[0001] The invention concerns an inductor chip device, and in particular, but not exclusively,
an inductor chip device for mounting on a multichip module, a direct-chip-attach assembly,
or a surface-mount assembly.
[0002] There is a growing requirement in the construction of very compact, low-cost radios
and other RF communications circuits, for small, high-performance and cost-effective
inductor components.
[0003] Surface-mountable chip inductors have recently become available that measure 2×1.25
mm in area (a standard surface-mount "0805" format), and that offer inductance values
up to some 20 nH, with self-resonant frequencies of 1-2 GHz and quality (Q) factors
that peak at about 80 at about half the self-resonant frequency.
[0004] Very compact inductors have also been realised in an integrated form within the upper
metallisation layers of a multichip module substrate structure. Such inductors can
provide inductance values between 1 and 100 nH within a 1mm square footprint, with
self-resonant frequencies between 20 GHz and 500 MHz. Q-factor in these multichip-module
inductors is determined by the inductor resistance at low frequencies, while the peak
Q-factor is related to the nature and dielectric structure of the substrate employed.
High-resistivity silicon substrates with inductors defined in an aluminium-polyimide
multichip-module structure can provide Q-factors between about 5 and 20, depending
upon the inductor structure and inductance value. The peak Q-factor occurs at a frequency
between one-quarter and one-half the self-resonant frequency.
[0005] In accordance with the invention, there is provided an inductor chip device for mounting
on a multichip module, on a direct-chip-attach assembly or on a surface-mount assembly,
comprising a dielectric substrate, a spiral metallisation structure defined on one
major surface of the substrate, and a plurality of solder bumps for effecting said
mounting defined on the other major surface of the substrate, the spiral structure
being electrically connected to the plurality of solder bumps by means of metal-filled
vias.
[0006] By arranging the inductor metallisation structure on one face of the substrate and
the connections to the inductor structure on the opposite face, the capacitance of
the inductor to ground is minimised and therefore the self-resonant frequency is maximised.
Thus, where the closest ground structure in actual use of the inductor chip device,
e.g. in a multichip-module assembly, is located on the plane of the solder bumps,
the capacitance to ground will be determined essentially by the dielectric constant
of the substrate and its thickness, a greater thickness giving a lower capacitance
and a higher self-resonant frequency.
[0007] The use of solid, metal-filled through-vias to form the connections to the inductor
structure minimises the resistance and inductance of the inductor feed structure,
thereby maximising the Q-factor and self-resonant frequency of the inductor. This
effect is further enhanced in that attachment of the inductor chip device to the next
level of assembly is effected by flip-chip solder bonding, through the solder bumps.
[0008] The geometry of the inductor structure may be a circular, square, rectangular or
polygonal spiral configuration. For mounting to a multichip-module assembly, a square
format is to be preferred, whereas if the inductor device is to be mounted onto a
direct-chip-attach or surface-mount device, a rectangular format is preferable.
[0009] The substrate may be either a pre-fired alumina or aluminium nitride substrate with
vias formed by the laser-drilling of the substrate when in its pre-fired state and
by the filling of the holes thereby produced by a copper-tungsten composite material
produced by a liquid phase impregnation process. Alternatively, the substrate may
be a co-fired alumina substrate, the vias being sintered tungsten or molybdenum vias
formed in the substrate during co-firing.
[0010] The spiral structure may be a copper metallisation structure, and may be photolithographically
defined. Use of copper for the metallisation minimises the series resistance of the
inductor and therefore maximises the Q-factor. Defining the inductor geometry by means
of a photolithographic patterning process enables very accurate inductor dimensions
to be achieved, and ensures therefore a high reproducibility of inductance value from
device to device.
[0011] The spiral metallisation structure may comprise a thin, metallic adhesion layer,
a plating seed layer and a thick copper layer, deposited in that order on the substrate.
The adhesion layer may be a chromium, titanium or nichrome adhesion layer, and the
plating seed layer may be a copper plating seed layer.
[0012] The solder bumps are preferably formed on a multilayer metallisation structure comprising
a chromium adhesion layer, a chromium-copper layer, and a copper or copper-plus-gold
layer deposited in that order on the substrate.
[0013] The inductor chip device may include at least one further metallisation layer and
dielectric layer, the spiral structure being defined in the metallisation layer formed
on the major surface of the substrate and in the at least one further metallisation
layer. The at least one further dielectric layer may be a polyimide layer.
[0014] The advantage of employing a multilayer structure is that the inductance value may
be significantly increased. Thus, by providing a two-layer structure, an approximately
fourfold increase (2 squared) in inductance per unit area may be achieved for a given
number of turns, n.
[0015] At least one further solder bump may be included to ensure mechanical stability when
the inductor device is mounted on a multichip-module device, etc. The minimum number
of solder bump connections that is required is that number which effects the necessary
electrical connections between the inductor structure and the multichip-module device,
etc. The electrical connections may include not only end connections to the inductor,
but also one or more tap connections inbetween. However, where the total number of
electrical connections in the inductor chip device is insufficient to maintain mechanical
stability when the chip device is mounted, extra non-electrical solder bump connections
may be needed. The absolute minimum number of solder bump connections may therefore
be said to be three, but four such connections would be a normal practical minimum.
[0016] An embodiment of the invention will now be described, by way of example only, with
reference to the drawings, of which:
Figure 1 is a side view of an inductor chip device according to the invention, and
Figure 2 is a plan view corresponding to the side view of Figure 1.
[0017] Referring to Figures 1 and 2, one embodiment of an inductor chip device for mounting
to a multichip-module assembly according to the invention is shown comprising an alumina
substrate 12, on one major surface 13 of which is disposed a metallisation layer 14
configured as a square spiral. On the other major surface 15 is disposed a further
metallisation layer 16 configured as a number of pads or islands, upon which are formed
a corresponding number of solder bumps 17. The two ends 20, 21 of the spiral inductor
structure are electrically connected to two of the solder bumps 17 through two metal-filled
via structures 18 and 19, respectively. The two bumps to which the via structures
18, 19 are connected are shown as 22 and 23, respectively in Figure 2. In addition
to the bumps 22, 23 there are additional bumps 24-26 which serve not to electrically
connect the inductor to the multichip-module assembly (not shown), but to ensure a
firm seating of the inductor chip on the assembly.
[0018] The inductor chip device is flip-chip solder-bonded to mating solder-bump, or other,
structures on the multichip-module device.
[0019] The substrate 12 and metal-filled vias of the inductor chip device may be fabricated
by a process such as that described in patent specification GB 2,226,707 in the name
of Micro Substrates Inc. Thus, a high-density, polycrystalline alumina ceramic wafer
with a well defined surface finish and a uniform thickness may be first processed
to provide an array of narrow holes through the thickness of the wafer. This may be
achieved by a laser drilling process, using a CO₂ or neodinium-YAG laser source, to
produce narrow (∼ 125 µm diameter) holes which are slightly tapered from front to
back. The disposition of the holes will follow the pattern shown in Figure 2.
[0020] The array of through wafer holes may then be filled with a conducting, metallic plug
via structure using a copper-tungsten liquid-phase-impregnated composite structure.
In this structure, the tungsten material is employed in a powder form in an organic
binder and solvent carrier system and the via holes are filled with the tungsten powder
"ink" by a screen-printing process, assisted by vacuum or applied pressure.
[0021] After hole-filling, any excess tungsten ink may be removed and the wafer fired in
a partially oxidising atmosphere (for example, wet hydrogen) to burn off the organic
binder and to partially sinter the remaining tungsten powder without the formation
of tungsten oxides. A peak process temperature of 1400-1600 °C may be required for
this process stage. A partially oxidising atmosphere is also important in this step
to ensure good adhesion between the tungsten plug metallisation and the alumina ceramic.
The resulting structure comprises a porous tungsten via in every laser-drilled through-via
in the alumina wafer. A fully dense plug via structure may then be achieved by the
infiltration of liquid copper into the porous tungsten via. The copper may be applied
by the screen-printing of a copper "ink" material or by a similar process, followed
by the heat treatment of the wafer to remove the binder, if one is employed, and then
to melt the copper (typically in an hydrogen atmosphere at above 1083 °C). The liquid
copper naturally wets the tungsten, particularly in the presence of certain trace
element (surface) additives such as nickel, and thus fills the porous plug via structure
to create a solid plug via on cooling to ambient temperature. Any excess material
may then be removed by a final lapping and polishing operation that also provides
the required surface finish and wafer thickness. Wafers may conveniently be produced
at dimensions identical to those of silicon IC wafers, for example 100 mm diameter,
380 µm thickness or 150 mm diameter, 525 µm thickness, to facilitate processing. The
use of a solid via structure greatly facilitates subsequent processing of the wafer
since a polished, planar surface is provided that can be readily coated with resist,
polyimide or other processing materials by IC-like spin-coating processes. A planar
surface also aids low defectivity and low particulate levels in wafer processing.
[0022] Having obtained the ceramic substrate with its appropriately located metal-filled
vias, the inductor structure is then defined on one face of the wafer. A thin metallic
adhesion layer (using a reactive metal system such as chromium, titanium or nichrome)
and a plating seed layer (in the form of a thin copper layer) are sputter-deposited
onto the surface in succession. These layers are typically between 0.05 and 0.5 µm
thick. The adhesion layer provides a strong bond to the alumina surface and also a
good electrical connection to the copper-tungsten solid plug vias. The copper layer
provides a compatible surface onto which to plate additional copper. A thick layer
of a photoresist material is applied to this face of the wafer and patterned to define
a spiral opening structure into which the copper metallisation that forms the inductor
itself is then plated. The copper plating should be of the maximum thickness consistent
with good control of the plated structure and small spacing between the inductor turns
(for lowest resistance at a given inductor pitch and hence maximum Q-factor). The
inductor geometry will be limited by the resolution and feature aspect ratios that
can be defined in the resist, together with the properties of the plating solution
(throwing power and plating efficiency). Materials exist that can allow at least 25
µm copper thicknesses at feature separations of as little as 10 µm. This should lead
to inductor peak Q-factors of at least 100 in inductors of mm dimensions. After electroplating,
the resist mask and the plating seed layer and adhesion layer are stripped by suitable
solvent and etchant treatments from all surface areas apart from where the inductors
are defined.
[0023] In order to protect the wafer from damage in subsequent process steps, the completed
plated copper inductor structure is coated with a suitable passivation, e.g. a polyimide.
[0024] The wafer is then inverted and an array of solder bump structures are defined, some
over the exposed copper-tungsten plug vias that make the connections between the inductor
input and output points on the other face of the wafer, and some at other points on
the surface, as will be explained below.
[0025] The solder bump structure requires a solderable metallisation layer to which the
solder bump itself is wetted and which defines the area of the solder bump. Chromium-copper
or chromium-copper-gold multilayer metallisation structures are suitable for this
requirement. The first, chromium, layer provides adhesion and an ohmic connection
to the underlying copper-tungsten plug via surface, while an alloyed chromium-copper
layer follows which provides solderability without layer dissolution (for multiple
solder bump melting operations). The final copper or copper-plus-gold layers provide
initial solderability, these metals dissolving into the solder on bump reflow and
reprecipitating on cooling as intermetallic compounds of tin. The gold, if employed,
allows the solderable layer to be exposed to the atmosphere without oxidation prior
to solder deposition. Solderable metallisation layers of this type may be defined
by sequential vapour deposition through an etched metal foil or similar physical masking
structure.
[0026] The solder itself is a tin-lead eutectic composition (63Sn-37Pb by weight, melting
point 183 °C) for direct-chip-attach or surface-mount applications, or is a 95Pb-5Sn
composition (melting point 310 °C) for multichip-module applications. The solder may
be applied by electrodeposition using a seed layer and photoresist masking scheme
similar to that described for the plating of the copper inductor structure on the
first face of the wafer. Alternatively a physical masking structure may be employed
with a vapour deposition process similar to that described for the solderable metallisation
deposition. The solder may be deposited as separate layers of lead and tin, or as
an alloy.
[0027] After deposition and patterning of the solderable metallisation and solder layers,
the solder bumps are reflowed by heating to above the solder liquidus temperature
under inert or reducing atmosphere conditions. Solder bump diameters close to those
of the copper-tungsten plug vias are appropriate for the flip-chip inductor structure,
i.e. 125 µm diameter. Bump heights between 30 and 100 µm are suitable, depending upon
whether the inductor is to be used in a direct-chip-attach, surface-mount or multichip-module
application. Such bump geometries are also typical for flip-chip solder-bonded integrated
circuits for multichip-module and direct-chip-attach applications.
[0028] A total of five solder bumps are defined, two over the input and output plug via
connections to the inductor, and a further three distributed across the alumina surface
to provide mechanical support for the mounted flip-chip inductor.
[0029] As far as size of the inductor chip is concerned, preferred dimensions are 0.5, 1.0,
1.25, 1.5 or 2.0 mm square. This gives some consistency with the trend in discrete
surface-mount component sizes, which are currently following a reduction from 0805
(2.0 × 1.25 mm) to 0603 (1.5 × 0.75 mm) to 0402 (1.0 × 0.5 mm).
[0030] After processing of the inductor and solder bump faces of the wafer, the temporary
passivation layers which were employed to protect the inductor face are removed, the
wafer is inspected and dimensional and electrical measurements are made to determine
inductor yield and to ensure that the component characteristics are within the prescribed
limits. Any defective inductors are identified, for example by ink marking, and the
individual inductor components are separated by mechanical sawing or laser scribing.
[0031] In a second embodiment (not shown) of the inductor chip device according to the invention,
a co-fired alumina substrate is used instead of a pre-fired substrate, the vias being
then of sintered tungsten or molybdenum. The steps for obtaining the copper spiral
metallisation and the solder bumps are as described for the first embodiment.
[0032] In a third embodiment (not shown) of the inductor chip device according to the invention,
a multilayer inductor is formed in a pre-fired substrate by the plating of copper
layers separated by a polyimide dielectric. Vias are defined through the interlayer
polyimide material by dry etching, in order electrically to link the inductor spirals
situated in the inner metallisation layers and that in the outer metallisation layer.
[0033] Dielectric materials other than polyimide may be used for the intermediate layer
or layers, and means other than dry etching may be employed for defining the interlayer
vias.
[0034] Although a total of five solder bumps are shown in Figure 2, more or less may be
required, depending mainly on the size of the chip being produced. Also, there may
be more than two solder bumps in electrical contact with the inductor metallisation
structure. Thus, for instance, an inductor spiral may be tapped at some point or points
along its length, or a multilayer inductor may be tapped at its interlayer junctions.
These variations require an appropriate number of metal-filled vias to form the electrical
connections, and correspondingly fewer solder bumps may then be required purely for
maintaining mechanical stability.
1. An inductor chip device for mounting on a multichip module, on a direct-chip-attach
assembly or on a surface-mount assembly, comprising a dielectric substrate (12), a
spiral metallisation structure (14) defined on one major surface of the substrate,
and a plurality of solder bumps (17) for effecting said mounting defined on the other
major surface of the substrate, the spiral structure being electrically connected
to the plurality of solder bumps by means of metal-filled vias (18, 19).
2. An inductor chip device according to Claim 1, in which the substrate is a pre-fired
alumina or aluminium nitride substrate with vias formed by the laser-drilling of the
substrate when in its pre-fired state and by the filling of the holes thereby produced
by a copper-tungsten composite material produced by a liquid phase impregnation process.
3. An inductor chip device according to Claim 1, in which the substrate is a co-fired
alumina substrate and the vias are sintered tungsten or molybdenum vias formed in
the substrate during co-firing.
4. An inductor chip device according: to any one of the preceding claims, in which the
spiral structure is a copper metallisation structure.
5. An inductor chip device according to Claim 4, in which the spiral structure is photolithographically
defined.
6. An inductor chip device according to Claim 5, in which the spiral metallisation structure
comprises a thin, metallic adhesion layer, a plating seed layer and a thick copper
layer, deposited in that order on the substrate.
7. An inductor chip device according to Claim 6, in which the adhesion layer is a chromium,
titanium or nichrome adhesion layer, and the plating seed layer is a copper plating
seed layer.
8. An inductor chip device according to any one of the preceding claims, in which the
solder bumps are formed on a multilayer metallisation structure comprising a chromium
adhesion layer, a chromium-copper layer, and a copper or copper-plus-gold layer deposited
in that order on the substrate.
9. An inductor chip device according to any one of the preceding claims, including at
least one further metallisation layer and dielectric layer, the spiral structure being
defined in the metallisation layer formed on the one major surface of the substrate
and in the at least one further metallisation layer.
10. An inductor chip device according to Claim 9, in which the at least one further dielectric
layer is a polyimide layer.
11. An inductor chip device according to any one of the preceding claims, in which at
least one further solder bump is included to ensure mechanical stability when the
inductor device is mounted.