1. 1
Copper Electrodeposition and the Effects of Organic Additives
on Deposit Growth
Nicholas Sullivan
Email: npp8@wildcats.unh.edu
Department of Chemical Engineering, University of New Hampshire, Durham,
New Hampshire 03824
Introduction
Copper electrodeposition is commonly used
in the electronics industry for the plating of electrical
components such as wires, semiconductors, and
printed circuit boards (PCB’s) due to the high thermal
conductivity and low electrical resistance of the metal.
Copper is also known for its strong resistance to
corrosion and its capability to expand and contract
with the thermal expansion of plastics [1]. More
recently, the use of copper electrodeposition in
through-silicon-via technology has become a hot
topic due to the many advantages three-dimensional
packaging of chips provides. Through-silicon-vias
are used in the integration of three-dimensionally
packed chips, which enables linear contact between
chips as opposed to external connections resulting in
much faster signal transmission as well as a
significant decrease in the packaging area [2]. The
challenges of copper electroplating in through-
silicon-via technology include the need for bottom-up
superfilling of the vias, which is especially
challenging because of their high aspect ratio and
depth. A combination of levelers, suppressors, and
accelerants is required to effectively fill all vias, free
of voids, while maintaining a uniform deposit surface.
Electrochemical deposition with copper
plating baths containing organic additives is
commonly used when surface complexity makes line-
of-sight deposition methods like PVD or CVD
processes impractical. Exceptionally smooth plated
surfaces can be obtained with the proper mixture of
additives, even over very rough surfaces or in deep
vias. The deposition process can be controlled
through the use of levelers (inhibitors), suppressors,
and accelerants (brighteners). The process of leveling
occurs when the rate of deposition of metal in
recessed areas increases relative to the deposition rate
on surface peaks and edges. Leveling works to
reduce surface roughness by inhibiting growth on
prominent peak features, therefore directing mass
transfer to low-set features enabling them to catch up
and promote layer growth as opposed to nucleation
coalescence, or column growth [1]. Levelers arrive
rapidly at the surface of exposed features and are
consumed in the process, thereby rendering their use
mass transfer controlled. Suppressors act very
similarly to levelers except that they are physically
adsorbed to surface features, are not consumed and
inhibit deposit growth where they are adsorbed.
Accelerants work to selectively remove suppressors
from the surface, enabling growth to progress. For
reasons that are not very well understood, accelerants
tend to accumulate in recessed areas and therefore
promote bottom-up filling of trenches and deep vias.
Brighteners act in much the same way as levelers, but
will smooth surfaces at a smaller scale where visible
light interactions are affected [3].
SPS (bis-(3- sulfopropyl)disulfide) is a
commonly used accelerant in copper deposition. In a
copper plating solution, SPS is reduced to its
monomer MPS which acts as an accelerant in the
deposition of copper. MPS reacts with copper cations
(Cu2+
) in solution to reduce them to copper (Cu1+
),
which forms a thiolate accelerant complex. The
catalytic reaction supports the bottom-up filling of
trenches [4].
𝟏
𝟐
𝐒𝐏𝐒 + 𝐇!
+ 𝐞!
→ 𝐌𝐏𝐒
𝟐𝐌𝐏𝐒 + 𝐂𝐮 𝟐!
→ 𝐂𝐮 𝐈 𝐭𝐡𝐢𝐨𝐥𝐚𝐭𝐞 +
𝟏
𝟐
𝐒𝐏𝐒 + 𝟐𝐇!
𝐂𝐮 𝐈 𝐭𝐡𝐢𝐨𝐥𝐚𝐭𝐞 + 𝐇!
+ 𝐞!
→ 𝐂𝐮 + 𝐌𝐏𝐒
Copper electrodeposition from acid solutions
is the most common technique used commercially,
specifically with solutions of cupric sulfate. Cupric
sulfate is relatively low in cost and simple to use
compared to other acid copper plating solutions and
has been extensively studied since its first use by
Bessemer in 1831[1-p63]. Cupric sulfate (copper (II)
sulfate) in its pentahydrate form (𝐂𝐮𝐒𝐎 𝟒 ∙ 𝟓𝐇 𝟐 𝐎) is
dissolved in sulfuric acid to create the acid copper
2. 2
plating solution.
Electrodeposition occurs through a plating
process where the desired component to be deposited
onto is placed in a solution of dissolved metal salts.
The component should ideally have a seed layer of
the metal already in place as is common in the
Damascene process so as to make initiation of the
deposition easier. This seed layer can be deposited in
many different ways including physical or chemical
vapor deposition (PVD/CVD), both of which are
more expensive than electroplating. During the
electroplating process, a current is supplied to the
solution by an anode composed of the metal being
deposited. The component being deposited onto acts
as the cathode in the process. The current supplied at
the anode oxidizes the metal and dissolves it into
solution. The rate at which metal is dissolved from
the anode is equal to the rate at which metal is
deposited onto the cathode [1].
Copper deposit growth is influenced by
several factors including molecular transport of ions
in solution to the surface of the deposit/plated device.
The transport of ions can either proceed by
concentration driven diffusion, or electrical migration
[5]. Electrochemical potential plays an important role
in electrodeposition; electrical potential differences in
a deposition process will serve as driving forces
and/or deposition limiting factors. The three major
categories of potential difference in electrochemical
deposition are concentration overpotential, activation
overpotential, and crystallization overpotential.
These potential differences occur in the solution at
the surface of the electrode, between an ion in
solution and an adsorbed ion, and between an
adsorbed ion and an ion introduced to the crystal
lattice. Concentration overpotential is the phenomena
described by a significant reduction in metal ions at
the electrode surface as compared to the bulk solution.
This potential difference results in deposition limited
by the transport of cations to the plating surface. The
activation overpotential is a result of the required
activation energy for an ion to react with the surface
of the metal. Crystallization overpotential is the
potential difference due to any hindrance of ions
becoming a part of the crystal lattice in a deposit [5].
A paper written by Schilardi, Marchiano,
Salvarezza, and Arvia in 1995 shows how through the
analysis of aggregate patterns and radial growth
velocity, it can be shown that large branch growth is
mainly controlled by the electric field between the
cathode and anode, while small branch growth is
controlled by diffusion [6]. In other words, large
branch growth is driven by the supply of current to
the anode, while small branch growth is limited by
the diffusion of ions to the metal surface.
Diffusion-limited-aggregation (DLA) is a
model describing the physics at the interface of the
deposition surface and the plating solution.
According to DLA branch growth is dependent on the
mass transfer (diffusion) of ions to the plating surface.
This model holds true initially in an electroplating
process, but as the branch growth proceeds, other
factors begin to play major contributing roles. The
DLA model assumes that branch growth is controlled
by the diffusion of ions through the solution to the
plating surface because the motion of ions through
the solution is more rapid than the growth velocity of
the deposit through the solution. When the branches
become more driven by potential, they are essentially
pushing through the solution rather than receiving
ions by diffusion. When the advance of the growth
tips into the solution becomes a larger driving force
for deposit growth than diffusion, the DLA model not
longer holds. This phenomenon has been shown to
occur after the initial growth stage.
Experimental
Copper electrodeposition was first tested
without any additives in an acid copper plating
solution of sulfuric acid (1M) and cupric sulfate
(0.2M) in a glass dish 50mm in diameter and 15mm
deep. A copper anode was placed in the solution at
the edge of the glass dish while a 0.24mm copper
wire was centered in the dish to act as the cathode.
The copper wire was insulated by 0.25mm of Teflon
shielding which was stripped away at the tip of the
wire to allow a deposit to grow. The cell was
operated in potentiostatic mode, meaning that the
potential difference between the anode and cathode
was held constant. First a short potential pulse of (-1)
volt was applied for five seconds. Then the potential
was stepped down to (-0.5) volts for the remainder of
the plating process. The larger initial potential
difference helps to initiate growth, but can also result
in hydrogen bubbles forming at the cathode. The
decreased potential stops hydrogen formation and
allows only deposit growth. Copper was deposited for
3,000 seconds.
Results
Shielding around the wire inhibited the
growth of copper crystals by forcing branch growth
outwards along the shielding, or out into the solution.
3. 3
Not until approximately 1,200 to 1,800 seconds of
deposition did the deposit reach the edge of the
shielding and full three-dimensional growth was able
to occur. When viewed macroscopically, the deposit
seemed only to grow spherically, but when viewed at
a more microscopic level, small bunched dendritic
branching was very prominent throughout the entire
deposit.
(a)
(b)
(c)
Figure 1. Copper deposit growth in acid copper
plating solution at (a) 120 seconds, (b) 1,800
seconds, and (c) 3,000 seconds.
Figure 2. Copper deposit growth as a function of
its charge accumulation.
Plotting the natural log of the deposit radius
as a function of the natural log of the deposit charge,
as is shown in figure 2, linearizes the relationship
between deposit growth and charge and expresses the
geometry of the deposit. The inverse of the slope of
this line should be approximately three for spherical
growth. Figure 1 shows how the growth of the
deposit is limited by the presence of the Teflon
shielding until nearly 1,800 seconds when it surpasses
the majority of the shielding. The Teflon shielding
limits spherical growth and therefore the inverse of
the slope in figure 2 is approximately six for the first
1,200 seconds. Since the deposit growth became
more spherical as it grew past the Teflon shielding
the inverse of the slope after 1,800 seconds is three as
was expected.
Figure 3. Charge accumulation as a function of
deposition time.
Charge in the deposit increases sharply at
first and slowly levels to a linear logarithmic increase
4. 4
with time after again approximately 1,200 seconds.
The conclusion that can be drawn from the data is
that the charge accumulation increases in a
logarithmically linear fashion with time for spherical
deposit growth. Figures 1 and 2 in combination with
figure 3 show the connection between charge, deposit
growth and geometry, and time.
Next Steps
A new electrochemical cell will be put
together to create a planar growth field. A glass dish
will contain plating solution and a much thinner
copper wire with less/no insulation, then a glass plate
will cover the cell so as to promote two-dimensional
growth which will be easier to photograph and
measure. Deposition growth will continue to be
measured while the effects of organic additives (PEG,
SPS, and BV) will be monitored. The setup will look
much the same as the setup described in the paper by
Schilardi et al entitled ‘The Development of 2D
Copper Branched Aggregates’ [6].
References
[1] Mordechay Schlesigner, Milan Paunovic, Modern
Electroplating (4th
Edition), New York, NY: John Wiley
& Sons, INC., 2000.
[2] P. Dale, Barkey, N. Rohan Akolkar, Kazuo Kondo,
Masayuki Yokoi, Copper Electrodeposition for
Nanofabrication of Electronics Devices, New York, NY:
Springer Science+Business Media, 2014.
[3] M.A. Pasquale, D.P. Barkey, A.J. Arvia, “Influence of
Additives on the Growth Velocity and Morphology of
Branching Copper Electrodeposits,” Dept. Chem. Eng.,
Univ. NH, Durham, NH, Journal of The Electrochemical
Society, 152 (3) C149-C157, Jan. 2005.
[4] Philippe M. Vereecken et al, “The Role of SPS in
Damascene Copper Electroplating,” IBM, T.J. Watson
Res. Cen., Yorktown Heights, NY.
[5] C. Thomas Halsey, Michael Lebig, “Electrodeposition
and Diffusion-Limited Aggregation,” Univ. Chic., The J.
F. Inst. and Dept. of Phys., Chicago, IL, Bost. Univ., Dept.
Phys., Boston, MA, Dec. 1989.
[6] P.L. Schilardi et al, “The Development of 2D Copper
Branched Aggregates,” Univ. La Plata, La Plata,
Argentina, Chaos, Solitons & Fractals Vol. 6, pp. 525-
529, 1995.