CN103456650A

CN103456650A - Eyepoint training method for wire bond and related semiconductor processing operation

Info

Publication number: CN103456650A
Application number: CN2013103055389A
Authority: CN
Inventors: M·T·迪莱; 王志杰; P·M·利斯特; D·索德
Original assignee: Kulicke and Soffa Investments Inc
Current assignee: Kulicke and Soffa Investments Inc
Priority date: 2007-03-13
Filing date: 2007-03-13
Publication date: 2013-12-18
Anticipated expiration: 2027-03-13
Also published as: CN103456650B

Abstract

The invention provides an eyepoint training method for wire bond operation. The eyepoint training method for the wire bond operation comprises the steps of (1) selecting a set of shapes from an area of a semiconductor device, using the shapes as eyepoint, and (2) using at least one of (a) a sample semiconductor device or (b) preset data related to the sample semiconductor device to train the eyepoint for a seam welder. The training step comprises the step of defining the position, relative to another, of each shape.

Description

Eyespot training method for wire bond and relevant semiconductor processing operation

The application is that the application number of submitting on June 19th, 2008 is divided an application for " for the eyespot training method of wire bond and relevant semiconductor processing operation " for " 200780001638.5 ", denomination of invention.

Cross reference to related application

The application is the China national phase application of the application number submitted on March 13rd, the 2007 PCT application that is PCT/US2007/063850, here this international application is incorporated herein by reference.

Technical field

The present invention relates to the wire bond system, more particularly, relate to improved eyespot (eyepoint) training method for the wire bond system.

Background technology

U.S. Patent No. 5,119,436 and No.6,869,869 relate to wire bond system and associated vision systems, here by reference to being incorporated to its full content.

In the processing and encapsulation process of semiconductor device, usually use the training and operation that utilizes vision system.For example, for example, carry out the wire bond operation on a collection of semiconductor device (, resembling this device of the semiconductor chip be arranged on lead frame) before, usually need the eyespot (or a plurality of eyespot) of " training " sample device.By " training " described sample device, some physical data that will be relevant to this sample device stores (for example,, in the memory of wirebond machine).In processing the process of device in batch using these physical datas as benchmark, for example, the correct location of each device or aim in the semiconductor device in batch of processed to guarantee (for example,, by wire bond).

Therefore, in the wire bond operating environment, wirebond machine (is for example used vision system, pattern recognition system or PRS) find previous training pattern (for example, eyespot, reference point etc.), so that after semiconductor device being placed on to welding position place and before the welding filament (for example, at semiconductor device and support between the lead frame of this semiconductor device weld filament before) alignment of semiconductor devices.Traditional approach is to carry out taught eyepoint based on sample device on wirebond machine, and wherein operating personnel utilize the training window to aim at the target area on sample device.Use some routine techniques (for example algorithm) to scan the target eyespot in conjunction with vision system.

A kind of conventional training technique utilizes normalized gray scale related system (NGCS) to carry out scanned samples device (for example, the selected part of sample device).By this technology, in each position, based on the detected content of vision system, distribute gray value.For example, during the welded disc of scanning semiconductor device, gray value is distributed to and is scanned position.After desired sector scanning completes, storage library of grayscale values (being associated with corresponding scanning position).When such actual semiconductor device is carried out to wire bond, vision system detects each gray value that is scanned position, and these gray values and those gray values be stored in training process in storehouse are compared.

Another kind of conventional training technique relates to scanned samples device (for example, the selected part of sample device), and detects each edge (that is, the pattern match based on edge) of definition in being scanned zone.By this technology, the detected content based on each position vision system, logarithm value is defined.For example, when scanning the welded disc of semiconductor device, marginal value is distributed to and is scanned position.After desired sector scanning completes, storage marginal value storehouse (being associated with corresponding scanning position).When such actual semiconductor device is carried out to wire bond, vision system detects each marginal value that is scanned position, and those values that are stored in these numerical value and training process in storehouse are compared.

Use any one in these conventional methods, divide the function that mark is wherein compared with the practical devices of wanting wire bond by the training sample device for each device of wanting wire bond provides weighting.If this mark surpasses certain threshold value, this device is acceptable so, and for example, by processed (wire bond); If but this mark is not just proceeded automatic operation usually lower than described threshold value.For example, can tell operating personnel by this low mark.In addition, can attempt follow-up location or eyespot and obtain the acceptable mark.Further, can also attempt alternate algorithm or recover sequence.

Unfortunately, all there are a lot of problems in each technology in these routine techniquess.The actual conditions of semiconductor device are that the different components from same batch (or different batches) can have different perceptual property, even they are identical devices and have identical electric function characteristic.For example, device may be different from surface color or the texture of device.This variation may be derived from the slight change of the different suppliers of same device manufacturing process used.These variations usually cause device to show totally different (for example nonlinear change of reflectivity) aspect contrast and reflectivity.Therefore, for example, due to these variations (, due to the difference between the reflectivity of wanting the wire bond parts and training sample device), the weighting branch that utilizes the traditional mode matching technique to obtain is lower.Therefore, ignore the searching result of mistake although usually utilize mark, but normal mode matching technique (such as NGCS system, pattern matching system based on edge etc.) all can cause mark lower than threshold value, even for further processing, this device is acceptable.Another problem of for example, being brought by this variation between semiconductor device (, the surface difference between device) is that the average time between manpower intervention, (being MTBA) was very short, caused the productivity ratio of automatic wire welding equipment lower.

Therefore, need to improve the eyespot training method of processing for semiconductor device, and use this eyespot to process the method for semiconductor device.

Summary of the invention

According to exemplary embodiment of the present invention, be provided for the method for a kind of taught eyepoint of wire bond operation.This method comprises: (1) selects one group of shape from a zone of semiconductor device, as eyespot, (2) use (a) i type semiconductor device or (b) at least one in the tentation data relevant with this semiconductor device, for wirebond machine is trained this eyespot.For example, these tentation datas can be available data (such as cad data, historical data etc.), can be also this training step definite tentation datas before.This training step comprises the position of each shape of definition with respect to another.

According to another exemplary embodiment of the present invention, provide a kind of method that operates wirebond machine.This method comprises: (1) selects one group of shape from a zone of semiconductor device, as eyespot, (2) use (a) i type semiconductor device or (b) at least one in the tentation data relevant with this semiconductor device, for wirebond machine is trained this eyespot.This training step comprises the position of each shape of definition with respect to another.This method also comprises: (3) will be directed to by the first semiconductor device of wire bond the precalculated position of described wirebond machine, (4) use the vision system of this wirebond machine to scan the selected portion of this first semiconductor device, selected portion is corresponding to by taught eyepoint.

The accompanying drawing explanation

Read in conjunction with the drawings following specific descriptions, can be best understand the present invention.Should emphasize, according to convention, in figure, different parts is out-of-proportion.On the contrary, for clarity sake, the size of different piece is zoomed in or out arbitrarily.These accompanying drawings comprise:

Fig. 1 is the top view of a semiconductor device part in exemplary embodiment of the present invention;

Fig. 2 is the top view of second half conductor device part in exemplary embodiment of the present invention;

Fig. 3 is the top view that exemplary embodiment of the present invention comprises second half conductor device part of eyespot;

Fig. 4 is that a part of top view of semiconductor device shown in the covered Fig. 3 in specific region in exemplary embodiment of the present invention;

Fig. 5 is the top view that exemplary embodiment of the present invention comprises second half conductor device part of eyespot;

Fig. 6 is the top view that is selected as a semiconductor device part shown in Fig. 5 of eyespot in exemplary embodiment of the present invention;

Fig. 7 is the invulnerable release of the eyespot of semiconductor device shown in Fig. 5 in exemplary embodiment of the present invention;

Fig. 8 is that a part of top view of semiconductor device shown in the covered Fig. 7 in specific region in exemplary embodiment of the present invention;

Fig. 9 is the top view that exemplary embodiment of the present invention comprises second half conductor device part of eyespot;

Figure 10 is the top view that is selected as a semiconductor device part shown in Fig. 9 of eyespot in exemplary embodiment of the present invention;

Figure 11 is the invulnerable release of semiconductor device eyespot shown in Fig. 9 in exemplary embodiment of the present invention;

Figure 12 is that a part of top view of semiconductor device shown in the covered Figure 11 in specific region in exemplary embodiment of the present invention;

Figure 13 is the top view that exemplary embodiment of the present invention comprises second half conductor device part of eyespot;

Figure 14 is the top view that is selected as a semiconductor device part shown in Figure 13 of eyespot in exemplary embodiment of the present invention;

Figure 15 is the invulnerable release of the eyespot of semiconductor device shown in Figure 13 in exemplary embodiment of the present invention;

Figure 16 is that a part of top view of semiconductor device shown in the covered Figure 15 in specific region in exemplary embodiment of the present invention;

Figure 17 is the top view that exemplary embodiment of the present invention comprises second half conductor device part of eyespot;

Figure 18 is the top view of the part of semiconductor device eyespot shown in the covered Figure 17 in specific region in exemplary embodiment of the present invention;

Figure 19 is the top view that exemplary embodiment of the present invention comprises the semiconductor device of two eyespots;

Figure 20 is the top view that is welded to the semiconductor device part of lead frame in exemplary embodiment of the present invention;

Figure 21 is in exemplary embodiment of the present invention, to be the flow chart of the method for wire bond operation training eyespot; And

Figure 22 is the flow chart of wirebond machine method of work in exemplary embodiment of the present invention.

Embodiment

According to exemplary embodiment of the present invention, a kind of method is provided, for generating the pattern based on geometric properties, this pattern is for locating and/or alignment of semiconductor devices at wire welding equipment.The method defines a kind of technology, for the invariant features that utilizes pattern on semiconductor chip surface, generates eyespot.This method can improve the robustness of for example, finding described pattern on certain semiconductor device that changes (variation of surface reflectivity) having.In addition, can for example, with synthesising pattern (using available data as this as cad data), build the eyespot generated, also can from actual parts of images, extract generated eyespot.Be different from traditional PRS system (for example grayscale pattern coupling, geometrical pattern coupling), method proposed by the invention can be caught the invariant features in image effectively.Compare with routine techniques, for tackling the variation of automatic wire bond production environment device, use model (for example being stored in the memory of wirebond machine) that this invariant features (and the correlation between these invariant features) sets up robust more.

The eyespot that uses different illustrative methods disclosed herein to generate, can, as dominant eye point or the standby eyespot in PRS, for example, be used the pad information (or other geological information or shape information) obtained in the eyespot training process.Can select the eyespot (and zone of semiconductor chip) generated according to following manner: (1) for example,, by selecting a zone (zone that comprises pad vertical row and pad horizontal line) on the semiconductor device having good pad geological information, is selected automatically by PRS; And/or (2) are selected by operating personnel's intervention at least in part when selecting this zone with desired pad information.

Automatically generate in the example of eyespot at PRS, can for example, obtain by off-line programing instrument etc. the relative position of shape (for example, the bond pad shapes of any type, as rectangle, octagon, circle etc.) and shape from described device data (cad data).Also can after observing sample device, use vision system, automatically extract these shapes (and relative position of these shapes).If can obtain the numeral of described device chip sufficiently complete, (i.e. predetermined device data) described, just can select for covering the suitable shape/feature of described eyespot with a kind of algorithm, then with described algorithm, the shape/feature of selection is bound together (relative position that for example, uses described shape/feature).Use the method, can provide " without the vision training " to process (being the off-line programing method), less needs of this processing or do not need the wirebond machine self study.Certainly, may need to confirm training eyespot out with the sample device on this off-line procedure and wirebond machine.This without the vision training system in, can reduce the demand to the wirebond machine optical system.This is that multiplication factor can change sometimes, is trained because wirebond machine usually adopts complicated vision system; Yet, if use the data that are pre-existing in, to be trained without the vision mode, just can reduce the complexity (and cost) of this vision system.For example, in the conventional wirebond machine of using two multiplication factors (two optical systems, as " magnification at high multiple system " and " low power amplification system "), can from wirebond machine, save the magnification at high multiple system.This is because the magnification at high multiple system is generally used for obtaining the image detail regional by training, for example less (for example, size is equivalent to sub-pixel) gap between adjacent pad.According to some exemplary embodiment of the present invention, use " without vision " training method can not need these details cost reason of magnification at high multiple system (for example due to).

Utilizing during operating personnel intervene the example that generates eyespot, can and calibrate PRS and assist operating personnel by kinematic system accurately.For example, the parameter that kinematic system and PRS system can be based on being provided by operating personnel, carry out training managing (for example pad training managing).For example, when (1) can't obtain the suitable description of semiconductor device, and/or (2) can adopt this method when eyespot comprises while there is no the geometric properties of better definition.This training managing may need, and for example, moves to successively the shape/feature that each needs, and obtains the data that need from the image of the shape/feature of needs.Once complete the training to whole shape/feature, just can utilize the combination of trained shape/feature, and their area/location in the infrastructural frame of having set up respectively, set up eyespot.

No matter how eyespot generates, after training managing, in the PRS internal system, can obtain shape information, for example pad size, pad locations, shape, polarity and other attribute.In addition, for example, except described single shape information (single bond pad shapes information), from described kinematic system and/or use the calculating that described PRS system carries out, can also obtain the geometrical relationship of (for example, between these pads) between these shapes.

As described above, for example, in the training zone (eyespot), the synthetic shapes/pads created can replace the information (the namely image based on the dish model) of relevant true shape/dish.More particularly, for example, by using the data that relate to semiconductor device that can obtain (available data, as cad data), provide the shapes/pads of this synthetic establishment.The shapes/pads of this synthetic establishment can provide the better expression of average shape/dish, because it comprises whole common characteristics (and may there is no some defect of sample device, otherwise just can utilize these defects to carry out taught eyepoint) of a plurality of shapes/pads.In addition, can use suitable covering technique, to remove some zone of the selected scope of semiconductor device, for different devices, these zones are different.In addition, can also use a kind of technology, with respect to other zone of device, this technology gives the specific region (for example peripheral weld disk area of chip) of semiconductor device more to weigh, so that eyespot robust more.

The shape be included in the eyespot generated according to the present invention comprises, for example, bond pad shapes, the pad group that forms pattern, traditional eyespot shape (such as cross, circle, square etc.), there are the circuit of definition shape and/or the arbitrary surfaces feature on different chip etc.In addition, can also be by two or more these (or other) exemplary shape combined, in the eyespot generated to be included in.

Can comprise a lot of different schemes for the algorithm of setting up eyespot, for example: (1) can pre-determine the shape/feature of minimum number in algorithm, and when the shape/feature of this predetermined quantity is completed to training, algorithm can be transferred to another part of process, for example calculates the shape/feature of being trained with respect to infrastructural frame and/or mutual position; (2) be defined in the given shape/feature needed in training process, thereby if this feature location failure can cause defining the eyespot failure; (3) adopt weighting system, importance or the predictability separately based on it like this, for given shape/feature provides power, this weighting system is for defining the score value of processed semiconductor device; (4) for example, if select whole shape/feature that will be included in the needs in eyespot can, in the single visual field of vision system (image camera), can operate to locate eyespot with single searching; (5) if the feature needed can not be all in visual field, can locate each feature (for example each pad of eyespot), afterwards, by combining each data of location feature, can access whole training pattern (relative position that for example comprises each feature); And/or (6) apply multiple (may be redundancy) training algorithm in the training stage, to improve reliability.Within the scope of the invention, can adopt these schemes (etc.) combination in any.

Fig. 1 is the top view (the diagram part is to see from top to bottom, the lower left of semiconductor chip) of semiconductor device 100 parts.Semiconductor device 100 comprises a plurality of pads that are arranged in around device.For example these pads comprise

pad

100a, 100b, 100c, 100d, 100e, 100f, 100g, 100h, 100i, 100j, 100k, 100l and 100m.Semiconductor device 100 also comprises circuit region 102 and circuit region 104.As is known to the person skilled in the art, the circuit region on semiconductor device (such as semiconductor chip) can comprise wiring diagram, wire etc.

For process semiconductor device 100(on wirebond machine, be for example the pad of welding device 100 and the wire loop between the second welding position, wherein the second welding position can be the lead-in wire on the lead frame of installing device 100), before needing to use, be the eyespot of wirebond machine training, on wirebond machine, alignment of semiconductor devices 100.Certainly, this aligning need to be applied various other semiconductor processing techniques, comprises such as short column salient point, sheet to sheet wire bond, the end of a thread/wire loop inspection etc.

As mentioned above, in actual semiconductor device, different components can present different qualities, even they are identical devices.For example Fig. 2 is the top view of another semiconductor device 200 parts.Although be to be provided by the manufacturer different from device shown in Fig. 1 100, from the angle of wirebond machine, semiconductor device 200 remains the device identical with semiconductor device shown in Fig. 1 100.Similar to the pad of semiconductor device 100, semiconductor device 200 comprises a plurality of pads, for

example pad

200a, 200b, 200c, 200d, 200e, 200f, 200g, 200h, 200i, 200j, 200k, 200l and 200m.In addition, similar with 104 to the circuit region 102 of semiconductor device 100, semiconductor device 200 comprises circuit region 202 and 204.From the angle of vision system, the

circuit region

102 and 104 of semiconductor device 100 looks like single component or the zone with total physics presentation; Yet, because the surface of two devices (namely device 100 and device 200) is different, same vision system can be seen the circuit region of semiconductor device 200 202 and 204 to comprise many independent sectors as.For example as shown in Figure 2, circuit region 202 is regarded as and comprised

part

202a, 202b and 202c etc.Equally, circuit region 204 comprises

part

204a, 204b and 204c etc.

Therefore, when semiconductor device 100 and semiconductor device 200 are carried out to same operation, due to one of a plurality of potential reasons (such as surface color, superficial makings etc.), vision system can be regarded them as diverse.Now, suppose to use the part semiconductor device 100 that comprises

circuit region

102 or 104, the eyespot of wire bond operation is trained.For example, when use is processed (wire bond) semiconductor device 200 by the eyespot of semiconductor device 100 training, (for example the mark of semiconductor device 200, lower than threshold value, allows to accept this device and is further processed can to cause a lot of problems; The MTBA of semiconductor device 200 this devices is undesirably low etc.).

According to some exemplary embodiment of the present invention, can not change between device with eyespot and select eyespot into purpose, and not consider the difference such as surface color/texture.For example can select one group of shape as eyespot from a zone of semiconductor device.When using the i type semiconductor device for the wirebond machine taught eyepoint, define the position of each shape with respect to another.

For example Fig. 3 illustrates according to the eyespot of exemplary embodiment of the present invention and selects.Fig. 3 illustrates semiconductor device 30, comprises a plurality of pads, for

example pad

300a, 300b, 300c, 300d, 300e, 300f, 300g, 300h, 300i, 300j, 300k, 300l and 300m.In addition, similar with the circuit region 102/104 and 202/204 of semiconductor device 100/200 respectively, semiconductor device 300 comprises circuit region 302 and 304.Because the

circuit region

302 and 304 of different components may change (as confirmed by comparator device 100 and 200), therefore, in selected eyespot 310, ignore these zones.Eyespot 310 comprises pad group 306(mono-row pad, comprises

pad

300h, 300i, 300j, 300k, 300l and 300m) and pad group 308(a line pad, comprise

pad

300a, 300b, 300c, 300d, 300e, 300f, 300g).Some feature of eyespot 310 (contour shape of each pad for example, pad and the interval between another etc.) can between device, not change, therefore, can provide the training managing of robust more (with practical devices to the wanting wire bond follow up scan of robust more).

It should be noted that, although above take the zone in dotted line in eyespot 310(Fig. 3) eyespot is described as example, but should be appreciated that eyespot 310 can only comprise some shape/feature in zone, for example bond pad shapes and some pad relevant position in zone 310.The shown of the present invention a plurality of exemplary embodiments with description of this paper are exactly such.

In addition, although exemplary eyepoint 310 comprises the bond pad shapes shape of circuit region 302 and 304 (and do not comprise) of the pad of group in 306 and 308, can comprise one of

circuit region

302 and 304 or both contour shapes in eyespot.In this eyespot, potential inside difference between the circuit region that may exist between device and device (for example in Fig. 1, circuit region 102 is compared in Fig. 2 different between circuit region 202), can not affect training managing (and follow-up searching is processed), because only comprise the contour shape of described circuit region in this eyespot.

In some applications, solder pad of semiconductor device can comprise " probe mark ", and they are visible for the vision system of wirebond machine.This probe mark is not unified.For example, due to typical semiconductor wafer testing mode, the part probe mark may for example, extend along a direction (horizontal direction), and other probe mark may for example, extend along other direction (vertical direction).These variations of probe mark on pad, also can cause conventional eyespot training system to go wrong (such as finding problem that mark is lower, MBTA problem etc.).

According to some exemplary embodiment of the present invention, ignore/get rid of the interior zone of pad from eyespot, can fully eliminate like this problem for example, to some variation (probe mark) relevant.For example these zones can (1) be scanned, but do not consider in the processing procedure of scoring, (2) scanning, but give lower power in the processing procedure of scoring, and/or (3) are not scanned.Therefore, very clear, for example, when (shadow region, some zone of mentioning device, as pad inside or other shadow region) time, " do not comprise ", the saying (or similar saying) of " ignoring ", " eliminating " and/or " covering ", needn't mean that from scan process (scan process in training process for example, processing will by the scan process in the practical devices process of wire bond) get rid of described zone.

Therefore, Fig. 4 show some is selected (such as predetermined, dynamically determine in real time etc.) zone compares eyespot 310 from eyespot 310a(, with eyespot, 310a compares, eyespot 310 has some difference, and for example eyespot 310 is not ignored the interior section of pad from eyespot) in the semiconductor device 300 that neglects.These selected areas in Fig. 4, be shade/oblique line (namely from the eyespot training managing by shadow region " covering ", for example wherein (1) by algorithm configuration for not scanning these zones, and/or (2) be configured algorithm, so that edge/feature that " coverings " intra-zone exists does not affect gross score calculating).As shown in Figure 4, the inside of each pad is marked as shade, so that described inside is not included in taught eyepoint.More particularly, the inside 300a1 that ignores pad 300a from eyespot 310a.Inside 300b1 for pad 300b, the inside 300c1 of pad 300c, the inside 300d1 of pad 300d, the inside 300e1 of pad 300e, the inside 300f1 of pad 300f, the inside 300g1 of pad 300g, the inside 300h1 of pad 300h, the inside 300i1 of pad 300i, the inside 300j1 of pad 300j, the inside 300k1 of pad 300k, the inside 300l1 of pad 300l and the inside 300m1 of pad 300m, be also like this.

Also comprise the zone outside actual pad in eyespot 310a, for example, in order to be the pad (shape of pad and position separately in eyespot 310a) of (1) taught eyepoint 310a, (2) follow up scan eyespot 310a, reduce error, and the zone of these shapes of location is provided.More particularly, provide non-hatched area 306a around pad row 306, non-hatched area 308a is provided around pad rows 308.By these zones on every side are provided, can increase the possibility of the accurate outline position (with they mutual outline position) of each pad in taught eyepoint.

Fig. 5 is the top view of semiconductor device 500 parts in exemplary embodiment of the present invention, and semiconductor device 500 comprises the selected zone 510 that will comprise eyespot.Semiconductor device 500 comprises a plurality of pads.As shown in Figure 5, around semiconductor device 500, provide the first assembly welding dish (to comprise

pad

500a, 500b, 500c, 500d, 500e, 500f, 500g, 500h, 500i, 500j, 500k, 500l, 500m, 500n, 500o, 500p, 500q, 500r, 500s, 500t, 500u and 500v), first group of inside, provide the second assembly welding dish (to comprise pad 502a simultaneously, 502b, 502c, 502d, 502e, 502f, 502g, 502h, 502i, 502j, 502k, 502l, 502m, 502n, 502o, 502p, 502q, 502r, 502s, 502t, 502u, 502v, 502w, 502x, 502y, 502z and 502aa).Semiconductor device 500 also comprises reference point 504.

Fig. 6 is the selected eyespot that will comprise semiconductor device 500, the detailed view in the zone 510 of semiconductor device 500.Zone 510 comprises

pad

500a, 500b, 500c, 500d, 500e, 500f, 500k, 500l, 500m, 500o, 502g, 502h, 502i, 502j, 502k, 502l, 502m, 502n, 502o, 502p, 502q and reference point 504.It should be noted that pad 502n and 502o comprise for the vision system that carries out pad scanning visible certain not normal (such as stain, different reflectivity, visible flaw etc.).If semiconductor device 500(comprises that pad 502n and 502o's is not normal) be used as sample device, eyespot with the training wirebond machine, for example use so, while by the eyespot of semiconductor device 500 training, processing (wire bond) actual semiconductor device, can cause a lot of problems (for example by the mark of processed semiconductor device lower than threshold value, even can accept this device, be further processed; The MTBA of device such as semiconductor device 500 is undesirably low etc.).

According to some exemplary embodiment of the present invention, can be by the given data about semiconductor device (such as the computer scanning of cad data, drawing, drawing, available data etc.) for the auxiliary definition eyespot.For example can define eyespot by these data, otherwise, if train this eyespot with sample device, this eyespot may comprise some defect of i type semiconductor device.More particularly, given data (such as the computer scanning of cad data, drawing, drawing, available data etc.) can be for definition eyespot as shown in Figure 7.The zone 510 that it should be noted that semiconductor device 500 does not as shown in Figure 7 comprise that pad 502n and 502o's is not normal.Eyespot by the shape to comprise the zone 510 shown in Fig. 7 starts (described eyespot is used given data to be defined, and on wirebond machine before hands-on sample device 500), eyespot 510 by original definition for there is no some defect.After this, for example, when on wirebond machine, defined eyespot 510 being carried out to hands-on (use as shown in Figure 6 have in pad 502n and 502o not normal sample device 500), can ignore some defect in training managing, because in the training process on wirebond machine, the total information of confirmatory sample device (pad that for example eyespot 510 comprises and shape and the relative position of reference point) only.Therefore, by utilizing (1) given data about semiconductor device, the sample device of training on wirebond machine in conjunction with (2), can provide improved eyespot.

Fig. 8 illustrates the further improved eyespot 510a of semiconductor device 500.More particularly, the selected area of ignoring this part of semiconductor device 500 from eyespot 510a.These selected areas are marked as shade (namely can from the eyespot training managing " covering " described shadow region) in Fig. 8.As shown in Figure 8, the inside of each pad is marked as shade, so that described inside is not included in the eyespot of being trained.More particularly, the inside 500a1 that ignores pad 500a from eyespot 510a.For the inside 500b1 of pad 500b, the inside 500c1 of pad 500c, and the inside of each

pad

500d, 500e, 500f, 500k, 500l, 500m, 500n, 500o, 502g, 502h, 502i, 502j, 502k, 502l, 502m, 502n, 502o, 502p and 502q, too.Equally, the inside 504a1 of reference point 504 is also neglected from eyespot 510a.Other shadow region means for example between device and device, have the zone of inconsistent texture/reflectivity.Therefore, can not affect will be by the mark of the device of wire bond in described shadow region.

In addition, also comprise the zone outside actual pad (with the reference point shape) in eyespot 510a, for example, in order to be the shape/relevant position (shape of the pad of eyespot 510a and reference point and position separately) of the feature of (1) taught eyepoint 510a, (2) follow up scan eyespot 510a, reduce error, and the zone of the described shape in location is provided.More particularly, provide non-hatched area 500a2 around pad 500a, non-hatched area 500b2 is provided around pad 500b.For each

pad

500c, 500d, 500e, 500f, 500k, 500l, 500m, 500n, 500o, 502h, 502i, 502j, 502k, 502l, 502m, 502n, 502o, 502p and 502q provide (shown in Fig. 8) this non-hatched area.Equally, provide non-hatched area 504a2 around reference point 504.By the zone around these is provided, can increase the possibility of the accurate profile (with they mutual outline position) of each pad and reference point in taught eyepoint.

Fig. 9 is the top view of semiconductor device 600 parts in exemplary embodiment of the present invention, and semiconductor device 600 comprises the selected zone 610 that comprises eyespot.Semiconductor device 600 comprises a plurality of pads.As shown in Figure 9, in the first row on semiconductor device 600, the first assembly welding dish (comprising

pad

600a, 600b, 600c, 600d, 600e, 600f, 600g, 600h, 600i, 600j, 600k, 600l and 600m) is provided, the second assembly welding dish (comprising

pad

602a, 602b, 602c, 602d, 602e, 602f, 602g, 602h, 602i, 602j, 602k, 602l, 602m, 602n, 602o and 602p) is provided simultaneously in the second row.

Figure 10 is the selected eyespot that comprises semiconductor device 600, the detailed view in the zone 610 of semiconductor device 600.Zone 610 comprises

pad

600a, 600b, 600c, 600d, 600e, 602a, 602b, 602c, 602d and 602e.It should be noted that

pad

602c and 602d comprise certain not normal (such as stain, different reflectivity, visible flaw etc.), this is not normal is visible to the vision system of carrying out the scanning of this pad.If semiconductor device 600(comprises that

pad

602c and 602d's is not normal) be used as the eyespot of sample device with the training wirebond machine, for example, when using the eyespot of being trained by semiconductor device 600 to process (wire bond) actual semiconductor device, can cause a lot of problems (for example by the mark of processed semiconductor device lower than threshold value, allow to accept this device and be further processed; Undesirably low etc. such as the MTBA of semiconductor device 600 this devices).

According to some exemplary embodiment of the present invention, can be for the auxiliary definition eyespot about the given data of semiconductor device (such as the computer scanning of cad data, drawing, drawing, available data etc.).For example these data can comprise the eyespot for some defect of the i type semiconductor device of taught eyepoint for correction.More particularly, described given data (such as the computer scanning of cad data, drawing, drawing, available data etc.) can be for definition eyespot as shown in figure 11.The zone 610 that it should be noted that semiconductor device 600 does not as shown in figure 11 comprise that described pad 602c and 602d's is not normal.By starting with eyespot shown in Figure 11 610 (described eyespot is used given data to be defined, and on wirebond machine before hands-on sample device 600), eyespot 610 by original definition for there is no some defect.After this, for example, when on wirebond machine, defined eyespot 610 being carried out to hands-on (use as shown in figure 10 at pad 602c and 602d place, there is not normal sample device 600), can ignore some defect in training managing, because in the process of training on wirebond machine, the total information of confirmatory sample device (shape and the relative position of the pad that for example eyespot 610 comprises) only.Therefore, by utilizing (1) given data about semiconductor device, the sample device of training on wirebond machine in conjunction with (2), can provide improved eyespot.

Figure 12 is the schematic diagram of the further improved eyespot 610a of semiconductor device 600.More particularly, the selected area of ignoring this part of semiconductor device 600 from eyespot 610a.These selected areas are marked as shade (namely can from the eyespot training managing " covering " described shadow region) in Figure 12.As shown in figure 12, the inside of each pad is marked as shade, so that described inside is not included in the eyespot of being trained.More particularly, the inside 600a1 that ignores pad 600a from eyespot 610a.For the inside 600b1 of pad 600b, the inside 600c1 of pad 600c, and the inside of each

pad

600d, 600e, 602a, 602b, 602c, 602d and 602e, be also like this.

In addition, also comprise the zone outside actual pad in eyespot 610a, for example, in order to be the shape/relevant position (shape of the pad of eyespot 610a and position separately) of the feature of (1) taught eyepoint 610a, (2) follow up scan eyespot 610a, reduce error, and be provided for locating the zone of described shape.More particularly, provide non-hatched area 600a2 around pad 600a, and provide non-hatched area 600b2 around pad 600b.For each

pad

600c, 600d, 600e, 602a, 602b, 602c, 602d and 602e provide (shown in Figure 12) non-hatched area.By these zones on every side are provided, can increase the possibility of the accurate profile (with they mutual outline position) of each pad in taught eyepoint.

Figure 13 is the top view of semiconductor device 700 parts in exemplary embodiment of the present invention, and this semiconductor device 700 comprises the selected zone 710 that comprises eyespot.Semiconductor device 700 comprises a plurality of pads.As shown in figure 13, around semiconductor device 700, provide the first assembly welding dish (to comprise

pad

700a, 700b, 700c, 700d, 700e, 700f, 700g, 700h, 700i, 700j, 700k, 700l, 700m, 700n, 700o, 700p, 700q, 700r, 700s, 700t, 700u and 700v), first group of inside, provide the second assembly welding dish (to comprise pad 702a simultaneously, 702b, 702c, 702d, 702e, 702f, 702g, 702h, 702i, 702j, 702k, 702l, 702m, 702n, 702o, 702p, 702q, 702r, 702s, 702t, 702u, 702v, 702w, 702x, 702y, 702z and 702aa).

Figure 14 is the selected eyespot that comprises semiconductor device 700, the detailed view in the zone 710 of semiconductor device 700.Zone 710 comprises

pad

700a, 700b, 700c, 700d, 700l, 700m, 700o, 702a, 702b, 702c, 702m, 702n and 702o.It should be noted that pad 702n and 702o comprise certain not normal (such as stain, different reflectivity, visible flaw etc.), this is not normal is visible to the vision system of carrying out the scanning of this pad.If semiconductor device 700(comprises that pad 702n and 702o's is not normal) be used to train the eyespot of wirebond machine, for example, when using the eyespot of being trained by semiconductor device 700 to process (wire bond) actual semiconductor device, can cause a lot of problems (for example by the mark of processed semiconductor device lower than threshold value, allow to accept this device and be further processed; The MTBA of semiconductor device 700 this devices is undesirably low etc.).

According to some exemplary embodiment of the present invention, can be for the auxiliary definition eyespot about the given data of semiconductor device (such as the computer scanning of cad data, drawing, drawing, available data etc.).For example these data can be for revising eyespot, and described eyespot comprises some defect of the i type semiconductor device for training this eyespot.More particularly, described given data (such as the computer scanning of cad data, drawing, drawing, available data etc.) can be for definition eyespot as shown in figure 15.The zone 710 that it should be noted that semiconductor device 700 does not as shown in figure 15 comprise that described pad 702n and 702o's is not normal.By to comprise that eyespot shown in Figure 15 710 starts (described eyespot is used given data to be defined, and on wirebond machine before hands-on sample device 700), eyespot 710 by original definition for there is no some defect.After this, for example, when on wirebond machine, defined eyespot 710 being carried out to hands-on (use as shown in figure 13 at pad 702n and 702o place, there is not normal sample device 700), can ignore some defect in training managing, because in the process of training, only confirm the total information (shape and the relative position of the pad that for example eyespot 710 comprises) of described sample device on wirebond machine.Therefore, by utilizing (1) given data about semiconductor device, the sample device of training on wirebond machine in conjunction with (2), can provide improved eyespot.

Figure 16 is the schematic diagram of the further improved eyespot 710a of semiconductor device 700.More particularly, the selected area of ignoring this part of semiconductor device 700 from eyespot 710a.These selected areas are marked as shade (namely can from the eyespot training managing " covering " described shadow region) in Figure 16.As shown in figure 16, the inside of each pad is marked as shade, so that described inside is not included in the eyespot of being trained.More particularly, the inside 700a1 that ignores pad 700a from eyespot 710a.For the inside 700b1 of pad 700b, the inside 700c1 of pad 700c, and the inside of each

pad

700d, 700l, 700m, 700n, 700o, 702a, 702b, 702c, 702m, 702n and 702o, be also like this.

In addition, also comprise the zone outside actual pad in eyespot 710a, for example, in order to be the shape/relevant position (shape of the pad of eyespot 710a and position separately) of the feature of (1) taught eyepoint 710a, (2) follow up scan eyespot 710a, reduce error, and the zone of the described shape in location is provided.More particularly, provide non-hatched area 700a2 around pad 700a, and provide non-hatched area 700b2 around pad 700b.For each

pad

700c, 700d, 700l, 700m, 700n, 700o, 702a, 702b, 702c, 702m, 702n and 702o provide (shown in Figure 16) described non-hatched area.By these zones on every side are provided, can increase the possibility of the accurate profile (with they mutual outline position) of each pad in taught eyepoint.

Figure 17 is the top view of semiconductor device 800 parts in exemplary embodiment of the present invention, and semiconductor device 800 comprises the selected zone 810 that comprises eyespot.Semiconductor device 800 comprises a plurality of pads.As shown in figure 17, semiconductor device comprises

pad

800a, 800b, 800c, 800d, 800e, 800f, 800g, 800h, 800i, 800j and 800k.Semiconductor device 800 also comprises circuit element 804.

As shown in the empty frame line in Figure 17, the zone 810 of semiconductor device 800 comprises

pad

800a, 800b, 800c, 800d, 800h, 800i and 800j, and circuit element 804.Therefore, should be understood that, outside pad and reference point (as circuit element 804, it can be wiring diagram, wire figure etc.), can consider the part as eyespot by the shape of the part of semiconductor device (with position separately).

Figure 18 is the schematic diagram of the further improved eyespot 810a of semiconductor device 800.More particularly, the selected area of ignoring this part of semiconductor device 800 from eyespot 810a.These selected areas are marked as shade (namely can from the eyespot training managing " covering " described shadow region) in Figure 18.As shown in figure 18, the inside of each pad is marked as shade, so that described inside is not included in the eyespot of being trained.More particularly, the inside 800a1 that ignores pad 800a from eyespot 810a.For the inside 800b1 of pad 800b, the inside 800c1 of pad 800c, and the inside of each

pad

800d, 800h, 800i and 800j, and the inside 804a1 of circuit element 804a, be also like this.

In addition, in eyespot 810a, also comprise outside actual pad and circuit element 804 outside zone, for example, in order to be the shape/relevant position (shape of the pad of eyespot 810a and circuit element 804 and position separately) of the feature of (1) taught eyepoint 810a, (2) follow up scan eyespot 810a, reduce error, and the zone of the described shape in location is provided.More particularly, provide non-hatched area 800a2 around pad 800a, and provide non-hatched area 800b2 around pad 800b.For each

pad

800c, 800d, 800h, 800i and 800j, and provide (shown in Figure 18) this non-hatched area for circuit element 804.By these zones on every side are provided, can increase the accurate profile of (with the profile of circuit element 804) of each pad in the taught eyepoint possibility of (with they mutual outline position).

Figure 19 is the top view of semiconductor device 900.As shown in figure 19, two different eyespots (namely eyespot 910a and eyespot 910b) have been defined.Therefore, should be understood that described training of the present invention disclosed herein, by its a plurality of exemplary embodiments, is to use the mode of a plurality of eyespots, to be applied in the wire bond operation.

Although a plurality of accompanying drawing provided in this article just shows separately semiconductor chip, but be understood that, in this processing procedure of wire bond, semiconductor chip has been welded to supporting construction usually, for example, as welding (utilizing the chips welding of adhesive and so on) framework or substrate.Figure 20 is the semiconductor device 1000(semiconductor chip 1000 for example that is welded to lead frame 1050) top view of a part.

Semiconductor device 1000 comprises pad 1000a, 1000b, 1000c, 1000d, 1000e, 1000f, 1000g, 1000h, 1000i, 1000j and 1000k.Eyespot 1010 is chosen as and comprises pad 1000a, 1000b, 1000c, 1000d, 1000h, 1000i and 1000j.Lead frame 1050 comprises a plurality of " lead-in wires ", and described " lead-in wire " comprises lead-in wire 1050a, 1050b, 1050c, 1050d and 1050e.For example, in the process of processing semiconductor device (wire bond), the device that occurs sometimes to complete wire bond is placed on the situation on wirebond machine.For example this may be a kind of error.A part that is likely also this device has completed wire bond, but another part of this device still needs to carry out wire bond.Be starved of the device of knowing on wirebond machine and whether completed wire bond.If for example do not know, may make the device that has completed wire bond again suffer the wire bond of not expecting, in this case, may damage this device, or cause useless processing.

According to exemplary embodiment of the present invention, provide a kind of definite device whether to complete the method for wire bond.As shown in figure 20, show a plurality of wire loop 1002a, 1002b, 1002c, 1002h and 1002i.More particularly, wire loop 1002a(comprises the ball bond 1002a1 be soldered on pad 1000a) provide electrical connection between pad 1000a and lead-in wire 1050a; Wire loop 1002b(comprises the ball bond 1002b1 be soldered on pad 1000b) provide electrical connection between pad 1000b and lead-in wire 1050b; Wire loop 1002c(comprises the ball bond 1002c1 be soldered on pad 1000c) provide electrical connection between pad 1000c and lead-in wire 1050c; Wire loop 1002h(comprises the ball bond 1002h1 be soldered on pad 1000h) provide electrical connection between pad 1000h and lead-in wire 1050d; And wire loop 1002i(comprises the ball bond 1002i1 be soldered on pad 1000i) provide electrical connection between pad 1000i and lead-in wire 1050e.Use mode identification method and/or system, describe for example, during above for example, about the relative position (relative position of bond pad shapes and bond pad shapes) of training shapes and training shapes, can identify the shape of the part of wire loop.More particularly, can be by algorithm configuration for searching the pad of semiconductor chip, filament is welded on described pad determining whether.In the example shown in Figure 20, can be the shape of searching ball bond on pad (such as circular, oval etc.) by algorithm configuration, in order to know whether filament has been soldered on described pad.

The further feature of this aspect of the present invention also is provided in addition.For example can select eyespot (for example eyespot shown in Figure 20 1010), to be included in the wire bond processing procedure, will be by the pad of wire bond for the first time.With reference to Figure 20, pad 1000a can be by first pad by wire bond.Therefore, eyespot 1010 is chosen as and comprises this pad.Equally, using pattern recognition, to know whether filament has been soldered in the processing procedure of pad (for example, by find the shape of ball bond on pad), because selected eyespot comprises first pad that may complete wire bond, for example, so the possibility of the filament (wire loop 1002a) that discovery has been welded is very large.

In addition, can will determine that whether filament has been soldered to the method on the pad of semiconductor device, is integrated in the processing of taught eyepoint.For example in first step, can realize the training managing according to eyespot of the present invention for training.In second step, can train for training the different training managings of the second eyespot (for checking the eyespot of bonding wire silk).Next, when will be by the device of wire bond on described machine the time, use the first taught eyepoint can completing place to confirm and/or device is aimed at.Then, once complete position and/or the aligning of confirming device, can complete the second scanning, to check whether there is bonding wire silk.In this embodiment, described the first eyespot can be by inner hide (as shown in the Fig. 4,8,12,16 and 18) of described pad, simultaneously, described the second eyespot is not by the inner covering of described pad (down to the inside that can scan described pad, to obtain the ball bond shape).

In another optional embodiment, can train single eyespot, and complete will be by the single sweep operation of the practical devices of wire bond.In this embodiment, the inner at least a portion of pad described in eyespot can be not covered, so that can scan the inside of described pad, and to obtain the ball bond shape.

Although in conjunction with wire loop, to determining the method whether pad has completed wire bond, be described, should be appreciated that checked wire bond can have different forms, such as conductive salient point (such as the short column salient point etc.).

Figure 21～22nd, the flow chart of some exemplary embodiment according to the present invention.As understood by one of ordinary skill in the art, can ignore some step comprised at described flow chart; Can increase some additional step; And can according to shown in order, change the order of described step.

More particularly, the flow chart of Figure 21 shows in exemplary embodiment of the present invention, and training is for the method for the eyespot of wire bond operation.At step 2100 place, select one group of shape as eyespot from a zone of semiconductor device.Can comprise bond pad shapes, reference point shape, circuitry shapes etc. such as described shape group.At step 2102 place, use (A) i type semiconductor device or (B) at least one in the tentation data relevant with this semiconductor device, for wirebond machine is trained described eyespot.This training step comprises the position of each described shape of definition with respect to another shape.Described position can be defined as comprise in this group shape each around zone.Described eyespot can be defined as and comprise described shape group, hide the preset range (namely step 2104) in described zone simultaneously, so that remove described preset range from described eyespot.For example described covering step can comprise a kind of algorithm of use, to remove the position of the preset range in described zone from described eyespot.Described covering step can comprise the interior section of the selected pad of covering (1), and at least one in the part in the zone between (2) pad.At optional step 2106 places, for wirebond machine is trained the second eyespot.The shape of the part of the wire loop on the pad of the corresponding described semiconductor device of this second eyespot.The shape of the ball bond that for example shape of the part of described wire loop can corresponding described wire loop.

The flow chart of Figure 22 shows a kind of method that operates wirebond machine.Shown in the step 2200,2202, the step 2100,2102,2104 and 2106 of the flow chart shown in 2204 and 2206 corresponding Figure 21 of method.At step 2208 place, be configured to the first semiconductor device by wire bond, be directed into the precalculated position (for example solder joint) of wirebond machine.At step 2210 place, the vision system of use wirebond machine scans the selected portion of described the first semiconductor device, wherein the corresponding described eyespot of having been trained of selected portion (eyespot of for example training at step 2202 place).At step 2212 place, by described taught eyepoint is compared with the selected portion of described the first semiconductor device, for described the first semiconductor device is specified a percentage.After step 2212, before the wire bond operation, can determine the position that should adjust described the first semiconductor device.At step 2214 place, the result based on described taught eyepoint is compared with the selected portion of described the first semiconductor device at least in part, adjust the position of described the first semiconductor device.At step 2216 place, between described the first semiconductor pads and another welding position, create (for example wire bond) wire loop.Thereby, a kind of improved operation of wirebond machine is provided.

As understood by one of ordinary skill in the art, the first semiconductor device as described above is first device of mark after training managing not necessarily.Using term " first ", is only for this device and described sample device are distinguished mutually, for example, at step 2102 place.Should be appreciated that a plurality of devices can be labeled, scanning etc., in order to determine the acceptable mark, and the mark that cannot accept.

Although relate generally to the eyespot in the position of (1) definition/training sample device, and then (2) scanning will be by the corresponding position of the device of wire bond, to confirm that this will be can be received by the device of wire bond, invention has been described, but be not limited to this.In some applications, can occur, device (for example, at the solder joint place) is moved, or is unknown in the direction of solder joint place device.In this case, will, by the device of wire bond, can find eyespot (eyespot that namely according to arbitrary exemplary embodiment of the present invention, defines/train) by scanning.If for example eyespot comprises five (5) pads, described pad has reservation shape/size and the predetermined position with respect to another (for example being scheduled to by training managing),, by scanning this device and locating the feature of corresponding institute taught eyepoint, can determine will be by the position of the semiconductor device of wire bond.Certainly, train this eyespot to complete by arbitrary illustrative methods described herein, for example: (1) generates this eyespot (for example using available data to generate) automatically, (2) utilize operating personnel to intervene to generate this eyespot, and/or (3) scanning device comprises the zone (for example use the single scope of checking, use a plurality of set of checking scope etc.) of this eyespot.

Although relate generally to eyespot, can be used at device wire bond (or other processing, as short column plush copper, device inspection etc.) before, suitably location and/or alignment of semiconductor devices, present invention is described, is not limited to this.Training provided in this article also can be applied to the eyespot of training for multiple other application.For example typically, before the semiconductor chip wire bond is arrived to lead frame or other substrate, train this lead frame or substrate.In addition, in some applications, the semiconductor chip sheet can received between lead frame, trained this semiconductor chip or lead frame.Therefore, method described herein (for example shown Part Methods in Figure 21～22), also can be applied to the aligning of doing in chip then operates.Equally, a plurality of exemplary embodiment of the present invention also can be applied to about described and eyespot other application.

Although for specific embodiment, the present invention is carried out to example and description in this article, the present invention is not intended to be limited to shown details.And, do not departing from situation of the present invention, in described claim and equivalent scope, can carry out in detail multiple modification.

Claims

1. the method that is wire bond operation training eyespot said method comprising the steps of:

(1) from a zone of semiconductor device, select one group of shape as eyespot, wherein said one group of shape is selected from the invariant features of the pattern on described semiconductor device; And

(2) use and one of at least following to train described eyespot for wirebond machine: (a) i type semiconductor device, or (b) tentation data relevant with described semiconductor device; This training step comprises each position with respect to another in the described shape of definition.

2. the method for claim 1, wherein step (1) comprises and selects described group of shape, so that it comprises a plurality of bond pad shapes.

3. the method for claim 1, wherein step (1) comprises and selects described group of shape, so that it comprises a plurality of bond pad shapes and at least one reference point shape.

4. the method for claim 1, also comprise the steps:

(3) described eyespot is defined as and comprises described group of shape, hide the preset range in described zone simultaneously, so that described preset range is removed from described eyespot.

5. method as claimed in claim 4, wherein said covering step comprises that covering is one of at least following: the interior section of (1) selected pad, and the some parts in the zone between (2) pad.

6. the method for claim 1, wherein step (2) comprises the described position of each described shape is defined as to the zone comprised around each described shape.

7. the method for claim 1 also comprises step:

(3) train the second eyespot for described wirebond machine, described the second eyespot is corresponding to the shape of the part of the wire loop on the pad of described semiconductor device.

8. method as claimed in claim 7, the shape of the described part of wherein said wire loop is corresponding to the shape of the ball bond of described wire loop.

9. the method for claim 1, the eyespot of wherein training by described method is configured to the standby eyespot of wire bond operation.

10. a method that operates wirebond machine, described method comprises step:

(1) from a zone of semiconductor device, select one group of shape as eyespot, wherein said one group of shape is selected from the invariant features of the pattern on described semiconductor device;

(2) use and one of at least following for described wirebond machine, to train described eyespot: (a) i type semiconductor device, or (b) tentation data relevant with described semiconductor device; This training step comprises each position with respect to another in the described shape of definition;

(3) will be directed to by the first semiconductor device of wire bond the precalculated position of described wirebond machine; And

(4) use the vision system of described wirebond machine to scan the selected portion of described the first semiconductor device, selected portion is corresponding to the described eyespot of being trained.

11. method as claimed in claim 10, wherein step (1) comprises described group of shape of selection, so that it comprises a plurality of bond pad shapes.

12. method as claimed in claim 10, wherein step (1) comprises described group of shape of selection, so that it comprises a plurality of bond pad shapes and at least one reference point shape.

13. method as claimed in claim 10, also comprise the steps: afterwards and in described step (3) before in described step (2)

Described eyespot is defined as and comprises described group of shape, hide the preset range in described zone simultaneously, so that described preset range is removed from described eyespot.

14. method as claimed in claim 13, wherein said covering step comprises, hides one of at least following: the interior section of (1) selected pad, and the some parts in the described zone between (2) pad.

15. method as claimed in claim 10, wherein step (2) comprises the described position of each described shape is defined as to the zone comprised around each described shape.

16. method as claimed in claim 10, also comprise the steps:

(5) train the second eyespot for described wirebond machine, described the second eyespot is corresponding to the shape of the part of the wire loop on the pad of described semiconductor device.

17. method as claimed in claim 16, the shape of the described part of wherein said wire loop is corresponding to the shape of the ball bond of described wire loop.

18. method as claimed in claim 10, wherein in step (2), the described eyespot of training is configured to the standby eyespot that wire bond operates.

19. method as claimed in claim 10, also comprise the steps:

(5) by being compared with the selected portion of described the first semiconductor device by taught eyepoint described, for described the first semiconductor device distributes a percentage.

20. method as claimed in claim 10, also comprise the steps:

(5) at least in part based on by described result of being compared with the selected portion of described the first semiconductor device by taught eyepoint, adjust the position of described the first semiconductor device.

21. method as claimed in claim 10, also comprise the steps:

(5) create wire loop between the described first semi-conductive pad and another welding position.