KR20020027483A - Membrane probing system - Google Patents

Abstract

Preferably the substrate 200 is made of a soft material and the tool 210 having the desired shape of the resulting device for contacting the contact pads on the test device is in contact with the substrate. The tool is preferably made of a harder material than the substrate so that the recess 216 can be easily made in the substrate. Preferably, the dielectric (insulator) layer formed in a pattern is supported by the substrate. The conductive material is placed in the recess and then preferably wrapped to remove excess from the top surface of the dielectric layer to provide an overall flat surface. Traces are formed in a pattern on the dielectric layer and the conductive layer. The polyimide layer is then preferably formed in a pattern over the entire surface. The substrate is then removed by a suitable process.

Description

Membrane Probing System {MEMBRANE PROBING SYSTEM}

The trend of electronics is increasingly towards smaller geometries, particularly in the field of integrated circuit technology, where a large number of discrete circuit elements are fabricated on a single substrate or "wafer". After fabrication, the wafer is divided into " dies, " in which a plurality of rectangular shaped chips or respective dies represent different regular arrays of metallized contact pads through rectangular or input / output connections. Although each die is packaged separately as a result, for efficiency, testing of the circuits formed on each die is preferably performed while the dies are still bonded to each other on the wafer. One representative method is to support a wafer on a flat stage or "chuck" and move the wafer in the X, Y, and Z directions relative to the head of the probing assembly, thereby bringing the contacts on the probing assembly into a die so as to continuously engage each die. Move. Individual signal, power, and ground lines extend from the test instrument to the probing assembly and thus allow each circuit to be sequentially connected to the test instrument.

One conventional type of probing assembly used to test an integrated circuit provides a contact that is shaped like a needle-shaped tip. These tips are mounted around a central opening formed in the probe card to concentrate radially inward and downward through the opening. When the wafer is raised beyond the point where the pads on the wafer first contact these tips, the tips are bent upwards to slide forward across the individual pads thereby removing oxides formed on the pads.

The problem with this type of probing assembly is that the needle-shaped tips exhibit high inductance due to their narrow geometry, which results in large signal distortion in high frequency measurements made through these tips. In addition, these tips can operate in the manner of a planing tool when rubbed across individual pads, thereby causing excessive pad damage. This problem extends to the extent that, during use, the shape of the probe tip bends or fails to terminate in the common plane, causing the tip in front to be supported too severely down on the individual pad. In addition, it is impractical to mount these tips with a spacing of less than 100 microns or in a multi-row mesh pattern in order to accommodate the pad arrangement of more recent high density dies. In addition, this type of probing assembly has a scrub length of needle tip of 25 microns or more, which increases the difficulty of staying within the allowed probing area.

To reduce induction losses, to reduce pad wear, and to accommodate smaller device geometries, the second type of probing assembly developed uses a flexible membrane structure to support the probing contacts. In this assembly, the contoured lead lines are formed on one or more layers of flexible insulation film such as polyimide or MYLAR ™. If separate layers are used, these layers are combined with one another, for example, to form a multilayer transmission line structure. In this flexible structure or central portion of the membrane, each conductive line is terminated by individual probing contacts formed on the outer surface of the membrane and projecting outward from the outer surface of the membrane. These probing contacts are arranged in a predetermined pattern that matches the pattern of the device pad and are generally formed like raised bumps for probing a flat surface that is typically defined by the pad. The inner surface of the membrane is supported by the supporting structure. This structure has, for example, the shape of a pyramid cut off at the top, in which case the inner surface of the central part of the membrane is supported on the cut end of the support while the edge part of the membrane surrounds the pad on the device. It is tilted to remove the vertical component and pulled from the center portion.

In the membrane probing assembly just described, excessive line inductance is eliminated by careful selection of the geometry of the lead line, and preferably the photolithographic process is characterized by the size of the probing contact, Used to allow control over gaps and arrays. However, although several different forms of such probing assemblies have been proposed, these may be used to reduce pad wear and to achieve reliable removal of the oxide layer from each device pad to ensure proper electrical connection between the assembly and the device under test. We have experienced difficulties with this type of assembly.

One conventional form of membrane probing assembly is, for example, the apparatus described in EP 259,163 A2. The device has a central portion of the sheet-like membrane mounted directly against the rigid support. In turn, this rigid support is connected to the main body of the assembly by an elastic member comprising an elastomer or a rubber block, so that the membrane can tilt to match the slope of the device. US Pat. No. 4,918,383 to Huff discloses a closely related device, in which radially extending leaf springs allow for vertical axis movement of the rigid support, while preventing tilting, thereby preventing the "misalignment" of the contact or bumps on the pads. And also to remove surface oxides from these pads, the entire membrane moves slightly in the horizontal plane where the contacts "scrub" across the individual pads.

However, for both of these devices, due to manufacturing tolerances, some of the contact bumps are placed in concave positions relative to the adjacent ones and these concave bumps are displaced from the pad by the action of the adjacent one on the rigid support, which is satisfactory for engaging the pad. You don't have a chance. Moreover, even when the "scrub" motion is provided in the manner of Huff, the contact shows a tendency to frictionally adhere to the device, such as when performing a scrubbing motion. That is, in order to eliminate the effect of contact motion, the pads of the device tend to move in synchronization with the contact. Some scrubbing action actually depends on how far the pad can move, which depends on the degree of lateral clearance that exists as a result of the usual tolerance between the chuck and the individual support surface of the probe head. Therefore this type of membrane probing assembly does not guarantee a reliable electrical connection between each contact and pad.

A second conventional form of membrane probing assembly is disclosed in EP 304,868A2 to Barsotti. The device provides a flexible base for the contact support or center portion of the flexible membrane. In Barsotti's, the membrane is supported directly by the elastomeric member which in turn is supported by the rigid support, so that some height change between the contacts or pads can be tolerated. U.S. Patent 4,649,339 to Gangroth, U.S. Patent 4,636,772 to Ardezzone, Reed, Jr. As disclosed in US Pat. No. 3,596,228 to et al and US Pat. No. 5,134,365 to Okubo et al, respectively, using constant pressure air, negative pressure air, liquid or unsupported elastomers to provide a flexible support for the membrane. It is also possible. However, these modified devices do not have enough pressure between the probing contacts and the device pads to reliably penetrate the oxide formed on the pad surface.

In this second type of membrane probing assembly, the contacts can be limited to movement along the Z axis to prevent slippage between the contacts and pads and resulting misalignment during engagement, as pointed out in Okubo's patent. Therefore, although it is also possible to mount the support for Z-axis motion in the manner disclosed in Huff's U.S. Patent 4,980,637, the rigid support placed under the elastomeric member in Barsotti is fixed in place. In general, however, there is a certain amount of tilting between the contact and the device, and the closest incline to the device produces a much higher contact pressure than the inclined away, so pad damage does not occur in this type of design. do. Although the property of the elastomeric member in Garretson's device is to press the contacts in lateral motion when these contacts are placed in pressure engagement with the pad, the same problem occurs with the related assemblies disclosed in Garretson's EP 230,348A2. Another related assembly is disclosed in US Pat. No. 4,975,638 to Evans, which uses a pivotally mounted support to support the elastomeric member to receive a tilt between the contact and the device. However, Evans's device experiences the friction sticking problem already described in that the pad of the device easily sticks to the contact as the support pivots to move the contact laterally.

Another form of conventional membrane probing assembly is disclosed in U.S. Patent 5,395,253 to Crumly, U.S. Patent 5,059,898 to Barsotti et al, and U.S. Patent 4,975,638 to Evans et al. In the Crumly patent, the central portion of the stretchable membrane is elastically pressed in a fully pulled state using a spring. When the contacts engage individual pads, the pulled center portion contracts in a partially relaxed state against it, pulling the contacts toward the center of the membrane in the radial scrub direction. In Barsotti's, each row of contacts is supported by the ends of the individual L-shaped arms so that when the contacts line up with the individual pads, the corresponding arms deform upwards and individual rows of contacts simultaneously simultaneously. Scrub across the pads. But in Crumly and Barsotti's case, if there is some tilt between the contact and the device at engagement, this tilt makes the scrub closest to the device more scrub than tilting away. In addition, shorter contacts are forced to move in the scrub direction before the contacts have the opportunity to engage individual pads by controlling the scrub motion of adjacent contacts. Another disadvantage of the Crumly device is that the scrub effectiveness varies with contact location, especially since contacts closer to the center of the membrane scrub less than those closer to the perimeter.

In US Pat. No. 5,355,079 to Evans et al, each contact constitutes a spring metal finger, and each finger is mounted so as to extend in a cantilever manner away from the underlying membrane at an angle to the membrane. Similar forms are disclosed in US Pat. No. 5,521,518 to Higgins. However, especially where high density patterns are required, it is difficult to initially place these fingers so that the fingers all terminate in a common plane. In addition, these fingers are easily bent out of position in use and do not easily bend back to their original position. Therefore, certain fingers are touched down before other fingers, and the scrub pressure and distance are different for different fingers. At least in Evans', there is no suitable mechanism to allow some tilting between the finger and the pad. Although Evans proposes roughening the surface of individual fingers to improve the quality of the electrical connections, this roughness can cause unsuitable wear and damage to the pad surface. Another disadvantage of the contact fingers disclosed in both Evans and Higgins is that these fingers experience fatigue and fracture due to repeated bending and stress after a relatively low number of "touchdowns" or duty cycles.

Referring to FIG. 1, Cascade Microtech Incorporated, located in Beaverton, Oregon, developed a probe head 40 for mounting a membrane probing assembly 42. In order to measure the electrical performance of the particular die area 44 contained on the silicon wafer 46, the high speed digital line 48 and / or coated transmission line 50 of the probe head is tested by a suitable cable assembly. Connected to the input / output port of the equipment, the chuck 51 supporting the wafer moves in the X, Y, and Z directions perpendicular to each other to press-fit the pad of the die area with the contacts contained on the lower contacts of the membrane probing assembly. do.

The probe head 40 includes a probe card 52 on which data / signal lines 48 and 50 are arranged. 2 and 3, membrane probing assembly 42 includes a support element 54 formed of an incompressible material, such as a hard polymer. This element is removably connected to the upper side of the probe card by four allen screws 56 and corresponding nuts 58 (each screw penetrates the respective attachment arm 60 of the support element and is separated The backing element 62 distributes the clamping pressure of the screw evenly throughout the back of the support element). With this detachable connection, different probing assemblies with different contact arrangements can be quickly replaced with each other when necessary to probe another device.

3 and 4, the support element 54 includes a rear base portion 64 to which the attachment arm 60 is integrally coupled. Also included on support element 54 is an anterior support or plunger 66 projecting outward from the flat base portion. This anterior support has an inclined side 68 that concentrates toward a flat support surface 70 to provide an anterior support in the form of a truncated pyramid at the top. Also referring to FIG. 2, the flexible membrane assembly 72 is attached to the support after being aligned by an alignment pin 74 included on the base portion. This flexible membrane assembly is formed by one or more layers of insulating sheet, such as KAPTON ™ or other polyimide film sold by Du Pont de Nemours, with a flexible conductive layer or strip provided between or over the layers. To form data / signal line 76.

As shown in FIG. 3, when the support element 54 is mounted on the upper side of the probe card 52, the front support 66 is flexible membrane assembly in a suitable position for pressure engagement with the pad of the device under test. It protrudes through the central opening 78 of the probe card to represent the contacts arranged on the central area 80 of the probe card. With reference to FIG. 2, the membrane assembly includes radially extending arm segments 82 separated by inwardly curved edges 84 that provide an intersecting form to the assembly, these segments being inclined sides 68. It extends in an oblique manner along which removes any vertical component surrounding the pad. A series of contact pads 86 terminate the data / signal line 76 so that when the support element is mounted, these pads are electrically connected to the contacts on the center area so that the data / signal line 48 on the probe card is electrically connected. It is electrically engaged with a corresponding termination pad provided on the upper side.

The nature of the probing assembly 42 is a rather dense arrangement of contact pads over a number of contact cycles in such a way as to provide a generally reliable electrical connection between the contact and the pad in each cycle despite the generation of oxides on the pad. Is the ability of the assembly to probe. This capability is a function of the support element 54, the configuration of the flexible membrane assembly 72, and the manner of interconnection. In particular, when the membrane assembly is constructed and connected to the support elements and is in pressure engagement with these pads, the contacts on the membrane assembly preferably wipe or scrub across the pads in a locally controlled manner. Preferred mechanisms for exhibiting this scrubbing action are described in connection with the construction and interconnection of the preferred membrane assembly as shown in FIGS. 6 and 7A-7B.

6 shows an enlarged view of the central region 80a of the membrane assembly 72a. In this embodiment, the contacts 88 are arranged in a square pattern suitable for engaging a square arrangement of pads. Referring also to FIG. 7A, which shows a cross-sectional view taken along lines 7a-7a of FIG. 6, each contact points a relatively thick rigid beam 90 at one end where a rigid contact bump 92 is formed. It is included. The contact bump includes a contact 93 composed of a nub of rhodium fused thereon. Using electroplating, each beam is formed with a connection that overlaps an end of the flexible conductive trace 76a to form a junction. The conductive traces associated with the back conductive layer 94 effectively provide controlled impedance data / signal lines to the contacts because their dimensions are set using photolithographic processes. The back layer preferably comprises openings, for example, to assist with gas evacuation during manufacturing.

The membrane assembly is interconnected to the flat support surface 70 by an intervening elastomer layer 98, which has the same width as the support surface and is made by ELMER'S STICK-ALL ™ or Dow Corning Corporation manufactured by Borden Company. It can be formed by a silicone rubber compound such as Sylgard (182). This compound may be conveniently applied in the form of a paste that cures when set. As mentioned earlier, the flat support surface is made of an incompressible material and is preferably a rigid dielectric such as polysulfone or glass.

According to the above-described configuration, when one of the contacts 88 is in pressure engagement with the respective pads 100 as shown in FIG. 7B, the result of off centering of the rigid beam 90 and bump 92 structures The normal force causes the beam to pivot or tilt against the elastic recovery force provided by the elastomeric pad 98. This tilting motion is locally concentrated in view of the forward portion 102 of the beam moving a greater distance than the rear portion 104 of the same beam towards the flat support surface 70. As shown in FIG. 7B with dashed lines and solid lines indicating the start and end positions of contact on the pad, this effect is to move the contact to lateral scrubbing motion across the pad. In this form, insulating oxides on each pad are removed to ensure proper electrical connection of the contacts and pads.

The dashed line in FIG. 8 shows the relative position of the contact 88 and the pad 100 at the moment of initial engagement or touchdown, and the solid line is shown in a vertical direction in a straight line towards the flat support surface 70. The same element is shown after the "overtravel" of the pad by the distance 106. As shown, the distance 108 of the lateral scrubbing operation depends entirely on the vertical deflection of the contact 88 or the overtravel distance 106 moved by the pad 100. Therefore, since the overtravel distance for each contact point on the center area 80a is substantially the same (the difference resulting from the change in contact height), the distance of the lateral scrubbing operation by each contact point on the center area is substantially uniform. In particular, it is not affected by the relative position of each contact on the central area.

Since the elastomer layer 98 is retained by the incompressible support surface 70, the elastomer layer exerts resilience for each tilting beam 90 and each contact 93 so that the pressure of the contacts and pads during scrubbing Keep it. At the same time, the elastomer layer accommodates slight height changes between the individual contacts. Thus, referring to FIG. 9A, when a relatively short contact 88a is located between a pair of relatively large contacts 88b immediately adjacent and these large contacts are engaged with individual pads, as shown in FIG. 9B, The deformation by the elastomer layer allows the small contact to engage the pad after slightly overtraveling by the pad. It can be seen that in this example the tilting action of each contact is locally controlled and in particular the larger contact can be tilted independently of the smaller contact, so that the smaller contact is not forced into lateral motion until it is actually touched down on the pad. have.

Referring to FIGS. 10 and 11, an electroplating process for constructing a beam structure as schematically shown in FIG. 8 defines a substrate material attached thereon, such as support surface 70 and elastomer layer 98. An incompressible material 68. Using flexible circuit construction techniques, flexible conductive traces 76a are formed in a pattern on the sacrificial substrate. Next, a polyimide layer 77 is formed in a pattern to cover the entire surface of the sacrificial substrate and trace 76a, except for a predetermined position of the beam 90 on a portion of the trace 76a. Beam 90 is then electroplated in the opening of polyimide layer 77. Thereafter, a photoresist layer 79 is formed in a pattern on the surface of the polyimide 77 and the beam 90, leaving an opening for the predetermined position of the contact bump 92. Contact bumps 92 are then electroplated in openings in photoresist layer 79. The photoresist layer 79 is removed and a thicker photoresist layer 81 is formed in a pattern to cover the exposed surface except for a predetermined position for the contact 93. Contact 93 is then electroplated in the opening of photoresist layer 81. The photoresist layer 81 is then removed. The sacrificial substrate layer is removed and the remaining layers are attached to the elastomer layer 98. As illustrated more precisely in FIG. 12, the resulting beam 90, contact bumps 92, and contacts 93 provide independent tilting and scrubbing functions of the device.

Unfortunately, the construction techniques mentioned above result in structures with many unwanted properties.

First, several beams 90, contact bumps 92, and contacts 93 (each of which may be referred to as devices) in close proximity to one another exhibit locally different current densities in the electroplating bath, which in turn means that multiple beams ( 90), the difference in height of the contact bump 92 and the contact portion 93. In addition, different densities of ions in the electroplating bath and “random” changes in the electroplating bath also result in differences in the heights of the multiple beams 90, contact bumps 92, and contacts 93. The different heights of the multiple beams 90, contact bumps 92, and contacts 93 are three times greater than the overall height of the multiple devices. Thus, many devices will have significantly different heights than other devices. Using a membrane probe with variable device height requires a higher pressure than required if all devices have the same overall height to ensure that all contacts 93 are in proper contact with the test device. do. For high density membrane probes, such as 2000 or more devices in a small area, the additional pressure accumulation effect required for each device may exceed the total force allowed for the probe head and probe station. Excessive pressure may also result in bending or breakage of the probe station, probe head, and / or membrane probing assembly. In addition, because of the increased pressure required to make proper contact with the device with the lowest height, the device with the largest height may damage the pad on the test device.

Second, the ability to reduce the pitch (gap) between devices is limited by the "machine" effect of the electroplating process beyond the edges of the polyimide 77 and photoresist layers 79 and 81. The “machine” effect is difficult to control and results in variable widths of the beam 90, contact bumps 92 and contacts 93. If the height of the beam 90, contact bump 92 or contact 93 is increased, the "machine" effect generally increases, thus increasing the width of the individual portions. In general, the increased width of a portion results in an overall wider device which in turn increases the minimum spacing between the contacts 93. Alternately, reducing the height of the beam 90, contact bump 92 or contact 93 generally reduces the "machine" effect, which in turn reduces the minimum spacing between the contacts 93. However, if the height of the contact 93 relative to the individual beam 90 is sufficiently reduced, during use the rear end of the beam 90 will sufficiently tilt the test device in an acceptable position, ie outside the contact pad. And contact.

Third, it is difficult to plate the second metal layer directly on top of the first metal layer, such as the contact 93 on the contact bump 92, especially when using nickel. In order to provide a bond between contact bump 92 and contact 93, an interface seed layer, such as copper or gold, is used to achieve improved interconnection. Unfortunately, the interface seed layer reduces the lateral strength of the device due to the lower strength of the interface layer.

Fourth, attaching the photoresist layer to the non-uniform surface essentially results in non-uniform thickness of the photoresist layer itself, which tends to exhibit quasi-matching. With reference to FIG. 13, the photoresist layers 79, 81 on the raised portion of the beam 90 tend to be thicker than the photoresist layers 79, 81 on the bottom of the polyimide 77. Moreover, the thickness of the photoresist 79, 81 is likely to change depending on the density of the beam 90. Thus, the photoresist layers 79 and 81 in the region of the membrane probe with more compact device spacing are on average thicker than the region of the membrane probe with less compact device spacing. During the exposure and etching process of the photoresist layers 79, 81, the duration of the process depends on the thickness of the photoresist 79, 81. In the case of having a variable photoresist thickness, it is difficult to properly process the photoprocess to provide uniform openings. In addition, thinner regions of the photoresist layer 79 or 81 are susceptible to overexposure resulting in openings of varying sizes. Also, the larger the photoresist layer thickness 79 or 81, the larger the change in thickness. Thus, the use of photoresists presents a number of process issues.

Fifth, a separate alignment process requires aligning the beam 90 on the trace 76a, the contact bump 92 on the beam 90, and the contact 93 on the contact bump 92. Do. Each alignment process has its own variables that must be considered in creating the size of each part. The minimum size of the contact 93 is basically defined by the lateral strength requirements and the maximum allowable current density. Because of the tolerance of the alignment, the minimum size of the contact 93 in turn defines the minimum size of the contact bump 92 so that the contact 93 is limited on the contact bump 92. In view of the contact 93 and because of the tolerance of the alignment, the contact bump 92 is limited on the beam 90 because the minimum size of the contact bump 92 defines the minimum size of the beam 90. Thus, the sum of the tolerances of the contact bump 92 and the contact 93 together with the minimum size of the contact 93 defines the minimum device size and thus the minimum pitch between the contact pads.

Therefore, preference is given to membrane probe construction techniques and structures that exhibit more uniform device height, reduced spacing between devices, maximized lateral strength, desired shape, and proper alignment.

FIELD OF THE INVENTION The present invention relates to probe assemblies of the type commonly used for testing integrated circuits (ICs), and in particular the present invention provides good electrical connection between the probing assembly and each device by reliably removing surface oxides present on the conductors. Membrane probing assemblies have contacts that scrub across individual input / output conductors of each device in a locally controlled manner to ensure.

1 is a perspective view of a membrane probing assembly bolted to a probe head and a wafer supported on a chuck in a suitable position for probing by the assembly.

FIG. 2 is a bottom view showing various parts of the probing assembly of FIG. 1 including a support element and a flexible membrane assembly, and a cross-sectional view of a probe card with data / signal lines connected with corresponding lines of the membrane assembly. FIG.

3 is a side view of the membrane probing assembly of FIG. 1 with the portion of the membrane assembly cut away such that the invisible portion of the support element is exposed.

4 is a plan view of an exemplary support element.

5A-5B are schematic side views illustrating how the support element and membrane assembly tilt to match the orientation of the device under test.

FIG. 6 is an enlarged plan view of the central region of the membrane assembly structure of FIG. 2. FIG.

7A-7B are cross-sectional views taken along line 7a-7a of FIG. 6, first showing the contacts before touchdown and then showing the scrub motion across the individual pads and the same contacts after touchdown. to be.

FIG. 8 is a schematic side view showing the contacts of FIGS. 7A-7B in dashed lines at the first touchdown moment, and in solid lines the contacts after vertical overtravel by the pads.

9A and 9B illustrate deformation of an elastomeric layer that makes contact with a pad.

10 is a longitudinal cross-sectional view of the device of FIG. 8.

11 is a cross-sectional view of the device of FIG. 8.

12 is a more precise pictorial view of the device shown in FIGS. 10 and 11.

FIG. 13 is a detailed view of the device shown in FIG. 11 showing a non-uniform layer generated during the process.

14 is a drawing diagram of a substrate.

15 is a pictorial view of an exemplary embodiment of the tool of the present invention, in particular the dampling tool.

FIG. 16 is a drawing illustrating the tool of FIG. 15 in contact with the substrate of FIG. 14.

17 is a drawing of the substrate of FIG. 14 after the tool of FIG. 15 contacts the substrate.

18 is a cross-sectional view of the substrate of FIG. 14 with a polyimide layer supported on the substrate.

19 is a pictorial view of the tool of FIG. 16 with a Z-axis stop.

20 is a cross-sectional view of the substrate of FIG. 14 with traces, recessed conductive material, and additional polyimide layers.

21 is a drawing of the device of FIG. 20 with the substrate removed and inverted.

22 is a cross-sectional view of the contact portion of FIG. 21.

Fig. 23 is a schematic diagram showing an arrangement of the device of the present invention.

24 is a schematic diagram showing oxide layers of solder bumps and contacts of conventional contacts.

25 is a top view of another device with an elongated probing portion.

FIG. 26 is a side view of the device of FIG. 25 with an elongated probing portion. FIG.

27 is a pictorial view of a solder bump with a mark as a result of the device of FIGS. 25 and 26.

28 is a pictorial view of another probing device.

29 is a pictorial view of another probing device suitable for solder bumps.

30 is a side view of a Kelvin connection using the device of the present invention.

The present invention preferably overcomes the above mentioned disadvantages of the prior art by providing a substrate composed of a flexible material. A tool with the desired shape of the resulting device for contacting the contact pad with the test device is brought into contact with the substrate. The tool is preferably made of a material stronger than the substrate so that depressions can be easily made in the substrate. Preferably the dielectric (insulating) layer on which the pattern is formed is supported by the substrate. The conductive material is located in the depression and is then planarized to preferably remove excess from the top surface of the dielectric layer and provide an overall flat surface. Traces are formed in a pattern on the dielectric layer and the conductive material. The polyimide layer is then preferably formed in a pattern on the entire surface. The substrate is then removed by a suitable process.

The foregoing and other objects, features and advantages of the present invention will be more readily understood upon consideration of the detailed description in conjunction with the accompanying drawings.

The construction techniques currently employed for membrane probes include starting with a flat, rigid substrate to support additional layers assembled thereon. To reduce pitch and provide devices with increased uniformity, increasingly more complex and costly process technologies are required. In contrast to current technology of layering "upwardly" on a supporting substrate, the inventors have recognized that by using suitable tools, the substrate can be made to produce the desired beams, contact bumps and contacts. The remaining layers are then configured "top down" on the beam. The substrate itself is then removed.

Referring to FIG. 14, the substrate 200 is preferably made of a soft material such as aluminum, copper, lead, indium, brass, gold, silver, platinum or tantalum and preferably has a thickness of between 10 millimeters and 8/1 inch. Have As described below, the upper surface 202 of the substrate 200 is preferably polished for flat and optical transparency to enhance viewing.

Referring to FIG. 15, the tool, and in particular the “dimpled” tool 210, is configured with a head 212 having the desired shape of the resulting device for contacting the contact pad on the test device. The dimple tool 210 includes a protrusion 214 for connecting to a dimple machine (not shown). The tool 210 is supported by the dimple machine with the head 212 oriented to contact the upper surface 202 of the substrate 200. Preferably, the tool 210 is made of a harder material than the substrate 200, so that a dimple may be easily made on the substrate. Suitable materials for the tool 210 are, for example, tool steel, carbides, chromium and diamond. Preferred dimpled machines are probe stations capable of precise X, Y and Z control. It will be appreciated that any other dimpled machine can be used as well. Referring to FIG. 16, the tool 210 is pressed against the top surface 202 of the substrate 200 to remove the tool from the substrate 200 as shown in FIG. 17. There will be a matching depression 216. The tool 210 is used to create a plurality of depressions 216 in the substrate 200 that match the desired pattern, such as the pattern shown in FIG. 6. Conversely, the same depression 216 coincides with the shape of the tool 210 as the tool 210 remains stationary and the top surface 202 of the substrate is removed from the substrate 200 as shown in FIG. 17. The substrate 200 may be moved in the Z direction until it is pressed in contact with the tool 210 to make.

Referring to FIG. 18, the polyimide layer 220 is formed in a pattern around the depression 216. Likewise, it will be appreciated that any other suitable insulating or dielectric layer may be used. In the process of forming the polyimide layer 220 in a pattern, it is rather difficult to remove the polyimide from the depression 216 during the exposure and etching process for the polyimide layer 220. This is especially true when recess 216 is relatively deep with steeply inclined sides. Alternatively, polyimide layer 220 may be formed in a pattern on top surface 202 of substrate 200 having an opening located where depression 216 is desired. The tool 210 is then used to create a depression 216 in the substrate 200 through an opening provided in the polyimide layer 220. This alteration technique is applied to the polyimide layer 220 from the depression 216. Rule out a difficult process to fully remove).

In order to expose the polyimide layer 220, it is costly to manufacture a mask with sufficient tolerance to precisely align the openings for the depressions 216. The tool 210 in combination with the dimpled machine may be aligned with the actual location of one of the openings to expose and etch the polyimide layer 220 with a relatively inexpensive and somewhat inaccurate mask. The inventors have recognized that the locally concentrated area of the mask and the openings thereby created can be relatively well aligned for the purpose of dimpled. In addition, the areas of the mask that are spaced apart from each other are relatively poorly aligned for the purpose of dimpled. Thus, with a precise dimpled machine, automatically dimple the substrate 200 to match the expected pattern with many depressions 216 away from each other, ensures that the dimple tool accurately aligns with the opening in the area away from the initial alignment point. Will not be. In order to improve the accuracy of the alignment process, the inventors realigned the actual openings in the polyimide layer 220 at different locations, where the dimpled machine is far away, so that each locally concentrated area is relatively precisely aligned while the overall alignment is achieved. Recognized that it could be slightly off. In this way a relatively inexpensive mask can be used.

Preferably the dimpled machine comprises a precise Z axis motion so that the depth of each depression is the same or substantially the same. Referring to FIG. 19, if a sufficiently precise Z-axis movement is not possible, a modified dimpled toll 240 with a Z-axis stop 242 may be used. Z-axis stop 242 is a protrusion that extends outwardly from head 244 that lies on top surface 202 of polyimide 220 or top surface 202 of substrate 200. The Z-axis stop 242 is positioned relative to the head 244 to obtain an appropriate depth considering whether or not the polyimide layer 220 has been previously formed in a pattern before using the dimple tool 240.

Referring to FIG. 20, conductive material 250 is electroplated onto polyimide 220 and substrate 200, thereby filling recess 216 with conductive material 250, such as nickel and rhodium. It will be appreciated that any other suitable technique may be used to position the conductive material in the depression 216. The conductive material 250 is then preferably wrapped to remove excess from the top surface of the polyimide layer 220 and provide an overall flat surface. Preferred lapping processes are chemical-mechanical planarization processes. Trace 252 is formed in a pattern on polyimide layer 220 and conductive material 250. Trace 252 is preferably a good conductor such as copper, aluminum, or gold. Polyimide layer 254 is then formed in a pattern over the entire surface. Still other metal and dielectric layers can be formed. Substrate 200 is then removed by any suitable process, such as etching with hydrochloric acid (HCL 15%) or sulfuric acid (H ₂ SO ₄ ). Hydrochloric acid and sulfuric acid do not react with the polyimide layer 220 or the conductive material 250, such as nickel or rhodium. It will be appreciated that the polyimide layer 254 may alternatively be any suitable insulator or dielectric layer.

Referring to FIG. 21, the contact 260 of the resulting device is preferably selected to have a low contact resistance so that a good electrical connection can be made with the test device. Nickel has a relatively low contact resistance, while rhodium has a much lower contact resistance and is more wear resistant than nickel. Thus, depression 216 is preferably applied in a layer of rhodium. Using conventional processing techniques, the thickness of rhodium is limited to approximately 5 microns. The resulting device includes an outer layer of rhodium, and in particular a contact 260 which is then filled with the remaining conductive material, such as nickel or nonconductive filler. The conductive material need not fill the entire depression.

The aforementioned "bottom-up" construction process offers a number of advantages over conventional "bottom-up" processing techniques for layering on a support substrate. These advantages also give you the ability to configure devices with improved characteristics.

First, there is no limit on the resulting device height that has been imposed in advance by the limitation of photoresist processing. The ability to construct a device with any suitable height also alleviates the restrictions imposed by attempting to electroplat in the deep and narrow openings of the photoresist.

Second, the height of the contact portion 260 of the device is very uniform since the height is entirely limited only by the mechanical tooling process. Locally different current densities of the electroplating bath, different ion densities in the electroplating bath, and "random" changes in the electroplating bath are removed so as not to damage the overall height and shape of the resulting device. Having a substantially uniform device height, less force is required for the device to make proper contact with the test device, which in turn reduces the tendency of bending and breakage of the probe station, probe head and / or membrane probing assembly. In addition, the substantially uniform height of the device reduces the tendency of contact pads on the test device to be damaged by excessive pressure.

Third, the contacts 260 of the device are more robust because the device is constructed of a single homogeneous material during one adhesion process that does not require any boundary layer such as requiring multiple processing steps in advance. This allows to reduce the size of the contact not by the minimum shear force of the boundary layer but by the limitation of the maximum current density allowable during the test.

Fourth, the shape of the resulting device can be tailored to efficiently probe different materials. The shape of the device may have a steep sidewall angle, such as 85 °, while still providing mechanical strength, stability and integrity. The steep sidewalls allow narrower devices to be constructed that allow for higher densities of the device for progressively denser arrangement of contact pads on the test device. In addition, the angle of the sidewalls does not depend on (ie, is independent of) the crystal structure of the substrate.

Fifth, the shape of the contacts is known in advance and uniform between the devices to enable uniform contact with the contact pads of the test device.

Sixth, since each device was configured using the same tooling process, the alignment of the different portions of the resulting device is completely uniform between the devices. By precise alignment of the bottom of each device (beam and contact bump) to the contacts, there is no need to provide additional margin to accommodate the processing variables inherent in the photoresist process and the electroplating process. In addition, the "machine" effect of the electroplating process is eliminated and also reduces the required size of the device. The reduction and substantial removal of alignment variability of the different device 300 allows a significantly reduced pitch, which is suitable for contact pads on test devices with increased density, to be obtained.

Seventh, the shape of the resulting device can be long tailed to provide the best mechanical performance. In order to provide a scrubbing function as described in the prior art section, the device must have a beam and bump structure that tilts upon contact. The device 300 may include an inclined surface 304 between the tail 302 and the contact 260. The inclined surface 304 may have the tail 302 thinner than the head 306 to provide increased strength along the length portion of the device 300. The torque force applied to the device 300 during the tilting process of the device 300 is reduced across the length of the device 300 with the correspondingly thinner material defined by the inclined surface 304. For the thinner tail 302 and the material adjacent to the tail 302, there is less chance that the tail 302 of the device 300 impacts the test device if excessive tilting occurs. The improved form of the device 300 also reduces the amount of metal material needed.

Eighth, a "look up" camera is used to obtain an image of the lower portion of the membrane probe to determine the exact location of the device 300 relative to the contact pad on the test device. The use of a "look up" camera allows for automatic alignment of the membrane device with respect to the contact pads so that an automatic test can be executed. In order to obtain an image of the device 300 on the membrane probe, a "look up" camera generally uses light to illuminate the device 300. Unfortunately, conventional planarization processing techniques present relatively flat surfaces on beams, contact bumps and contacts, reflecting light back to the "lookup" cameras in a direction perpendicular to the "lookup" cameras, respectively. Light reflecting back from all surfaces to the "look up" camera often results in some confusion about the exact location of contact 260. The inclined surface 304 of the device 300 tends to reflect incident light away from the "lookup" camera disposed below, while the contact 306 reflects the incident light back to the "lookup" camera located below. There is a tendency. Light returned primarily from the contact 306 to the "look up" camera represents less potential confusion about the exact location of the contact.

Ninth, initial polishing of the upper surface 202 of the substrate 200 results in a smooth lower surface for the polyimide layer 220 formed in a pattern thereon. After etching or otherwise removing the substrate 200, the bottom surface of the polyimide layer 220 is smooth and the resulting polyimide layer 220 is generally optically transparent. Thus, the space between the trace and the metalized device 300 is relatively optically transmissive so that the operator positioning the device can easily see through the device between the trace and the device. This aids the operator in manually positioning the membrane probe on a device that is not otherwise visible. In addition, the pyramid shape of the device 300 allows the operator to more easily determine the exact location of the contact with respect to the contact pad on the test device that has been previously hidden by the wide beam structure (relative to the contact).

Tenth, referring to FIG. 22, the contacts 260 of the device are preferably configured to have an outer surface of rhodium 340, which may typically be effectively plated to a thickness of approximately 5 microns. The plating process of rhodium is quasi-matched, so the resulting layer is approximately 5 microns thick in the direction perpendicular to the outer sides 352,354. The width of the top 350 of the contact portion and the angle of the sides 352 and 354 of the tool 210 are selected such that the rhodium 340 plated on both sides 352 and 354 is preferably bonded to each other to form a V shape. The remainder of the device is preferably nickel. The thickness of rhodium 340 is only 5 microns in the vertical direction, while the thickness of rhodium 340 in the vertical direction from the top 350 of the device is greater than 5 microns. Thus, in a direction generally perpendicular from the top 350, the abrasion contact in use will last longer than if the top portion had been plated with rhodium of 5 microns thick.

Eleventh, the tissue of contact 260 may be selected to provide the described scrubbing effect on the contact pad of the test device. In particular, the tool may include a rough surface pattern on the corresponding contact to provide uniform tissue for all devices.

Twelfth, because the number of processing steps is reduced, the configuration technique of the present invention allows for relatively quick device configuration and substantial cost savings.

The above described construction techniques also provide several advantages over device types that are not impossible but difficult to configure in other ways.

First, if no scrubbing action is required, the tool can provide any shape desired, such as a simple bump.

Second, compared to the metal portion simply supported by the larger contact bumps, the inclined support side of the test device up to the contact 260 provides good mechanical support for the contact 260. By receiving this support from the inclined side, the contact can be smaller without the risk of falling off the device. When the device is tilted to penetrate the oxide formed on the surface of the contact pad, the smaller contact provides improved contact with the contact pad of the test device. In addition, the tail 302 of the device can be substantially thinner than the rest of the device, reducing the tendency of the tail 302 portion to impact the contact pad of the test device when the device is tilted during testing.

Third, the pressure exerted by the contacts of the device, given the predetermined pressure exerted by the probe head, can be changed by changing the center of rotation of the device. The center of rotation of the device may be selected by selecting the location / height of the contact relative to the length and length of the device. Thus, the pressure can be selected as desired to match the properties of two different contact pads.

Fourth, referring to Figure 23, the triangular shape of the device footprint provides high lateral stability of the device while allowing for a reduction in pitch between the devices. The contacts 403 of the device preferably provide many contacts of the test device. Aligned in a linear array for the pads. The triangular portions of the device are alternately aligned in opposite directions.

Fifth, the ability to construct a raised contact from the bottom surface of the device while still maintaining the uniformity of device height and structural strength allows the device to provide scrubbing motion while the bottom surface of the device requires less movement. do. Less movement of the bottom surface of the device to make good electrical contact during testing reduces the stress on the layer below the device bottom surface. Thus, the tendency to crack the polyimide layer and the contact traces is reduced.

When probing an oxide layer on solder bumps, such as solder bumps on a printed circuit board or solder balls on a wafer used with "flip-chip" packaging technology, the resulting oxide layer is difficult to penetrate effectively. Referring to FIG. 24, upon contact with conventional contacts of membrane probes on solder bumps, oxide 285 tends to press into solder bumps 287 with contacts 289 resulting in poor interconnection. When using conventional needle probes on solder bumps, the needles tend to slip on the solder bumps, bend down in the solder bumps, debris collect on the needles, and debris on the surface of the test device Cleaning of the probe takes time. In addition, the needle probe leaves uneven probe marks on the solder bumps. When probing the solder bumps used for flip-chip, the probe marks left in the upper part of the solder bumps tend to trap fluxes in them that are prone to explosion when heated, which destroys or degrades the interconnects. 25 and 26, an improved device configuration suitable for probing solder bumps is shown. The upper portion of the device includes a pair of steeply inclined sides 291, 293, preferably with polished sides, such as 15 ° away from the vertical. The inclined sides 291 and 293 preferably form a sharp ridge 295 at the top of the side. Since the angles of the sides 291 and 293 are chosen in terms of the coefficients of friction between the oxides on the solder bumps and the sides, the oxides covered on the surface are mainly slipped or sheared along the surfaces of the sides 291 and 293, and the device is exposed to the solder bumps. It does not remarkably move on the side when penetrating. Referring to Figure 27, the substantially sharp ridges also provide marks (lower regions) that extend across the solder bumps after contact. Subsequent heating of the solder bumps with the flux causes the flux to escape from the sides of the solder bumps, thereby avoiding the possibility of explosion. In addition, the resulting marks left on the solder bumps are uniform and allow the manufacturer of the solder bumps to consider the resulting marks in the design. It also requires less force to be applied to the device because it tends to slice through the solder bumps rather than in press contact with the solder bumps. The flatter surface 405 prevents slices too deep into the solder balls (bumps).

Referring to FIG. 28, a waffle pattern may be used to provide a larger contact area for testing solder bumps.

Referring to FIG. 29, the modified device preferably includes a pair of protrusions 311, 313 at the end of the arch 315. The spacing between the protrusions 311 and 313 is preferably smaller than the diameter of the solder bumps 317 to be tested. With this arrangement, the protrusions 311 and 313 strike the sides of the solder bumps 317, thereby leaving no mark on the upper portion of the solder bumps 317. Since it has a mark on the side of the solder bump 317, the flux used for the next time is trapped in the mark, and the tendency to explode becomes small. In addition, if the alignment of the device is not centered on the solder bumps 317, one of the protrusions 311 and 313 is still likely to hit the solder bumps 317.

Previous device configuration techniques have resulted in devices with contacts that are rather large and difficult to ensure alignment. Referring to FIG. 30, in an improved construction technique, the inventors have recognized that a membrane probe can be used to make a "true" Kelvin connection to a contact pad on a test device. The pair of devices 351 and 353 are aligned with the contacts 355 and 357 adjacent to each other. In this arrangement, one of the devices may be the "pressing" part while the other device may be the "detecting" part of the Kelvin test arrangement. Both contacts 355 and 357 contact the same contact pads on the test device. A more detailed analysis of Kelvin connections is described in Fink, D.G., ed., Electronics Engineers' Handbook, 1st ed., McGraw-Hill Book co., 1975, Sec. 17-61, pp. 17-25, 17-26, "The Kelvin Double Bridge", and US Patent Application No. 08 / 864,287, which are incorporated herein by reference.

Depending on the technology used, the desired use and the structure achieved, it should be noted that the advantages described above may appear at all or not all in the device constructed according to the invention.

Claims (1)

In the method of constructing a membrane probe, (a) providing a substrate; (b) creating a depression in the substrate; (c) placing a conductive material in the depression; (d) connecting a conductive trace to the conductive material; And (e) removing the substrate from the conductive material.