DE9218423U1 - Device for producing a three-dimensional object by stereolithography - Google Patents

Description

EP 92912072.3 17 December 1993

3D Systems Inc. ; ' D~17339EP F/Sh/La/dw

Device for producing
of a three-dimensional object by stereolithography

The invention relates to the field of stereolithography, which is a technique for fabricating solid, three-dimensional objects (or "parts") from solidifiable materials (e.g., fluid or fluid-like materials such as photopolymers, sinterable powders, and bindable powders) on a layer-by-layer basis. More particularly, this invention relates to an apparatus for successively exposing layers of stereolithographic buildup materials to form the layers of stereolithographic parts with less distortion.

In recent years, stereolithography systems such as those described in U.S. Patent No. 4,575,330 have come into use. Basically, stereolithography is a technique for automatically building complex three-dimensional parts by successively and selectively solidifying and adhering a plurality of thin cross-sectional layers, each thin cross-sectional layer formed representing a corresponding cross-sectional area of the three-dimensional object. These layers are composed of a material that can be selectively transformed from a fluid-like state to a solid, or cohesive, state upon exposure to a synergistic stimulus. The layers are typically sequentially solidified on top of one another until all of the thin layers are bonded together to form an entire part.

The material is typically composed of liquid polymer resins, e.g. photopolymers or thermosetting polymers, powdered materials or similar. Typical powders include waxes, thermoplastics, metals, ceramics or similar. The solidification

s mechanism may be sintering or melting followed by bonding of particles after cooling and resolidification. The sintering or melting may be induced by synergistic stimulation in the form of heat. For example, the heat may be IR radiation generated by a laser or it may be induced by

&iacgr;&ogr; Ultrasonic vibration may be induced by friction.

Another consolidation mechanism uses an adhesive bonding agent to form thin layers of a solid. The bonding agent is selectively dispensed, for example by an ink jet or the like, into a thin layer of powder, thereby bonding the powder particles together. The particle diameter is preferably less than the desired formation resolution, more preferably less than half the desired resolution, and most preferably less than one-tenth the desired resolution. Many photopolymers have a photospeed (the rate of transformation from liquid to solid) upon exposure to ultraviolet (UV) light that is fast enough to make them practical model making materials, such as acrylic or urethane based materials, for example XB 5143 manufactured by Ciba Geigy, Switzerland. In a preferred system, a radiation source (e.g., a HeCd or argon ion ultraviolet laser) produces a beam focused to a small intense spot that is moved across the liquid photopolymer surface by servo-type galvanometer or mirror x-y scanners. The scanners are driven by computer-generated vectors. In another embodiment, masks and flood exposure are used.

to cause selective solidification of the material. The material that is not polymerized when a part is made is reusable and remains in the vat for use when subsequent parts are made.

Stereolithography provides an effective way to quickly produce complex or simple parts without tooling. Because this technology depends on the use of a computer to generate its cross-sectional patterns, there is a natural data link to computer-aided design and manufacturing (CAD/CAM). However, such systems have presented challenges related to structural stress, shrinkage, curl and other distortions, as well as resolution, speed, accuracy and difficulty in producing specific object shapes.

Additional stereolithographic information is provided in the following patent publications and applications, all of which are incorporated herein by reference: U.S. Patent Nos. 4,575,330; 4,929,402; 4,996,010; 4,999,143; 5,015,424; 5,058,988; 5,059,021; 5,059,359; 5,076,974 and 5,104,592; PCT Publications Nos. WO89/01801; WO89/10256; WO89/11085; WO89/10255; WO89/10254; PCT/US89/01557; WO90/03255; WO90/15674; WO91/04798; WO91/06378 and WO92/08200.

One aspect of this invention relates to exposing the build material to synergistic stimulation to cause solidification of the build material, whereby exposure of solid lines, e.g. vectors, of material in a regular or irregular pattern is interrupted so that the formation of long lines of material is avoided. This reduces long line pulling effects, resulting in

excessive distortion of the three-dimensional object being formed. Preferably, the lines are formed with a cure depth that does not cause adhesion with the previous layer, at least on initial exposure.

A second aspect of the invention relates to closing gaps between panels by exposing at least some of the areas above the gaps in a manner that allows the material above the gap to shrink at least partially before causing the solidifying material to adhere to the panels on both sides of the gap. The solidifying material can be adhered to the panel on one side of the gap before being allowed to shrink, followed by causing adhesion to the panel on the other side only after a substantial amount of shrinkage has occurred.

A third aspect of the present invention relates to reducing distortion of stereolithographically manufactured parts by alternating the curing sequence from layer to layer to evenly distribute any imbalances in curing that may otherwise result. An additional embodiment of the third aspect of this invention is tailoring the drawing order and vector types for a given part geometry to mitigate curl along a given axis. An additional embodiment of the third aspect of this invention relates to forming at least a portion of a three-dimensional object using the "star weave" build-up method using a weave build-up piece with the addition of using at least two non-parallel hatch types on each layer, offsetting the hatch vectors from layer to layer, retracting a

End of each hatching vector by a border and changing the order of the drawing.

A fourth aspect of the invention relates to the use of modified vector scanning techniques which allow most, if not all, of the significant decelerations and accelerations of the scanning beam to occur when the beam is apertured so that constant exposures can be achieved along the entire length of the vectors and thus uniform cure depths can be achieved.

A fifth aspect of the invention relates to combinations of the above four aspects to further improve part manufacturability. For example, the modified vector scanning techniques of the fourth aspect of the invention can be used to improve the results obtainable from the other three aspects. In particular, limiting significant decelerations and accelerations to non-exposure times results in more uniform cure depths that can be used to ensure that the floating vectors of the retracted starweave buildup technique do not adhere unintentionally or prematurely to the previous layer. This allows them to shrink freely without causing puckering in a previous layer.

The invention includes devices for the above aspects.

Fig. la and lb are plan views of vectors sampled, respectively, according to a first pass of a square plate-shaped exposure pattern of a staggered weave and according to both passes of a

·

square panel-shaped exposure pattern of a staggered weave.

Fig. 2a, 2b, 2c are plan views of vectors each sampled according to a first pass of a

hexagonal-plate exposure pattern of a staggered weave and according to both passes of a hexagonal-plate exposure pattern of a staggered weave and according to both passes &iacgr;&ogr; of an alternative hexagonal-plate exposure

pattern of a staggered weave.

Fig. 3 shows the currently used conventional scanning

sequence for fill and hatch vectors.

Figures 4a to 4d respectively illustrate each of the scanning patterns of a four-layer sequence of cross sections using the character order of Example #2 for alternating sequencing.

Figures 5a to 5h respectively illustrate each of the scanning patterns for an eight-layer sequence of cross sections using the character order of Example #3 of alternating sequencing.

Fig. 6a,6b,6c show the gaps between panels on a first layer, which are filled with slightly offset panels on a second layer, followed by grouting between the offset panels.

Fig. 7 shows the closing of gap intermediate plates on

a first layer with an at least partially floating member on a second layer, wherein the floating part is riveted to an adjacent plate on the first layer

Figures 8a to 8g illustrate different orientations of vectors η and n+1 together with different virtual sampling phases that can be followed to sample from the vector η to the vector n+1.

The first aspect of the present invention relates to methods and apparatus for improving part accuracy by providing improved stereolithographic vector scanning procedures known as "interrupted scanning".

"Intermittent scanning" refers to scanning at least a portion of a layer to solidify material to form a portion of a three-dimensional object by exposing at least one vector with one or more gaps along its length to relax transmitted stress. The portions of the vector intended to cause solidification of material may be formed sequentially along the length of the vector, interleaved between exposed portions of other vectors, or by non-sequentially forming all portions of one vector before forming any portions of a second vector.

Long vectors can cause enormous amounts of frizz when they are cured and adhered to a previous layer while still shrinking. This stress-inducing adhesion

at the previous layer can occur along the length of the vector. In the case of a floating vector, if a retracted vector design is not used, it can occur where the ends of the vector join at a boundary.

In the "intermittent scanning" technique, vectors are drawn with periodic interruptions along their length. The interrupted vectors may be formed in a single pass design, wherein adhesion to the previously formed layer of cured material occurs after exposure to the first pass. Alternatively, and more preferably, a multiple pass design is used, wherein adhesion to the previous layer does not occur after the first pass, but instead occurs on a later pass. In the multiple pass design, the multiple exposures may occur by re-sampling the same vector multiple times, or alternatively by scanning different paths that overlap at one or more points, resulting in multiple exposures at those points. As in slab laying, the exposed length is considerably longer than the gap between interruptions. The length of solids between interruptions is preferably about 2.5 mm to 120 mm, with the length of the gap being 0.025-0.250 mm . The breaks in the vectors may be in a random pattern over the length of a vector, or the breaks in adjacent vectors may form a random pattern. Alternatively, and currently preferred, are methods that use breaks formed in a regular or at least predictable manner.
The individually broken vectors may either be cured to a depth that causes adhesion, or they may be cured to a lesser depth and thereby remain floating, or one end may be adhered to the previous layer,

with most of the vector floating. The last two options are most preferred.

A first "interrupted scanning" method superimposes the interrupted scanning over a web exposure pattern. The web build or exposure technique is described in PCT publication WO91/06378, published May 16, 1991. The web build technique exposes cross sections with at least two types of non-parallel hatch vectors, where effective adhesion (capable of transmitting significant curl) occurs only at the overlapping points between the vectors of two hatch types. In addition, the hatch vectors of each type are spaced as closely together as possible without being spaced so closely that they can induce curl in neighboring vectors, or that curl is induced in them by neighboring vectors.

A first embodiment of woven-interrupted scanning is to scan a first pass over a grid of plates (the floating pass) followed by a second pass over a second grid of plates to complete the weaving. The second grid of plates is displaced from the first grid and is exposed after a sufficient delay to allow the fractured vectors of the first pass to shrink before adhering to each other or to the previous layer. This is particularly advantageous when hexagonal plate patterns are used. To eliminate the possibility of preferred axes, the hexagonal grid can be rotated between passes, as well as offset, or alternatively the hexagonal grid can be rotated or translated from layer to layer.

• · * •

·

- 10 -

Fig. la represents a top view of a cured material after a first pass over a square plate grid. Fig. lb represents the same top view after a second pass over an offset square plate grid. As can be seen from Fig. lb, this offset woven plate grid potentially still contains two weak axes.

Fig. 2a, 2b and 2c are similar to Fig. la and 1b, but they use a hexagonal plate lattice. As can be seen from Fig. 2b

“, the hexagonal staggered woven plate pattern has no long straight weak axes. Fig. 2c illustrates an alternative staggering technique where the hexagonal plate lattice is both rotated and staggered. Since the length and width of the hexagons are not equal, the process illustrated in Fig. 2c does not form a simple recursive pattern.

There are several ways to implement the discontinuous scanning aspect of the present invention. One implementation involves controlling the scanning system as if no discontinuity is anticipated in conjunction with the placement of a physical barrier, e.g. a mask, between the scanning mirrors and the surface of the build material. For example, this physical barrier may take the form of a grid pattern that selectively blocks the beam, thereby causing the formation of discontinuities in the cured material. This barrier may be selectively moved in and out of place as needed.

A second implementation involves providing control information to the scanning mirror and shutter system to cause selective exposure of the medium. The control information may be from a

program that generates vectors with interrupts, or that inserts interrupts into vectors generated and supplied by a second program. Alternatively, vector information can be supplied without interrupts and software or hardware additions to the mirror control system can insert interrupts into the vectors simultaneously with the sampling of the vectors. This last alternative is the preferred approach because it dramatically reduces the number of vectors that must be stored and processed. Another implementation would use a pulsed

&iacgr;&ogr; Use lasers to expose the material, where the laser duty cycle (pulse pattern) would be modulated by control signals to cause the formation of discontinuities.

The second aspect of the present invention relates to methods

is and an apparatus for improving part accuracy by providing improved stereolithographic "plate laying" procedures. The formation of stereolithographic parts wherein at least portions of one or more layers are exposed in a plate pattern is disclosed in PCT Publication No. WO91/067378, published May 16, 1991.

"Platter laying" is an interrupted exposure or scanning technique that applies to relatively wide areas as opposed to narrow areas, for example lines or vectors. The wide areas can be built up of vectors. The "platter laying" exposure or scanning results in different shapes that fit together very well, but are not stuck together. The intention of this technique is to maximize the percentage of cured material in the build process while keeping curl at a reasonable level by reducing transmitted stress that causes curl.

- 12 -

A major disadvantage of slab laying is that it may provide parts with lower green strength. This disadvantage can be partially overcome by curing the material between the slabs, i.e. cementing it, after the slabs have been formed. Green strength can be increased by this technique, but at least some curl distortion is reintroduced.

After forming one or more layers with slab laying and without cementing, it is generally desirable to move the

“Starting panels from layer to layer, or starting cementing between panels, or stopping the panel laying process, all to ensure proper structural integrity of the part. However, if panel laying is to be interrupted, cemented, or staggered, it is important to minimize any tendency toward reintroduction of curl that may result from closing the gaps. When the gaps are closed, any shrinkage of material that occurs across the gap while the shrinking material adheres to both ends of the gap can cause curl distortion by tending to bring the tops of the gaps closer together (closer than the distance between the bottom edges of the gap). Since gaps can result in relatively weak axes, reintroduction of curl is very likely. In addition, it has recently been experimentally verified that shrinkage of curing material can still occur several seconds after exposure to an area has been stopped. This means that closing a gap with a line of hardening material, adhered to the first side, and extending from the first side to the second side, and adhered to the second side, can, within a few seconds, induce stresses in the part that can lead to distortion.

A preferred technique for closing gaps, and thereby ensuring adequate structural integrity of a part while avoiding the reintroduction of curl, is based on ensuring that at least a substantial portion of the material being cured over the gap is allowed to shrink on both sides of the gap prior to adhesion of the cured material over the gap.

A first embodiment of this technique slightly offsets the plates

β on a second layer from the corresponding positions of plates on the previous (first) layer. With the offset, the plates on the second layer substantially cover the gaps between the plates on the first layer without completely bridging the gaps. This avoids the simultaneous adhesion of a single cured shrinking mass to both sides of the gap on the first layer. After allowing sufficient time for the plates and material spanning the gap to complete their shrinkage, cement is formed between the plates on the second layer. This cement completes the bridging of the gap with only minimal shrinkage of material across the gap while adhesion exists on both sides. Because this cement is offset from the gaps on the first layer, and because shrinkage of material on the second layer across the gaps occurs before the cement is formed, significantly less curl is introduced. Any desired exposure may be used to form the cement without necessarily being limited to cementing at a lower exposure than the exposure used to cure the panels.

This first embodiment is shown in Figures 6a-6c. Figure 6a shows a side view of a portion of an article having a first layer formed with plates 800 and a second layer with slightly offset plates 804 and offset cement. The plates 804 of the second layer are sufficiently offset from the plates 800 on the first layer to substantially cover the gap area on the first layer without being offset enough to adhere to the adjacent plates on the first layer. The cement 808 between the plates 804 on the second layer is solidified after the plates 804 on the second layer have been allowed to shrink (for example, generally at least a 3 to 5 second delay between completing adjacent plates and beginning to cement, but the delay can vary with the material). An acceptable delay time can be determined for a given material by constructing one or more test parts consisting of floating lines of material secured at only one end and observing the time required for the lines to complete a substantial portion of their shrinkage. Figure 6b shows a top view of the first layer panels 800. Figure 6c shows a top view with first layer panels 800 and second layer panels 804 superimposed and second layer cement 808.

A second embodiment of this technique forms the second layer, on which the gaps are closed by floating at least one end of the consolidated material spanning the gap, until after at least a substantial portion of shrinkage has occurred. After shrinkage is allowed to occur, the floating end(s) can be tacked down with rivets,...or by multiple passes or the like.

Rivet and multi-pass are disclosed in US Patent 5,104,592 and PCT Patent Publication No. WO89/01801.

Figure 7 shows a side view of a gap 810 between panels 800 on a first layer that are close to a second layer by floating at least one end of bridging material 814 until the bridging material has finished shrinking. Rivet 818 completes the closure point.

&iacgr;&ogr; Additional embodiments of this technique involve progressively partially closing the gap across a plurality of layers. For example, a gap may be partially narrowed from one or both sides on a second layer, followed by additional narrowing or complete closing on a third or

is higher class.

As will be apparent to those skilled in the art, these embodiments can be modified and/or combined with each other or other frizz reduction techniques to effectively close gaps without reintroducing significant frizz.

As with the interrupted scanning aspect of this invention, the implementation of the plate shutter aspect can be implemented by software or hardware; by controlling the scanning system and/or shutter; or by automatically or manually placing an exposure interrupting grid between the scanning mirrors and the surface of the build material and moving or changing the grid when it is desired to close the gaps formed on a previous layer.

The third aspect of this invention for improving stereolithographically manufactured parts is known as "alternating sequencing" or "reciprocal sequencing". This technique is applicable to the various skintinuous build methods, the hatched build styles, and the original fully solid part build methods. These various build methods are disclosed in PCT Publication No. WO91/06378, published May 16, 1991. In conventional stereolithography, the vector string is essentially identical from

γ layer to layer. This is especially true when considering fill and hatch vectors. The fill vectors on each layer are processed according to the same rules and output in the same order on each layer. For example, this conventional approach may sample a first set of X-hatch vectors, starting with the X-hatch vector closest to the X-axis (and origin), followed by sampling successive X-hatch vectors that are farther from the axis. After all the X-hatch vectors have been processed, the Y-hatch vectors may then be sampled, starting with the Y-hatch vector closest to the Y-axis (and origin), and followed by the vectors that are successively farther away. This conventional approach results in the same hardening and shrinking forces reinforcing each other as successive layers are drawn. This is especially true for identical or nearly identical layers. Therefore, if there are unbalanced forces associated with a particular cure order, even if these forces are small, layer-to-layer reinforcement may eventually cause enough buildup to induce significant levels of stress and distortion in a part. This distortion may occur in a partially completed

part as it is being built up, or it may be seen shortly after post-curing, or after the part is allowed to settle after post-curing.

This problem is well known in the art. For example, some symmetrical parts tend to curl in the same area for a previously unexplained reason. An example of this is a part called a SLAB-6. This is a 6-inch by 6-inch by 1/4-inch high square diagnostic part. When this part is repeatedly built using conventional sequencing rules, the same corner of the part tends to pull away from the substrate. The primary difference between the corners of the part is in the order in which they are formed.

is One solution to this problem is alternate sequencing. Alternate sequencing is a method of sampling where the sampling pattern is deliberately changed from layer to layer, resulting in more uniform spread of stresses that may build up during the curing process. The pattern of alternate sequencing can be random or predictable. It can be periodic or non-periodic.

The currently preferred stereolithographic machine, specifically the SLA250 manufactured by 3D Systems, Incorporated of Valencia, California, USA, uses a conventional sequence pattern. Currently, the conventional sequence pattern samples all hatch and fill vectors on all wafer layers in the same sequence. For purposes of description, we will assume that the machine is oriented so that the front of the tub is considered south and the back of the tub is considered north, and that when facing the front of the tub (facing north), the right side of the tub is east and the left side is west. Specifically

in current SLAs, Y vectors are drawn first (they run according to :": ·": ··' ·**.: .··. ····

North and South), starting at the east side of the tub and proceeding towards the west side of the tub. The X vectors are drawn second (running east and west) and spread from north to south. This drawing pattern is shown in Fig. 3, with numbers 1 to 8 indicating the drawing order of the vectors.

A first example of alternating sequencing uses X-hatching and Y-hatching, where the X-hatch vectors are sampled followed by sampling the Y-hatch vectors on a first layer. Then the Y-hatch vectors are sampled, followed by sampling the X-hatch vectors on a second layer, and repeating this pattern on alternating layers.

is A second example extends the two-layer patterns of the first example to a four-layer pattern. On the first layer, the X-hatch is sampled first, starting with the vectors closest to the X-axis (and the origin), followed by sampling the Y-hatch vectors starting with the vectors closest to the Y-axis (and the origin). On the second layer, X-hatch vectors are drawn first, starting with the vectors farthest from the X-axis (and the origin), followed by sampling the Y-hatch vectors starting with the vectors farthest from the Y-axis (and the

origin). On the third layer, the Y hatch vectors are sampled first, starting with those closest to the Y axis (and origin), followed by sampling the X hatch vectors, starting with those closest to the X axis (and origin). On the fourth and final layer, before repeating the pattern, the Y hatch vectors are sampled first, starting with those farthest from the Y axis (and origin), followed by sampling the X hatch vectors, starting with the vectors closest to the

furthest from the X-axis (and the origin). An example of this drawing order for the four characteristic layers of the pattern is shown in Figs. 4a to 4d, where numbers 1 to 8 represent the relative drawing order of the vectors.

A third example is described by the following table, and is shown in Figs. 5a to 5h, where the numbers 1 to 8 indicate the relative drawing order of the vectors:

&iacgr;&ogr; Example 3: An 8-scnon-X/Y pattern with reciprocal sequencing

Fig. # Layer number Vector type Propagation direction

5a 1 -· Y
X East to West
North to South 20 5b 2 X
Y South to North
West to East 25 5c 3 • · ·**&phgr;
"··: .·" Y
X East to West
South to North 5d 4 X
Y North to South
West to East 30 5e 5 Y
X West to East
South to North 5f 6 X
Y North to South
East to West 35 5g 7 Y
X West to East
North to South 40 5 hours 8th X
Y South to North
East to West 45 ·: .*·. : .·
1 ." * ·
, ! : ««j·

- 20 -

Similar scanning patterns can be developed based on X/60/120 hatching, or other hatching and filling types, and other alternative orderings will be apparent to those skilled in the art. For example, if only one vector type is used, the propagation direction can be sequenced alternately. Additionally, as with other aspects of this invention, paths other than vector-defined can be used, for example, scanning paths based on complex curves or the like can be used. Similarly, the drawing order can be based on the part geometry, where

&iacgr;&ogr; a fixed character order can be used to strengthen or weaken the tendency of an object feature to curl or not curl along a particular axis. For example, if a boom section of each stand is oriented along the x-axis (the length along the x-axis is greater than the length along

is the y-axis), curl along the x-axis can be reduced by solidifying the cantilever using only vectors oriented along the y-axis. Alternate sequencing is a powerful tool that results in more uniform internal stress distributions, reduced distortion during build, and improved overall part accuracy.

Currently, "Starweave", or "Star Weave", is the most preferred construction method. This embodiment uses an X/Y weaving technique in which the weave pattern is staggered from layer to layer, using the 8-layer alternating sequencing pattern described above along with the hatch vectors being retracted from the boundary at their endpoints.

In conventional weaving, X and Y hatching vectors on the &eegr;-th layer, which have a width of the cured line L,

-21

drawn with their respective center lines at a distance H from the center lines of their nearest neighbors. The cured lines are separated from each other by a distance S = &EEacgr; - L. On the next layer, the procedure is repeated with the center line of a vector on the (n + l)-th layer lying directly above the center layer of a corresponding vector on the &eegr;-th layer. This procedure is then continued on all layers. The result is that at the intersection points of the X and Y vectors, the doubly cured points sit directly on top of each other, and that the spaces between the

η doubly cured points sit directly on top of each other. Thus, conventional weaving is analogous to building a wall with bricks stacked directly on top of each other. On the other hand, in the offset weaving method, the center line of a given vector on the (n + l)th layer is offset or staggered with respect to the corresponding vector on the ηth layer. Moreover, the amount of offset from layer to layer is approximately H/2. Thus, a doubly exposed point on the (n + l)th layer is located midway between the four adjacent doubly exposed points on the nth layer. Therefore, offset weaving is analogous to building a wall with the positions of the bricks offset from those on the previous layer. When weaving is used with some photopolymeric resins, microcracks have been observed between adjacent vector lines. These microcracks have been eliminated by using offset weaving. Of course, other patterns of staggering are possible, including for example 3- or 4-layer patterns.

The retracted hatching used in "Star Weave" concerns drawing vectors for which each is attached only to the boundary of its start point, and by a small distance r from the

boundary to its opposite or terminal end. The optimal retraction distance may be build material dependent (e.g. resin), but can be easily determined by building different diagnostic parts with different amounts of retraction with a given build material, and then determining which retraction values are associated with the most accurately formed parts. In addition, the retraction itself should alternate between adjacent vectors and possibly between layers. For example, the first X-vector could be placed at a right edge

&iacgr;&ogr; and be retracted from the left edge by a suitable amount (for example, 0.025 to 1.3 mm, and preferably 0.38 mm to 0.76, and more preferably about 0.50 mm or less). As previously noted, the adjacent vector (or an alternating pattern of vectors) may be attached to the left edge and

is retracted to the right edge. A similar retraction technique is used for the Y vectors. Benefits of retracted hatching include reduced internal stresses during part build, reduced stresses on beams with a corresponding reduction in the likelihood of catastrophic beam failure, reduced time-dependent post-cure distortion known as "creep," and improved overall part accuracy.

A first implementation of this aspect of the invention is based on software generation and output of hatch and fill vectors according to the rules of any of the above embodiments or similar. The software may form the hatch and/or fill vectors and output them to a file after correctly arranging them, or alternatively the software may essentially form the hatch and/or fill vectors when they are needed and output these vectors directly to a scanner system for controlling

the solidification of the material. Vector generation and sequencing can also be handled by suitably configured hardware.

The fourth aspect of the present invention includes improved stereolithographic exposure techniques involving a combination of a fast shutter (e.g., an acousto-optic shutter) and creative scanning techniques. This aspect of the invention is applicable to both rotary scanning systems and translation scanning systems. This aspect of the invention has several advantages over conventional stereolithography scanning techniques: 1) Improved generation of sharp object features,
2) Ability to use less expensive scanners to produce high resolution parts, 3) improved part accuracy through improved

is exposure, 4) smaller laser beam spot size on the work surface, 5) longer scanner lifetime, 6) improved part accuracy through improved positional accuracy of a scanner encoder system due to constant mirror rotation speeds during part exposure, and 7) significantly reduced scan times compared to those for raster scanning techniques.

Disadvantages of conventional stereolithography exposure systems relate to limited maximum angular acceleration of the scanning mirror system. The first disadvantage relates to the inability to uniformly expose a line of material when the laser scanning system must accelerate or decelerate during the exposure process. This need for acceleration during exposure and limits on the actually achievable acceleration lead to a maximum scanning speed. This maximum scanning speed is based on the need to achieve a relatively uniform exposure over the length of a

vector, which limits one to periods and acceleration lengths that are relatively short.

The second disadvantage with conventional scanning techniques concerns the ability to form sharp features. When drawing sharp features with a vector scanning system, the maximum angular acceleration capability and/or the capability of the servo/scanner system limits the accuracy of the sharp feature. These features exhibit a limited radius and sometimes overshoot or

&iacgr;&ogr; decay behavior behind the sharp feature, which corresponds to unwanted exposure. To limit these inaccuracies, scanners with high acceleration capability must be used and redesigned servo drives must be developed and optimized for both regular and sharp feature execution. Scanning systems with

is higher acceleration are more expensive and dual-function servo optimization (for both regular and sharp features) limits the effectiveness on regular features.

Spot size is also compromised because the rotational inertia (moment of inertia) of the mirror must be reduced to obtain the necessary system acceleration capability. This prevents the use of larger mirrors in existing scanners that would allow a larger beam size (smaller f/number) and a smaller focused spot size on the resin surface.

The lifetime of the scanner is also affected by the peak-bearing stresses generated during these maneuvers on sharp features.

A primary feature of the aspect of the invention relates to dimming of the beam when the beam must make a significant change in scanning speed, i.e., scanning direction and/or velocity, or must travel over an area that is not exposed

s should be. By stopping the beam when these target points of non-exposure to the change of direction or change of speed are reached, the scanning system is allowed to overshoot the target point and decelerate or accelerate (including constant speed changes of the scanning direction),

Î while behind the target point and outside the area to be exposed. After the scanning system and hence the beam has been brought up to or down to the appropriate exposure speed along the appropriate exposure path and the beam has passed over the target point, the shutter is reopened so that the beam can expose the work surface to appropriately transform the build medium.

A particular embodiment of this aspect of the invention relates to the following steps:
20

1.) Before drawing, or during drawing of vector η at a particular drawing speed and direction, calculate vector η +1 to determine any necessary modifications to the scanning parameters that should be made. These scanning parameter changes concern determining whether or not vector η and vector η +1 meet head to tail, whether vector η and η +1 should be exposed using the same scanning speed, and whether or not there is a change in scanning direction between the two vectors that requires a significant acceleration. If the

- 26 -

Vector &eegr; and &eegr; +1 meet head to tail, go to step 2a, if they do not meet head to tail, go to step 2b.

2a) In terms of vectors that meet head to tail, a suitable laser path radius is calculated at an up-coming sharp feature. This calculation may involve the scanning speed or changes thereto and the change of scanning direction (angle). The purpose of the calculation is to determine a radius

&iacgr;&ogr; based on the desire to predict possible sampling inaccuracies

inaccuracies at a tolerable level. These inaccuracies include speed inaccuracies as well as position inaccuracies for this connection. The speed inaccuracies are preferably kept below a level

is maintained, which results in hardening depth errors of less than 0.050 mm and

more preferably 0.025 mm. The positional inaccuracies are preferably kept below a level that will result in solidifying material that is more than 0.050 or 0.076 mm out of position, and more preferably below a level that would result in solidifying material that is more than 0.025 mm out of position, and most preferably below a level that would result in an error.

3a) Determine whether the predicted radius is less than or greater than a preset maximum radius limit. If it is less than the preset radius limit, the transition from vector &eegr; to vector &eegr; +1 is made by the conventional scanning method. If the predicted radius is greater than the preset maximum limit, the shutter at the junction of the sharp corner is closed and the scanning system continues scanning over the

- 27 -

connection in the same direction as that which was scanned for the vector η. Before starting scanning or while approaching or turning away from the connection, a trajectory is determined that will bring the scanning direction and speed into line with the desired position, scanning direction and speed for the vector η +1. When the scanner directs the beam back toward the connection point while traveling along the desired path at the desired speed,

Î&ogr; the shutter is opened and the vector Î+1 is sampled. The

particular trajectory can be made to minimize the arc of rotation with which the beam would scan the work surface if it were not gated. Alternatively, the beam can simply be controlled to jump to a point of

is on the line of the vector &eegr; +1 and at such a distance of

the starting point of the vector η +1 so that the ray is directed towards the starting point, and during the time it takes to go there, it travels stably at the desired speed.

4a) Continue this process by storing necessary information or exposing vector η +1 and evaluating vector η +2. Therefore, in the above steps, replace η +1 with η +2 and replace η with η +1.

2b) With respect to vectors that do not meet head to tail, the shutter is closed when the connection point is crossed. The scanning system continues to scan across the connection in the same direction as that scanned for vector η. Before scanning or while approaching the

connection or moving away from it, a trajectory is determined that will bring the scanning direction and speed into line with the desired starting position, scanning direction and speed for vector η +1. When the scanners direct the beam toward the starting point of vector η +1 as it travels along the desired path at the desired speed, the shutter is open and vector η +1 is scanned.

&iacgr;&ogr; 3b) Continue this process by storing information or exposing vector &eegr; +1 and evaluating vector &eegr; +2. Therefore, in the above step, replace &eegr; +1 with &eegr; +2 and replace &eegr; with &eegr; +1.

is Figures 8a through 8g show various orientations of vector η and η +1, 900 and 904 respectively, along with gated (virtual) scan paths while maintaining a constant scan speed and implementing a maximum tolerable acceleration indicated by 908. Alternatively, these gated paths 912 may be modified by allowing a change in scan speed before and after the change in scan direction (angular acceleration of the beam on the working surface). The scan speed may be reduced after the gated occurs and before the attempted angular acceleration and then increased before reopening the shutter. As indicated, this may allow shorter virtual scan paths by allowing a tighter turning radius. Figures 8a, b, c and e show virtual paths at constant speed for different types of connections using a maximum allowed value of angular acceleration. Figures 8d and e show virtual paths where the scanning speed

- 29 -

is linearly decreased and accelerated before the angular acceleration involved in changing directions (smaller turning radii). Alternatively, Figures 8d and e can be viewed as representing virtual paths where there is a change in velocity during virtual scanning to accommodate a desired change in scanning velocity between the two vectors.

Other alternative approaches include:

ββ 1.) Any combination of movements behind the link that reduces the maximum mirror acceleration. For example, as indicated above, as a sharp feature becomes sharper and approaches a 180° turn, the constant speed and minimal turning technique may result in unacceptable time delays (e.g.

is Figure 8e). Therefore, a more computationally complex embodiment can be implemented in which there is no attempt to minimize the amount of angular rotation. An example of this type of embodiment is shown in Figure 8g, where the virtual scan begins with a slight angular acceleration in the direction opposite to that which would directly bring the scan path parallel to the appropriate direction. This type of virtual scan is known as an "airplane style" rotation. The advantage of using opposite angular acceleration in this way is the reduced scan path length and the resulting reduced virtual scan time. This airplane style rotation is suitable for scanning hatch and fill type vectors where there is a short gap between the vectors and a required 180° rotation.

- 30 -

2.) Use of two-stage scanning, with coarse scanners limited to a preset acceleration, while fast scanners with limited range "sharpen" the corners. Acousto-optic crystal modulators act as frequency-controlled diffraction gratings by varying standing waves within the crystal, and can be used to slightly deflect a beam for fine position adjustments, or to deflect it sufficiently into a beam trap, thereby acting as a shutter. Therefore, a pair of acousto-optic modulators or similar can function as both shutters and fine-tuning scanners. An example of another candidate for fine-tuning scanners is piezoelectric crystal mirrors, which can be mounted in front of or on the path of the main scanning mirrors.

A first implementation of this aspect of the invention is based on the program that generates the vector data for exposing a layer also generating and providing additional non-exposure vector data indicating when dimming should occur, the path the beam should follow if no exposure occurs to align the path and velocity for the next exposure vector, and indicating when the shutter should be reopened.

A second implementation of this embodiment is based on the use of vector exposure data combined with scanning mirror, control software or hardware that compares the velocity, end point and scanning direction of a first vector to the velocity, start point and scanning direction of a second vector, and determines whether dimming and/or virtual scanning is necessary.

In each implementation, the virtual scanning can be fully or partially controlled after leaving the first vector and before starting exposure of the second vector. In the shutter-locked mode, the exact travel path of the beam is not critical as long as the path and speed are adjusted to hit the starting point of the second exposure vector along the appropriate direction and at the appropriate speed. In other words, after crossing the end point of the first vector and closing the shutter, the scanning system can

&iacgr;&ogr; to simply jump to a new virtual vector start point that is close enough to the start point of the second vector so that little time is wasted during the virtual scan, just aligned with the second exposure vector, and far enough from the start point of the second vector so that sufficient time is available to stabilize the beam position and velocity before the start point is reached.

Alternative embodiments include any combination of exposure vectors generated during scanning as needed or prior to scanning and the generation of non-exposure vectors during scanning. For computational simplicity, all vectors may be followed by virtual scanning, followed by realignment to scan the next exposure vector. However, it is more preferred to allow at least continuous exposure scanning when vectors meet head to tail along the same line. It is even more preferred to allow continuous scanning when vectors meet head to tail with only a slight acceleration and (an acceleration resulting in a minimal change in the cure depth of the material exposed by the beam, preferably less than 2 mil change in cure depth and even more preferably

typically less than 1 mil change in cure depth) required to change speed or direction.

It is currently preferred to maintain maximum beam power to use the zero-order mode of the acousto-optic modulator for exposure and first-order deflection in the non-exposure mode. However, when the beam is in the first-order mode, the zero-order mode does not completely disappear, thus exposure can still occur. For a given application, if the attenuation coefficient is high enough when going from zero-order "on" to first-order "off" at the expense of reduced beam intensity, the on/off sequences and optical paths can be reversed, thereby allowing an infinite attenuation coefficient.

While four aspects of the invention have been presented separately, a fifth aspect of this invention encompasses any combination of the above four aspects. It will be apparent to those skilled in the art that numerous combinations of the groups can be made, resulting in further improvements in the manufacture of stereolithographic parts. In particular, it is noted that aspect 4 can be combined with aspects 1 to 3 for better control of exposure, and that aspects 1 to 3 can be combined in pairs or as a trio.

The four aspects of the invention should be used to the extent necessary or desired to achieve sufficient reduction in distortion.

·

·

In addition, all four aspects of the invention are equally applicable to top-down and right-side build techniques, as well as to different types of build materials including liquid polymers, sinterable powders, and bondable powders. All aspects of the invention are also applicable to an exposure technique that uses scan paths controlled by other than vectors. It will also be apparent that the three-dimensional objects to be formed by this invention can be used in many ways. For example, they can be used as end products for particular applications, or as patterns for reproductions of the object in other materials, or as patterns for forming necessary tools for high volume reproductions of an object.

Claims (55)

EP 92912072.3 1 February 1993 3D Systems, Inc. D 17339EPDEGbm F/La/is Claims 1. An improved stereolithographic device for constructing a three-dimensional object from a medium that is transformable after exposure to synergistic stimulation, which comprises a device for producing synergistic stimulation, means for forming successive layers of a medium in preparation for forming layers of the article and a device for forming successive layers of the article by selectively exposing successive layers of the medium to synergistic stimulation in patterns corresponding to successive cross-sections of a three-dimensional object, to build the three-dimensional object layer by layer, with the patterns showing exposure paths, marked by a device for forming - at least sections - of a plurality of layers by exposing a first layer along at least a set of exposure paths and means for changing the order in which the at least one set of exposure pathways is exposed during formation of a subsequent layer. 2. Device according to claim 1, wherein the propagation direction of the at least one set of paths is inverted by the changing device between the first and a subsequent layer. 3. Device according to claim 1, wherein the at least one set of paths in the changing device comprises first and second - and optionally further - sets of paths. 4. The apparatus of claim 3, wherein the first set of paths comprises a plurality of substantially parallel vectors, and wherein the second set of paths comprises a plurality of substantially parallel vectors, and the vectors of the first and second sets are not parallel to each other. r- 5. The device of claim 4, wherein the vectors of the first and second sets are substantially perpendicular to each other. 6. Device according to one of claims 3 to 5, wherein by the changing device on a first layer the first set of paths is exposed, before the exposure of the second set of paths, and on the following layer the second set of paths is exposed, before the exposure of the first set of paths. 7. The device of claim 6, wherein the following layer is a next layer after the first layer. 8. The device of claim 7, wherein the order of exposing the first and second paths is repeated every other layer. 9. Device according to claim 7 or 8, wherein the propagation direction of the first set of paths is inverted by the changing device between a first layer and a following layer. 10. Device according to claim 9, wherein the propagation direction of the second set of paths is inverted between a first layer and a subsequent layer. 11. The device of any of claims 7 to 10, wherein the combination of the order of exposure of the paths and the propagation direction of each set of paths is repeated every fourth layer. 12. The device of any of claims 7 to 10, wherein the combination of the order of exposure of the paths and the propagation direction of each set of paths is repeated every eighth layer. 13. Device according to claim 3, wherein the first and second and
further set of paths of the change device have at least first, second and third sets of paths. 14. The apparatus of claim 13, wherein the first, second and third sets of paths each comprise a plurality of substantially parallel vectors that are not parallel to the vectors in the other sets. is 15. Apparatus according to claim 13 or 14, wherein the first, second and third sets of paths are oriented at an angle of 60° to each other. 16. The device of any of claims 13 to 15, wherein the order of exposing the at least first, second and third sets of paths on a first layer is different from the order of exposing on a subsequent layer. 17. Device according to one of claims 13 to 16, wherein the propagation direction of at least one of the first, second and third paths is inverted between a first and a subsequent layer. 18. Apparatus according to any one of claims 1 to 7 or 13 to 17, wherein the changing means changes the order of exposure of the quantities. 19. Apparatus according to any one of claims 1 to 18, wherein the order of exposure is varied by the varying means according to the object geometry so as to reduce the tendency of an object feature to curl along a particular axis. 20. Apparatus according to any one of claims 3 to 12, wherein the at least two sets of paths of the alteration device each comprise individual exposure lines separated by adjacent exposure lines by more than the width of cure of each individual line, and wherein each set of paths experiences individually insufficient exposure for adhesion to the previous layer, but sufficient exposure such that doubly exposed areas adhere. 21. Apparatus according to claim 20, wherein the positions of each of the lines in each of the two paths are offset from layer to layer by the amount of the distance between the individual lines. 22. Device according to claim 20 or 21, wherein at least one end of each individual line ends before a contour boundary. 23. Apparatus according to claim 22, wherein only one end of each individual line terminates beforehand and, for consecutive lines, opposite ends are correspondingly retracted. 24. The device of claim 23, wherein the retracted end of each individual line alternates between successive layers. 25. Improved stereolithographic apparatus to construct a three-dimensional object from a medium, &iacgr;&ogr; that is transformable after exposure to synergistic stimulation , which is a source of synergistic stimulation, means for forming successive layers of a medium in preparation for forming layers of the article, and means for forming successive layers of the article by selectively exposing the successive layers of the medium to synergistic stimulation in patterns corresponding to successive cross-sections of the three-dimensional object, to build the three-dimensional object layer by layer, marked by a device for forming - at least a portion - at least one layer of the three-dimensional object of plates separated from each other by gaps; and means for solidifying a substantial portion of the medium over at least one gap on at least one subsequent layer &iacgr;&ogr; including a device for exposing the medium via the gap and means for preventing at least one side of the gap from adhering to the medium solidified over the substantial portion of the gap, until the solidified medium above the gap has at least partially shrunk, and means for subsequently causing adhesion between the previously unadhered side of the gap and the solidified medium above the gap. 26. The device of claim 25, wherein the portion of the at least one layer formed from plates comprises more than one layer. 27. Device according to claim 25 or 26, wherein plates on the subsequent layer are slightly offset so that a substantial portion of the gap is covered without completely covering the gap to bridge, and wherein, after the solidified medium spanning the gap has been allowed to shrink for a period of time, the displaced plates are cemented together by solidifying the medium. 28. Apparatus according to claim 25 or 26, wherein at least one end of the solidified medium spanning the gap is suspended above a plate on one side of the gap until at least a substantial portion of the solidified medium spanning the gap &iacgr;&ogr; has shrunk. 29. Apparatus according to claim 28, wherein the floating solidified medium is riveted to the plate. 30. The apparatus of claim 28, wherein the suspended solidified medium adheres to the plate after at least one additional exposure. 31. Device according to one of claims 25 to 30, wherein the time available for shrinking is at least three seconds. 32. Apparatus according to claim 31, wherein the time available for shrinking is at least five seconds. 33. Improved stereolithographic apparatus to construct a three-dimensional object from a medium, which is transformable after exposure to synergistic stimulation , comprising means for forming successive layers of a medium in preparation for forming layers of the article and means for forming successive layers of the article by selectively exposing the successive layers of the medium to synergistic stimulation in patterns corresponding to successive cross-sections of the three-dimensional object, is to build the three-dimensional object layer by layer, wherein the patterns have exposure pathways defined by a plurality of vectors, marked by, means for identifying an end point of a first vector and a start point of a second vector; means for deflecting the synergistic stimulation along the first vector at a substantially constant velocity; means for terminating the synergistic stimulation when it reaches the end point of the first vector without significantly changing the substantially constant velocity; means for directing the synergistic stimulation, while dimmed, in a path toward the starting point of the second vector such that a suitable velocity for exposing the second vector is achieved when the starting point of the second vector is reached; means for terminating the fading of the synergistic stimulation when it reaches the starting point of the second vector; and is a device for diverting the synergistic stimulation along the second vector at a substantially constant speed. 34. The apparatus of claim 33, wherein the end point of the first vector coincides with the start point of the second vector. 35. The apparatus of claim 33, wherein the end point of the first vector does not coincide with the start point of the second vector. 36. Apparatus according to any one of claims 33 to 35, wherein there is a substantial difference between the scanning speed of the first vector and the second vector, and at least a significant part of the difference is due to a scanning direction difference. 37. Apparatus according to any one of claims 33 to 36, wherein there is a substantial difference between the scanning speed of the first vector and the second vector, and at least a significant portion of the difference is due to a scanning speed difference. 38. An improved stereolithographic apparatus to construct a three-dimensional object from a &iacgr;&ogr; medium, which is transformable after exposure to synergistic stimulation , is a device for forming successive layers of the Medium in preparation for forming layers of the object, and means for forming successive layers of the article by selectively exposing the successive layers of the medium to synergistic stimulation in patterns corresponding to successive cross-sections of the three-dimensional object to build up the three-dimensional object layer by layer, wherein the patterns have exposure pathways defined by a plurality of vectors, marked by, means for sampling a first vector at a first velocity and for sampling a second subsequent vector at a second velocity, the second vector having a starting point coincident with the ending point of the first vector; means for determining a speed difference between the vectors; &iacgr;&ogr; a device for specifying a maximum acceptable speed difference; a device for comparing the determined speed difference with the maximum acceptable speed difference; means for terminating the synergistic stimulation when the end of the first vector is reached if it is determined that the predetermined velocity difference is greater than the maximum acceptable velocity difference, the scanning velocity of the first vector being maintained substantially constant over its length; means for directing the synergistic stimulation, while dimmed, along a path toward the starting point of the second vector such that the appropriate velocity for exposing the second vector is achieved when the starting point of the second vector is reached; •-VB ⁹ · means for terminating the fading of the synergistic stimulation when it reaches the starting point of the second vector; and means for sampling the synergistic stimulation along the second vector at a substantially constant rate. 39. Apparatus according to claim 38, additionally comprising sampling third &iacgr;&ogr; and fourth consecutive vectors having a coincident end point and start point, the speed difference between the third and fourth vectors being determined to be less than the maximum acceptable speed difference, and wherein the synergistic is stimulation is not dimmed when switching from the third to the fourth vector. 40. Apparatus according to claim 38 or 39, wherein the determined velocity difference between the first and second vectors is based at least in part on a difference in the scanning direction. 41. Apparatus according to any one of claims 38 to 40, wherein the determined speed difference between the first and second vectors is based at least in part on a difference in scanning speed. 42. Apparatus according to any one of claims 38 to 39, wherein an acceptable cure depth difference caused by the difference in scanning speed is less than 0.050 mm if the transition from the first speed to the second speed occurs at the maximum possible acceleration, resulting in an error of less than 0.076 mm in the sampling position. 43. Apparatus according to claim 42, wherein the acceptable error in the cure depth is 0.025 mm. 44. Apparatus according to claim 42 or 43, wherein the acceptable error in the scanning position is less than 0.050 mm. 45. Apparatus according to claim 42 or 43, wherein the acceptable error in the scanning position is less than 0.025 mm. 46. In a stereolithographic device for constructing a three-dimensional object from a medium that is solidifiable after exposure to synergistic stimulation, comprising means for forming successive layers of the medium in preparation for forming layers of the article and a device for forming successive layers of the article by selectively exposing the successive layers of the medium through synergistic stimulation in patterns according to · '-15*- successive cross-sections of the three-dimensional object, to build the three-dimensional object layer by layer, with the patterns showing exposure paths, shows an improvement: means for exposing at least a portion of a &iacgr;&ogr; layer to a beam of synergistic stimulation along a first plurality of paths; means for releasing small discontinuities of unexposed and unconsolidated material along the first plurality of paths; means for exposing at least a portion of a layer to a beam of synergistic stimulation along a second plurality of paths; means for releasing small breaks in exposure along the second plurality of paths, within at least a portion when the second plurality of paths is exposed to at least a second pass of a beam of synergistic stimulation, wherein a substantial number of breaks in exposure along the second plurality of paths occur at locations different from the locations of the breaks in exposure along the first plurality of paths; and means for re-exposing, in a second pass, the material cured in the first pass after the material to be re-exposed has undergone a substantial portion of the shrinkage induced by the first pass. 47. A device according to any preceding claim, wherein the transformable medium solidifies after exposure to synergistic stimulation. 48. A device according to any preceding claim, wherein the medium is a polymer. 49. The device of claim 48, wherein the polymer is a photopolymer. 50. Device according to one of claims 1 to 47, wherein the medium is a powder. 51. Device according to one of the preceding claims, wherein the source of synergistic stimulation is a source of UV radiation. 52. Device according to claim 51, wherein the UV radiation source is a UV laser. 53. Apparatus according to any preceding claim, wherein the source of synergistic stimulation is an IR radiation source. 54. Apparatus according to any preceding claim, wherein the means for selectively exposing comprises X and Y scanning mirrors. 55. The apparatus of claim 50, wherein the synergistic stimulation means delivers a binder that is selectively distributed onto the powder.