Process for collectively curving a set of electronic chips

Abstract

A process includes providing electronic chips, the chips having been diced beforehand and each including a stack including a matrix-array of pixels, an interconnect layer, first layer, joining the electronic chips to a carrier substrate, so as to leave a spacing region between the chips; forming a redistribution layer having lateral ends extending into each spacing region; forming metal pillars on the lateral ends; moulding a material including first segments, facing the first layers, second segments which are separate from the first segments, and which extend around the metal pillars; the first and second segments being coplanar; applying a heat treatment, the formed material being chosen so that the stack is curved with a convex shape; the second segments remaining coplanar at the end.

Claims

1. A process for collectively curving a set of electronic chips, comprising the steps: a) providing the set of electronic chips, the chips having been diced beforehand and each comprising a stack comprising: a matrix-array of pixels, an interconnect layer, electrically connected to the matrix-array of pixels, and a first layer, comprising through-vias that are electrically connected to the interconnect layer; the stack possessing a first thickness and a first coefficient of thermal expansion; b) joining the set of diced electronic chips to a carrier substrate, via a side opposite the first layers, so as to leave a spacing region between adjacent electronic chips; c) forming a redistribution layer comprising metal tracks that are electrically connected to the through-vias, the metal tracks having lateral ends extending into each spacing region; d) forming metal pillars on the lateral ends of the metal tracks; e) moulding a material on the redistribution layer, the material comprising: first segments, facing the first layers of the electronic chips, and second segments, which are separate from the first segments, and which extend in the spacing regions around the metal pillars; the first and second segments of the material being coplanar; the material possessing a formation temperature; and f) applying a heat treatment at the formation temperature of the material, the first and second segments of the material being mechanically attached to the first layers and to the carrier substrate, respectively, the material possessing a second thickness, and a second coefficient of thermal expansion strictly higher than the first coefficient of thermal expansion; a ratio between the first and second coefficients of thermal expansion, a ratio between the first and second thicknesses, and the formation temperature being adapted so that, at the end of step f), the stack of each electronic chip is curved with a preset convex shape, when viewed from the matrix-array of pixels, at a given corresponding operating temperature of the electronic chip; the second segments of the material formed around the metal pillars remaining coplanar at the end of step f).

2. The process according to claim 1, wherein step f) is preceded by a step e.sub.1) of planarizing the material so as to free the metal pillars.

3. The process according to claim 1, comprising a step g) of forming solder bumps on the second segments of the material at the end of step f), the solder bumps making contact with the metal pillars.

4. The process according to claim 3, wherein said step g) is followed by a step h) of dicing the carrier substrate to separate the electronic chips.

5. The process according to claim 1, wherein step b) is executed so that the carrier substrate is transparent in the spectral domain of operation of the electronic chips.

6. The process according to claim 1, wherein step b) is executed so that the carrier substrate comprises spacers to which the diced electronic chips are joined.

7. The process according to claim 1, wherein step c) comprises the steps: c.sub.1) forming a dielectric layer that fills the spacing regions so as to lie flush with the first layers of the electronic chips, the dielectric layer being mechanically attached to the carrier substrate; and c.sub.2) forming the redistribution layer, the latter extending over the first layers of the electronic chips and over the dielectric layer, the redistribution layer being mechanically attached to the first layers and to the dielectric layer.

8. The process according to claim 1, wherein step a) comprises the steps: a.sub.1) providing a first substrate, on which is formed the set of electronic chips; a.sub.2) joining a temporary substrate to the set of electronic chips; a.sub.3) thinning the first substrate until the first layer is obtained; a.sub.4) forming through-vias in the first layer; and a.sub.5) dicing the electronic chips individually.

9. The process according to claim 1, wherein step a) is executed so that: the interconnect layer is formed on the first layer, the matrix-array of pixels is formed on the interconnect layer, and the stack comprises a matrix-array of focusing lenses, formed on the matrix-array of pixels.

10. The process according to claim 1, wherein step a) is executed so that: the matrix-array of pixels is formed on the first layer, the interconnect layer is formed on the matrix-array of pixels, and the stack comprises a matrix-array of focusing lenses, formed on the interconnect layer.

11. The process according to claim 1, wherein step e) is executed so that the material is a thermosetting polymer, chosen from an epoxy resin and a polysiloxane resin.

12. The process according to claim 1, wherein step f) is executed so that the first and second coefficients of thermal expansion respect: $\frac{{\alpha }_2}{{\alpha }_1}\ge 4,$ where: α.sub.1 is the first coefficient of thermal expansion of the stack, and α.sub.2 is the second coefficient of thermal expansion of the material.

13. The process according to claim 1, wherein step f) is executed so that the material possesses a Young's modulus higher than 100 MPa.

14. The process according to claim 1, wherein step f) is executed so that the first and second coefficients of thermal expansion respect: $\frac{{\alpha }_2}{{\alpha }_1}\ge \frac{{\alpha }_2}{{\alpha }_1}\ge 6,$ where: α.sub.1 is the first coefficient of thermal expansion of the stack, α.sub.2 is the second coefficient of thermal expansion of the material.

15. The process according to claim 1, wherein step f) is executed so that the material possesses a Young's modulus higher than 1 GPa.

16. The process according to claim 1, wherein step f) is executed so that the material possesses a Young's modulus higher than 3 GPa.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) Other features and advantages will become apparent in the detailed description of various embodiments of the invention, the description being accompanied by examples and references to the appended drawings.

(2) FIGS. 1a to 1d are schematic cross-sectional views illustrating the preparation of FSI image sensors in step a) of a process according to the invention.

(3) FIGS. 2a to 2d are schematic cross-sectional views illustrating the preparation of BSI image sensors in step a) of a process according to the invention.

(4) FIGS. 3a to 3j are schematic cross-sectional views illustrating a mode of implementation of a process according to the invention for an FSI image sensor.

(5) FIGS. 4a to 4j are schematic cross-sectional views illustrating a mode of implementation of a process according to the invention for a BSI image sensor.

(6) FIG. 5 is a partial schematic cross-sectional view of an FSI image sensor provided in step a) of a process according to the invention.

(7) FIG. 6 is a partial schematic cross-sectional view of a BSI image sensor provided in step a) of a process according to the invention. The dashed line represents the bonding interface.

(8) FIG. 7 is a partial schematic cross-sectional view of a 3D stacked BSI image sensor provided in step a) of a process according to the invention. The dashed line represents the bonding interface.

(9) It should be noted that, for the sake of legibility and ease of understanding, the drawings described above are schematic and not to scale.

DETAILED DISCLOSURE OF THE EMBODIMENTS

(10) For the sake of simplicity, elements that are identical or that perform the same function in the various embodiments have been designated by the same references.

(11) One subject of the invention is a process for collectively curving a set of electronic chips P, comprising the steps:

(12) a) providing the set of electronic chips P, said chips having been diced beforehand and each comprising a stack comprising: a matrix-array of pixels 2, an interconnect layer 3, electrically connected to the matrix-array of pixels 2, a first layer 1, comprising through-vias 100 that are electrically connected to the interconnect layer 3; the stack possessing a first thickness and a first coefficient of thermal expansion;

(13) b) joining the set of diced electronic chips P to a carrier substrate SS, via the side opposite the first layers 1, so as to leave a spacing region ZE between adjacent electronic chips P;

(14) c) forming a redistribution layer 4 comprising metal tracks 40 that are electrically connected to the through-vias 100, the metal tracks 40 having lateral ends 400 extending into each spacing region ZE;

(15) d) forming metal pillars PM on the lateral ends 400 of the metal tracks 40;

(16) e) moulding a material 5 on the redistribution layer 4, the material 5 comprising: first segments 50, facing the first layers 1 of the electronic chips P, second segments 51, which are separate from the first segments 50, and which extend in the spacing regions ZE around the metal pillars PM; the first and second segments 50, 51 of the material 5 being coplanar; the material 5 possessing a formation temperature;

(17) f) applying a heat treatment at the formation temperature of the material 5, the first and second segments 50, 51 of the material 5 being mechanically attached to the first layers 1 and to the carrier substrate SS, respectively, the formed material 5 possessing a second thickness, and a second coefficient of thermal expansion strictly higher than the first coefficient of thermal expansion; the ratio between the first and second coefficients of thermal expansion, the ratio between the first and second thicknesses and the formation temperature being adapted so that, at the end of step f), the stack of each electronic chip P is curved with a preset convex shape, which is oriented toward the matrix-array of pixels 2, at a given corresponding operating temperature of the electronic chip P; the second segments 51 of the material 5 formed around the metal pillars PM remaining coplanar at the end of step f).

(18) Step b) is illustrated in FIGS. 3a and 4a. Step c) is illustrated in FIGS. 3b-3c and 4b-4c. Step d) is illustrated in FIGS. 3d and 4d. Step e) is illustrated in FIGS. 3e and 4e. Step f) is illustrated in FIGS. 3g and 4g.

(19) First Layer

(20) The first layer 1 comprises opposite first and second surfaces 10, 11. The first layer 1 is advantageously obtained from a first substrate 1′ that is thinned to promote the curvature of the electronic chips P. The first layer 1 advantageously has a thickness smaller than 500 μm, preferably smaller than 100 μm, and more preferably smaller than 50 μm. Such a thickness range allows the curvature at the end of step f) to be promoted. The first layer 1 is advantageously made of a semiconductor material, preferably silicon. In the case where the first layer 1 is made of silicon, the through-vias 100 are through-silicon vias (TSV).

(21) FSI Image Sensor

(22) Step a) may be executed so that: the matrix-array of pixels 2 is formed on the first layer 1, the interconnect layer 3 is formed on the matrix-array of pixels 2, the stack comprises a matrix-array of focusing lenses 6, formed on the interconnect layer 3.

(23) More precisely, the matrix-array of pixels 2 is formed on the first surface 10 of the first layer 1. By “formed on the surface”, what is meant is formed on or through the surface. Thus, when the electronic chips P are front-side-illuminated (FSI) image sensors, the incident light penetrates via the interconnect layer 3, and the matrix-array of pixels 2 is located downstream of the interconnect layer 3.

(24) It will be noted that the matrix-array of focusing lenses 6 may be optional in certain types of sensors, and notably in cooled infrared sensors.

(25) BSI Image Sensor

(26) Step a) may be executed so that: the interconnect layer 3 is formed on the first layer 1, the matrix-array of pixels 2 is formed on the interconnect layer 3, the stack comprises a matrix-array of focusing lenses 6, formed on the matrix-array of pixels 2.

(27) More precisely, the interconnect layer 3 is formed on the first surface 10 of the first layer 1. By “formed on the surface”, what is meant is formed on or through the surface. Thus, when the electronic chips P are back-side-illuminated (BSI) image sensors, the incident light penetrates into the matrix-array of pixels 2, which is located upstream of the interconnect layer 3, this preventing energy loss, and thereby increasing the sensitivity of the image sensor.

(28) It will be noted that the matrix-array of focusing lenses 6 may be optional in certain types of sensors, and notably in cooled infrared sensors.

(29) Preparation of the Electronic Chips

(30) As illustrated in FIGS. 1a to 1d and 2a to 2d, step a) advantageously comprises the steps: a.sub.1) providing a first substrate 1′, on which is formed the set of electronic chips P; a.sub.2) joining a temporary substrate 7 to the set of electronic chips P; a.sub.3) thinning the first substrate 1′ until the first layer 1 is obtained; a.sub.4) forming through-vias 100 in the first layer 1; a.sub.5) dicing the electronic chips P individually.

(31) In the case of an FSI image sensor illustrated in FIGS. 1a to 1d, step a.sub.2) is executed so that the temporary substrate 7 is joined to the interconnect layer 3. By way of nonlimiting example, the temporary substrate 7 joined in step a.sub.2) may be an adhesive film, arranged on the side of the interconnect layer 3, suction being applied to this film via a vacuum chuck. The use of a temporary substrate 7 permits the first substrate 1′ to be thinned to ensure the stack has the required mechanical strength.

(32) Step a.sub.3) is preferably executed by chemical-mechanical polishing, preceded by grinding.

(33) As illustrated in FIG. 1d, step a.sub.4) consists in forming through-vias 100 through the first layer 1, said vias being electrically connected to the interconnect layer 3.

(34) By way of nonlimiting example, step a.sub.5) may be executed using a precision circular saw, with a resin- or metal-bonded diamond blade.

(35) In the case of a BSI image sensor illustrated in FIGS. 2a to 2d, step a.sub.2) is executed so that the temporary substrate 7 is joined to the matrix-array of pixels 2. By way of nonlimiting example, the temporary substrate 7 joined in step a.sub.2) may be an adhesive film, arranged on the side of the matrix-array of pixels 2, suction being applied to this film via a vacuum chuck. The use of a temporary substrate 7 permits the first substrate 1′ to be thinned to ensure the stack has the required mechanical strength.

(36) As illustrated in FIGS. 6 and 7, the set of electronic chips P may be formed before step a.sub.1) within a substrate S that is then bonded to the first substrate 1′. The dashed line shows the bonding interface IC. As illustrated in FIG. 7, the first substrate 1′ may be equipped with a read circuit capable of simultaneously reading each pixel of a matrix-array formed in the substrate S.

(37) Step a.sub.3) is preferably executed by chemical-mechanical polishing, preceded by grinding.

(38) As illustrated in FIG. 2d, step a.sub.4) consists in forming through-vias 100 through the first layer 1, said vias being electrically connected to the interconnect layer 3.

(39) By way of nonlimiting example, step a.sub.5) may be executed using a precision circular saw, with a resin- or metal-bonded diamond blade.

(40) Interconnect Layer

(41) The interconnect layer 3 is a stack of interconnect levels comprising metal tracks embedded in a dielectric material. By way of nonlimiting example, the metal tracks may be made of copper, tungsten, titanium or of aluminium, and the dielectric material may be organic (a polymer such as a polyimide, or ALX (sold by ASAHI GLASS)) or inorganic (SiO.sub.2, SiN, etc.).

(42) Matrix-Array of Pixels

(43) The pixels 2 may be photosensitive cells (also called photosites) in the case of an electronic chip P of an image sensor. The pixels 2 may be the light-emitting (or emissive) cells in the case of an electronic chip P of a display.

(44) As illustrated in FIGS. 5 to 7, when the electronic chip P is an image sensor, the matrix-array of pixels 2 is advantageously equipped with photodiodes 21 that are linked to the surface 20 that receives the incident light. The matrix-array of pixels 2 is advantageously equipped with a complementary metal-oxide-semiconductor (CMOS) circuit configured to process the electrical signal generated by the photodiodes 21 (amplification of the signal, pixel selection, etc.).

(45) When the electronic chip P is a display, the matrix-array of pixels 2 is advantageously equipped with light-emitting diodes (not illustrated). The light-emitting diodes may be organic or inorganic. The matrix-array of pixels 2 is advantageously equipped with a CMOS circuit configured to control the light-emitting diodes.

(46) As illustrated in FIGS. 5 to 7, the matrix-array of pixels 2 is advantageously equipped with colour filters 60. When the electronic chip P is an image sensor, the colour filters 60 are advantageously arranged in a Bayer pattern. The colour filters 60 are interposed between the surface 20 that receives the incident light and the matrix-array of focusing lenses 6.

(47) Matrix-Array of Focusinq Lenses

(48) In the case of an electronic chip P of an image sensor comprising a matrix-array of focusing lenses 6, the focusing lenses 6 are convergent so as to concentrate the incident light toward the matrix-array of pixels 2. Each focusing lens 6 is associated with one pixel. The focusing lenses 6 are preferably micro-lenses.

(49) Carrier Substrate

(50) Step b) is advantageously executed so that the carrier substrate SS is transparent in the spectral domain of operation of the electronic chips P. By way of nonlimiting example, the carrier substrate SS may be made of glass.

(51) The electronic chips P, which are diced beforehand, are advantageously transferred to the carrier substrate SS in step b) using a pick-and-place machine.

(52) Step b) is advantageously executed so that the carrier substrate SS comprises spacers 9 to which the diced electronic chips P are joined. The electronic chips P are preferably bonded to the spacers 9 using a thixotropic adhesive.

(53) Redistribution Layer

(54) Step c) advantageously comprises the steps: c.sub.1) forming a dielectric layer 8 that fills the spacing regions ZE so as to lie flush with the first layers 1 of the electronic chips P, the dielectric layer 8 being mechanically attached to the carrier substrate SS; c.sub.2) forming the redistribution layer 4, the latter extending over the first layers 1 of the electronic chips P and over the dielectric layer 8, the redistribution layer 4 being mechanically attached to the first layers 1 and to the dielectric layer 8.

(55) Step c.sub.1) is illustrated in FIGS. 3b and 4b. Step c.sub.2) is illustrated in FIGS. 3c and 4c.

(56) The dielectric layer 8 formed in step c.sub.1) may be a resist, for example a SINR™ resist as sold by Shin-Etsu MicroSi, or a polyimide. Step c.sub.1) is executed so that the dielectric layer 8 is formed on the spacers 9.

(57) More precisely, step c.sub.2) is executed so that the redistribution layer 4 extends over the second surface 11 of the first layers 1 of the electronic chips P and over the dielectric layer 8.

(58) The redistribution layer 4 comprises metal tracks 40 embedded in a dielectric material. By way of non-limiting example, the metal tracks 40 may be made of Cu, Al, W, Ti. The dielectric material may be a resist, for example a SINR™ resist, or a polyimide.

(59) Metal Pillars

(60) The metal pillars PM are preferably made of aluminium or of copper.

(61) For example, a metal seed layer may be deposited on the redistribution layer 4.

(62) The seed layer allows the wafer-level electrical contact required for the future electrodeposition of the metal pillars PM to be obtained. The seed layer may comprise a plurality of sublayers, including a sublayer suitable for achieving mechanical coupling. The seed layer may have a thickness of the order of 300 nm. By way of nonlimiting example, the seed layer may comprise a first sublayer made of Ti of 100 nm thickness and a second sublayer made of Cu of 200 nm thickness. Next, a photoresist may be deposited on the seed layer, then exposed to ultraviolet radiation through a mask so as to form patterns delineating the future metal pillars PM. The thickness of the photoresist is chosen so as to be equal to the height of the future metal pillars PM. The seed layer is then biased in a dedicated bath allowing the metal pillars PM to be electrodeposited. Lastly, the photoresist is removed, and the portion of the seed layer lying under the photoresist during the electrodeposition is etched.

(63) Step f) is advantageously preceded by a step e.sub.1) consisting in planarizing the material 5 so as to free the metal pillars PM. Step e.sub.1) is illustrated in FIGS. 3f and 4f. By way of nonlimiting example, step e.sub.1) may be executed by chemical-mechanical polishing or by plasma etching.

(64) Material

(65) The material 5 formed in step f) is mechanically attached to the redistribution layer 4. Since the redistribution layer 4 is mechanically attached to the first layers 1 of the electronic chips P, the first segments 50 of the material 5 formed in step f) are mechanically attached (indirectly) to the first layers 1 of the electronic chips P. Furthermore, the second segments 51 of the material 5 that are formed in step f) are mechanically attached (indirectly) to the carrier substrate SS. Specifically, the spacers 9 are mechanically attached to the carrier substrate SS, the dielectric layer 8 is mechanically attached to the spacers 9, and the redistribution layer 4 is mechanically attached to the dielectric layer 8.

(66) Step f) is advantageously executed so that the first and second coefficients of thermal expansion respect:

(67) $\frac{{\alpha }_2}{{\alpha }_1}\ge 4,$
and preferably

(68) $\frac{{\alpha }_2}{{\alpha }_1}\ge 6,$
where: α.sub.1 is the coefficient of thermal expansion of the stack, α.sub.2 is the coefficient of thermal expansion of the formed material 5.

(69) It is possible to measure the coefficient of thermal expansion of the stack using a technique known to those skilled in the art, as described in chapter 2 of the document “ASM Ready Reference: Thermal Properties of Metals», ASM International, 2002, or even in the document B. Cassel et al., «Coefficient of Thermal Expansion Measurement using the TMA 4000», PerkinElmer, Inc., 2013.

(70) To a first approximation, the coefficient of thermal expansion of the stack is substantially equal to the coefficient of thermal expansion of the first layer 1 in so far as the thickness of the first layer 1 is predominant in the stack. When the first layer 1 is made of silicon, α.sub.1 is about 2.5×10.sup.−6 K.sup.−1. The material 5 will therefore be chosen to have an α.sub.2 such that α.sub.2≥10.sup.−5 K.sup.−1, and preferably such that α.sub.2≥1.5×10.sup.−5 K.sup.−1. The radius of curvature obtained at the end of step f) may be computed depending on α.sub.1 and α.sub.2 by virtue of Stoney's formula, which is known to those skilled in the art.

(71) Step f) is advantageously executed so that the formed material 5 possesses a Young's modulus higher than 100 MPa, preferably higher than 1 GPa, and more preferably higher than 3 GPa.

(72) The formed material 5 advantageously has a second thickness comprised between 120 μm and 600 μm. The material 5 may be monolayer or multilayer. To a first approximation, the ratio between the first thickness (of the stack) and the second thickness (of the formed material 5), which has an influence on the curvature of the convex shape, may be considered to be governed by the ratio between the thickness of the first layer 1 and the second thickness in so far as the thickness of the first layer 1 is predominant in the stack. The formed material 5 will preferably be chosen to have a second thickness about 2.5 times larger than the thickness of the first layer 1 when the first layer 1 is made of silicon in order to optimize the curvature of the convex shape, depending on the envisaged application. When the material 5 is polished in step e.sub.1), the second thickness of the polished material 5 is advantageously at least 2 times larger than the thickness of the first layer 1 when the first layer 1 is made of silicon. Thus, the curvature of the convex shape is little affected at the end of step f).

(73) Step e) is advantageously executed so that the material 5 is a thermosetting polymer, preferably chosen from an epoxy resin and a polysiloxane resin. Where appropriate, the second coefficient of thermal expansion is the coefficient of thermal expansion of the set polymer. By way of non-limiting example, the thermosetting polymer may be: an epoxy resin, with a Young's modulus of about 9 GPa, α.sub.2 comprised between 3.1×10.sup.−5 K.sup.−1 and 1.14×10.sup.−4 K.sup.−1, and a cross-linking temperature of about 71° C.; a polysiloxane resin, with a Young's modulus of about 3.3 GPa, α.sub.2 comprised between 2×10.sup.−5 K.sup.−1 and 9.1×10.sup.−5 K.sup.−1, and a cross-linking temperature of about 180° C.

(74) The material 5 possesses a formation temperature (e.g. cross-linking temperature) strictly above the given operating temperature of the electronic chips P. Step e) is preferably executed using an injection mould. The injection mould is made of a material that is refractory at the formation temperature of the material 5, the cross-linking temperature of the thermosetting polymer for example. The injection mould is preferably made of silicone.

(75) Preset Convex Shape

(76) The preset convex shape may have a constant or variable radius of curvature (of given sign). The preset convex shape may be aspherical. The (constant or variable) radius of curvature is preset depending on the envisaged application.

(77) Finalization

(78) The process advantageously comprises a step g) consisting in forming solder bumps BS on the second segments 51 of the material 5 at the end of step f), the solder bumps BS making contact with the metal pillars PM. Step g) is illustrated in FIGS. 3h and 4h.

(79) Step g) is advantageously followed by a step h) consisting in dicing the electronic chips P. Step h) is illustrated in FIGS. 3i and 4i. By way of nonlimiting example, step h) may be executed using a precision circular saw, with a resin- or metal-bonded diamond blade.

(80) The invention is not limited to the described embodiments. Those skilled in the art will be able to envisage technically workable combinations thereof and to substitute equivalents therefor.