The present invention relates to a method and a system for attaching particulate material onto a receiving unit with affinity for said particulate material. The particles are applied in conformity with given two-dimensional patterns, and can possibly be applied in several successive layers on top of each other. However, it is an important condition that the particulate material and the material of the receiving unit in question have a reciprocal affinity, i.e. that bonding forces exist which make the two materials adhere to each other when they are brought sufficiently close to each other.

More generally, one object of the present invention is to achieve a dynamic transfer of patterns of material particles to a receiving unit in an arbitrary number of layers. The patterns may be any two-dimensional constructions, like text, figures and geometry. The meaning of the expression "dynamic" is that the patterns can be changed very rapidly from layer to layer without any restrictions. The information regarding single layers can be handled in a digital format. The particulate material may be any material which can be attracted by means of electrostatic fields, and which can be received and held by the receiving unit.

As for fields of utilization, the simplest embodiment of the invention is a process for rapid digital one-colour printing. A document stored in a diskette can for example be duplicated in any number and without any present original in the usual sense of the word. The same speed as in the best copying machines can be obtained.

Since the layers to be applied have patterns which can change dynamically, and since any number of layers is possible in principle, the system may also be used as a digital multi¬ colour printing machine. It is even possible to print pictures with a certain three-dimensional structure, as is often the case with a painted picture. The material can be

freely selected. The need for an original, in the usual sense, disappears also in this case, i.e. the costly production of films and production originals is no longer necessary.

One further promising use of the invention is the production of multi-layer printed circuit cards. Instead of coloured particles used in an ordinary printing process on paper or similar material, one uses particles which are respectively electrically conductive and electrically insulating. It is then possible to build up any number of layers directly from a CAD model.

The final and most advanced utilization of the present invention, is layered construction of objects. Objects may be produced in any material and can be used as design and test models, high precision moulds, special tools and also products to be produced in small numbers.

Thus, in general terms the final object of the present invention is to transfer a product definition existing in a computer to a physical product in a rapid and cost efficient manner. This implies that the time spent from the design process itself in a CAD system, and up to the presence of a physical first version of the product, may become very short, possibly down to a few hours. This opens new possibilities of an interactive process regarding product design and product testing. The result is a better design and a substantial reduction of the development period of the product. The invention also opens the possibilities for rapid and cost efficient finishing, and as previously mentioned, also for producting objects or products in small numbers.

Regarding the prior art, certain features will be recognized from the so-called xerography. The related feature within the art of xerography is the accumulation of particulate material on a substrate by means of electrical fields, and thereafter transfer of these particles to a receiving unit. However, in the xerographic process the receiving unit is a paper or a polyester sheet, with assistance from an electrical field applied to the back side

of the receiving unit to pull the particles over from the substrate to the receiving unit. The most important difference in relation to the present invention is that the substrate used in the start, is a so-called photoconductive substrate, which is first illuminated to provide differently charged areas of a surface, depending on the incident light.

In the present invention no such photoconductive substrate appears, and there is no use of light in order to generate charges or electrical fields. However, electrical fields are generated from a plurality of separate and separately controlled capacitors distributed over the surface of a "substrate". Thus, the starting point is completely different from the xerographic process, which is a photo¬ copying process.

The objects of the invention as defined above, are obtained by providing a method and a system as precisely defined in the appended patent claims.

A more detailed description of the invention shall now be given, using certain preferred embodiment examples and with reference to the enclosed drawings, where fig. 1 shows an example of a pattern matrix of the type included in the invention, with an array of silicon chips, as well as an enlarged detail with one single silicon chip, fig. 2 shows a section through one part of such a silicon chip with capacitor plates near the surface, and fig. 3 shows an outlined method in accordance with the invention for production of an object directly from a CAD- model.

An embodiment regarding the central feature of the present invention is shown in fig. 1. Generally this central feature is constituted by a substantially two-dimensional pattern matrix which comprises embedded memory elements having associated respective capacitor plates distributed substan¬ tially along a main surface of the matrix. The state of the memory elements is controllable via underlying electronic control circuitry, and in this embodiment the pattern matrix is constituted by a number of silicon chips placed in a matrix

configuration on a substrate, and with mutual interconnection. This is shown in fig. lb. The electronic control circuitry is incorporated in the silicon chips. A multiplexer and control logic (transputer technology) deals with the programming of a complete such substrate. Control signals are received from a computer or a CAD system.

One single silicon chip is shown enlarged in fig. la. Uppermost toward the surface, i.e. as shown in fig. la, lies a capacitor with the ability to generate an electrical field reaching out beyond the chip surface. Below every such capacitor lies a corresponding memory element which, as indicated in the drawing, is governed to adopt one of two different states, "0" and "1". As a typical value, a matrix of capacitors deposited on the surface may have a size of the order 50 x 50 μm. A thin, insulating layer is placed on top (see fig. 2) .

Inputs Ii, ∑2 . . . , IN ^are used for serial loading of one "pattern" per row. The pattern, indicated by 0 or 1 in fig. la, makes it possible to program each capacitor individually to a voltage V ^" o or Vi. V ^" ι is then chosen in such a manner that charged particles of the opposite polarity are attracted to the capacitor, while V ^" o is selected in such a manner that the particles are not attracted. Values are chosen in such a manner that Vi holds the particles in a gravitational and centrifugal field.

The actual loading is made by using shift registers which are able to transport programming information (0/1) . The charging to voltage Vi, respectively Vo takes place via a power supply. A complete row of silicon dice can be programmed via outputs Oχ, O2, ..., ON, the programming merely continuing on the next chip as appears from fig. lb, there being conductors connected from 0 in one chip to the corresponding I of the neighbouring chip.

The pattern matrix, or the "dynamic printing matrix" which it will also be designated, plays an important part in all applications mentioned previously in this description. The matrix may consist of a planar (as indicated in fig. lb)

or cylindrical matrix comprising a plurality of "points" where a memory element and a capacitor plate define one "point". The memory element defines the "on" state (or "1") or "off" state (or "0") for a matrix point. If the memory element is in state "on", a voltage potential is delivered to the capacitor plate, creating an electrostatic field in the close surroundings of the plate. If the memory element is in the "off" state, no voltage potential is delivered to the capa¬ citor plate. In this manner any pattern of local, electro¬ static fields can be formed, which fields are determined by the state of the memory elements in the matrix. These electrostatic fields will attract particles brought into their near surroundings. The result is then an arbitrarily chosen pattern of particles which are held by the electrostatic fields. The particulate material itself is fed from a reservoir, and is brought to the area close to the pattern matrix, i.e. close to the electrostatic fields possibly existing, by means of sprinkling, blowing or pouring.

As mentioned above, the memory elements are serially connected in the same line in one of the matrix directions. The pattern which it is desirable to print, is shifted inward in the matrix under control from the attached computer, through serially connected memory elements lying in parallel rows. After having shifted the whole pattern completely into the matrix, the states of the memory elements describe a raster picture of the pattern. When thereafter particles have been brought in and are attached due to the electrostatic fields, the whole matrix is moved toward a receiving unit, see fig. 3. (Optionally, if the matrix has a cylindrical configuration, it may roll over such a receiving unit.) When thereafter the matrix, including electrically attached particles, engages the receiving unit (which is actually not shown in fig. 3), all of the memory elements are switched to state "off", and hence the pattern matrix does no longer hold the particles. On the other hand the receiving unit will take over the particles while maintaining the same (or rather the inverse) pattern, then assuming that sufficient affinity

forces exist between the particulate material and the receiving unit material. Especially these forces must be clearly stronger than possible similar forces between the particles and the surface material of the pattern matrix. Hence the particles are deposited in the desired pattern on the receiving unit.

"The building rate" for a process as the one just described, can rather simply be brought to a level of one complete layer per second. This is a magnitude order 10-20 times faster than any previously known layer production process.

In order to enhance the affinity forces between the particles and the receiving unit, the receiving unit must possibly be prepared between successive layer applications, and such intermediate processes may comprise heat curing, glue/adhesive application etc. Such processing make the just deposited layers stick together during the production process. After having brought all layers in place, the product is treated in a subsequent curing prosess in order to render the object sufficiently strong for permanent use. This after-cure will vary dependent on what sort of material is used.

As previously mentioned, the dynamic pattern matrix is formed by using semiconductor die technology in the embodiment just exemplified. The semiconductor die elements are constructed with a planar or curved configuration, depending on whether the matrix is intended to be planar or cylindrical. (In the configuration shown in fig. 3 is shown something in between, namely an approximately cylindrical configuration of several flat matrixes.) Then, the semiconductor die elements are gathered together to provide a complete surface of the desired size. The elements are connected electrically in series with each other. The mesh width between memory elements is only limited by the available state of the art regarding chip production, and for the time being this limitation is approximately 0,6 μm, but in many applications such a high resolution is not necessary. Therefore the mesh width can be adapted to the actual need in a particular

application field. Practical pattern matrices may typically have a mesh width in the range 0,1 - 0,01 mm. Such a mesh width is simple to implement in the present semiconductor die technology.

This implies that a 500 x 500 mm matrix (i.e. relatively large) with a mean resolution of 0,05 mm will then contain 10.000 x 10.000 elements. That is, the matrix will consist of 10.000 lines, each line containing 10.000 serially connected memory elements. The information in each line is shifted in in parallel using the so-called "transputer technology". The in-shifting of input signals can be executed with a frequency of 1-10 MHz, dependent on the voltage necessary to support or hold the particles. Each input line can therefore be shifted in (the example above) from one hundred to thousand times per second, or the whole matrix can be filled within one second by using ten to one hundred parallel processes, if several lines are filled from the same input channel.

The computer with the control system which is used, contains a geometric modelling system which is capable of receiving a product model from an external CAD system via a standardized format (such as IGES, VDA-FS, STEP) . The control system may then optionally change the incoming model to a certain degree in order to adapt this model to the purpose. This may be done to provide slip angles for moulding pieces, correction for shrinking etc. In workshops lacking an external CAD system, also the product design can possibly be made directly in the control system itself.

The next steps of the control process are to divide the model in a set of two-dimensional layers. The content of each layer describes the patterns to be transferred into the printing matrix, but so far this content is available in a vector or text format.

In the third step the information regarding the layers is converted to a raster picture adapted to the dimensions of the pattern or printing matrix to be used.

In the next step the generated raster picture is shifted into the pattern matrix, at the moment when this is necessary,

using parallel processing and transputer technology.

Thus, to sum up, the chain of actions is as follows in the present concept:

A product model is generated in a computer.

- The model is transferred to the control system.

The model is automatically divided into two-dimensional layers or units.

The two-dimensional information in each respective layer is transferred, one at a time, into the pattern matrix (printing matrix) , thereby creating corresponding electrostatic fields on the main surface of the matrix. Particulate material is supplied to the region near the matrix surface, and is attracted to the electrostatic field in such a manner that the particles represent a physical image corresponding to the digital information which has been fed into the matrix.

- The particles are thereafter deposited on the receiving unit, and partially bonded thereto.

Subsequent to deposition of all desired layers, the object is after-cured and is ready to be used.

It must be noted that in this system one actually uses the pattern matrix which in principle is a semiconductor die, as a printing original, i.e. an original in the usual sense does not exist.

As previously mentioned, electrically charged particulate material can very well be used, but it is also possible to use electrically neutral particulate material, however, such material must be of a polarizable type in order to be attracted and held against the capacitor plates.

It may be of special interest to use the so-called "mono- disperse polymer particles", i.e. polymer particles with a very homogeneous size, as particulate material.

Regarding the use of a cylindrical matrix, it should be noted that during the transfer process where the matrix is rolled over a receiving unit to deposit particles thereon, the electrostatic holding fields must be switched off in synchro¬ nism with successive capacitor rows going into engagement

against the receiving unit.