WO2012120245A1

WO2012120245A1 - Connection component provided with hollow inserts

Info

Publication number: WO2012120245A1
Application number: PCT/FR2012/050502
Authority: WO
Inventors: François Marion
Original assignee: Commissariat A L'energie Atomique Et Aux Energies Alternatives
Priority date: 2011-03-10
Filing date: 2012-03-09
Publication date: 2012-09-13
Also published as: FR2972569A1

Abstract

The invention relates to an electro-mechanical connection component (10) provided with conductive inserts (16) to be inserted into respective conductive pads of another connection component for a flip-chip-type hybridisation. Each insert (16) comprises: a metal core (24) that is non-oxidised on at least part of its surface and has a higher hardness than that of the pads; a first metal layer (26) that is non-oxidised on at least part of its surface, and covers at least said non-oxidised portion of the core (24), the first layer having a higher plasticity than that of the core; and a second layer (28) covering at least the first layer (26), over its non-oxidised part, and having a lower plasticity than that of the first layer.

Description

CONNECTING COMPONENT HAVING HOLLOW INSERTS

FIELD OF THE INVENTION The invention relates to the field of the connection of two components according to the technique of face-to-face hybridization, called "flip chip", and more particularly the connection of two electronic components by insertion of the metal type in metal inserts in pads at room temperature. The invention thus finds particular application in assemblies known as "chip on chip", "chip on wafer" and "wafer on wafer".

STATE OF THE ART To replace "flip chip" hybridizations by solder balls, it is known to provide on one face of a first electronic component inserts made of a hard metal, for example titanium nitride, and on one side a second electronic component pads made of a ductile metal, for example silver, and then hybridize the two components by cold inserting the inserts in the pads, thereby creating mechanical and electrical interconnections between the components.

However, a recurring problem in this type of "metal-in-metal" insertion hybridization is that, without particular measurement, the surface of the inserts oxidizes, which creates poor electrical connections between the inserts and the inserts. studs in which they are plugged.

A first solution used to avoid the insertion of oxidized inserts is to cover the inserts before their oxidation of a noble metal layer, and therefore non-oxidizable, such as gold or platinum. The inserts and their noble metal layer are thus inserted together in the studs without the appearance of oxide likely to affect the quality of the electrical connections.

However, this solution has a certain number of disadvantages, among which: ■ a high cost, on the one hand, because of the cost of the noble metals and, on the other hand, because of the complex and numerous steps to be taken in to cover only the inserts with a layer of such a metal; ^■ an impossibility of reducing the interconnection steps, that is to say the minimum space between two interconnections inserts / pads, if low-cost manufacturing techniques of the noble metal layer covering the inserts are used. In fact the low-cost manufacturing techniques consist in making a full-plate deposit of a noble metal layer on the face of the component comprising the inserts, then in then etching the noble metal layer present between the inserts. However, the only low-cost etching technique applicable to noble metals such as gold and platinum is a liquid-phase etching that currently does not allow the etching of surfaces smaller than 10 microns. Only etching or ionic machining now makes it possible to engrave interconnection steps of less than 10 micrometers. However, this technique has a very low efficiency especially due to the necessary cleaning between each deposit, and is therefore expensive;

^■ inter-diffusion and electro-migration of the noble metal covering the inserts.

Thus the interconnections end up being constituted by a complex multilayer formed of the material of the inserts, the noble metal and the material of the pads, which makes the interconnections very sensitive to solid / solid type diffusion, to the creation of holes of the type Kirkendall and holes at the interfaces of areas made of different metals; and

■ cross-contamination of gold. Indeed, gold is a very doping material for silicon usually present in electronic components. All manufacturing steps using gold must therefore be performed in manufacturing areas different from those where silicon is exposed. Another solution to prevent the inserts are introduced into the pads with a native oxide layer is to perform the hybridization under a deoxidation flow that removes the native oxide inserts before their introduction into the pads.

However, this solution is complex to implement, in particular because of the spreading and the cleaning of the residual flows after the insertion. Indeed, when a deoxidation flow is used to remove the native oxide, part of this flux, which also includes liquid and material residues, remains stuck between the interconnections. We must clean these residues. In fact, this technique also limits the reduction of the interconnection step. Indeed, at present, it is not possible to clean the residual flows present between the interconnections, and therefore forming short circuits between them, when the interconnections are spaced less than 10 micrometers. SUMMARY OF THE INVENTION

The object of the present invention is to propose a "flip chip" hybridization by inserting metal inserts in metal pads providing an electrical connection without the use of noble metals or deoxidation flux and allowing the realization of a hybridization at room temperature .

To this end, the subject of the invention is a process for producing an electromechanical connection component provided with conductive inserts each formed of a hollow cylinder intended to be inserted at ambient temperature into respective aluminum conductive pads of another connection component for face-to-face hybridization.

According to the invention, the method consists for each insert:

■ to achieve a metal core, not oxidized on at least a portion of its surface, and plasticity greater than that of aluminum;

^■ to produce a first aluminum metal layer, on at least a portion of its surface, covering at least said unoxidized portion of the core; and

^Performing oxygen oxidation of the aluminum of the first layer so as to create a second layer of native aluminum oxide on only a portion of the thickness of the first layer, and said second layer having a plasticity less than that of the first layer.

By "native oxide layer" is meant an oxide layer obtained by the natural oxidation of the metal when in contact with oxygen.

By "ambient temperature" is meant here a temperature which is distant from the melting temperature of the material constituting the first layer and the pads, for example temperatures of the order of 300 ° K, for which there is therefore not observed a important softening of this first layer and pads in which are intended to be inserted inserts. The "cold" or "room temperature" hybridization of the invention is thus distinguished from "thermo-compression" type hybridizations during which both a pressure and a heating are exerted, the heating being intended to soften or melt the pads to facilitate insertion of the inserts. In other words, under the effect of the insertion in a pad, the different regions of an insert undergo a deformation. Having a plasticity greater than that of the core, the first layer will therefore undergo a greater deformation than the latter during penetration into a stud. Since, on the other hand, the second native oxide layer has a plasticity lower than that of the first layer, this second layer can not deform as much as the first layer without breaking. Not being able to conform to the deformation undergone by the first layer, the second layer is "cracking".

In addition, the native oxide layer has the dual property of being very brittle and very little adhere to the metal from which it is derived. Under the effect of the penetration of Γ insert in the pad, there is thus observed a phenomenon of "ice on mud", that is to say that the native oxide layer crackles in the form of plates that slide on the first layer during insertion. The second layer is "peeled" and remains outside the pad. In addition, the native oxide layer has the advantage of having a very thin thickness, of the order of a few nanometers, completely defined by the nature of the metal. Thus whatever the time of exposure of the first layer to oxygen, the thickness of the second layer remains constant. In addition, the electronic component provided with inserts can therefore be stored under oxidizing conditions, such as air for example, without special precautions.

Thus, not only is the first layer not a noble metal but also in a preferred manner, oxidizable metals are sought for it so as to form a native oxide layer.

Aluminum has the advantage of being a material both ductile while maintaining a constant hardness for a wide temperature range because of its high melting temperature above 500 ° C. In addition, aluminum is a low cost material.

The advantage of the implementation of a hollow insert lies in the reduction of the bearing surface of the insert on the pad, and thus to facilitate insertion, or even to allow a cold insertion at room temperature . Because of the reduced bearing surface, the surface pressure exerted on the surface of the first and second layers bearing on the pad is also increased, which facilitates the deformation of the first layer, and corollary the cracking of the second layer. This also increases the effect of shearing and peeling aid of the second layer in case of poor adhesion thereof on the first layer. We note that the cylinder is the shape that optimizes these effects.

According to one embodiment, the pressure exerted on a bearing surface of each insert during their insertion into the pads is greater than 1800 megaPascals, which allows effective peeling of the oxide layer.

According to one embodiment,

^■ the core is made under a non-oxidizing atmosphere;

In that the first layer consists of a metal film of greater ductility than the metal of the core deposited in a non-oxidizing atmosphere.

According to one embodiment, the first layer covers substantially the entire core, and in that the first layer is oxidized over substantially its entire surface.

The invention also relates to a method for producing an electromechanical connection component provided with conductive inserts each formed of a hollow cylinder intended to be inserted at ambient temperature in respective conductive pads of another component of connection for face-to-face hybridization.

This process consists for each insert:

^■ to achieve a metal core, non-oxidized on at least a portion of its surface;

^■ to achieve a first metal layer, not oxidized on at least a portion of its surface, covering at least said unoxidized portion of the core, the first layer having a plasticity greater than that of the core; and

^■ to produce, under a non-oxidizing atmosphere, a second layer covering at least the first layer on its non-oxidized portion, the second layer being made of a material different from the material of the first layer or oxides of this material, the second layer having plasticity lower than that of the first layer.

In other words, under the effect of the insertion in a pad, the different regions of an insert undergo a deformation. Having a plasticity greater than that of the core, the first layer will therefore undergo a greater deformation than the latter during penetration into a stud. Since, on the other hand, the second layer has a plasticity lower than that of the first layer, this second layer can not deform as much as the first layer without breaking. Not being able to conform to the deformation undergone by the first layer, the second layer is "cracking". If the adhesion of the second layer to the first layer is weak, the second layer "peels" by sliding on the first layer during insertion and remains outside the pad, then fully discovering the first layer which is not oxidized and therefore a good conductor of electricity.

If the adhesion of the second layer to the first layer is strong, the second layer also penetrates into the stud while having cracks due to the plasticity differential with the first layer. The cracks thus define so many unoxidized electrical "paths" towards the first non-oxidized layer, and therefore good conductors of electricity, thus ensuring good electrical conduction of the interconnection formed by the insert and the pad.

This result is obtained independently of the oxidizable nature of the first layer which is therefore advantageously chosen from non-noble materials. It is also not necessary to use a deoxidizing flux during insertion since the electrical paths are formed, even if the second layer is oxidized.

According to one embodiment of the invention, the first layer is made of a metal selected from the group consisting of aluminum, tin, indium, lead, silver, copper, zinc and alloys. based on these metals. These materials are advantageously very plastic, and can be implemented in inexpensive etching processes exploitable for low interconnection steps of less than 10 micrometers, or even 5 micrometers. Aluminum also has the advantage of being a material both ductile while maintaining a constant hardness over a wide temperature range because of its high melting temperature above 500 ° C.

According to one embodiment of the invention, the adhesion of the second layer to the first layer is low so that the second layer slides on the first layer under the effect of a shear applied to the stack of first and second layers. In this way, during insertion of the insert in a pad, the second layer peels and remains outside the stud.

According to one embodiment, the second layer consists of a photosensitive epoxy resin, a polymer layer, such as parylene for example, a hard metal layer or a hard and brittle insulating layer such as SiO 2 or SiN According to one embodiment, the first layer has a ductility substantially equal to that of the pads, which facilitates the deformation undergone by the first layer during insertion, and thus also facilitates the cracking of the second layer. According to a preferred embodiment, the first layer substantially covers the entire core. The formation of the second layer thus uses conventional manufacturing methods, such as, for example, a full-plate deposit followed by etching of the metal deposited between the inserts. According to one embodiment, the insert is hollow.

The invention also relates to a method of electromechanical connection between a first micro-electronic component provided with inserts made according to one or the other of the aforementioned type of method and a second micro-electronic component provided with pads , in which the inserts of the first microelectronic component are able to be respectively inserted, said method comprising inserting the inserts of the first component into the pads of the second component.

In particular, the subject of the invention is also an electromechanical connection method for face-to-face hybridization between a first microelectronic component obtained according to the above-mentioned first method and a second micro-electronic component provided with aluminum conductive pads. wherein the inserts of the first micro-electronic component are respectively insertable, said method comprising inserting the inserts of the first component at ambient temperature into the pads of the second component.

The invention also relates to a microelectronic system obtained according to this method. BRIEF DESCRIPTION OF THE FIGURES

The invention will be better understood on reading the following description, given solely by way of example, and made with reference to the accompanying drawings, in which identical references designate identical or similar elements, and in which:

^■ Figures 1 and 2 are schematic sectional views of the hybridization of a first and a second microelectronic components by inserting inserts into pads; ^■ Figure 3 is a sectional view of an insert according to an embodiment of the invention;

^■ figure 4 is a schematic sectional view of Γ insert of Figure 3 along the line AA;

FIG. 5 is a diagrammatic sectional view illustrating the penetration of the insert of FIG. 3 into a ductile stud; and

^■ 6 to 13 are schematic sectional views illustrating an F insert manufacturing method of Figure 3. DETAILED DESCRIPTION OF THE INVENTION

FIGS. 1 and 2 schematically illustrate the "flip-chip" hybridization of a first and a second microelectronic component 10, 12. The first component 10 comprises, on one of its faces 14, a set of inserts 16 electrically conductive, intended to penetrate into respective electrically conductive pads 18, the pads 18 being arranged on a face 20 of the second component 12.

To carry out the hybridization preferably cold, the electronic components 10 and 12 are aligned so as to present each insert 16 in front of a stud 18, and an appropriate pressure, illustrated by the arrows, is for example exerted on the first component which is mobile (Figure 1). The inserts 16, which have a hardness greater than that of the pads 18, then penetrate into them. Interconnections 22 between the first and the second microelectronic components 10, 12 are thus produced (FIG. 2). The interconnections 22 mechanically join the two components 10, 12, while creating electrical connections therebetween.

By way of example, the first component 10 is a detection matrix consisting of a plurality of detection sensitive elements, and the second component 12 is a circuit for reading the sensitive elements. The interconnections 22 thus make the electrical connection of the read circuit with each of the sensitive elements of the first component 10.

The inserts 16 can take any shape, although the inserts having a reduced bearing surface, such as hollow cylinders for example, are preferred to reduce the pressure required for their insertion into the studs 18. In what follows, cylindrical and hollow inserts 16 will be described, this form constituting a preferred embodiment. However, it will be understood that the considerations relating to the constituent materials of the inserts 16 and pads 18 are independent of the form adopted for them. For example, the inserts may be solid and / or triangular, square, and more generally polygonal, star-shaped, etc.

Referring to FIGS. 3 and 4, the inserts 16 each comprise a central core 24 covered by a metal layer 26, the latter being itself covered by a protective layer 28.

The central core 24 is metallic, cylindrical and hollow, of U-shaped cross-section. It has a hardness greater than that of the pads 18 to be able to be inserted therein. For this purpose, the central core 24 preferably has a Young's modulus greater than 1.5 times the Young's modulus of the material of the studs 18. Advantageously, the central core 24 is made of a hard metal, such as nitride titanium (Ti), tungsten nitride (TiW), copper (Cu), vanadium (V), molybdenum (Mo), nickel (Ni), titanium tungstenate (TiW), WSi, or tungsten (W) for example, and the pads 18 are made of a ductile metal, for example aluminum, tin, indium, lead, silver, copper, zinc , or an alloy of these metals. Moreover, the central core 24 is not oxidized.

The metal layer 26, in addition to its function of being electrically conductive and strongly adhering to the central core 24 because of the metal-metal interface that it forms with the core 24, has the function of deform, while remaining attached to the core 24, during the penetration of Γ insert in a stud. For this purpose, it has a plasticity greater than that of the core 24. The layer 26 may thus consist of a ductile metal. In particular, a ductile metal having a Young's modulus greater than 1.5 times that of the material of the core 24 has a suitable plasticity.

Preferably, the layer 26 has a ductility substantially equal to that of the pads 18 so as to allow penetration of the hard core 24 without breaking and obtain relative deformations of the layer 26 and the pad 18 substantially equal. The layer 26 is thus advantageously made of aluminum, tin, indium, lead, silver, copper, zinc or an alloy of these metals. Moreover, the metal layer 26 is not oxidized. The layer 26 is preferably made of aluminum, this metal having the advantage of having a very high melting temperature greater than 500 ° C.

The primary function of the protective layer 28 is to protect the metal layer 26 from oxidation, and for the second function to release at least a portion of the metal layer 26 during insertion of the insert 16 into a stud 18 of the to create an electrical connection between the material of the pad 18 and the central core 24. To do this, the protective layer 28 is chosen to crack under the effect of the deformation of the metal layer 26. The protective layer 28 thus has a plasticity lower than that of the metal layer 26.

Preferably, the protective layer 28 is chosen so as to have a breaking point under very low deformation stresses, in other words it is very "brittle". The protective layer 28 may be a protective film attached to the metal layer 26, for example a photosensitive epoxy resin or a polymer layer such as parylene for example, or a hard metal layer or a layer of hard insulation and brittle such as SiO 2 or Si.

Preferably, the protective layer 28 consists of the native oxide of the constituent metal of the metal layer 26, which has the triple advantage:

to have a very thin protective layer 26 of the order of a few nanometers, to be hard and brittle, and in particular of plasticity and ductilities much lower than those of metal 26 itself, and

• adhere very weakly to the metal layer 26.

In addition, this embodiment has the advantage that it is not necessary to take special measures to prevent oxidation of the inserts during storage, since the inserts 26 are voluntarily allowed to be oxidized. FIG. 5, during the penetration of the insert 16 in the stud 18, a deformation, even a small one of the metal layer 26, breaks the oxide layer into plates, and under the effect of shearing, the plates of Native oxides slide on the metal layer 26 remaining outside the stud 18. The oxide layer 28 is thus "peeled" during insertion by exposing the metal layer 26, thereby creating a quality electrical connection, including without oxide. As mentioned above, the inserts 16 are hollow cylinders, preferably cylinders of revolution. The inserts have very small bearing surfaces S (FIG. 3 and 4), so as to be able to perform a cold insertion under ambient atmosphere, that is to say at an ambient temperature much lower than the melting temperature. pads 18, for example a temperature of about 300 ° K, and at atmospheric pressure. In addition to this, very low bearing surfaces have the effect of increasing the stresses exerted on the different regions of the inserts and therefore in particular the deformation and shearing forces, which facilitates the cracking of the protective layer 28 as well as its peeling in the case of a protective layer slightly adherent to the metal layer 26. It is advantageous to refer to document FR 2 928 033 for calculating the bearing surface for cold insertion under ambient atmosphere.

Advantageously, the pressure exerted on the bearing surface S during insertion of the insert comprising a first layer 26 of aluminum covered with a layer of native oxide 28 (alumina Al ₂ O ₃ ) in pads 18 Aluminum is greater than 1800 megaPascal (MPa). The inventors have indeed observed that for lower pressure values, the interconnections formed inserts 16 in the pads 18 have a high electrical resistance, which means that the peel of the oxide layer 28 is not sufficient. The inventors have, on the other hand, observed that for the above configuration of inserts and pads, pressures greater than 1800 megaPascal produce good quality interconnections, that is to say having an electrical resistance close to that of aluminum. which means that the oxide layer has been peeled optimally. Advantageously, the global insertion force, or equivalently the global insertion pressure, exerted on the circuits 10 and 12 to hybridize them, for example that exerted on the circuit 10 as illustrated by the arrows. in FIG. 1, and the bearing surface S of the hollow cylinders are thus chosen so as to obtain the said minimum pressure.

For example, a hollow cylinder with a diameter equal to 4μιη, with a wall thickness equal to 0.2 μιη has a bearing surface S equal to 2.512 μιη ² . When such an insert undergoes an insertion force of 5mN, the pressure exerted on its bearing surface S is equal to 1990 MPa. Knowing the overall insertion force and the number of interconnections between circuits

10 and 12, we therefore deduce the unitary insertion force experienced by each insert 16. Knowing the unitary insertion force, it is therefore possible to deduce a maximum support area to obtain at least the minimum pressure of 1800 MPa. Finally, the bearing surface S of a hollow cylinder is given by the relation S = 2 x π x (R ₂ - x R ₂ , where R ₂ - R _x is the thickness of the walls of the inserts 16 and 2 x R ₂ is the external diameter of the inserts 16 (Figure 4), it is easy to deduce pairs of thickness and diameter.The particular value choice for the thickness and the diameter can then be made according to other considerations, including considerations however thicknesses achievable depending on the manufacturing process used or considerations of the mechanical strength of the inserts.

11 will now be described in connection with the schematic figures in section 6 to 13, an example of a method of manufacturing inserts 16 just described.

The process starts with the deposition of a sacrificial layer 40 of a thickness e on the face 14 of the component 10, for example a resin layer of the polyimide type, followed by photolithography to produce circular holes 42 in the sacrificial layer. 40 to the face 14 of the component 10 (Figure 6). The thickness e corresponds to the desired height for the core 24 of the inserts and the diameter of the circular holes 42 corresponds to the outside diameter of the core 24.

The process is followed by the full-plate deposition of a hard metal layer or multilayer 44, for example titanium nitride or a titanium nitride-based alloy, of a thickness corresponding to the thickness of the metal. 24. The deposit is for example a chemical vapor deposition, or deposit "CVD" (for the acronym "Chemical Vapor Deposition") made at a temperature compatible with the electronic micro elements of the component 10, in particular a temperature below 425 ° C for a component 10 implementing a CMOS technology (FIG. 7).

A withdrawal of the portion of the hard metal layer 44 deposited between the holes 42 is then performed, for example using a "damascene" etch or "gap wire" well known per se. For example, according to the etching "gap-wire", a fluid resin layer 46 is deposited full plate and thus fills the holes 42 and planarize the assembly obtained in the previous step (Figure 8). Once solidified, the resin layer 46 is then etched uniformly, for example by mechanical or mechanochemical polishing, up to to reach the metal layer surface 46. The holes 42 remain filled with resin 46 to protect the metal covering them in the subsequent steps (Figure 9). An etching of the metal 44 arranged between the holes 42 is then implemented in a manner known per se (FIG. 10).

The process is then continued by the removal of the resin 46 included in the holes 42, for example using a deletion based on a plasma O ₂ followed by removal of the sacrificial layer 40, for example by means of a deletion based on a 0 ₂ plasma (FIG. 11). The webs 24 of the inserts 16 are thus made.

A layer 26 of ductile and naturally oxidizable metal is then deposited full plate, for example a layer of aluminum, tin, indium, lead, silver, copper, zinc, or an alloy of these metals for example by means of a CVD deposit (FIG. 12), then the layer portion 26 arranged between the cores 24 is removed, for example by means of the technique of a conventional photolithography technology (FIG. 13).

Finally, the process ends with the oxidation of the layer 26, for example leaving the component in the open air. It has been described a layer 26 completely covering the core 24. Of course this layer may cover only part of the core 24 if the application requires it.

Likewise, a non-oxidized central core 24 has been described on its entire surface. Alternatively, only a portion of the surface of the central core 24 is unoxidized. The central core 24 is then covered with the layer 26 at least on this non-oxidized portion, and the layer 28 covers at least the portion of the layer 26, covering the non-oxidized portion of the core 24, this portion of the layer 26 being unoxidized.

Claims

A method of producing an electromechanical connection component (10) having conductive inserts (16) each formed of a hollow cylinder for insertion at ambient temperature into respective aluminum conductive pads (18) of a another connection component for face-to-face hybridization, characterized in that the method consists for each insert:

^■ to achieve a metal core (24), not oxidized on at least a portion of its surface, and plasticity greater than that of aluminum;

^■ to produce a first layer (26) of aluminum metal, on at least a portion of its surface, covering at least said non-oxidized portion of the core (24); and

^■ to perform oxidation by oxygen of the aluminum of the first layer (26) so as to create a second layer (28) consisting of native aluminum oxide on only a portion of the thickness of the first layer, and said second layer having a plasticity lower than that of the first layer.

Method of producing an electromechanical connection component according to Claim 1, characterized:

^■ in that the core (24) is in a nonoxidative atmosphere;

^■ in that the first layer (26) consists of a metal film ductility exceeding that of the metal of the core (24) deposited in a non-oxidizing atmosphere.

Process for producing an electromechanical connection component according to any one of the preceding claims, characterized in that the first layer (26) substantially covers the entire core (24), and in that the first layer (26) is oxidized over substantially its entire surface.

Electro-mechanical connection method for face-to-face hybridization between a first microelectronic component (10) obtained according to any one of claims 1 to 3 and a second microelectronic component (12) provided with conductive pads ( 18) in which the inserts (16) of the first microelectronic component (10) are respectively insertable, said method comprising the insertion at ambient temperature of the inserts (16) of the first component (10) in the pads. (18) of the second component (12).

5. A method of connection according to claim 4, characterized in that the pressure exerted on a bearing surface (S) of each insert (16) when inserted into a stud (18) is greater than 1800 megaPascals. Microelectronic system obtained according to claim 4 or 5.