Electricity: electrical systems and devices – Electrostatic capacitors – Fixed capacitor
Reexamination Certificate
2001-01-26
2002-08-06
Reichard, Dean A. (Department: 2831)
Electricity: electrical systems and devices
Electrostatic capacitors
Fixed capacitor
C361S306100, C361S312000, C361S313000, C361S320000
Reexamination Certificate
active
06430030
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates in general to multilayer ceramic capacitors. In particular, the present invention describes a multilayer ceramic capacitor with low dielectric constant material in the vicinity of signal vias. The present invention also pertains to an interposer between a semiconductor chip and the chip carrier.
BACKGROUND OF THE INVENTION
Multilayer ceramic capacitors for decoupling applications require the use of high k (dielectric constant) dielectric to provide the necessary capacitance. However, for high performance applications, in an interposer configuration, this high dielectric capacitance between signal vias can cause excessive parasitic capacitance
oise, thus detracting from the performance of the decoupling capacitor or “decap”. Therefore, while a high k ceramic is required for the capacitor, it is a detriment in the immediate vicinity of signal vias.
Multilayer ceramic (MLC) capacitors are frequently used to suppress the voltage noise generated by circuits in semiconductor chips. This noise arises as a result of voltage fluctuations generated across power bus lines during the simultaneous switching of circuits. This noise can degrade chip performance. Further, as circuit densities increase and rise times decrease, and supply voltage levels decrease, the simultaneous switching problem gets exacerbated. This can seriously impede chip performance at high frequencies of operation.
Decoupling capacitors are used as a means to reduce the extent of this problem. Typical low inductance MLC discrete decaps are made using relatively high permittivity ceramic materials which enable the delivery of high capacitance. These may be made in the parallel plate configuration with a plurality of termination points on either side which act as the two electrodes, which may be then connected to power and ground respectively. Such low inductance decaps can provide 30 to 200 nF per discrete MLC capacitor of less than 2 mm×2 mm×2 mm size. However the inductance associated with these discrete decaps is still too high to handle high frequency, high power, chip requirements.
The ideal location for decoupling capacitance is between the chip and the power supply system, on its carrier. An interposer decap or an integrated capacitor structure are both ways of achieving this ideal location. The main problem with this type of capacitor structure is that signal traces must also pass through high permittivity layers. This can contribute to unacceptably high signal to signal coupling or signal capacitance.
In an ideal capacitor structure, signals should move through low permittivity regions, while voltage and ground transmissions can move through areas of high permittivity. Such a structure enables decoupling capacitance to be delivered at the required site without the excess load of high parasitic capacitance and signal to signal coupling which might arise if signal transmission were to occur through regions of high electrical permittivity.
As semiconductor logic chips improved in performance by decreasing the cycle time and hence increasing the execution rate, it became desirable to incorporate decoupling capacitors into the circuit to reduce the effect of the noise generated at these increased switching rates and to otherwise improve overall performance. Initially these decoupling capacitors were constructed using discrete components and were mounted on a convenient location of the ceramic substrate or printed circuit board to which the chips were attached. However, as chip performance continued to improve, it became advantageous to provide low inductance paths between the capacitors and the chips.
Arnold et al. describe a low inductance electrode design for discrete capacitors in U.S. Pat. No. 4,831,494. These discrete capacitors would preferably be mounted as close to the chips as possible, on areas of the substrate adjacent to the chips for example, to further limit inductance and otherwise improve performance.
In U.S. Pat. No. 4,800,459, Takagi et al. suggest placement of discrete capacitors in substrate cavities under the chip site.
Bajorek et al. (U.S. Pat. Nos. 4,328,530 & 4,349,862) also suggested placing the decoupling capacitors in the ceramic substrate under the chip sites, but their design integrates the capacitor structures into the construction of the substrate. That is, specific regions of the ceramic multilayer structure are constructed using high dielectric constant materials and electrodes are placed on opposing sides of the high dielectric constant material to form capacitors. Bajorek, et al. also suggested that these integrated capacitors utilize specially designed low inductance electrode geometries.
Placement under the chip is the preferred location for the decoupling capacitors as it provides the shortest electrical path. Under-the-chip integrated capacitors have been described for ceramic substrate applications where the chips are wire bond connected to electrical connection pads on the substrate (U.S. Pat. Nos. 5,099,388, 5,521,332 & 5,371,403). The chips can be attached to the flat upper surface or within cavities. For substrates where chip attach is by C4 or flip-chip solder attach and the signal and power I/O (pins, ball or column grid, land grid, etc.) are on the other side; under-the-chip integrated capacitors can also be constructed.
Bajorek, et al. (U.S. Pat. Nos. 4,328,530 & 4,349,862) demonstrated capacitor placement in the ceramic multilayer substrate (which could also have thin-film wiring on the chip side surface), as did Hiroichi, et al. (U.S. Pat. No. 5,177,670).
Alternately the under-the-chip integrated capacitors can be constructed within thin-film layers on the chip side surface of the MLC substrate as described by Chance, et al. in U.S. Pat. No. 5,177,594 and Hiroichi, et al. in U.S. Pat. No. 5,177,670.
With today's extremely fast chips it is not only important to place the integrated capacitors very close to the chips with low inductance connecting paths, but to also insure that the signal wiring passes through low dielectric constant materials. With many wire bond designs, the chip and substrate I/O are on the same side of the ceramic substrate and high dielectric constant integrated capacitor layers can be placed under the chip without contacting the signal lines, as in U.S. Pat. Nos. 5,099,388 & 5,521,332. But even some wire-bond substrate designs are preferably constructed with low dielectric layers next to the chip to minimize degrading signal line performance (U.S. Pat. No. 5,371,403). However, with multilayer ceramic substrates (with or without thin-film layers) where chip connection is by C4 or flip-chip type connections, it is usually not possible to construct the substrates with layers completely consisting of high dielectric constant material(s) without having the signal lines undesirably passing through the high dielectric material. It is therefore preferable to construct capacitor containing layers using both high and low dielectric constant materials such as described in U.S. Pat. No. 5,177,670.
FIG. 1
shows the general arrangement of an interposer capacitor
10
. In this figure the chip
14
is attached to the interposer
16
by standard flip chip techniques which are well known in the industry, i.e. C4 bonding as illustrated by circles
15
. The interposer
16
is in turn connected using connections
17
to the carrier
18
using similar technology. Both signal and voltage connections are made from chip
14
to interposer
16
and from interposer
16
to carrier
18
. Items identified as
19
are module IO, and connectors such as pins, solder, balls, etc.
FIG. 2
a
shows a conventional interposer capacitor
20
.
FIG. 2
a
shows a multilayer structure (a possible embodiment could be using ceramic layers) having dielectric layers
21
,
22
,
23
. In one embodiment layers
21
,
22
,
23
are made of a high k material to enhance the value of the capacitance between metal electrodes
27
and
28
. Passing through this structure are signal vias
25
. Also passing through the dielectric
Farooq Mukta S.
Hamel Harvey C.
Rita Robert A.
Stoller Herbert I.
Reichard Dean A.
Thomas Eric W
Townsend Tiffany L.
LandOfFree
High k dielectric material with low k dielectric sheathed... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with High k dielectric material with low k dielectric sheathed..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and High k dielectric material with low k dielectric sheathed... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2889626