying area that is to support an object cross section defined in a subsequent layer, compared to voxels outside the underlying area; and

[0045] (ii) non-fusing energy beams of lower non-fusing energies are provided to voxels within a covering area that is positioned above an object cross section defined in a previous layer, compared to voxels outside the covering area; and

[0046] (iii) non-fusing energy beams of higher non-fusing energies may be provided to voxels within a peripheral area of periphery of the object cross section compared to voxels further removed from the periphery. Selective heating may be applied during or after providing fusing energy to object voxels.

[0047] The different energies of the non-fusing energy beams may be predetermined based on slice data that defines the location of each cross section within the plurality of layers. The different energy levels for the non-fusing energy beams 420 to be applied by each slice may for example be defined based on one or more of the slices of the plurality of slices, the shape of the object cross sections, the area of the object cross section, a closest distance to an adjacent cross section, and a location of the object cross section with respect to a boundary of the build area. The different non-fusing energies may be based on slice data for at least one of: one or more previous layers, the present layer, and one or more subsequent layers.

[0048] The controller may be configured to determine the different energy levels by applying a computational routine that determines the thermal impact of each cross section and of the object(s) throughout the build volume, e.g. based on a thermal model. Additionally, or instead, the different energies may be determined based on a set of predefined rules developed to condition each non-fusing area based on slice data. For example, the rules may define the extent of a periphery around each object cross section and how the non-fusing energy beams are to differ in energy. For example, the peripheral region may extend over a few 100pm or more, or over 1 mm or 2mm, or over 5mm, and the non-fusing energy may be increased gradually towards the object periphery over several degrees Celsius, e.g. over 1-10 °C. Alternatively, the non-fusing energy beams may be applied selectively only to the peripheral region to boost a preceding heating step.

[0049] Slice data is conventionally generated by a computational process that divides the one or more object models within a virtual build volume into a sequence of two dimensional slices, each slice corresponding to a layer to be processed in the apparatus. The non-fusing slice data may be generated based on the fusing slice data since it considers the regions outside the object model cross sections. The non-fusing slices may simply identify the non-fusing pixels outside the model object cross sections in each slice. A further routine may be applied to determine the different energy levels for each non-fusing pixel on a thermal model to for example determine each energy level for each non-fusing pixel to prepare for a subsequent fused cross section, to compensate for a previous fused cross section, or to condition the peripheral non-fusing area of am object cross section. The routine may thus define adjusted non-fusing slices defining the different energy levels for each slice.

[0050] Directionality, scrolling and delays between thermal event steps

[0051] The build material at the point of distribution is at a lower temperature than the top most processed layer it is to cover. The step of distributing may therefore be referred to as a ‘cold’ thermal event. The steps of heating and fusing meanwhile are ‘hot’ thermal events after each of which the layer starts to cool again. As described above, it may be advantageous to control the directionality of, and preferably also the timing between, thermal events on the build area. This improves repeatability and uniformity of the parts resulting from a build, by ensuring that each voxel experiences the same thermal history in the same or similar manner within the build area, and for each layer.

[0052] Distribution of each new layer preferably occurs before the fused, hot, object cross sections of the (previous) layer cool to become too viscous to allow adequate adhesion with the new layer. The timing of the distribution step may thus be carefully controlled with respect to the temperature of the fused object cross section. As a result, the step of distributing a new layer is initiated before the step of fusing (e.g. the traverse of a fusing energy source over the layer) is complete. For example, the distributor may immediately follow a scanning fusing energy source moving ahead of the distributor. The layer cycle, which refers to the process sequence for each layer, may include multiples of each of the heating and / or fusing steps in the same layer cycle (for example multiple heating steps as disclosed herein).

[0053] In variants of the apparatus, the micromirror array may be statically arranged above the working surface. The step of distribution of build material to form each new layer is a directional step, and it is preferable that the following thermal event steps are applied with the same sequence and with the same directionality. Optionally, some or all thermal event steps may also be applied with the same delay duration between them. The controller may be arranged to control the micromirror array to progressively scan across the build area along the direction of distribution of each layer. For example, the DMD array 240 may be mounted to a galvanometer arranged to vary the angle between the build area and the non-fusing energy source 250. Alternatively, the non-fusing energy source 250 may be moveable such that the location of incidence of each non-fusing energy beam on the build area is adjustable. In this way, the non-fusing energy beams may be varied in location to operate in a “scrolling” or “scanning” mode while providing non-fusing energy beams across the layer, so that for example the non-fusing energy beams irradiate sequential arrays of voxels synchronised with the motion of the distributor as it distributes each layer.

[0054] The controller 300 may be arranged to provide the steps of distribution and heating and, where present, the step of fusing, along the same direction and such that each voxel experiences the same delay between the steps of distribution and heating, heating and distribution, and, where present, fusing and distribution. Optionally, the respective delays between the steps of distribution and heating, heating and distribution, and, where present, fusing and distribution, are the same for each voxel of each layer and may be controlled by the controller 300.

[0055] In variants of the apparatus 10, the micromirror array 240 may be arranged to move across the build area. The controller 300 may thus arranged to control the micromirror array 240 to progressively provide non-fusing energy beams to the build area along the direction of distribution of each layer. Any of the DMDs that provide non-fusing energy beams 420 herein may be operated in a scrolling mode. For example, the DMD providing the postdistribution non-fusing energy beams 420 in Fig. 2 may scroll to follow the distributor 290 along the same direction and at the same speed. The DMD and the distributor may be arranged on the same carriage, for example.

[0056] Slice data may be used to determine locations at which an underlying object cross section contributes to the heating of a subsequent layer. Additionally, it may be used to determine how best to condition a new layer for optimal fusing of object cross sections in a subsequent layer. This is illustrated in Figs. 6A to 6D, which are schematic cross sections through a side view of the apparatus 10 of Fig. 1.

[0057] Figs. 6A to 6D illustrate a non-fusing heating step that is applied to the layer after fusing an object cross section 150 using a scanning infrared fusing lamp 260. The DMD 240 is arranged to scroll across the layer 160 following the fuse lamp 260 and providing non-fusing energy beams 420 to voxels outside the object cross section 150. The fuse lamp in this example is operating throughout its traverse emitting fusing energy beams 460, such as infrared radiation. The cross section in this example comprises infrared radiation absorber which selectively absorbs the fusing energy beams from the fusing lamp 260 such that the build material heats up sufficiently to fuse. The fusing energy beams do not cause fusing in any other regions. Alternative to the fuse lamp and radiation absorber, a scrolling DMD emitting fusing radiation may be provided that selectively provides fusing energy beams only to voxels within the object cross section.

[0058] In this example, the delay applied between the steps of fusing and post-fuse heating is a constant delay delta t for all voxels traversed by the fuse lamp 260 and the DMD 240. The DMD may for example be mounted to the same carriage as the fuse lamp 260. The delay may not be fixed in other examples. In Fig. 6A, the fuse lamp has just completed the fuse step of the cross section 150. Peripheral regions 130P surrounding the periphery of the cross section 150 are indicated to either side of the cross section 150. Peripheral regions 130P are to receive non-fusing energy beams of higher energy than regions 130N outside the peripheral region. Regions 130N outside the peripheral region receive non-fusing energy beams 420L of same energy as indicated by the same arrow thickness.

[0059] In Fig. 6B, the DMD has begun to traverse the fused cross section 150 and provides nonfusing energy beams to areas outside the cross section 150. Directly above the fused cross section, corresponding mirrors may be put in an inoperative state so that the fused cross section does not receive any non-fusing energy beams. The peripheral region 130? is provided with non-fusing energy beams 420H1 and 420H2, where non-fusing energy beam 420H2 is of higher energy than non-fusing energy beam 420H1, i.e. the non-fusing energy is increased toward the periphery of the object cross section. A relatively lower temperature may be applied to the remaining area, which increases the number of times the unfused material may be reused after recovering it from the build bed. Fig. 6C shows the DMD finishing its traverse over the fused cross section 150 and applies non-fusing energy beams 420 to the peripheral region 13 OP on the far side of the fused cross section 150 according to the same considerations as in Fig. 6B, such that the energy beams 420H2 are of higher energy than energy beams 420H1, decreasing away from the periphery of the object cross section, and both of which are of higher energy than non-fusing energy beams 420L provided to the unfused region 130N outside the peripheral region. As can be seen, some or all of the mirrors of the DMD may be in an inoperative state during its traverse over the object cross section. The fuse lamp 260 has completed its pass over the layer and in Fig. 6D the DMD is shown to continue its traverse at the same speed vl, while its mirrors are progressively put into an operative state as it finishes its traverse over the object cross section 150. It can be seen how different sections or mirrors of the DMD address the same voxels during the traverse. In this way, each voxel of unfused regions 130?, I 30N may be receive a heating boost following the step of fusion for a duration defined by the speed vl and the extent along the scrolling direction of the DMD.

[0060] Fig. 7A is a variant of Fig. 6A illustrating additional features and methods of operation for providing fusing energy beams. Fig. 7A illustrates the energy beams provided by a scrolling DMD as it begins its traverse over the layer 160, and being controlled by controller 300 to provide non-fusing energy beams 420 of different non-fusing energies. The DMD itself is not shown for simplicity. The same features and operations of control of the DMD may apply to the unfused region as described for Fig. 6A to 6D. Furthermore, peripheral regions may be defined and heated as described for Fig. 6A to 6D. For simplicity any variation in non-fusing energy beams is not shown but equally applies according to any of the examples described herein.

[0061] Next, in Fig. 7B, based on slice data, the DMD (or a fusing DMD scrolling downstream of a non-fusing DMD) provides fusing energy beams 410 to begin fusing an object cross section 150 while non-fusing energy beams 420 are provided concurrently to voxels upstream of the object cross section 150, with respect to the scrolling direction. Fig. 7C illustrates how the scrolling DMD may further apply fusing energy beams of different energies to corresponding voxels within the object cross section 150. For example, the fusing energy may be reduced towards the centre of the cross section 150 as indicated by the thickness of the fusing energy beams 410. Figs. 7D and 7E illustrate further stages of the DMD during its traverse and the different energy beams provided depending on its position over the layer. The additional application of fusing energy beams of variable energy provides an advantage over the use of a scanning infrared lamp 260 of Fig. 6A to Fig. 6D. In Fig. 6A to Fig. 6D, selective energy variation may only be achieved in combination with absorber printed at different densities per voxel, or by use of a laser. The example configuration of Fig. 7A to 7D allows fast and convenient formation of objects uncoloured by absorber, which is often black and prevents or makes it difficult to produce objects with lighter or vibrant colour appearances. The scrolling direction may preferably be the same as the direction in which the layer 106 was distributed. Furthermore, the DMD(s) may begin their traverse before the distributor has completed its traverse over the layer along the same direction (e.g. along x). The velocities at which each traverse the layer may be the same, constant velocity.

[0062] Thus the fusing energy beam source may be comprised within the digital light projector and the controller may be arranged to control the individual one or more mirrors of the micromirror array based on slice data. Optionally, the micromirror array may be controlled such that within the same layer non-fusing energy beams are provided while, or concurrent with, fusing energy being provided to voxels within the object cross section. The provision of non-fusing energy beams may be limited to regions of the layer peripheral to object cross sections that are to be formed. For example, some non-fusing areas which are not in the vicinity of an object cross section to be formed may not receive any preheating or non-fusing energy beams from the DMD, thereby prolonging the quality and recyclability of the unfused powder.

[0063] Fig. 5 is a flow diagram illustrating a method 500 of operating the apparatus of Fig. 1 or any of its variants described herein, the method comprising:

[0064] - at block 510, distributing a layer over a build area;

[0065] - at block 520B, providing non-fusing energy beams of different non-fusing energies to voxels that are not to fuse; and repeating the steps at blocks 510 and 520B to form a plurality of layers, wherein for one or more layers, the method further comprises,

[0066] - at block 550, providing fusing energy to object voxels that are to fuse within a defined cross section of an object.

[0067] The step at block 520B of providing non-fusing energy beams may be carried out while scrolling the DMD array across the build area, preferably along the same direction of distributing each layer. Optionally, the step of providing fusing energy beams at block 550 may be provided by the DMD array while scrolling along the same direction during or after providing non-fusing energy beams to the build area. The fusing energy beams may be provided at different energies to voxels within the object cross section, for example as described with reference to Fig. 7A to 7E.

[0068] The application of the DLP is however not limited to a unidirectional thermal cycle; improved results may be obtained in which thermal events are applied in an opposing or orthogonal direction to other thermal events. In multi-directional processes, which may generate larger thermal deviations from the predefined thermal profile across each layer, fine tuning of non-fusing energy and / or fusing energy may be advantageously provided by the DMD.

[0069] The step of heating at block 520B may be based on slice data. In addition, or instead, it may be based on temperature measurements of the layer made after the step of distribution at block 510. In variants of the method therefore, at block 530A, the thermal sensor may be operated to measure temperatures of the build area following the step of distributing the layer. Optionally, a thermal map may be provided that is based on a plurality of measured temperatures of the layer. From comparison of the measurements or of the thermal map to the target layer temperature, different non-fusing energies for a plurality or all of the mirrors of the DMD are determined. The DMD may then be operated at block 520B based on the determined different non-fusing energies to fine tune the temperature at voxel resolution on the build area and reduce the range over which the temperature varied from the target layer temperature.

[0070] The thermal sensor 280 may be a thermal camera of sufficiently high resolution to capture the temperature of each layer at or near voxel resolution. The thermal sensor may be arranged stationary above the layer as shown in Fig 1. Alternatively, the thermal sensor may be a line scan camera having an elongate array of sensor pixels arranged to move above the layer, in a direction perpendicular to the direction of elongation of the array, to measure the temperature of the layer surface line by line. The line scan camera may for example follow the distributor and / or the fusing radiation source. A thermal map may be reconstructed from the multiple line scans taken to represent the temperature of the entire layer. The thermal map generated may be used to control the DMD to selectively heat the layer.

[0071] As described above with respect to Fig 2, optionally a first preheat step at block 520A may be provided by a non-selective preheat lamp, for example by a scanning infrared lamp that extends along the width of the layer (along y in Fig. 1 for example) and follows the distributor. This first heating step immediately follows distribution to elevate the average temperature of the layer towards the target layer temperature. After the first preheating step, the method proceeds to blocks 530A and 520B as described above. In this way, the temperatures may be fine tuned. Due to the first heating step using a non-selective, and cheaper, first preheat source, the DMD may be operated to fine tune the temperature uniformity of the layer without having to operate the DMD at high power for each layer. This prolongs the lifetime of the DMD and reduces risk of damage due to overheating.

[0072] The method may further comprise applying a constant respective delay between the steps of distributing and heating at blocks 510 and 520B (and optionally also at block 520A), heating and distributing at blocks 520B and 510, and where present heating and fusing at blocks 520B and 550, for each layer as experienced by each voxel on the build area. Modification of slice data by applying masks

[0073] The inventors have further found that masks may be applied to slice data according to the method of the invention to generate modified slice data, based on which the improved apparatus may be controlled to provide improved control over the non-fusing thermal profile. Such masks and a method of providing modified slice data based on masks to operate the DLP will now be described with reference to Figs. 8A to 12.

[0074] Typically, in static DLP projection, the energy applied to each 2D control area, which is the area associated with a single pixel of the image or slice being projected, is controlled through a combination of projection duration (global), light source intensity (global), and grayscale value (local to pixel). To locally vary the amount of energy applied (duration * intensity * greyscale %) a combination of exposing sequential images or adjusting greyscale value may be carried out. For a scrolling DMD, each control area is instead associated with an entire row (along x) of pixels, where the time each control area is associated with an individual pixel is equal to the projected pixel length divided by the scrolling speed. In the case above where the DMD is moving along the x direction, the resolution of each control area is defined by the projected pixel width (along y) and a combination of scroll speed, image patterning frequency, and projected pixel length (along x). Typically, the images or slices being projected are synchronized with the scroll speed such that while the projection moves along the x axis, the projected geometry remains in a fixed position on the material bed for the duration the projected area overlaps with the control area. Under such conditions, when the reference frame is fixed on the control area, the control area spends an equivalent amount of time being addressed by each pixel. Thus, when generating a series of images and compensations that control the object forming process, they will largely be fixed to one of two reference frames:

[0075] (i) The reference frame fixed to the control area: Position of the features within the projected image are synchronized with the scrolling to maintain a consistent position on the layer surface. In this way, part geometries and compensations related to part geometry can be controlled.

[0076] (ii) The reference frame fixed to the projection system: Position of the image features within the image fixed to the pixel location regardless of scrolling. This allows application of e.g. calibrations such as uniformity masks, scaling, etc. which are independent of part geometry. Regarding option (i), in which the reference frame is fixed to the control area, this allows compensating for persistent deviations from the target layer temperature. Such deviations may be determined empirically, and may be due to the arrangement of the apparatus, differential heat leakage paths, and / or process directionality. Process directionality of providing one or more thermal steps in the same direction, may lead to increased energy input in that direction. Another example may be creation of thermal inhomogeneity due to heat radiation from specific parts within the build bed. This may lead to predictable and / or dynamically measurable regions within the build bed that deviate from a global average, or to regions of significant size within a plurality of successive layers that deviate consistently from the target layer temperature. In these cases, one or more masks may be defined to compensate for thermal deviations. Such a mask may be fixed with respect to the reference frame of the layer. The mask may be determined based on a thermal model that analyses the slice data to determine for example locations of large thermal masses within the build bed, or based on empirical measurements that may determine a consistent directional variation from the target layer temperature for each layer. These may be apparatus independent but result from the directionality of a process. In addition, or instead, masks may be generated dynamically throughout a build process, based on one or more temperature measurements made by the thermal sensor, or resulting thermal maps, of the build area, determined for one or more instances of the layer process. The non-fusing energy supplied by the DMD may thus be a combination of slice data modified by one or more masks, either as a further step of generating modified slice data outside the apparatus using object model packing software, and / or dynamically onboard the apparatus during the build process.

[0077] A simple example of a slice 6000 comprising a plurality of fusing pixels defining an object model cross section 6500 is shown in Fig. 8 A. The area surrounding the object model cross section 6500 is the non-fusing slice area 6300 defining “non-fusing areas” over which voxels in the layer are not to fuse. The slices are prepared typically from object models that have been placed or nested within a virtual build volume which is based on the real build volume of the apparatus. The virtual build volume is a 3D volume within which certain compensations may be applied to account for shrinkage and deformation during processing od the object. After this, the virtual build volume is divided into a sequence of slices, which are the 2D images defined to map to a corresponding layer of the build process. In this example, the non-fusing pixels define a certain non-fusing energy value, for example in terms of duty cycle for each mirror. The initial duty cycle is the same for all pixels as indicated by the solid grey shading of the non-fusing pixel area 6300. To provide a predetermined non-fusing profile to each layer using the DMD, for example to compensate for a hotter left hand side of the layer compared to the right hand side, a non-fusing mask 5000 as illustrated by example in Fig. 8B is provided and applied as follows.

[0078] The non-fusing mask comprises data to modify that of the non-fusing pixels of each slice. In Fig. 8B, the shading from light to dark, from left to right, along the x-direction indicates a linear variation of weights increasing from low to high. The weights may be factors varying from 0 to 1 as the shading changes from white to black. Therefore, a constant non-energy level defined for the non-fusing pixels of the non-fusing pixel area 6300 would be modified by the mask by multiplying each pixel energy level with a corresponding weight. The modified slice 7000 of Fig. 3C illustrates the modified non-fusing pixel area 7300 over which the modified energy level now gradually increases from left to right to the original energy level. In other words, the left hand side of the slice defines a reduced non-fusing energy to compensate for the hotter layer region. Such a compensation may be applied for non-uniformities in temperature caused by the design of the apparatus or the build process, for example, and which persists over a plurality of layers.

[0079] This modification of the original pixel data by mask weights is further illustrated in Fig. 4, in which weights a increase linearly from left to right (along x) from a = 0.4 to 1.0, for example along a mask row 5000R defining a sequence of weights for a corresponding pixel row Pn along the non-fusing pixel area 6300. The pixels Pn are mapped below the graph, separated by a gap for visual clarity. Their solid grey shading is the same for all, i.e. defining a constant initial non-fusing energy level for non-fusing. Mapped underneath are modified pixel energy levels of modified slice 7000. For example, if non-fusing pixels Pi , P2, and P3 define a duty cycle dci = dc2 = des of 50% beam energy output from the DMD, and the weights for a sequence of pixels is Q.i = 0.5, a.2 = 0.6, a.3 = 0.4, the modified duty cycles for modified non-fusing pixels would be a.i x dci = 25%; a.2 x dc2 = 30%; a.2 x dc3 = 20%.

[0080] The energy level data to control the DMD to provide different non-fusing energy levels to the layer as modified by a mask may be prepared according to a method 100 illustrated by the flow chart of Fig. 11. At a first block 110, slice data comprising non-fusing slice data is provided with is associated with the reference frame of the layer. Optionally, the slice data may be provided separately in the form of non-fusing slices and fusing slices. The nonfusing slice data defines non-fusing pixels outside of object model cross sections of one or more 3D object to be formed.

[0081] At block 130, a non-fusing mask is provided comprising modification data for each nonfusing pixel of the non-fusing slice data. The non-fusing mask may be associated with the reference frame of the layer so as to apply compensations to recurring deviations from a predefined thermal profile. The deviations may have been determined empirically over a substantially non-changing area within the layer.

[0082] At block 140, a modified slice, or a modified non-fusing slice, is generated by modifying the non-fusing data according to the non-fusing mask, by applying to each non-fusing pixel data the corresponding mask value.

[0083] The modified slice data may be generated in a single step to modify the slice data based on the fusing mask and based on the non-fusing mask. For example, the pixels of each slice may comprise an identifier, or address, that identifies them as a fusing or non-fusing pixel. To generate the modified slice, the non-fusing mask may be applied to the non-fusing energy levels defined by corresponding fusing pixels and the non-fusing mask may be applied to non-fusing energy levels defined by corresponding non-fusing pixels according to the identifier. Any calibrations relating for example to the brightness and / or duty cycle “grey level” for individual mirrors may be performed on a combined data slice. This may further save memory and lead to faster image processing operations. The modified data may be generated within the apparatus upon provision of the mask and slice data, or it may be generated during the image processing stage and provided in modified form to the apparatus. The apparatus then receives and applies the modified slice data to manufacture the object based on the modified slice data.

[0084] As described herein, different non-fusing energies may be defined within the non-fusing slice data. Different non-fusing energies may further take into account the location and shape of the object model cross sections, for example, and any of the considerations for varying the non-fusing energy level within the non-fusing area described herein. Different nonfusing energies may be defined based on the fusing-slice data, such that the initial non-fusing slice data may be adjusted to take into the account the fusing slice data. Different non-fusing energy levels may for example be based on a thermal model, or predefined rules developed to prevent heat leakage from each object cross section, as described herein. Fig. 11 shows an optional block 120 at which the non-fusing slices are adjusted to include spatially different fusing energy levels for the non-fusing pixel. The mask provided at block 130 may then be applied to the adjusted non-fusing slice data such that at block 140, modified adjusted non-fusing slice data is generated.

[0085] Dynamic corrections to slice data

[0086] The modified slice data may require dynamic corrections as may be determined from time to time during the build process 500 of Fig. 5. The apparatus 10 may comprise a thermal sensor such as thermal sensor 280 shown in Fig. 2, arranged above the layer and configured to measure temperatures of the layer. During the build process 500, the thermal sensor may be used to measure the temperature of, for example, the non-fusing area and optionally the fused object cross sections; for examples at one or both of blocks 530A and 530B. A thermal map may be generated to represent the thermal profile of the non-fusing area which may be compared to a predefined non-fusing thermal profile (or predefined non-fusing thermal map). The predefined non-fusing thermal profile may be generated from a model-based simulation generating “greyscale” thermal information of the adjusted data slices. From the comparison, deviations from the predefined map that exceed a predefined threshold may be determined, and a correction to the modified slice data may be determined that is applied to a subsequent layer. The correction may be in the form of a correction mask that is applied to the modified non-fusing slice data of a subsequent layer so as to compensate for such excessive deviations in non-fusing temperature and to converge towards the predefined nonfusing area topography for the subsequent slice. The step of processing the modified slice data at block 520B may further follow a decision flow involving dynamic corrections based on in-process thermal measurements.

[0087] A dynamic correction routine 1700 is illustrated in Fig. 7, which may be applied to the method 500 of Fig. 5. The dynamic correction routine 1700 may be applied for some or all layers. The correction routine 1700 may be applied after the step of preheating at block 520A for example, as part of block 530A, such that at a block 1710 temperatures of the layer are measured, and a thermal non-fusing profile is generated for the non-fusing area. By comparison of the generated thermal profile against a predefined non-fusing thermal profile, it may be determined whether the measured thermal profile deviates by more than a predefined threshold from the predefined thermal profile. If the threshold is not exceeded, the modified slice data is selected, or provided, at block 1720 to heat the non-fusing areas using the DLP at block 520B. If the threshold is exceeded, the modified slice data is not used, and a correction is determined at block 1730 instead, for example in the form of a correction mask. The correction is applied to the modified non-fusing slice data at block 520B using the DLP to compensate for the determined deviation, or optionally at block 520C concurrent with block 550 of fusing the object cross section. The correction may be determined for a subsequent layer that is not the immediate next layer and applied to the relevant modified slice data for that subsequent layer. The correction determination at block 1730 may only be triggered once a deviation has been determined repeatedly and with a consistent trend for a predefined number of successive layers. Before it is triggered, the modified slice data may be applied. In other words, block 1710 may comprise measuring layer temperature and determining over a predefined number of layers that the measured layer temperature repeatedly exceeds a predefined threshold in the same or similar manner, e.g. over the same region and of similar or growing magnitude in temperature, for the predefined number of layers. Upon determining that the measured layer temperature repeatedly exceeded the predefined threshold over the predefined number of layers, the routine proceeds to block 1730 and a correction for the modified slice data for a subsequent layer is determined. The correction mask may be associated with or fixed to the reference frame of the layer. For example, it may be applied to fine tune the spatial energy profile to be provided by the non-fusing energy beams in a subsequent layer.

[0088] Dynamic correction data may optionally be provided for the modified fusing slices in a manner similar to that described above, from comparing a measured fusing thermal profile of the object cross sections against a predefined fusing thermal profile. The measurement may be carried out after block 550 and the correction applied to the fusing slice data of a subsequent layer according to similar decision flows as described for the non-fusing slice data corrections.

[0089] A validation routine may be applied to the modified adjusted non-fusing slice data to determine whether a predefined minimum non-fusing energy level required to be provided to each voxel of the layer is still achieved, following application of the fusing mask, and optionally of the correction mask. If the validation routine determines that the lowest modified energy level is too low, the non-fusing mask and / or adjusted non-fusing slice data may be adjusted to raise the lowest modified energy level to the predefined minimum nonfusing energy level.

[0090] The adjusted non-fusing slice data may define different energies of non-fusing energy beams for pixels for an overlap or cover area overlapping with a preceding or subsequent object model cross section in a preceding or subsequent slice compared to the remaining non-fusing area not overlapping, and / or for pixels within an overlap or cover area. For example, the energy levels at the centre of the overlap or cover area may be higher than the energy levels near the boundary of the overlap or cover area. In addition, or instead, the energy levels of the remaining area may be lower than the energy levels of the overlap or cover area.

[0091] Once the apparatus has commenced the build process, it may be determined that the temperature profile over the non-fusing areas of the layer deviates from an expected temperature profile and that the modified slice data requires a correction.

[0092] Such different levels may be dynamically defined by a further, or different, mask to the one shown, and that is dynamically updated every layer or every number of layers. With reference to Fig. 5, from temperature measurements made by the thermal sensor at block 530A, or made at any other suitable times over the duration of the layer processing sequence, a deviation of temperature from a target temperature profile may be determined over one or more regions of the build area, for example in the form of one or more “hot spots”. A correction mask may be generated to correct the modified non-fusing slice data over the region over which the deviation was determined.

[0093] Processing the modified slice data by the apparatus during the build process 500 of Fig. 5 may follow block 140 of Fig. 9. The build process 500 of Fig. 5, comprising the steps of measuring the thermal profile of the non-fusing area at block 530A, may further include a correction routine 1700 applied before block 520B, which is illustrated by the flowchart of Fig 10. The correction routine 1700 comprises a decision flow involving dynamic corrections based on in-process thermal measurements, and may be applied for some or all layers, for example after the step of fusing of the previous layer. Following the step of measuring temperature of the layer at block 530A, the routine comprises at block 1710 determining, by comparison against a target temperature profile for the non-fusing area, whether the measured temperature profile deviates by more than a predefined threshold. If the threshold is not exceeded, the modified non-fusing slice data is selected, or provided, at block 1720 to heat the layer at block 520B. If the threshold is exceeded, the modified slice data is not used, and a correction is determined at block 1730 instead, for example in the form of a correction mask. The correction is applied to the modified non-fusing slice data to compensate for the determined deviations in at least one of the next layers at the step of heating at block 520B. The correction mask may be determined for a subsequent layer that is not the immediate next layer and applied to the relevant modified slice data for that subsequent layer. The correction determination may only be triggered once a deviation has been determined repeatedly and with a consistent trend for a predefined number of successive layers. Before it is triggered, the modified slice data may be applied. In other words, block 1710 may comprise measuring layer temperature and determining over a predefined number of layers that the measured layer temperature repeatedly exceeds a predefined threshold in the same or similar manner, e.g. over the same region and of similar or growing magnitude in temperature, for the predefined number of layers. Upon determining that the measured layer temperature repeatedly exceeded the predefined threshold over the predefined number of layers, the routine proceeds to block 1730 and a correction for the modified slice data for a subsequent layer is determined.

[0094] Masks associated with the reference frame of the DMD in scrolling mode

[0095] In conventional print and fuse processes, fusing energy is provided by non-selective heat sources. Selectivity is achieved by printing absorber over each object cross section with one or more printheads which causes enhanced heating of the cross section compared to the surrounding area. The non-selective heat source is typically an elongate infrared lamp that spans the width of the layer and that is operated to fuse the object cross sections as it traverses the layer. The shape of the intensity profile of the fusing radiation experienced by each voxel cannot be controlled finely - it is of a gaussian shape provided typically over the order of seconds, as determined by the speed and duty cycle of the lamp. Fine control in intensity is not possible along the direction over which the elongate lamp extends - all voxels extending along that direction are heated in the same way. Furthermore, during the fusion step, the surrounding non-fusing powder is also heated by the non-selective fusing lamp which may be an unwanted heating step. In laser sintering, the laser traces the shape of each object cross section and has to follow a two dimensional route. The intensity delivered to each voxel is is designed to overheat over a very short timeframe of the order of milliseconds or less. The particles typically reach full consolidation over a duration of of order of seconds. Since this process therefore needs to continue beyond the application of the laser beam for the particles to reach full consolidation, the material needs to effectively be overheated to ensure it remains long enough in a sufficiently low viscous state. The resulting high temperatures can lead to material degradation in the form of chain scission and / or oxidation within the fused cross section, leading to weaker parts. Neither application allows irradiation intensity control over the duration each voxel is heated. The inventors have developed a process by which the DLP may provide such control.

[0096] Turning now to option (ii) above, instead of using a mask associated with the reference frame of the layer to modify the slice data, the mask may be associated with the reference frame of the DMD. The weights of the mask map onto a corresponding micromirror. To illustrate this, the row Pn of Fig. 9 may instead be thought of as representing a row of n mirrors. For a given mirror, the mask data value remains the same irrespective of the slice data. When the DMD is operated in a scrolling mode to sweep the layer along the x-direction, such that it heats successive bands of voxels extending along the y-direction, the mask may be defined to provide control over the fusing energy profile provided to each voxel.

[0097] This is illustrated in more detail in Figs. 11 A to 1 ID which show a variant of the apparatus 10 of Fig. 1 comprising a scrolling DMD 240 at four different stages of its traverse over the layer 160. During its traverse at a constant speed v, the DMD addresses an object cross section 150 progressively with six mirrors of the array, aligned one after the other in the direction of the traverse. For a row of different mirrors extending along the x-direction, the mask 5000 defining the weights a inset at the top left of Fig. 11 A, each mirror along x is operated with a specified weight for that mirror. This weight does not change for that mirror as the DMD scrolls along the layer since the fusing mask is fixed to reference frame of the DMD, or micromirror array. As a result, the right-most mirror at the leading edge of the array may be operated at the highest modified duty cycle, for example at 90% (a=0.9), and the left-most mirror at the trailing edge is operated at the most reduced duty cycle, for example at 40% (a=0.4). Each Figure further shows modified fusing pixel data of the modified slice 700 corresponding to the position at which the DMD is operated above the layer. The modified slice 700 is identified by the time step at which is it applied, where the time difference between successive mirrors providing fusing energy beams to a given voxel remains constant since the speed of the DMD is constant. The scale of the x-axis of the layer 160 is not the same as that of the modified slice 7000 since modified slice 7000 shows only an extract to focus on the modified pixels. The unmodified fusing slice data 6000 may in this example have specified a constant duty cycle of 100% for each pixel. To illustrate the effect of the weights on each pixel in each slice, the variation in weight is superimposed on the modified slice, in which the y-axis of the slice is parallel to the vertical axis of the weight value. The modified energy level for each pixel of the cross section 150, which in this example is formed from four successive pixels Pl to P4, is indicated by its shading at each stage. As before, white indicates the lowest energy level of a modified pixel and black the highest energy level. The six mirrors may for example have been assigned weights 0.9, 0.8, 0.7, 0.6, 0.5 and 0.4, from right to left. The original pixel values may have been a constant 100% and are modified the least, by a= 0.9, and thus the modifed pixel values are the highest at the leading edge of the DMD. The energy level provided by the left most (sixth) mirror at the trailing edge of the DMD is the most modified, by a= 0.4 and provides the lowest modified pixel values.

[0098] In Fig. 11 A, the DMD 240 has just started at starting time tl to address the first voxel of the object cross section 150, with the first mirror. The first voxel is shown in black to indicate fusing, and the first mirror at the leading edge of the DMD provides a fusing energy beam 410 according to the first modified pixel Pl. The first voxel receives an energy beam at a duty cycle of 90% (a= 0.9) as defined by the modified fusing slice 7000(tl).

[0099] After a first time interval, at time t2 illustrated by Fig. 1 IB, the DMD 240 is addressing the first and the second voxel at the same time. The modified fusing slice 7000(t2) now defines by the first modified pixel energy Pl the energy level the second mirror provides for the first voxel, provided by the second mirror operated at 80% duty cycle (a= 0.8). The second modified pixel P2 defines the location and energy level of the second voxel to the right of the first voxel, and is fused by the first mirror at a duty cycle of 90%. The shading of the first modified pixel Pl is slightly lighter than that of the second modified pixel P2.

[0100] After a further two time intervals, at time t4 illustrated by Fig. 11C, the DMD 240 is addressing all four voxels at the same time as defined by the modified fusing slice 7000(t4). The first voxel is now heated at a 60% duty cycle by the fourth mirror, according to the first modified pixel Pl, and for which a= 0.6. The second and third voxels are addressed by the second and third mirror, for which a= 0.7 and 0.8, respectively, which are thus operated at duty cycles of 70% and 80%. The fourth voxel at right edge of the cross section 150 is heated by the first mirror, at 90% duty cycle.

[0101] After a further two time intervals, at time t6 illustrated by Fig. 11D, the DMD 240 is addressing all four voxels with a different set of mirrors, mirrors six to three from left to right, such that the first voxel is now heated by the sixth mirror modified at a= 0.4 at the trailing edge of the array, and operated at 40% duty cycle. The fourth voxel is now heated by the third mirror operated at a duty cycle of 70%. The second and third voxels are heated by respective fifth mirror at 50% and fourth mirror at 60% duty cycle.

[0102] A linear intensity profile may be provided to each voxel in this way during the traverse of the DMD over it. Assuming the DMD travels at a constant speed v, the first voxel is heated at a modified duty cycle of 90% at a starting time tl (a=0.9), at a modified duty cycle of 80% at t2 (a=0.8), at a modified duty cycle of 60% at t4 (a=0.6), and at a modified duty cycle of 40% at t6 (a=0.4). This is a simple illustration of how a linear intensity profile may be achieved applying the method according to the invention.

[0103] It should be noted that while the pixel numbering has been kept the same for clarity, the position of each pixel within the modified slice data is different to reflect the changing position of the DMD with respect to the layer. In other words, the slice and modified slice is fixed to the reference frame of the layer so that object cross section remains in the same position for each position of the DMD. Thus, the location of the first, second, third and fourth pixel changes in each slice when the DMD is provided in a scrolling mode. Instead of moving the DMD over the layer, the angle of a fixedly arranged DMD and / or the angle of light source 250 may be varied to achieve a similar scrolling effect.

[0104] The mask may be defined to achieve any desirable intensity profile for each non-fusing or fusing voxel over the duration of heating when applied in combination with a scrolling DMD. This provides a simple process that allows full control over the intensity profile for each voxel of the layer. Improved mechanical properties may be provided that are not achievable with conventional powder bed fusion apparatus. Different masks may be defined for different layers such that different energy levels can be individually assigned in any direction of the layer, per non-fusing area and / or per object cross section, either through varying the gradient between different y axis positions or by associating a specific mask to certain groups of the plurality of slices during the slicing operation to ensure that the variation of energy applied throughout the DMD pass is optimal for each individual object, or non-fusing area.

[0105] Control data for operating a DMD array comprised within an apparatus for the layerwise manufacture of 3D objects from build material may be provided as follows: Providing slice data defining one or more model object cross sections in one or more slices, each slice corresponding to a layer to be processed in the apparatus, and each model object cross section corresponding to an object cross section of a 3D object to be formed; determining from the slice data at least one of: non-fusing energy data for each slice outside object model cross sections, wherein the non-fusing energy data defines different non-fusing energies within one or more slices to control the temperature profile in corresponding layers; and different fusing data for one or more model object cross sections so as to control the temperature profile over the object cross section to be formed; optionally providing a mask to modify the non-fusing and / or fusing data of one or more slices; and generating modified slice data based on the determined non-fusing data and / or fusing data, optionally modified by the mask. The DMD may be controlled to provide non-fusing energy beams and / or fusing energy beams as described herein based on the modified slice data. Non-fusing masks and fusing masks may be applied in combination and applied depending on whether the pixel of the slice data is a non-fusing pixel or a fusing pixel. In addition, or instead, multiple masks may be defined that are be associated with different reference frames. For example, a first non-fusing mask may be defined fixed to the reference frame of the layer, and a second nonfusing mask may be defined fixed to the reference frame of the micromirror array of the DMD. The first mask may be applied to modify the non-fusing slice data and the second mask may be applied to modify the energy output of each mirror.

[0106] General considerations

[0107] The melting temperature may be defined by the onset of melting. The predefined temperature may further be above the solidification temperature. The light source 250 may be an infrared light source, such as an infrared laser, for example. The DMD 240 may comprise a single array arranged to simultaneously address each voxel of the entire layer surface from a static position. Alternatively, the DMD 240 may comprise an array of mirrors fewer in number than the voxels on the layer surface, and may be arranged on a scanning (or “scrolling”) gantry above the layer to successively address rows of voxels along the y- direction as the DMD is moved (scrolled or scanned) above the layer surface, as is illustrated in Figs. 6A to 8C. Alternatively, the DMD may be mounted to a galvanometer for example, and the controller may be configured to vary the angle of the array with respect to the layer surface. The digital light projector may comprise a group of DMD arrays, each receiving a portion of the slice, in the form of sub-slices, that are synchronised to project a reconstructed slice from the sub-slices. This mode of operation is also known as tiling. Typically the subslices partially overlap their nearest neighbours to avoid visible or measurable joints in the final object. Adjacent arrays may jointly irradiate the overlap section in any conventional manner known from “printhead stitching”, where neighbouring sections are addressed in any suitable pattern by both subarrays so as to obscure otherwise visible or measurable overlap joints.

[0108] Optionally, a non-selective non-fusing energy source may be provided in addition, to follow the distributor to preheat each layer immediately following distribution. The non-selective non-fusing energy source may be an elongate infrared lamp extending along the length of the distribution device, and / or the width of the layer (the width being along the y-axis). Such a non-selective non-fusing source may be used to provide a first non-fusing energy to a plurality of voxels, for example in the variant of Fig. 2. The temperature of the distributed layer 160n may be measured following the provision of this first non-fusing energy by the non-selective non-fusing energy source, and the deviation from the intended target layer temperature determined. Next, the DMD may be used to fine tune the temperature at voxel resolution to be within the intended range from the target layer temperature, based on the determined differences as described herein. The use of a two step (pre)-heating process allows the DMD to be operated at lower power which may reduce the risk of overheating.

[0109] Fig. 1 shows an apparatus 10 that may optionally comprise an overhead heater 270 arranged above the build area and configured to maintain the layer at or close to a predefined target layer temperature. Such an overheat heater typically has an array of individually controllable heating elements, each element addressing a plurality (or “zone”) of voxels over a relatively large region on the build bed. The heating elements may be ceramic heater tiles and not be able to respond rapidly, thus the DMD enables a fast-acting correction at high resolution while the overhead heater is continuously operated to maintain the layer at or close to (just below) the target layer temperature. As an example, the overhead heater 270 may comprise one or more, for example four or more heater elements, for example at least one above each comer of a rectangular build area, or one above and along each side of the build area. The controller 300 may be configured to control the heater elements of the overhead heater based on a temperature distribution of the layer as measured by the thermal sensor 280. For example, with reference to Fig. 5, the thermal sensor may measure temperatures of the layer at block 530B after the step of distributing and heating the layer at voxel resolution using the DMD. The overhead heater 290 may be controlled continuously to maintain the layer temperature at the average layer target temperature throughout the layer cycle. As described before, the micromirror array may provide non-fusing energy beams to the distributed layer based on temperature measurements made at block 530A measured after the step of distributing each new layer, such that the different energies of the non-fusing energy beams are determined from the measurements made at block 530A. Any and any combination of variants as described herein may further apply. In any of the described variants, the DMD may be applied to preheat or heat the layer during or after fusing and before a new layer is distributed. This is indicated at block 520C and may be applied as described with respect to Figs. 6A to 6D or concurrently with block 550 as described with respect to Figs. 7A to 7D.

[0110] In variants, the digital light projector may comprise a first micromirror array arranged to heat the build area at block 520B, behind the distribution device with respect to the direction of distribution, and a second DMD array arranged to fuse at block 550, and / or to heat at block 520C, the build area ahead of the next step of distribution at block 510.

[0111] The controller 300 may be a computer or microprocessor provided with a program that, when executed, causes the steps of the methods disclosed herein. A computer program may be provided comprising instructions which, when the program is executed by the controller 300, e.g. a computer, cause the controller or computer to carry out the method and its variants as described herein. The controller may be configured to control any one or any combination of components to provide the method steps for controlling the apparatus 10 as described herein.

Claims

CLAIMS1. An apparatus for the layerwise manufacture of 3D objects from particulate build material, the apparatus comprising: a distribution device arranged to distribute each layer; a digital light projector comprising a digital mirror device and a light source, wherein the digital mirror device comprises an array of individually addressable micromirrors, wherein each micromirror is arranged to redirect an energy beam incident on the array from the light source towards the layer to heat corresponding voxels of the layer; and a controller; wherein the controller is arranged to: control the distribution device to distribute a layer; control each micromirror to selectively provide non-fusing energy beams of different non-fusing energies to the layer to heat corresponding voxels to a predefined temperature below the melting temperature; repeat the steps of distributing and heating to form a plurality of layers; and for at least one of the plurality of layers, to control a fusing energy beam source to provide fusing energy to voxels within the object cross section of said at least one layer so as form a 3D object.

2. The apparatus of claim 1, wherein the controller is configured to provide the different non-fusing energies such that at least one of:(a) non-fusing energy beams of higher non-fusing energies are provided to voxels within a peripheral region surrounding the periphery of the object cross section compared to voxels outside the peripheral region and the object cross section, during or after providing fusing energy to the object cross section;(b) non-fusing energy beams of higher non-fusing energies are provided to voxels within an overlap region that is to overlap an object cross section of a subsequent layer, compared to non-fusing energy beams of lower non-fusing energies provided to surrounding the overlap region; and(c) non-fusing energy beams of lower non-fusing energies are provided to voxels within a covering region that overlaps an object cross section in a previous layer, compared to non-fusing energy beams of higher non-fusing energies provided to voxels surrounding the covering region.

3. The apparatus of claim 2, wherein option (a) comprises providing non-fusing energy beams of higher non-fusing energies to voxels within the peripheral area while providing fusing energy to the object voxels; optionally wherein the digital light projector comprises the fusing energy beam source.

4. The apparatus of any preceding claim, wherein the controller is configured to control individual one or more mirrors of the micromirror array to provide different non-fusing energies to voxels of the layer based on a predefined non-fusing thermal profile to be achieved across the layer within a predefined range of temperature variation.

5. The apparatus of claim 4, wherein the apparatus comprises a thermal sensor arranged to measure temperatures of the layer, wherein the controller is configured to receive a measured non-fusing thermal profile of the layer based on the measured temperatures, and to control each micromirror based on the measured non-fusing thermal profile to provide the different non-fusing energies.

6. The apparatus of claim 5 when dependent on claim 2, and comprising step (c), wherein the energies of the non-fusing energy beams to be provided by each micromirror are defined based on the measured non-fusing thermal profile of the layer determined after distributing the layer.

7. The apparatus of claim 5 or claim 6, further comprising a non-selective energy source arranged to provide non-fusing energy to each layer immediately following distribution, and wherein the measured temperature comprises the temperature of the layer following the provision of the non-fusing energy by the non-selective energy source.

8. The apparatus of any one of claims 4 to 7 when dependent on claim 2, and comprising step (a), wherein the measured temperature comprises the temperature of the layer following the provision of fusing energy to at least part of the layer, and wherein the different energies are defined based on the per-voxel temperature map of the layer following the provision of fusing energy.

9. The apparatus of any preceding claim, wherein the controller is configured to control each mirror based on slice data.

10. The apparatus of claim 9 when dependent on claim 2, and comprising step (b) and / or (c), wherein the different non-fusing energy beams are defined based on one or more of the slice data, the shape of the object cross section, the area of the object cross section, a closest distance to an adjacent cross section, and a location of the object cross section with respect to a boundary of the build area.

11. The apparatus of any preceding claim, wherein the controller is arranged to control respective delays experiences by each voxel of the layer between: distribution and heating; heating and distribution; and, where present, fusing and distribution; such that the delays are the same for each voxel.

12. The apparatus of any preceding claim, wherein the controller is arranged to provide the distribution and heating and, where present, fusing, along the same direction.

13. The apparatus of claim 11 or claim 12, wherein the digital mirror device is statically arranged above the layer and wherein the controller is arranged to control an angle of the array with respect to the layer and / or an angle of the the light source with respect to the array to progressively address all voxels of the layer along the direction of distribution of each layer.

14. The apparatus of claim 11 or claim 12, wherein the digital mirror device is arranged to move across the layer and wherein the controller is arranged to control the digital mirror device to progressively provide non-fusing energy beams to the layer along the direction of distribution of each layer in synchronisation with the motion of the distribution device.

15. The apparatus of any preceding claim, wherein the digital light projector comprises the fusing energy beam source and wherein the controller is arranged to control the individual one or more mirrors to provide fusing energy beams to the object cross section based on slice data.

16. The apparatus of claim 15 when dependent on claim 2, wherein the controller is arranged to control each micromirror such that step (a), and optionally step (b) and / or step (c), are provided while providing fusing energy to the object cross section within the same layer.

17. A method of operating an apparatus for the layerwise manufacture of 3D objects from particulate build material according to any one of claims 1 to 16, the method comprising the steps of: distributing a layer; selectively providing non-fusing energy beams of different non-fusing energies to voxels that are not to fuse based on slice data; repeating the steps of distributing and heating; and for at least one layer, providing fusing energy to object voxels within an object cross section that are to fuse so as to form a 3D object.

18. The method of claim 17, wherein the step of providing non-fusing energy beams is carried out while scrolling non-fusing energy beams provided by the digital mirror device across the layer along the same direction of distributing each layer; optionally wherein the step of providing fusing energy beams is provided by the digital light projector, while scrolling fusing energy beams over the object cross section along the same direction during or after providing non-fusing energy beams to the layer.

19. The method of claim 18, comprising providing fusing energy beams of different energies to voxels that are to fuse.

20. The method of any one of claims 17 to 19, comprising a step of measuring temperatures of the layer following the step of distributing the layer; providing a non-fusing thermal profile based on the measured temperatures; and determining a correction to the different non-fusing energies defined by the slice data based on a comparison of the measured nonfusing thermal profile to a predefined non-fusing thermal profile.

21. The method of claim 20, comprising non-selectively preheating each layer following the step of distributing the layer and before the step of providing different non-fusing energy beams; wherein the step of measuring temperatures of the layer is provided after the step of non-selectively preheating.

22. The method of any one of claims 17 to 21, comprising determining one or more nonfusing masks arranged to modify the slice data defining the different non-fusing energy beams; and applying the non-fusing mask to the slice data to modify the different non-fusingenergy levels to generate modified different non-fusing energy levels; wherein the nonfusing mask is associated with the reference frame of the layer.

23. The method of any one of claims 17 to 22, comprising determining the different nonfusing energies based on slice data for at least one of one or more previous layers, the present layer, and one or more subsequent layers and generating adjusted non-fusing slice data.

24. The method of any one of claims 17 to 23, comprising applying a constant respective delay between the steps of: distributing and heating; heating and distributing; and where present heating and fusing; for each layer as experienced by each voxel of the layer.

25. The method of any one of claims 17 to 24, wherein the different non-fusing energies to be applied to corresponding voxels of the layer are provided such that at least one of:(a) non-fusing energy beams of higher non-fusing energies are provided to voxels within a peripheral region surrounding the periphery of the object cross section compared to voxels outside the peripheral region and the object cross section, during or after providing fusing energy to the object cross section;(b) non-fusing energy beams of higher non-fusing energies are provided to voxels within an overlap region that is to overlap an object cross section of a subsequent layer, compared to non-fusing energy beams of lower non-fusing energies provided to surrounding the overlap region; and(c) non-fusing energy beams of lower non-fusing energies are provided to voxels within a covering region that overlaps an object cross section in a previous layer, compared to non-fusing energy beams of higher non-fusing energies provided to voxels surrounding the covering region.

Images