Formation of alpha particle shields in chip packaging

This application is a Continuation application claiming priority to Ser. No. 13/533,182, filed Jun. 26, 2012, which is a Continuation of Ser. No. 12/200,352, filed Aug. 28, 2008, U.S. Pat. No. 8,247,271, issued Aug. 21, 2012, which is a Divisional application claiming priority to Ser. No. 11/211,116, filed Aug. 24, 2005, Abandoned Dec. 22, 2008.

1. Technical Field

The present invention relates to integrated circuit packaging, and more specifically, to using alpha particle shields in integrated circuit packaging.

2. Related Art

In flip-chip technologies, solder bumps are typically formed on top of a semiconductor chip (i.e., integrated circuit IC). Each solder bump is formed directly on a bond pad of the chip. Then the chip is flipped face down and then aligned to a package/substrate so that the solder bumps are bonded directly, simultaneously, and one-to-one to the pads of the package/substrate (called package/substrate pads). However, for ceramic substrates, alpha particles (large subatomic fragments consisting of 2 protons and 2 neutrons) continuously emit from the substrate and enter the chip resulting in a large number of soft errors in the chip during the normal operation of the chip. Alpha particles are also generated from 210 Pb contained in the solder bumps.

Therefore, there is a need for a structure (and a method for forming the same) that reduces the number of alpha particles that enter the chip.

The present invention provides a structure fabrication method, comprising providing an integrated circuit including N chip electric pads, wherein N is a positive integer, and wherein the N chip electric pads are electrically connected to a plurality of devices on the integrated circuit; providing an interposing shield having a top side and a bottom side and having N electric conductors in the interposing shield, wherein the N electric conductors are exposed to a surrounding ambient at the top side but not being exposed to the surrounding ambient at the bottom side; bonding the integrated circuit to the top side of the interposing shield such that the N chip electric pads are in electrical contact with the N electric conductors; polishing the bottom side of the interposing shield so as to expose the N electric conductors to the surrounding ambient at the bottom side of the interposing shield after said bonding the integrated circuit to the top side is performed; and forming N solder bumps on the polished bottom side of the interposing shield and in electrical contact with the N electric conductors.

The present invention also provides a structure fabrication method, comprising providing an integrated circuit including N chip electric pads, wherein N is a positive integer, and wherein the N chip electric pads are electrically connected to a plurality of devices on the integrated circuit; providing a semiconductor interposing shield having a top side and a bottom side and having N electric conductors in the semiconductor shield, wherein the N electric conductors are exposed to a surrounding ambient at the top side but not being exposed to the surrounding ambient at the bottom side; bonding the integrated circuit to the top side of the semiconductor interposing shield such that the N chip electric pads are in electrical contact with the N electric conductors; polishing the bottom side of the semiconductor interposing shield so as to expose the N electric conductors to the surrounding ambient at the bottom side of the semiconductor interposing shield after said bonding the integrated circuit to the top side is performed; forming N solder bumps on the polished bottom side of the semiconductor interposing shield and in electrical contact with the N electric conductors; after said forming the N solder bumps is performed, bonding a ceramic substrate that includes N substrate pads such that the N substrate pads are bonded to the N solder bumps, wherein the semiconductor interposing shield comprises essentially only silicon, and wherein the semiconductor interposing shield has a thickness of at least 50 μm after said polishing the bottom side is performed.

FIGS. 1A, 1B, and 2-12 show the fabrication process for forming a structure, in accordance with embodiments of the present invention.

FIGS. 1A, 1B, and 2-12 show the fabrication process for forming a structure 700 ( FIG. 10 ), in accordance with embodiments of the present invention. More specifically, with reference to FIG. 1A , in one embodiment, the fabrication process starts out with an interposing shield 100 that comprises a semiconductor (e.g., silicon, germanium) layer 110 . Next, in one embodiment, annular trenches 112 a and 112 b are formed in the semiconductor layer 110 . Illustratively, the annular trenches 112 a and 112 b are formed using a photolithographic process. In one embodiment, the annular trenches 112 a and 112 b have a depth 113 of around 50-70 μm. FIG. 1B shows a perspective view of the interposing shield 100 of FIG. 1A .

Next, with reference to FIG. 2 , in one embodiment, a dielectric film 210 is formed on exposed-to-ambient silicon surfaces of the interposing shield 100 of FIG. 1A . As a result, the dielectric film 210 forms on, among other places, bottom walls and side walls of the annular trenches 112 a and 112 b . Illustratively, exposed-to-ambient silicon surfaces of the interposing shield 100 of FIG. 1A can be thermally oxidized so as to form silicon dioxide resulting in the dielectric film 210 .

Next, with reference to FIG. 3 , in one embodiment, an electrically conducting layer 310 is formed on top of the interposing shield 100 of FIG. 2 so as to completely fill the annular trenches 112 a and 112 b . Illustratively, the electrically conducting layer 310 comprises a metal (such as copper Cu) and is formed by CVD (chemical vapor deposition), ALD (atomic layer deposition), or electrochemical plating of the metal (i.e., Cu). It should be noted that if the metal used for the electrically conducting layer 310 is tungsten (W), a seed layer (not shown) of Ti or TiN needs to be formed first on top of the interposing shield 100 of FIG. 2 by PVD, CVD or ALD to serve as nuclei for the ensuing growth of tungsten to form the W electrically conducting layer 310 . Likewise, it should be noted that if the metal used for the electrically conducting layer 310 is copper (Cu), a seed layer (not shown) of TaN, Ta and Cu needs to be formed first on top of the interposing shield 100 of FIG. 2 by PVD, CVD or ALD to serve as nuclei for the ensuing growth of copper to form the Cu electrically conducting layer 310 .

Next, in one embodiment, a chemical mechanical polishing (CMP) step is performed on top surfaces 320 of the interposing shield 100 of FIG. 3 until the dielectric film 210 is exposed to the surrounding ambient. The resulting interposing shield 100 is shown in FIG. 4 (without the top layer 420 ). What remains of the electrically conducting layer 310 after the CMP step resides in the annular trenches 112 a and 112 b and can be referred to as the annular electric conductors 410 a and 410 b ( FIG. 4 ).

Next, with reference to FIG. 4 , in one embodiment, a dielectric layer 420 is formed on top of the dielectric film 210 and in contact with the annular electric conductors 410 a and 410 b . Illustratively, the dielectric layer 420 comprises silicon dioxide and is formed by CVD of silicon dioxide.

Next, with reference to FIG. 5 , in one embodiment, electric pads 510 a and 510 b are formed in the oxide layer 420 and in direct physical contact with the annular electric conductors 410 a and 410 b , respectively. Illustratively, the electric pads 510 a and 510 b comprise copper and can be formed using a conventional damascene process. More specifically, the damascene process starts with etching trenches (which the electric pads 510 a and 510 b later occupy) in the oxide layer 420 using a conventional lithographic process. Next, copper is deposited (e.g., by electroplating) to fill the trenches. Finally, excess copper outside the trenches is removed by a CMP step resulting in the electric pads 510 a and 510 b as shown in FIG. 5 .

Next, in one embodiment, the oxide layer 420 is recessed so that its top surface 422 is lower than the top surfaces 512 of the electric pads 510 a and 510 b as shown in FIG. 6 . In one embodiment, the oxide layer 420 is recessed by several thousand Å to 0.5 μm. Illustratively, the oxide layer 420 is recessed by a wet etch using a dilute hydrofloric acid solution (HF).

Next, with reference to FIG. 6 , in one embodiment, the interposing shield 100 is aligned with a semiconductor chip (integrated circuit IC) 600 such that the electric pads 622 a and 622 b of the semiconductor chip 600 are aligned with the electric pads 510 a and 510 b of the interposing shield 100 , respectively. In one embodiment, the semiconductor chip 600 is fabricated separately from the fabrication of the interposing shield 100 . Illustratively, the semiconductor chip 600 comprises a device region 610 and a back-end-of-line (BEOL) region 620 . The device region 610 can comprise devices such as transistors, resistors, and capacitors (not shown). The (BEOL) region 620 can comprise electrically conducting lines (not shown) running in a dielectric material so as to (i) electrically connect the devices of the device region 610 together and (ii) electrically connect the devices of the device region 610 to the electric pads 622 a and 622 b.

Next, with reference to FIG. 7 , in one embodiment, the interposing shield 100 and the chip 600 are bonded together to form a structure 700 such that the electric pads 510 a and 622 a ( FIG. 6 ) merge together to form an electric pad 510 a , 622 a and such that the electric pads 510 b and 622 b ( FIG. 6 ) merge together to form an electric pad 510 b , 622 b . In one embodiment, the bonding process is performed at 350-400° C.

Next, with reference to FIG. 8 , in one embodiment, the bottom side of the structure 700 is polished until the annular electric conductors 410 a and 410 b are exposed to the surrounding ambient. Illustratively, the bottom side of the structure 700 is mechanically ground by a mechanical grinding process only. Alternatively, the bottom side of the structure 700 is ground down by a mechanical grinding process until the annular electric conductors 410 a and 410 b are about to be exposed to the surrounding ambient. Then, a wet etch is performed on the bottom side of the structure 700 so as to expose the annular electric conductors 410 a and 410 b to the surrounding ambient.

Next, with reference to FIG. 9 , in one embodiment, solder bumps 910 a and 910 b are formed on bottom side of the structure 700 and in electrical contact with the annular electric conductors 410 a and 410 b , respectively, using a conventional solder bump formation process (also known as the flip chip technologies). The resulting structure 700 is shown in FIG. 9 . The solder bumps 910 a and 910 b are electrically connected to the annular electric conductors 410 a and 410 b via electric chip pads 920 a and 920 b , respectively. Illustratively, the electric chip pads 920 a and 920 b comprises aluminum. Although not shown, between the solder bumps 910 a and 910 b and the aluminum chip pads 920 a and 920 b is a ball limiting metallurgy (BLM) (illustratively comprising TiW/CuCr/Cu). The rest of the bottom side of the structure 700 is covered by a polyimide layer 930 which is a dielectric material.

Next, with reference to FIG. 10 , in one embodiment, a ceramic substrate 1010 is bonded with the structure 700 such that substrate pads 1010 a and 1010 b of the ceramic substrate 1010 are bonded with the solder bumps 910 a and 910 b , respectively. Illustratively, the substrate pads 1010 a and 1010 b comprises aluminum.

Next, in one embodiment, the structure 700 is placed in a package (not shown) having package pins (not shown) that are electrically connected to the substrate pads 1010 a and 1010 b via metal lines (not shown).

In summary, with reference to FIG. 10 , the interposing shield 100 is sandwiched between the ceramic substrate 1010 and the semiconductor chip 600 . As a result, the interposing shield 100 helps reduce the alpha particles that are generated by the ceramic substrate 1010 and enter the semiconductor chip 600 . The interposing shield 100 also helps reduce the alpha particles that are generated by the solder bumps 910 a and 910 b (i.e. Pb).

In one embodiment, the thickness 114 of the interposing shield 100 is sufficiently large such that at least a pre-specified percentage of alpha particles entering the interposing shield 100 from the ceramic substrate 1010 do not pass through the interposing shield 100 so as to reach the semiconductor chip 600 .

It should be noted that the thickness 114 of the interposing shield 100 is essentially the depth 113 ( FIG. 1A ) of the annular trenches 112 a and 112 b of FIG. 1A . As a result, with the depth 113 ( FIG. 1A ) of around 50-70 μm, the thickness 114 of the silicon interposing shield 100 is also around 50-70 μm and therefore is sufficiently thick to prevent most of the alpha particles generated by the ceramic substrate 1010 from entering the semiconductor chip 600 .

It should also be noted that the annular electric conductors 410 a and 410 b provide electric paths from the solder bumps 910 a and 910 b to the devices (not shown) of the semiconductor chip 600 (via the electric pads 510 a , 622 a and electric pad 510 b , 622 b , respectively). The annular shape is chosen for the electric conductors 410 a and 410 b so as to save metal material during the step of filling the trenches 112 a and 112 b ( FIG. 3 ) to form the electric conductors 410 a and 410 b . Moreover, because the trenches 112 a and 112 b ( FIG. 3 ) are filled fast, the excess metal outside the trenches 112 a and 112 b ( FIG. 3 ) are less, and therefore, the ensuing removal of the excess metal becomes easier. In general, the trenches 112 a and 112 b ( FIG. 3 ) can have any shape and size.

It should be noted that the solder bumps 910 a and 910 b may comprise a tin-lead alloy which itself generates alpha particles. Because the interposing shield 100 is sandwiched between the solder bumps 910 a and 910 b and the semiconductor chip 600 , the interposing shield 100 also helps reduce the alpha particles that enter the semiconductor chip 600 from the solder bumps 910 a and 910 b.

In one embodiment, the structure 700 comprises a dielectric layer (not shown) that electrically insulates the electric chip pads 920 a and 920 b from the silicon region of the silicon layer 110 such that there is no electrically conducting path between the electric chip pads 920 a and 920 b through the silicon region of the silicon layer 110 .

In the embodiments above, there are two trenches 112 a and 112 b ( FIG. 1A ) formed. In general, there can be N trenches formed, wherein N is a positive integer. As a result, there are N solder bumps (like the solder bumps 910 a and 910 b ) electrically connected one-to-one to N electric pads (like the electric pad 510 a , 622 a and 510 b , 622 b ) through N electric conductors (like the electric conductors 410 a and 410 b ).

In one embodiment, with reference to FIG. 11 , metal (e.g., copper) regions 1110 a , 1110 b , and 1110 c are formed in the semiconductor regions of the interposing shield 100 such that the metal regions are electrically insulated from the electric conductors 410 a and 410 b . Because copper is better than silicon in absorbing alpha particles, the interposing shield 100 with such embedded copper regions performs better in preventing alpha particles from reaching the semiconductor chip 600 . Illustratively, the copper regions can be formed by creating trenches (not shown) similar to the trenches 112 a and 112 b ( FIG. 1A ) and filling these trenches with copper.

In one embodiment, a metal (e.g., copper) layer 1210 ( FIG. 12 ) may be formed on the bottom side of the structure 700 of FIG. 8 . Then, the solder bumps 910 a and 910 b are formed as described above. Additional conventional fabrication steps are needed after the copper layer is formed and before the solder bumps 910 a and 910 b are formed so that the copper layer is sandwiched between, and electrically insulated from, the electric conductors 410 a and 410 b and the solder bumps 910 a and 910 b . The resulting structure 700 is shown in FIG. 12 . Because copper is better than silicon in absorbing alpha particles, the interposing shield 100 with the copper layer performs better in preventing alpha particles from reaching the semiconductor chip 600 . It should be noted that a dielectric layer (not shown) electrically insulates the copper layer 1210 from the silicon regions of the silicon interposing shield 100 . In one embodiment, the thickness of the copper layer 1210 is about one third of the thickness of the silicon interposing shield 100 . In one embodiment, the thickness of the copper layer 1210 is less than 15 μm and the silicon interposing shield 100 has a thickness in a range of 30 μm-70 μm. If the thickness of the copper layer 1210 is increased, the thickness of the silicon interposing shield 100 can be reduced. This means that the depth 113 ( FIG. 1A ) of the trenches 112 a and 112 b ( FIG. 1A ) can be reduced. In one embodiment, the copper layer has a thickness in a range of 10 μm-15 μm, which is sufficient by itself in blocking alpha particles, and therefore, the thickness of the silicon interposing shield 100 can be less than 1 μm or even zero (i.e., silicon interposing shield 100 can be omitted).

In one embodiment, the silicon regions of the semiconductor interposing shield 100 are doped with boron atoms (using, illustratively, ion implantation). This enhances the capability of the semiconductor interposing shield 100 in preventing cosmic thermal neutrons from passing through the semiconductor interposing shield 100 and reach the semiconductor chip 600 . The cosmic thermal neutrons undergo reactions with the B that emit <2 MeV alpha particles. Therefore it is advantageous to have this B doped region on the top of the Si interposer layer (on the opposite side from the semiconductor device).

While particular embodiments of the present invention have been described herein for purposes of illustration, many modifications and changes will become apparent to those skilled in the art. Accordingly, the appended claims are intended to encompass all such modifications and changes as fall within the true spirit and scope of this invention.