JOSELL DANIEL (US)
BRAUN TREVOR MICHAEL (US)
US20190093248A1 | 2019-03-28 |
S-K. KIMJ.E. BONEVICHD. JOSELLT.P. MOFFAT: "Electrodeposition of Ni in Sub-micrometer Trenches", J. ELECTROCHEM. SOC., vol. 154, 2007, pages D443 - D451
C.H. LEEJ.E. BONEVICHJ.E. DAVIEST.P. MOFFAT: "Magnetic Materials for 3-D Damascene Metallization: Void-free Electrodeposition of Ni and Ni70Fe30 Using 2-Mercapto-5-benzimidazole sulfonic Acid", J. ELECTROCHEM. SOC., vol. 155, 2008, pages D499 - D507
T.P. MOFFATD. JOSELL: "Extreme bottom-up superfilling of Through-Silicon-Vias by Damascene Processing: Suppressor disruption, positive feedback and Turing patterns", J. ELECTROCHEM. SOC., vol. 159, 2012, pages D208 - D216
D. JOSELLD. WHEELERT.P. MOFFAT: "Modeling Extreme Bottom-Up Filling of Through Silicon Vias", J. ELECTROCHEM. SOC., vol. 159, 2012, pages D570 - D576
D. WHEELERT.P. MOFFATD. JOSELL: "Spatial-temporal Modeling of Extreme Bottom-Up Filling of Through- Silicon-Vias", J. ELECTROCHEM. SOC., vol. 160, 2013, pages D3260 - D3265
D. JOSELLT.P. MOFFAT: "Superconformal Electrodeposition in Complexed Alkaline Electrolyte", J. ELECTROCHEM. SOC., vol. 161, 2014, pages D287 - D292
D. JOSELLT.P. MOFFAT: "Bottom-Up Electrodeposition of Zinc in Through Silicon Vias", J. ELECTROCHEM. SOC., vol. 162, 2015, pages D129 - D135
D. JOSELLT.P. MOFFAT: "Superconformal Bottom-Up Nickel Deposition in High Aspect Ratio Through Silicon Vias", J. ELECTROCHEM. SOC., vol. 163, 2016, pages D1 - D10
D. JOSELLM. SILIVAT.P. MOFFAT: "Superconformal Bottom-Up Cobalt Deposition in High Aspect Ratio Through Silicon Vias", J. ELECTROCHEM. SOC., vol. 163, 2016, pages D809 - D817
D. JOSELLT.P. MOFFAT: "Superconformal Bottom-Up Gold Deposition in High Aspect Ratio Through Silicon Vias", J. ELECTROCHEM. SOC., vol. 164, 2017, pages D327 - D334
D. JOSELLT.P. MOFFAT: "Superconformal Copper Deposition in Through Silicon Vias by Suppression Breakdown", J. ELECTROCHEM. SOC., vol. 165, 2018, pages D23 - D30
T.M. BRAUNS.-H. KIMH.-J. LEET.P. MOFFATD. JOSELL: "Superconformal Nickel Deposition in Through Silicon Vias: Experiment and Prediction", J. ELECTROCHEM. SOC., vol. 165, 2018, pages D291 - D300
D. JOSELLT.P. MOFFAT: "Bottom-up Filling of Damascene Trenches with Gold in a Sulfite Electrolyte", J. ELECTROCHEM. SOC., vol. 166, no. 1, 2019, pages D3022 - D3034
L.A. MENKD. JOSELLT.P. MOFFATE. BACAM.G. BLAINA. SMITHJ. DOMINGUEZJ. MCCLAINP.D. YEHA. E. HOLLOWELL: "Bottom-Up Copper Filling of Large Scale Through Silicon Vias for MEMS Technology", J. ELECTROCHEM. SOC., vol. 166, no. 1, 2019, pages D3066 - D3071
D. JOSELLL.A. MENKA.E. HOLLOWELLM. BLAINT.P. MOFFAT: "Bottom-Up Copper Filling of Millimeter Size Through Silicon Vias", J. ELECTROCHEM. SOC., vol. 166, 2019, pages D3254 - D3258
T. BRAUND. JOSELLM. SILVAJ. KILDONT.P. MOFFAT: "Effect of Chloride Concentration on Copper Deposition in Through Silicon Vias", J. ELECTROCHEM. SOC., vol. 166, 2019, pages D3259 - D3271
T.M. BRAUND. JOSELLJ. JOHNT.P. MOFFAT: "Simulations of Copper Electrodeposition in Through-Hole Vias", J. ELECTROCHEM. SOC., vol. 167, 2020, pages 013510
T. M. BRAUND. JOSELLT.P. MOFFAT: "Microelectrode Studies of S-NDR Copper Electrodeposition: Potentiodynamic and Galvanodynamic Measurements and Simulations", J. ELECTROCHEM. SOC., vol. 167, 2020, pages 082509
T. M. BRAUND. JOSELLS. DESHPANDEJ. JOHNT.P. MOFFAT: "Simulations of Copper Electrodeposition in Millimeter Size Through-Silicon Vias", J. ELECTROCHEM. SOC., vol. 167, 2020, pages 162508
T. BRAUND. JOSELLT.P. MOFFAT: "Simulating Cu Electrodeposition in High Aspect Ratio Features: Effect of Control Mode and Uncompensated Resistance in S-NDR Systems", ELECTROCHIMICA ACTA, vol. 375, 2021, pages 137925
What is claimed is: 1. A process for performing hysteretic current-voltage mediated void- free superconformal and bottom-up filling of recessed features of a substrate with a resistance member, the process comprising: providing an electrodeposition composition comprising: a metal electrolyte comprising a plurality of metal ions and a solvent; and a suppressor disposed in the solvent; and a hysteretic cyclic voltammogram; providing the substrate comprising: a field surface; and a recess disposed in the substrate, the recess comprising a distal position and a proximate position relative to the field surface of the substrate; exposing the recess to the electrodeposition composition; potentiostatically or potentiodynamically controlling an applied electric potential of the recess with a potential wave form; autonomously reducing, with the resistance member in presence of the electrodeposition composition with the hysteretic cyclic voltammogram, the deposition potential of the recess from that applied by the potential waveform; bifurcating the recess into an active metal deposition region and a passive region in response to the deposition potential and ohmic variations of the substrate; forming, in response to bifurcating the recess, a transition zone at an interface of the active metal deposition region and the passive region; progressively moving the transition zone closer to the field surface and away from the distal position through the metal deposition; and reducing the metal ions to form metal and depositing the metal in the active metal deposition region and not in the passive region; and forming a resistance enhanced superconformal filling in the recess of the substrate from the metal in the active metal deposition region, the resistance enhanced superconformal filling being: void-free, disposed in the recess in the active metal deposition region from the distal position to the transition zone, and absent in the passive region between the proximate position and the transition zone, such that forming the resistance enhanced superconformal filling occurs in consequence of autonomously reducing the deposition potential of the recess with the resistance member in a presence of the hysteretic cyclic voltammogram of the electrodeposition composition. 2. The process of claim 1, wherein the resistance member is selected from the group consisting essentially of a lumped resistor, a baffle, and a selected interelectrode separation distance between the substrate and a reference electrode in electrical communication with the electrodeposition composition. 3. The process of claim 2, wherein the resistance member is the lumped resistor, and the lumped resistor comprises a resistor in electrical communication with and electrically interposed between the substrate and a counter electrode in electrical communication with the electrodeposition composition. 4. The process of claim 2, wherein the resistance member is the baffle, and the baffle is in fluid communication with and fluidically interposed between the substrate and the reference electrode. 5. The process of claim 1, wherein the resistance member is the selected interelectrode separation distance between the substrate and a reference electrode in electrical communication with the electrodeposition composition, and the process further comprises adjusting the interelectrode separation. 6. The process of claim 1, further comprising terminating the depositing the metal before completely filling the recess to the field surface. 7. The process of claim 1, further comprising terminating depositing the metal after completely filling the recess to the field surface. 8. The process of claim 1, wherein the hysteretic cyclic voltammogram comprises an S-shaped negative differential resistance. 9. The process of claim 1, wherein the metal ions comprise Fe2+, Fe3+, Pt2+, Pt4+, Ir3+, Ir4+, Rh3+, Pd2+, Co2+, Ni2+, Au3+, Zn2+, Bi3+, Pb2+, Re7+,Au+, Ag+, Sn2+, W6+, Mo6+,Cu2+, Cu+, or a combination comprising at least one of the foregoing metal ions, and the metal comprises cobalt, gold, nickel, iron, silver, platinum, iridium, rhodium, palladium, rhenium, tungsten, molybdenum, tin, bismuth, zinc, lead, copper or a combination comprising at least one of the foregoing metals. 10. The process of claim 1, wherein the electrodeposition composition further comprises anions for the metal ions, the anions comprising sulfate, chloride, sulfite, perchlorate, bromide, citrate, tartrate, ethylenediamine, ethylenediaminetetracetic acid, or a combination comprising at least one of the foregoing anions. 11. The process of claim 1, wherein the suppressor comprises a polyether; a polyethylene oxide; a polyethylene glycol; a poloxamer; a poloxamine; alkylammonium cations, or a combination comprising at least one of the foregoing suppressors. 12. The process of claim 1, wherein the electrodeposition composition further comprises a leveler comprising an amine, a polyethyleneimine, a phenolphthalein, or alkylammonium cations; or a combination comprising at least one of the foregoing levelers. 13. The process of claim 1, wherein the electrodeposition composition further comprises chloride ,bromide, or iodide. 14. The process of claim 1, wherein the recess comprises: a depth from the field surface to the distal position that is from 10 nm to 900 μm, and an aspect ratio from 1 to 70. 15. The process of claim 1, wherein the recess comprises a through hole, a blind hole, or a combination comprising at least one of the foregoing recesses. 16. The process of claim 1, wherein the potential wave form comprises a single fixed applied potential. 17. A system for performing hysteretic current-voltage mediated void- free superconformal and bottom-up filling of recessed features of a substrate with a resistance member, the system comprising: a cell; an electrodeposition composition disposed in the cell and comprising: a metal electrolyte comprising a plurality of metal ions and a solvent; and a suppressor disposed in the solvent; and a hysteretic cyclic voltammogram; the substrate disposed in the cell in fluid communication with the electrodeposition composition and comprising: a field surface; and a recess disposed in the substrate, the recess comprising a distal position and a proximate position relative to the field surface of the substrate, such the recess is in contact with the electrodeposition composition, such that an applied electric potential of the recess is under potentiostatically or potentiodynamically control with a potential wave form; the resistance member in communication with the electrodeposition composition and the substrate, wherein the system is arranged and configured such that the resistance member in presence of the electrodeposition composition with the hysteretic cyclic voltammogram: autonomously reduces the deposition potential of the recess from that applied by the potential waveform, such that the recess is bifurcated into an active metal deposition region and a passive region in response to the deposition potential and ohmic variations of the substrate; whereby, in response to bifurcating the recess, a transition zone is formed at an interface of the active metal deposition region and the passive region, such that the transition zone progressively moves closer to the field surface and away from the distal position through the metal deposition to reduce the metal ions and form metal to deposit the metal in the active metal deposition region and not in the passive region; thereby forming a resistance enhanced superconformal filling in the recess of the substrate from the metal in the active metal deposition region, wherein the resistance enhanced superconformal filling is void- free, disposed in the recess in the active metal deposition region from the distal position to the transition zone, and absent in the passive region between the proximate position and the transition zone, such that forming the resistance enhanced superconformal filling occurs in consequence of autonomously reducing the deposition potential of the recess with the resistance member in a presence of the hysteretic cyclic voltammogram of the electrodeposition composition. 18. The system of claim 17, wherein the resistance member is selected from the group consisting essentially of a lumped resistor, a baffle, and a selected interelectrode separation distance between the substrate and a reference electrode in electrical communication with the electrodeposition composition. 19. The system of claim 18, wherein the resistance member is the lumped resistor, and the lumped resistor comprises a resistor in electrical communication with and electrically interposed between the substrate and a counter electrode in electrical communication with the electrodeposition composition. 20. The system of claim 18, wherein the resistance member is the baffle, and the baffle is in fluid communication with and fluidically interposed between the substrate and the reference electrode. 21. The system of claim 17, wherein the resistance member is the selected interelectrode separation distance between the substrate and a reference electrode in electrical communication with the electrodeposition composition, and the process further comprises adjusting the interelectrode separation. 22. The system of claim 17, wherein the hysteretic cyclic voltammogram comprises an S-shaped negative differential resistance. 23. The system of claim 17, wherein the metal ions comprise Fe2+, Fe3+, Pt2+, Pt4+, Ir3+, Ir4+, Rh3+, Pd2+, Co2+, Ni2+, Au3+, Zn2+, Bi3+, Pb2+, Re7+,Au+, Ag+, Sn2+, W6+, Mo6+,Cu2+, Cu+, or a combination comprising at least one of the foregoing metal ions, and the metal comprises cobalt, gold, nickel, iron, silver, platinum, iridium, rhodium, palladium, rhenium, tungsten, molybdenum, tin, bismuth, zinc, lead, copper or a combination comprising at least one of the foregoing metals. 24. The system of claim 17, wherein the electrodeposition composition further comprises anions for the metal ions, the anions comprising sulfate, chloride, sulfite, perchlorate, bromide, citrate, tartrate, ethylenediamine, ethylenediaminetetracetic acid, or a combination comprising at least one of the foregoing anions. 25. The system of claim 17, wherein the suppressor comprises a polyether; a polyethylene oxide; a polyethylene glycol; a poloxamer; a poloxamine; alkylammonium cations, or a combination comprising at least one of the foregoing suppressors. 26. The system of claim 17, wherein the electrodeposition composition further comprises a leveler comprising an amine, a polyethyleneimine, a phenolphthalein, or alkylammonium cations; or a combination comprising at least one of the foregoing levelers. 27. The system of claim 17, wherein the electrodeposition composition further comprises chloride ,bromide, or iodide. 28. The system of claim 17, wherein the recess comprises: a depth from the field surface to the distal position that is from 10 nm to 900 μm, and an aspect ratio from 1 to 70. 29. The system of claim 17, wherein the recess comprises a through hole, a blind hole, or a combination comprising at least one of the foregoing recesses. |
[00117] COMPUTATIONAL METHODS
[00118] Finite element method (FEM) computations are used to simulate copper electrodeposition in 2D axisymmetric annular and cylindrical through-silicon vias and 2D trench arrays. The dimensions of the annular TSV (Ri = 4 μm, Ro = 9.5 μm, and H = 56 μm) match those of prior experimental systems. Simpler cylindrical TSV were also examined having radii of R cyl = 5 pm and heights, H cyl , ranging from 50 μm to 200 μm corresponding to aspect ratios from 5 to 20. Trench array were simulated in a 4x1 configuration with dimensions of trench width 144 = 10 μm, trench height Ht = 50 μm, and pitch Pt = 20 μm between trench edges. More complex trench arrays having a 3x1 configuration with varying trench width from 10 μm to 30 μm and trench height from 50 μm to 100 μm were also examined. For all workpiece geometries, the reference and counter electrode were combined in a common plane opposite the working electrode at a distance L, and the electric potential is fixed at zero. For the base case, L was taken to be 0.25 cm, thereby specifying the uncompensated resistance between the working and reference electrode. The hydrodynamic boundary layer thickness 5 is set to 25 μm in all simulations and the concentration of each species (Cu 2+ , Cl; and polymer) is set equal to that of the bulk solution at these boundaries. The solution conductivity K for the 1 .0 mol/L CuSCO 4 - 0.5 mol/L H 2 SO 4 equals 15.26 S/m. Each simulation begins with a 2 min incubation period that emulates an experimental setup where patterned electrode fragments are pretreated with an ethanol wetting solution prior to insertion into the plating solution. Computationally, this is approximated with an applied potential of -0.40 V (or current equal to 1 % of the set value) for 2 min before stepping to the set value. Initial concentrations are in the electrolyte domain below the hydrodynamic boundary layer to emulate electrolyte exchange with the ethanol filled features. [00119] Variation in the uncompensated resistance between the reference and working electrodes was examined using three different schemes: (a) alteration of the distance between the reference and working electrodes, (b) insertion of a baffle into the electrolyte phase with fine scale porosity that effectively increases the resistance between the working and reference electrodes, and (c) insertion of an external resistor between the sensing point on the working electrode and the actual metal/electrolyte interface. All three approaches have been used by experimentalists for various ends. The first provides a simple method to evaluate the effect of the electrochemical cell time constant (t ~ RC) while a baffle has been used in commercial electroplating cells to counter the terminal effect that otherwise leads to non-uniform deposition on resistive seed layers. The last approach, namely the introduction of the external resistor in series with a 2-electrode cell, has been used to examine the effects of variation in the experimental control mode, from potentiostatic to galvanostatic, and its impact on pattern formation in bistable systems. [00120] The impact of uncompensated resistance on via filling was examined using the three different configurations outlined above. Given the symmetry of the cell and workpiece, despite differences in the location of the uncompensated resistance within the cell, the net effect on feature filling is the same, provided that the baffle is located sufficiently far from the working electrode to not influence the development of chemical gradients. Increasing the uncompensated resistance relative to that for the default geometry is investigated. Considering the uncompensated ohmic drop in the context of the electrolyte conductivity and the geometric configuration of the cell, with the working and counter electrode positioned at opposite end of circular tube of radius R c , allows the ohmic losses due to current flow in the electrolyte to be defined by
[00121 ] For a given workpiece, area is taken to define the cross sections of the ceil so that changes in Ωs are accomplished by adjusting the position of the RE/CE, L or through the solution conductivity, K. A series of calculations were performed where Ωs is increased by a factor of 6x, 9x, 11 x, 21 x, 41 x, 81 x, 161 x, or 401 x. This can be realized by an increase in L from 0.25 cm to 1.5 cm, 2.25 cm, 2.75 cm, 5.25 cm, 10.25 cm, 20.25 cm, 40.25 cm, and 100.25 cm, respectively. Alternatively, the same effect can be obtained, perhaps more conveniently, by holding L at 0.25 cm and increasing the effect of K by insertion of a porous but insulating baffle whose net effect on ohmic losses is a composite of the respective materials properties. This is realized by making a slice of the electrolyte adjacent to the RE/CE plane having a thickness of 25 pm have a conductivity smaller, e.g. 500x to 20,000x, than that of the actual electrolyte conductivity, K, effectively increasing the uncompensated resistance, e.g. by factor of 11 x to 201 x, greater than the default Ωs based on the homogenous K solution resistance and L = 0.25 cm. Going further, the same effects can be obtained, again perhaps more conveniently, by inputting an external resistor, that is some multiple of Ωs, between the WE and RE contacts (or CE/RE plane in the current construct). In fact, as long as the equivalent total cell resistance is the same, simulations produce identical results regardless of which method for adjusting resistance as outlined above Is implemented. Nevertheless, of the three methods, adjusting the position of the RE/CE plane is the most computationally demanding as it requires additional meshing that leads to longer computational times. From an experimental standpoint, integrating uncompensated resistance as a variable control parameter into an existing 2- or 3-electrode system is most easily realized by the external resistor approach. As such, all simulations presented herein use an external resistor to adjust the overpotential to account for the potential drop associated with the deposition current flowing the external series resistor in order to model the impact of a modified global cell resistance.
[00122] The concentration Ci and flux N, of each species in the electrolyte domain is described by the Nernst-Planck equation, capturing both diffusion and electromigration, such that the evolution of concentration is given by given the species’ charge z i , diffusion coefficient D i , Faraday’s constant F, and mobility u m,i calculated by the Einstein relationship [00123] The simulated electrolyte assumes full dissociation of CuSO 4 and NaCl, reasonable for the relevant concentrations Cu 2+ and Cl-, ignoring hydronium, sulfate, bisulfate, and sodium species. The poloxamine suppressor (subscript P) is assumed to be neutral in charge (z p = 0). Diffusion coefficients listed in FIG. 27 for Cu 2+ , Cl-, and poloxamine tetronic (TET) are taken or estimated from literature sources. [00124] Due to the high concentration of CuSO 4 and H2SO 4 supporting electrolyte, potential in the electrolyte (כ) is well approximated by Laplace’s equation which neglects potential variation in solution arising from ionic gradients. The current density ଔറ associated with the Cu 2+ flux through the electrolyte is given by Ohm’s law where ^ is the solution conductivity. A zero flux symmetry condition is imposed on the side of the cell (at r = R c in axisymmetric simulations and x = 0 and x = 12W t in 2D simulations) for gradients of solution potential and gradients of concentration [00125] Accumulation of adsorbates on the electrode follows Langmuir adsorption kinetics with deactivation of suppression related to metal deposition involving a combination of desorption and/or incorporation into the growing deposit. Evolution of the fractional chloride coverage θ Cl , defined as the surface concentration divided by the saturation coverage, is described by where is the adsorption rate constant, C Cl is the chloride concentration at the evolving metal/electrolyte interface, the deactivation rate constant and υ is the metal deposition rate. Evolution of the fractional poloxamine coverage θ P, is described by the analogous where the poloxamine is restricted to adsorption on top of the halide covered sites and thereby implicitly subject to the requirement that θ P cannot exceed θ Cl through adsorption. The fractional chloride and poloxamine coverages are further limited to values between 0 and 1. Values for listed in FIG. 27 are estimated from model fits to the S-NDR voltammetry with the fitting procedure focused on capturing the critical onset potential for suppressor breakdown as a function of suppressor and halide concentration. 20,44 In previous efforts the critical breakdown potential was shown to be dominated by the higher rate of Cl- adsorption and consumption relative to that of the polymer. Once a fully developed Cl-polyether adlayer exists, Cl- consumption into the solid by the second term in Eq. 7 can make the adsorption term for the suppressor in Eq.8 function effectively as a desorption term. [00126] The metal deposition rate is assumed to be a function of the suppressor coverage θ P (or equivalently, coverage of the polyether-chloride bi- layer), metal ion concentration C Cu , and overpotential η at the interface, thus [00127] The current densities on unsuppressed (j θ=0 ) and suppressed (j θ=1 ) surfaces for the two electron reduction of Cu 2+ to its metallic form are translated into growth velocity, u, using Faraday's constant (F = 96,485 C-mol" 1), the ionic charge n, and the molar volume Q of solid copper. This simple form captures suppression arising from the polyether coverage (as limited by chloride coverage). The current densities (j θ=0,01 ) are assumed to exhibit the conventional exponential dependence on overpotential q by
[00128] The applied potential V app is related to the working electrode through where the potential Φ within the electrolyte is evaluated at the electrolyte/deposit interface to capture the dissipative losses between the workpiece and the reference electrode associated with current flow through the electrolyte and external resistor layer. The overpotential driving electrodeposition is referenced to the reversible Nernst potential for the Cu 2+/ Cu reaction. The values of for the fully suppressed surface are obtained by fitting the negative- going voltammetric scans up to the onset of suppression breakdown. Although the kinetics of metal deposition on polymer-free surfaces are known to be a function of halide coverage, for simplicity, the present work uses a single set of values for deposition on the polymer-free surface.
[00129] The local current density at the electrode is equated to the Cu 2+ flux from the electrolyte onto the electrode interface (outward surface normal ) according to
Similarly, the normal fluxes of chloride and polyether from the electrolyte to the interface are equated to the rates of their adsorption yielding and with saturation coverages estimated. As stated previously, the ( θ Cl - θ p ) term captures the requirement that suppressor adsorption only occurs at chloride covered surface sites.
[00130] The full system of equations is solved numerically in 2D and 2D-axisymmetric geometries using a finite element method employed in the COMSOL Multiphysics version 5.5 software package with the default solver, implementing the following modules: tertiary current distribution, primary current distribution, separate coefficient form boundary partial differential equations for both chloride and suppressor, and deformed geometry. The 2D triangular mesh elements are more highly refined along the electrode interface, their dimensions initially equal to 20 % of the feature radius or width on each side. The mesh scales up to a maximum of 2.5 μm within the boundary layer and 26 pm outside of the boundary layer. The mesh in the thin resistive layer is also refined, having a maximum size of 5.2 μm. Automatic remeshing is enabled, prompting re-mesh when the maximum mesh distortion parameter exceeded 1 .56. A moving boundary smoothing parameter of 2, geometry shape order of 1 , and Laplace mesh smoothing type are used in the deformed geometry module (see COMSOL documentation for detailed explanation on how these parameters impact moving boundary convergence). The system of equations was solved so that the overall charge imbalance (the fractional difference between the total integrated currents at the counter and working electrodes) was less than 0.02 %. The numerical evaluation error, thus, is acceptably small for the present purposes. To give a sense of the computational expense, the smaller geometry simulations having 1800 domain and 400 boundary mesh elements take on the order of 5 minutes to compute. Larger simulations having 5000 domain and 800 boundary mesh elements takes on the order of 30 minutes to compute. All simulations used a Dell Precision 3630 desktop computer with an Intel Xeon E-2186G CPU @ 3.80 GHz and 64 GB RAM using a Windows 10 Enterprise 64-bit operating system. [00131 ] RESULTS AND DISCUSSION
[00132] Experimental copper deposition in through-silicon vias (TSV) in electrolytes having a single suppressing polyether additive for a range of chloride concentrations (2 μmol/L to 1000 μmol/L) and combinations of CuSO 4 and H 2 SO 4 concentrations were performed. In low chloride solutions (< 80 μmol/L), copper deposition initially occurs on the bottom of the via as well as the neighboring sidewalls up to a position marking a transition between active and passive plating regions. This transition point shifts upward in the via with lower chloride concentrations or more negative potentials. At fixed potential, deposition is eventually quenched at a position within the via determined by the balance between transport constrained adsorption of the suppressing additives and its disruption by the metal deposition reaction. Thus, for these low chloride electrolytes it is necessary to step or ramp the potential to more negative values in order to fully fill features. This approach requires tuning of the applied potential waveform to optimize filling. For stepped potentials, the discontinuous nature of the increase in available free energy might be expected to impact, or at least mark, the deposit microstructure. Galvanostatic deposition provides both operational simplicity and cost advantages in process control, congruent with its use in industrial electroplating practice. Deposition is sustained as long as the current is applied and with proper optimization void free filling is possible. If the applied current is too high void formation will occur while too low a value will result in uneven activation of deposition across the workpiece. Even with an appropriate value of applied current substantial under- or overfill will occur if the deposition time is not tuned appropriately for each substrate pattern. Alternatively, the spontaneous self-passivation associated with potential- controlled deposition offers an alternative path to feature filling that should be less sensitive to variations in pattern density on the work piece.
[00133] Simulations in annular TSV - FIG. 17 shows simulated final growth profiles and interface contours (in 6 min intervals) for potentiostatic and galvanostatic deposition in 1 mol/L CuSO 4 , 0.5 mol/L H 2 SO 4 , 40 μmol/L TET, and 80 μmol/L Cl- for the annular TSV configuration shown schematically in FIG. 17a. The cell and workpiece geometry has been previously detailed both experimentally and computationally for potential-controlled deposition with L = 0.25 cm. Simulations of potentiostatic deposition do not predict complete filling even under conditions yielding deposition localized to the via bottom, rather passivation occurs less than halfway up the via after ≈ 20 min at -0.54 V. Deposition at -0.52 V passivates even earlier, after ® 6 min, filling just the lowermost 4 pm of the 56 pm tall feature. At a more reducing potential of -0.56 V deposition is predicted to yield void formation after 8.5 min. Galvanostatic deposition, conversely, results in a nearly filled feature at -0.10 μA after 1 hour of deposition with the applied current appropriately tuned to the relevant electrochemically active surface area in the present case. A factor of 2 decrease in the applied current results in only 1 /3 of the feature being filled after 1 h before localized bottom-up deposition is lost and the applied current redistributes across the entire surface corresponding to a 7x decrease from peak current density during bottom-up fill. At the other limit, a doubling of the applied current produces a void after 9 min.
[00134] FIG. 17 shows (a) a schematic of the axisymmetric geometry used in the S-NDR model to simulate deposition in the annular TSV with dimensions of Ri = 4 pm, Ro = 9.5 pm, and Hann = 56 pm. Relevant domains and boundaries are indicated, (b) Simulated growth contours in 6 min intervals (left-hand via) and final deposit positions (right-hand via) for potentiostatic (top) and galvanostatic (bottom) copper electrodeposition after 1 h at the indicated operating conditions, (c) Current and (d) overpotential, η, transients for the indicated simulations presented in (b). Full seam-free and void-free filling can only be obtained for a narrow window of galvanostatic deposition conditions and is not obtainable at all for potentiostatic conditions.
[00135] Current and potential transients in FIG. 17c and FIG. 17d, respectively, provide insight into the growth dynamics under potentiostatic deposition at -0.54 V and -0.56 V for the base case with an uncompensated resistance of 1 W s and, for the same geometry, galvanostatic deposition at -0.10 μA and -0.20 μA. The appropriate area for current density scaling is somewhat of a mystery for bifurcating systems with a temporally varying total electrode area. That said, an applied current of -0.10 μA corresponds to -4.7 mA/cm 2 when scaling by projected electrode area, -42.9 mA/cm 2 when using only the via bottom, and -1.45 mA/cm 2 when using the entire electrode interface, signifying the available current density range depending on what is truly the electrochemicaily relevant surface area. Each simulation begins with a 2 min incubation period to emulate experiments where the electrolyte mixes with an ethanol wetting solution, as described in detail in the preceding section. After stepping to -0.54 V the current sharply rises to -0.17 mA followed by a slow decent associated with Cu 2+ depletion as deposition begins on the bottom surface and the adjacent sidewalls. This is followed by a current inflection near 10 min that correlates to the area reduction of the growth front during the transition from conformal deposition on the via bottom and adjacent sidewalis to bottom-up filling, evident in growth front profiles at 8 and 12 minutes shown in FIG. 17b. Following the brief rise in current to reach a local maximum near - 0.13 mA the onset of passivation begins as the suppressor phase reforms and the current descends to a final passive current plateau of ~ -0.02 mA for the workpiece. Stepping to a more aggressive applied potential of -0.56 V leads to an initial current in excess of -0.20 mA followed by a slow current decay for the first few minutes as depositions develops in the lower sidewalls and via bottom; however, the resulting Cu 2÷ depletion followed by sidewall collision results in void formation by 9 min which halts further simulation. In the galvanostatic simulations, the applied potential (and thus overpotential) of the working electrode varies with time. For deposition at -0.10 mA the overpotential transient exhibits a gradual increase as bottom-up filling of the via progresses to almost fill the via by 1 h. The increase in driving force is required to sustain the applied current in the face of increased suppressor flux as the unfilled via depth shrinks and gradient becomes steeper. A doubling of the applied current to -0.20 mA yields a high overpotential close to -0.16 V that remains constant during a period of conformal growth on the lower sidewalis and bottom surface until sidewall impingement and void formation halts the simulation, anaiogous to the case for poientiostatic deposition at an excessively negative applied potential of -0.56 V.
[00138] FIG. 18 shows simulated final growth profiles and positional contours (in 6 min intervals) for poientiostatic deposition in the annular TSV depicted in FIG. 17a but with an additional uncompensated resistance equal to (a) 10 Ωs, and (b) 40 Ωs (corresponding to 11 x and 41 x, respectively, higher total cell resistance relative to the base case). For both conditions, it is evident that the increase in resistance permits more complete filling of the annular TSV at a constant applied potential following 1 h of deposition. As with gaivanostatic deposition in FIG, 17, the applied potential must be chosen appropriate to the pattern density and geometry but, significantly, by appropriate engineering of the uncompensated resistance full bottom-up filling can be accomplished at a fixed applied potential, analogous to earlier TSV filling albeit at higher C!- concentrations. 18 At potentials that are too positive the system fails to activate and the current is distributed uniformly across the passivated surface; at the other extrema, when the applied potentials are too negative, a seam or void is formed. Simulations (not shown) reveal that applied potentials 20 mV more negative than those shown in FIG. 18 result in seams or voids. It is also noteworthy that the potential window between complete interface passivation and voided via is significantly widened with the increase in uncompensated resistance, shifting from 40 mV (for partial bottom-up filling as in Fig. 1) for the base case with no added resistance (1Ωs) to 140 mV and 420 mV with total uncompensated cell resistances equivalent to 11Ωs and 41 Ωs, respectively. Further still, one can readily envision tuning the resistance and/or applied potential such that deposition passivates just before the via outlet, a phenomenon difficult to achieve by a gaivanostatic approach.
[00137] In an embodiment FIG. 18 shows simulated growth contours in 6 min intervals (left-hand via) and final interface positions (right-hand via) for potentiostatic copper deposition in the annular via at the Indicated potentials with an increased system resistance equivalent to (a) 11 x and (b) 41 x the system resistance associated with the electrodeposition composition in the cell alone, Qs. Schematics show a 1-D representation of the resistances in the circuit. Corresponding (c) current and (d) overpotential, h, transients are included for the passive (-0.58 V, -0.84 V), partial fill (-0.62 V, -0.84 V) and full fill (-0.88 V, -1.01 V) characteristic deposition profiles. Use of sufficiently large system resistances permits seam-free and void-free filling at -0.68 V and -1.04 V under potentiostatic conditions. The increased system resistance may be obtained using an externai resistance, a semipermeable or permeable resistive baffle or membrane or expanding a distance (800) between the reference (500) and substrate (800) electrode, in FIG. 15 and FIG. 16,
[00138] The current and overpotentiai transients associated with the filling simulations are shown in FIG. 18c and FIG. 18d, respectively, categorized as passive, partial fill, or full fill according to the final deposition profiles (left, middle, and right, respectively, of FIG. 18b). In each case the current spikes to a more negative value after the initial 2 min incubation period. The current gradually falls as the via is filled; current reduction occurs faster for lower applied potentials and for the smaller value of uncompensated resistance. At less reducing applied potentials ( passive , shown as dotted lines ■ ■ ■ ), the more rapid current reduction ends with universal interface passivation and sustained deposition currents near -25 nA that are distributed over the remaining deposit surfaces for both values of uncompensated resistance. The small difference in the final current, for the -0.56 V and -0.64 V examples, reflects the more negative overpotentiai for the latter as evident in the respective transients shown in FIG. 18d; both profiles rise quickly to « -0.16 V and -0.165 V, respectively, at which value they remain constant for the duration of the 1 h simulation. The transients for the intermediate applied potentials (partial fill, shown as dashed lines — ) exhibit a more gradual reduction in current after the initial spike (FIG. 18c), plateauing after 35 min of deposition for the 11 Q S case and 55 min for the 41 Q s case (the latter evident in plots extending to longer times). The current in the simulation with a total uncompensated cell resistance of 1 1 Qs decreases (in magnitude) by about 0.06 mA while that for 41 Q s decreases by about 0.02 pA. The more stable current profile for a higher cell resistance reflects the transition towards gaivanostatic control. Interestingly, the current transients at intermediate ( — ) and more negative (full fill, shown as solid lines — ) applied potentials exhibit similar trends, with the latter shifted to larger current values. With 110 s of uncompensated resistance, the current plateaus at a similar time for both applied potentials. The higher current at -0.68 V yields more sustained localized deposition that results in a fully filled via. With 41 Q s of uncompensated resistance, continued motion explicit in the last two growth contours at -0.84 V suggests that the feature might fill completely given sufficient deposition time. The simulation at -1.04 V reaches the field (i.e., z = 0) sooner because of the higher overall current throughout the process (although it is clear that more material is also deposited on the field). It is notable that the overpotential profiles for both full fill simulations, i.e., at -0.68 V with 11Ωs and at -1.04 V with 41 Ωs, are nearly identical for the entirety of the simulations. With filling contours that are quite similar almost through full filling, it is understood that the trend reflects the increased driving force required to advance the growth front nearer the via opening due to the enhanced transport of the suppressing additives and despite the increased transport of metal ion.
[00139] Simulations of the impact of uncompensated resistance on the filling of the annular TSVs were expanded to explore the full range of applied potentials between passive and voided growth profiles and include an intermediate value for total uncompensated resistance of 21 Ωs. FIG. 19a shows the lowest position on the TSV interface after 1 h of deposition for the indicated uncompensated resistances as a function of applied potential. As seen in FIG. 17, potentiostatic control with 1 W s uncompensated resistance only achieves a fill height of 23 pm before the next -20 mV increment of applied potential results in void formation, increasing the uncompensated resistance expands the applied potential window between fully passive and voided filling while also permitting higher possible fill heights for constant potential simulations; the interval ranges from positive of -0.7 V to negative of -1.0 V, more than 0.3 V, with 41 Ωs of total uncompensated resistance as reflected in the data in FIG. 19a. The impact of each -20 mV increment decreases at more negative applied potentials particularly for the 41 Ωs data. The curve reflects the balance of transport, which scales with the concentration gradients and thus increasing with 1 /unfilled depth (i.e., increasing asymptotically as deposition height approaches via depth), with the potential dependent deposition kinetics that drives disruption of the suppressor layer. FIG. 19b plots the time needed to achieve 90 % fill of the annular TSV as a function of the applied potential for each resistance. Higher uncompensated resistance also enables faster filling of the TSV, reaching 50.4 pm of height as quickly as 38 min for 41 Ωs, 42 min for 21 Ωs, and 53 min for 11 Ωs. As in FIG. 18, decreased fill time with increased uncompensated resistance is associated with more stable operating currents throughout filling consistent with a trend towards galvanostatic operation,
[00140] FIG. 19 Panel (a) shows a chart of the lowest position (at the centerline due to symmetry) on the annular TSV interface after 1 h of deposition as a function of applied potential for the indicated increased system resistance. Simulations 20 mV more negative of the most negative potentials shown for each resistance result in a seam or void. The dashed line represents the deposit height for a fully filled via, FIG. 19 Panel (b) shows a chart of the time needed for filling along the centerline to reach 90 % of the via height plotted as a function of applied potential for the indicated uncompensated cell resistance. Simulations requiring more than 60 min deposition time to achieve 90 % fill are not included. The increased system resistance broadens the range of applied potentials yielding seam-free and void-free filling, including filling that passivates near the field surface. The increased system resistance can be obtained using an external resistance, a semipermeable or permeable resistive baffle or membrane or increased separation between the reference and substrate..
[00141 ] Simulations in cylindrical TSV - The simulations of deposition in annular TSV in the preceding section offer prediction that can be validated against prior experimental work. 20 This section explores the influence of uncompensated resistance on deposition in the more generic cylindrical TSV geometry depicted in FIG. 20a. Similar to the annular TSV, simulations of potentiostatic deposition with 1Ws uncompensated resistance (again, determined by L = 0.25 cm and R c by Eq. 1 ) in the 5 μm radius and 50 μm tall via in FIG. 20b depict a narrow operating window between passive (-0.52 V) and voided (-0.56 V) growth profiles. Achieving a fully filled cylindrical TSV again requires a potentiodynamic waveform that progressively shifts the location of the sidewall passive-active transition upwards like that detailed in previous work. As with the annular TSV, galvanostatic deposition enables void- free bottom up via filling that is almost complete after 1 h of deposition at an applied current of -0.05 μA as shown in FIG, 20c. However, closer inspection (not shown) reveals that the majority of the via is filled in the first 30 minutes of deposition before additive transport is sufficient to passivate deposition and shift much of the current towards deposition on the TSV field. Of course, galvanostatic methods still require tuning of the applied current to avoid conformal, passive deposition on the via (as at -0.02 μA) or voided deposits (as at -0.08 μA). FIGs. 20d and 20e show simulations of potentiostatic deposition with an increase in uncompensated resistance to 11 Ωs and 101 Ωs, respectively. As with the annular geometry, simulations with the additional uncompensated resistance predict higher filling within the cylindrical TSV for the given electrolyte chemistry. After 1 h, the deposit height reaches -22 pm for the 10 Ωs case at -0.80 V and -8 pm for the 100Ωs case at -1 .12 V. For each resistor, the simulations at potentials 20 mV more negative than these values predict a seam or void.
[00142] FIG, 20 shows (a) a schematic of the axisymmetric geometry used in the 8-NDR model to simulate deposition in a cylindrical TSV with dimensions of Rcyi = 5 pm and Hcyl = 50 pm. Relevant domains and boundaries are indicated. Simulated growth profiles for (b) potentiostatic and (c) galvanostatic copper electrodeposition after 1 h at the indicated applied potentials and currents with 1 Ωs of system resistance. Simulated growth profiles after 1 h of potentiostatic copper electrodeposition with system resistance of (d) 11Ωs and (e) 101Ωs. Only potentiostatic deposition with the increased system resistance permits seam-free and void-free filling that approaches near the field surface. The increased system resistance may be obtained using an externa! resistance, a semipermeabie or permeable resistive baffle or membrane or increased separation between the reference and substrate.
[00143] The interesting behaviors uncovered thus far motivated further exploration of the effect of uncompensated resistance on the filling of even higher aspect ratio features. FIG. 21a shows simulated growth contours in 8 min intervals for the indicated control-method, operating condition, and added uncompensated resistance for the cylindrical TSV geometry presented in FIG. 20a (having an aspect ratio of 5). The potentiostatic deposition at -0,54 V for a ceil with 1 Ws passivates shortly after 12 minutes of deposition, having achieved a deposit height of only 12 pm. Note, each condition presented in FIG. 21 is the most reducing applied potential or current before simulations produce a seam to a resolution of 20 mV and 2 nA intervals. Galvanostatic deposition at -72 nA permits active bottom-up deposition within the via for 24 minutes before shifting to a slower conformal deposition mode where the last 30 minutes of deposition only results in the addition of 3.5 μm to the 43 pm tail deposit. This tuned current corresponds to current densities of -3.39 mA/cm 2 when scaled by projected area, -91 .7 mA/cm 2 when scaled by area of the via bottom, and -1 .95 mA/cm 2 when scaled by the entire electrode interface. For potentiostatic deposition with 11 Ωs of uncompensated resistance an applied potential of -0.60 V yields maximum filling with active deposition for ≈ 21 min before full passivation occurs. Potentiostatic deposition with 101Ωs of total uncompensated resistance and an applied potential of -1.12 V yields maximum filling with a profile that is remarkably similar to that for galvanostatic deposition at -72 nA; a shift from bottom~up growth to conformal deposition occurs after 24 min.
[00144] FIG. 21 shows simulated growth contours for a cylindrical TSV with a 5 μm radius and heights of (a) 50 μm, (b) 100 μm, and (c) 200 μm for the indicated applied potentials/current and system resistance. Final contours are at 1 h, 2 h, and 8 h with contour spacing of 6 min, 10 min, and 40 min for TSV with heights of 50 μm, 100 μm, and 200 μm, respectively. The specified applied potentials are the most negative values that do not result in seam or void formation and thus reflect the highest filling along the via centerline obtained for the given value of system resistance (in increments of 20 mV). Neither galvanostatic nor potentiostatic deposition alone are capable of providing seam-free and void-free filling that approaches near to the field surface in the vias having the higher aspect ratios of 10 and 20, but higher system resistance enables higher filling and the broadest potential window and essentially matches the highest filling that can be obtained using a narrow window of currents. The increased system resistance may be obtained using an external resistance, a semipermeable or permeable resistive baffle or membrane or increased separation between the reference and substrate. [00145] A sequence of simulations are presented for deeper cylindrical TSVs having aspect ratios of 10 and 20 (R cyl = 5 μm ) in FIGs. 21b and 21c, respectively. In general, the trends in deposition profiles across the control- methods and uncompensated resistance match those in FIG.21a. Specifically, potentiostatic deposition with 1Ws results in early passivation deep within the via whereas a substantially larger uncompensated resistance shifts the final deposit height higher, eventually approaching the height achieved using a galvanostatic approach for the largest values of uncompensated resistance. In higher aspect ratio vias the applied potential/current must be less reducing, regardless of control method or uncompensated resistance, to avoid Cu 2+ depletion and formation of a seam or void. For galvanostatic deposition, the most reducing current is roughly halved for each doubling of the via depth, consistent with consideration of the gradient that underlies reactant and additive transport. In terms of current densities, the aspect ratio 10 and 20 features correspond to -1.51 mA/cm 2 and -0.75 mA/cm 2 when scaled by projected area, -40.7 mA/cm 2 and -20.4 mA/cm 2 when scaled by area of the via bottom, and - 0.61 mA/cm 2 and -0.19 mA/cm 2 when scaled by the total electrode interface area, respectively. Despite the aggressive selection of applied potentials and currents, passivation occurs before the deposit reaches the field for each geometry. Although not the focus of this study, fill height would be improved by reducing the flux of suppressor additives by either decreasing chloride concentration or reducing convective transport (i.e., increasing the boundary layer thickness in the model). The final deposit height (lowest point on interface which, by symmetry, lies at the via middle) as functions of applied potential for the cylindrical vias with aspect ratios of 5, 10, and 20 are presented in FIG.22a, 22b, and 22c, respectively, for the indicated values of total uncompensated resistance. As with the annular geometry, the fill height for each geometry increases and the potential window between fully passive and voided filling profiles widens as the uncompensated resistance is increased. Curvature reflecting the balance between transport, with its asymptotic dependence on via depth minus fill height, with deposition rate dependent disruption of the suppressor layer is observed again as well. [00146] FIG. 22 shows charts of the lowest centerline position on the cylindrical TSV interface as a function of applied potentiai for the indicated total ceil resistances and deposition times for 5 μm radius TSV. Panel (a) for 50 μm, Panel (b) for 100 μm, and Panel (c) for 200 μm height. Higher values of system resistance permit higher filling over a broader range of applied potentiai. The increased system resistance may be obtained using an external resistance, a semipermeable or permeable resistive baffle or membrane or increased separation between the reference and substrate.
[00147] Deposition in french arrays - The influence of uncompensated resistance on Cu deposition in high aspect ratio features was further explored in trench arrays as depicted in the 2D geometric configuration of FIG. 23a. Trenches are 10 μm wide, 50 μm tali, 1 mm deep (into the page), and are spaced 30 μm apart center-to-center ( Pt = 20 μm). Simulations of galvanostatic operation produce fully filled trench arrays at currents ranging from -6 μA to - 30 μA, the major difference being the time necessary to achieve complete fill. Currents in excess of -30 μA produce voids towards the bottom of the trench while currents smaller than -6 μA result in conformal deposition over the entire electrode Interface. For -30 μA of applied current, the current densities are -25 mA/cm 2 when scaling by projected area, -75 mA/cm 2 when scaling by the area of the trench bottom, and -5.77 mA/cm 2 when scaling by the entire electrode interface. Final deposit profiles and contours shown in FIG. 23b indicate the electrode interface position spaced in 10 min intervals for gaivanostatic conditions that enable full filling. For all three conditions the trench array is completely filled, with deposition at -8 μA requiring 80 min longer than -30 μA. For the two higher applied currents the resulting overburden is non-uniform across the array. The interface contours (colorized to indicate time) show that deposition at -30 μA occurs uniformly in all 4 trenches from 2 min to 32 min. After this time, deposition is localized to the 1 st trench and shifts across the array from left to right. The charts in FIG. 23d show the center-line position as a function of time for each trench depicted in FIG. 23b. The profiles for all 4 trenches at -30 μA are identical until ≈ 32 min; shortly after this time, deposition continues in trenches 1 and 2 while trenches 3 and 4 have passivated. Eventually, the 3 rd and 4 th trenches reactivate and achieve a similar height across the array at 52 min of deposition. At i ~ 41 min, all 4 trenches have reached the field (position of y = 0), although a difference in height of ~ 10 μm exists between trenches 1 and 4.
[00148] In an embodiment with multiple features on a single substrate FIG. 23 shows (a) a schematic of the 2~D geometry used in the S-NDR model to simulate deposition in trench arrays with dimensions of Wt = 10 μm, Ht = 50 μm, and Pt = 20 μm , and length of 1 mm (into the page). Simulated growth profiles for (b) galvanostatic and (c) potentiostatic copper electrodeposition at the indicated operating conditions and final times with a system resistance of 11 Ωs. Contour lines represent the individual trench centerline positions of the electrode interface, spaced in 10 min intervals and beginning after the pretreatment step, colorized to indicate time, (d-e) Charts showing the centerline position of the growth front for each individual trench (denoted as T1 , T2, T3, and T4 from left-to-right) as time progresses with annotations indicating the time when all trenches reach y = 0. It is seen that complete seam-free and void-free filling of all four features occurs sequentially under galvanostatic conditions of smaller current and with substantial deposit on the field surface because active deposition shifts between the recessed features because deposition in the S-NDR electrodeposition composition is unstable to small perturbations under galvanostatic conditions of higher currents. Potentiostatic deposition with a small system resistance of 1 Ωs yields a very narrow process window, iess than 20 mV, in which seam-free and void-free filling is obtained but with the larger system resistance of 11 Ωs shown here an increased potential range, 380 mV, an order of magnitude improvement, yields seam-free and void-free filling. The increased system resistance may be obtained using an external resistance, a semipermeable or permeable resistive baffle or membrane or increased separation between the reference and substrate.
[00149] Deposition at the lower applied current of -16 μA exhibits analogous behavior, although deviation of height among the trenches in the array begins at a lower height within the trench after ~ 35 min. In this case, deposition dynamically passivates and re-activates within various trenches from 35 min to 95 min, at which point the profiles all merge again. Similar to the -30 μA condition, there is an ≈ 10 μm difference in height between trenches 1 and 4 at t = 67 min although in this simulation trench 1 is the last to reach the field (y = 0). Deviation across the trench array occurs at even earlier times for a lower applied current of ~8 μA, after only 3 minutes of deposition, and progresses in almost discrete steps, not unlike the sequential activation seen in the case of microelectrode arrays under controlled current where the total current is a globally conserved quantity. At 35 min the difference in height between trenches 1 and 4 is 34 pm. After 110 min the individual trenches have all reached roughly the same height, breaking the y = 0 threshold at 124 min.
[00150] Inherent to the S-NDR system is competition between interface activation driven by disruption of the suppressor layer by metal deposition that is balanced against interface passivation driven by additive adsorption. In short, under galvanostatic control, higher applied currents for a given active area are associated with higher rates of deposition and increased suppressor disruption associated with halide incorporation at larger overpotentials. If the applied current is sufficiently high then transport limited passivation is insufficient to shut down deposition within the feature (suppressor flux being reduced deeper within a feature) and the deposition profiles across the array are uniform. However, the additive flux avallable to passivate active deposition in a feature increases as filling proceeds upward. At some point, variations in the geometry between the individual features, or its numerical simulation, occur such that a portion of the active interface can passivate and redistribute the current to other features. Subsequently the inverse can occur as well, where passive features reactivate and draw current from other sites. Such localized passivation and reactivation is the origin of the non-uniform profiles across the arrays in FIG, 23. Improved uniformity within the array at - 30 μA relative to that at -16 μA and -8 μA reflects the higher suppressor flux required to passivate the surface when operating at the higher deposition (and adsorbate consumption) rates and correspondingly higher overpotentials, subject to the limits on additive transport defined principally by depth within the trench. For the value of applied current that yields localized deposition within the trenches there will necessarily be a depth at which the transport limited suppressor flux, scaling as the inverse of that depth, is sufficient to induce such instabilities. Eventually, however, the profiles all converge, indicating that the increased transport limit on suppressor flux is balanced by increased overpotential in a manner that enforces activation of all features. Both the passivation/activation of individual features and the ultimate convergence of filling contours in all features are clear for the -8 μA profile in FIG. 23b and are also evident at longer times in the FIG. 23d charts for -16 μA and -30 μA.
[00151] As with the annular and cylindrical vias shown earlier, simulation of trench array filling under potentiostatic conditions in the presence of 1 Ws of uncompensated resistance has a narrow operating window between fully passive (-0.58 V) and voided deposition (-0.60 V). Deposition at -0.59 V (not shown) eventually passivates after 30 minutes at a height 13 μm below the field (y = 0). An increase in the uncompensated resistance to of 11 Ωs allows for complete filling of the trench arrays similar to gaivanostatic operation. The potentiostatic operating window is also much wider between full passivation (- 0.64 V) and voided deposition (-1.02 V), examined in 20 mV increments. Deposit profiles and interface contours for two potentiostatic conditions are shown in FIG. 23c, exhibiting similar behavior to the gaivanostatic profiles in FIG. 23b. For the more reducing condition of -0.98 V, deposition proceeds uniformly across the array until the interface is near the field (y = 0) at which point individual trenches begin to osciilate between passive and active states. It takes 45 min for all trenches to reach y = 0 with about a 10 μm difference in height between the highest and iowest (trenches 1 and 4). Deposition at the less reducing condition of -0.72 V, on the other hand, exhibits uniform deposition across the array until a depth of -34 pm. At this point deposition localizes into select trenches, dynamically passivating and re-activating at different heights. Deposition across the Interface becomes uniform when each trench approaches the field with a height of y ~ 0 achieved for all features after 102 min.
[00152] FIG. 24 shows the global constraint and response for select gaivanostatic and potentiostatic conditions, respectively, for deposition in the 4 trench array depicted in FIG. 23 and for 2 trench and 8 trench arrays as well. For gaivanostatic operation, similar filling behavior is achieved by scaling the applied current with the number of features or, more importantly, the surface area. Under potentiostatic control, the global current automatically adjusts for the increased surface area of larger arrays; the peak current at -0.98 V immediately after the pretreatment period shifts from -14.5 μA to -29 μA to -58 μA as trench quantity increases from an array of 2 to 4 to 8 trenches. In fact, the current profiles at -0.98 V fall directly on top of each other when scaling for the number of trenches, demonstrated by the dashed grey line in FIG. 24c representing the current profile at -0.98 V in a 4-trench array that has been doubled. Thus, a potentiostatic approach may be advantageous for depositing on wafers or dies with an unknown recessed surface area, dimensions, or number of features that vary experiment-to-experiment. Additionally, the noise associated with passivation of an individual trench at -720 mV is reduced for larger trench arrays; in a 2~trench array individual trench passivation accounts for 50 % of the surface area versus only 12.5 % in an 8-trench array.
[00153] The influence of the control mode on copper deposition behavior is further explored in trench arrays of varying widths as shown in FIG. 25. Trench height and spacing is fixed at 50 μm and 30 μm, respectively, in a repeating array of 10 μm, 20 μm, and 30 μm wide features. In contrast to the uniform array in FIG. 23 and FIG. 24, galvanostatic simulations in a trench array with varying widths indicate a smaller and discontinuous operational window that results in void-free filling. Simulations at -5 μA produce conformal deposition on the electrode interface and, in 2.5 μA increments, deposition at - 10 μA, -20 μA, and all currents greater than -30 μA result in void formation; two examples of voided deposition are shown in FIG. 25 at -10 μA and -40 μA. Void formation at the lower applied currents (i.e. , -10 μA and -20 μA) is caused by localization of the majority of applied current to the narrowest trench. For applied currents greater than -30 μA, deposition occurs across the trench array but at a rate sufficient to deplete metal ion in the narrowest trench leading to void formation. Interestingly, all other conditions between -5 μA and -35 μA (again, in 2.5 μA intervals) produce fully filled trench arrays: two examples are also shown in FIG. 25 at -7.5 μA and -25 μA. For the less aggressive applied current value filling of each trench progresses individually; deposition occurs in trench 1 from 2 min to 25 min, trench 2 from 26 min to 70 min, and trench 3 from 71 min to 132 min. The local deposition on all three trenches slows after 155 min as the filling profile approaches the free surface with slightly varying heights across the array. Deposition at -25 μA occurs evenly across all three trenches until ~ 10 min when the deposition rate accelerates in the smallest trench until it temporarily passivates at a height of -22 pm for about 20 min. This cycle of passivation and re-activation occurs once more before the deposit height reaches y = 0. Deposition in trench 2 behaves similarly, cycling between passive and active but lagging behind trench 1 , whereas trench 3 remains active until the deposit breaches the opening and passivates at a height of 8 pm. Interestingly, the smallest trench is the last to reach a height of y ~ 0, occurring after 62 minutes of deposition. The overfill on the field is non-uniform similar to that observed for galvanostatic deposition in an array of identical trenches.
[00154] In an embodiment FIG. 25 shows simulated growth profiles for galvanostatic and potentiostatic copper electrodeposition at the indicated operating conditions, final times, and system resistances (1Ωs, 60s, or 41Ωs) in an array of trenches having varying widths: 10 μm, 20 μm, and 30 μm from left-to~right. Contour lines represent the position of the growth front, spaced in 10 min intervals and beginning after the pretreatment step, colorized to indicate time. Charts show the centerline position of the growth front of each individual trench (denoted as T1 , T2, and T3 from left-to-right) as time progresses with annotations indicating the time at which all trenches reach y = 0. Potentiostatic deposition with a range of system resistance values permits full seam-free and void-free filling that is impossible with potentiostatic deposition with smaller resistance and possible only in limited and non-contiguous ranges of applied current under galvanostatic conditions. The increased system resistance may be obtained using an external resistance, a semipermeabie or permeable resistive baffle or membrane or increased separation between the reference and substrate.
[00155] Unlike galvanostatic deposition, or even potentiostatic deposition in a uniform array, simulations of potentiostatic deposition for the base case with 1 VV S results in either conformal deposition or voided filling in the varying width trench array in FIG. 25. At -0.57 V and applied potentials more positive, negligible deposition occurs and is largely conformal in nature. Deposition at applied potentials between -0.58 V and -0.805 V, in 5 mV intervals, produces a void in the narrowest trench while the 2 nd and 3 rd trenches remain passive. Simulations of deposition at -0.61 V and more negative potentials produce a fully activated trench array but with sufficient cupric ion depletion that a void forms in the narrowest trench. However, increasing the uncompensated resistance enables complete void-free filling of the varying width trench array. With a 8 Ωs of total uncompensated ceil resistance deposition at potentials of -0.60 V and more positive values result in conformal deposition. Decreasing the applied potential to -0.63 V leads to iterative trench filling similar to the -7.5 μA gaivanostatic condition except a transition to the passive state occurs for all 3 trenches after 100 min at heights of -15 μm, -20 pm, and -22 pm, respectively, from left to right. For a more reducing applied potential of -0.68 V, deposition occurs in trenches 1 and 2 initially while trench 3 remains passive, with all three trenches eventually reaching the field (y = 0) after 112 min. With 6Ωs of uncompensated resistance, voiding occurs at -0.64 V, similar to the -10 μA condition with localization of all current to trench 1 , and at potentials more negative than -0.82 V.
[00156] Increasing the uncompensated cell resistance makes potentiostatic deposition behave more like gaivanostatic deposition. With a 41 Ωs uncompensated resistance, the operating window between fully passive (-0.78 V) and voided fill (-2.38 V) is greatly widened. However, simulations within this range, in 100 mV increments, show intermediate potentials can produce voids similar to the gaivanostatic conditions seen in FIG. 25; specifically at -1.18 V and -1.68 V for the present geometry. Simulations at a less aggressive applied potential of -1.08 V predict individual filling similar to - 0.63 V with a 6Ωs uncompensated resistance, however, interface passivation occurs higher in the trench and all individual trench heights eventually reach y = 0 after 162 min with only minor height variation across the array. Deposition at a more reducing applied potential of -2.18 V initially exhibits more uniform growth across the trench array, with individual trench passivation occurring sequentially in order of increasing trench width. The time at which all trenches reach a height of y = 0 is also reduced, occurring after 55 min of deposition. Unlike deposition at -1 .08 V, there is about a 10 μm height variation across the array for this condition. The individual deposit height profiles for potentiostatic deposition at -2.18 V and galvanostatic deposition at -25 μA show similar characteristics. The global current at -2.18 V with a 41 Ωs uncompensated resistance shows little variation, peaking at -29.3 μA shortly after the pretreatment step and gradually decaying to -27.7 μA after 1 .5 h of deposition.
[00157] Copper deposition is further explored in an even more complex trench array of varying depths in FIG. 26; trench width and spacing is 10 μm and 20 μm, respectively, in a repeating array of 50 μm, 75 μm, and 100 pm deep features. Simulations of galvanostatic deposition predict fully filled trench arrays at conditions between -4 μA and -8 μA. Deposition at an applied current less than -4 μA results in conformal deposition or bottom-up fill that does not achieve a height of y = 0 after 5 h while conditions more aggressive than - 8 μA produce a void in the deepest trench. Simulations of deposition at the less reducing condition of -4.5 μA shows preferential filling of the deepest trench for the first 90 min before passivation occurs with a shift to deposition in the middle trench. After an additional 50 min a similar transition occurs with deposition shifting to the shallowest trench. After 220 min all three trenches are at approximately the same height, although still lower than the field at y = 0. The filling profile is similar at -7.5 μA, except the applied current is sufficient to begin activating deposition in the middle trench immediately after the pretreatment step. Deposition in trenches 2 and 3 is sustained for 45 min and 75 min, respectively, before passivation. The shallowest trench remains passive for the first 80 min before significant filling occurs. As the deposit nears the trench outlet the differences in height lessen across the array. By 145 min there is minimal variation in trench height and all features have reached the position of the field (y = 0). As expected, deposition at the more aggressive current of -7.5 μA achieves complete filling ≈ 100 min faster than -4.5 μA.
[00158] In another embodiment FIG. 26 shows simulated growth profiles for galvanostatic and potentiostatic copper electrodeposition at the indicated operating conditions, system resistances (10s, 90s, or 21Ωs), and final times in a trench array of varying heights: 50 μm, 75 μm, and 100 μm from left-to-right. Contour lines represent the position of the growth front, spaced in 10 min intervals and beginning after the pretreatment step, colorized to indicate time. Charts show the centerline position of the growth front for each individual trench (denoted as T1 , T2, and T3 from left-to-right) as time progresses with annotations indicating the time at which all trenches reach y = 0. Galvanostatic as well as potentiodynamic control with the noted system resistance values achieve full seam-free and void-free filling. Filling in both cases starts first in the deepest features and progressively activates the next deepest features. Higher system resistance yields seam-free and void-free filling of ail features nearly as fast as can be obtained with galvanostatic processing with a broad processing window. The increased system resistance may be obtained using an external resistance, a semipermeab!e or permeable resistive baffle or membrane or increased separation between the reference and substrate.
[00159] Similar to deposition in the other trench arrays (FIG. 23 and FIG. 25), potentiostatic deposition with only 1 W s in uncompensated resistance cannot achieve a full filling of the array. The heights of each trench in the array at the most negative applied potential before voiding occurs are -49 pm, -57 pm, and -53 pm for increasing trench depth (not shown). The smallest value of uncompensated resistance that predicts a fully filled profile at a single applied potential is 9 Ωs (evaluated in 1 Ωs intervals) where complete filling occurs within 4 h at -0.66 V as shown in FIG. 26. Filling under these conditions has similarities to galvanostatic deposition at -7.5 pA; namely, deposition is initially localized to trenches 2 and 3 while trench 1 remains passive for the first 75 min. A significant portion of the feature filling time is associated with slower conformal deposition that occurs after ≈ 130 min, apparent in the closely spaced contours and thick deposit on the field. The centerline of the trenches only reaches the field (y = 0) after 212 min. Increasing the total uncompensated cell resistance to 21 Ωs permits even faster filling of the trench array during potentiostatic operation at -0.84 V. Analogous to the 9 Ωs simulation, the shallowest trench remains passive for the first 75 min while deposition proceeds in the other two trenches. Finally, a shift to conformal deposition across the workpiece occurs after ≈ 130 min with a smaller height variation across the interface compared to that seen for the 90s simulation. The smaller remaining height for the final stage of conformal deposition enables the 21 Ωs simulation to achieve the y = 0 threshold with only an additional 22 min (t = 152 min) of deposition. For both the 9Q S and 21 Ωs uncompensated resistance examples shown in FIG. 28 deposition at potentials 20 mV more negative predict seam or void formation.
[00160] Bottom-up via and trench filling were shown for a variety of additive-derived S-IMDR metal deposition systems. The simulations in the present work indicate that with appropriate optimization of the applied current, galvanostatic Cu deposition from a polyether - Cl- suppressed CuSO 4 - H 2 SO 4 electrolyte can completely fill high-aspect ratio via and trench features for various dimensions. The same is true for potentiostatic conditions, however, filling under conditions where the uncompensated resistance is minimized often results in passivation before feature filling is complete. Different strategies have been explored to overcome this limitation that range from the use of potentiodynamic waveforms to increasing the uncompensated resistance of the electrochemical ceil. The latter can be implemented in a number of ways, from judicious positioning of the reference electrode to insertion of a baffle that increases the effective resistivity of the electrolyte, to the addition of an external series resistor. The latter can be envisioned as a resistive contact on the working electrode in a 3-electrode system and, more generally, a series resistor located anywhere in a 2-electrode circuit. The current response to potentiostatic deposition in the S-NDR system with a significant uncompensated resistance begins to approach that for galvanostatic control. This not only helps establish conditions where complete feature filling is possible but also significantly broadens the processing window making the method more robust to variations in the workpiece geometry, from trench dimensions to patterning effects. Particularly interesting characteristics are captured for feature arrays of both uniform and variable dimensions where sequential filling of different sized features are predicted as well as oscillation between passivation and reactivation during filling under different conditions. Even more complex behavior can occur with discontinuities appearing in the processing window where periodic transitions between complete filling and void formation occur as the control parameter (potential or current) is increased linearly. All of the above reflect the strong path-dependent behavior expected for non-linear bifurcation reactions and further highlights the complex interactions in systems where a rapid electric response is globally coupled to slower, locally non-uniform, mass transport constrained, mixed control reactions such as evidenced in additive induced S-NDR systems. [00161] The following are incorporated by reference herein in their entirety. S-K. Kim, J.E. Bonevich, D. Josell and T.P. Moffat, “Electrodeposition of Ni in Sub-micrometer Trenches,” J. Electrochem. Soc., 154, D443-D451, (2007). C.H. Lee, J.E. Bonevich, J.E. Davies and T.P. Moffat, “Magnetic Materials for 3-D Damascene Metallization: Void-free Electrodeposition of Ni and Ni70Fe30 Using 2-Mercapto-5-benzimidazole sulfonic Acid,” J. Electrochem. Soc., 155, D499-D507, (2008). T.P. Moffat and D. Josell, “Extreme bottom-up superfilling of Through- Silicon-Vias by Damascene Processing: Suppressor disruption, positive feedback and Turing patterns,” J. Electrochem. Soc., 159, D208-D216, (2012). D. Josell, D. Wheeler and T.P. Moffat, “Modeling Extreme Bottom-Up Filling of Through Silicon Vias, J. Electrochem. Soc., 159, D570-D576, (2012). D. Wheeler, T.P. Moffat and D. Josell, “Spatial-temporal Modeling of Extreme Bottom-Up Filling of Through- Silicon-Vias,” J. Electrochem. Soc., 160, D3260-D3265, (2013). D. Josell and T.P. Moffat, “Superconformal Electrodeposition in Complexed Alkaline Electrolyte, J. Electrochem. Soc.,, 161, D287- D292, (2014). D. Josell and T.P. Moffat, “Bottom-Up Electrodeposition of Zinc in Through Silicon Vias,” J. Electrochem. Soc., 162, D129- D135, (2015). D. Josell and T.P. Moffat, “Superconformal Bottom-Up Nickel Deposition in High Aspect Ratio Through Silicon Vias,” J. Electrochem. Soc.,163, D1-D10, (2016). D. Josell, M. Siliva and T.P. Moffat, “Superconformal Bottom-Up Cobalt Deposition in High Aspect Ratio Through Silicon Vias,” J. Electrochem. Soc., 163, D809-D817, (2016). D. Josell and T.P. Moffat, “Superconformal Bottom-Up Gold Deposition in High Aspect Ratio Through Silicon Vias,” J. Electrochem. Soc.,164, D327-D334, (2017). D. Josell and T.P. Moffat, “Superconformal Copper Deposition in Through Silicon Vias by Suppression Breakdown,” J. Electrochem. Soc., 165, D23-D30, (2018). T.M. Braun, S.-H. Kim, H.-J. Lee, T.P. Moffat, D. Josell, “Superconformal Nickel Deposition in Through Silicon Vias: Experiment and Prediction,” J. Electrochem. Soc., 165, D291-D300 (2018). D. Josell and T.P. Moffat, “Bottom-up Filling of Damascene Trenches with Gold in a Sulfite Electrolyte,” J. Electrochem. Soc., 166, 1, D3022- D3034 (2019). L.A. Menk, D. Josell, T.P. Moffat, E. Baca, M.G. Blain, A. Smith, J. Dominguez, J. McClain, P.D. Yeh and A. E. Hollowell, “Bottom-Up Copper Filling of Large Scale Through Silicon Vias for MEMS Technology,” J. Electrochem. Soc., 166, 1, D3066-D3071 (2019). D. Josell, L.A. Menk, A.E. Hollowell, M. Blain and T.P. Moffat, “Bottom- Up Copper Filling of Millimeter Size Through Silicon Vias,” J. Electrochem. Soc., 166, D3254-D3258 (2019). T. Braun, D. Josell, M. Silva, J. Kildon, T.P. Moffat, Effect of Chloride Concentration on Copper Deposition in Through Silicon Vias, J. Electrochem. Soc., 166, D3259-D3271 (2019). T.M. Braun, D. Josell, J. John, and T.P. Moffat, “Simulations of Copper Electrodeposition in Through-Hole Vias,” J. Electrochem. Soc.,167, 013510 (2020). Editors’ Choice T. M. Braun, D. Josell, T.P. Moffat” Microelectrode Studies of S-NDR Copper Electrodeposition: Potentiodynamic and Galvanodynamic Measurements and Simulations,” J. Electrochem. Soc., 167, 082509 (2020). T. M. Braun, D. Josell, S. Deshpande, J. John, T.P. Moffat, “Simulations of Copper Electrodeposition in Millimeter Size Through-Silicon Vias,” J. Electrochem. Soc., 167, 162508 (2020). T. Braun, D. Josell and T.P. Moffat, “Simulating Cu Electrodeposition in High Aspect Ratio Features: Effect of Control Mode and Uncompensated Resistance in S-NDR Systems,” Electrochimica Acta, 375, 137925 (2021). [00162] While one or more embodiments have been shown and described, modifications and substitutions may be made thereto without departing from the spirit and scope of the invention. Accordingly, it is to be understood that the present invention has been described by way of illustrations and not limitation. Embodiments herein can be used independently or can be combined. [00163] Reference throughout this specification to “one embodiment,” “particular embodiment,” “certain embodiment,” “an embodiment,” or the like means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Thus, appearances of these phrases (e.g., “in one embodiment” or “in an embodiment”) throughout this specification are not necessarily all referring to the same embodiment, but may. Furthermore, particular features, structures, or characteristics may be combined in any suitable manner, as would be apparent to one of ordinary skill in the art from this disclosure, in one or more embodiments. [00164] All ranges disclosed herein are inclusive of the endpoints, and the endpoints are independently combinable with each other. The ranges are continuous and thus contain every value and subset thereof in the range. Unless otherwise stated or contextually inapplicable, all percentages, when expressing a quantity, are weight percentages. The suffix “(s)” as used herein is intended to include both the singular and the plural of the term that it modifies, thereby including at least one of that term (e.g., the colorant(s) includes at least one colorants). “Optional” or “optionally” means that the subsequently described event or circumstance can or cannot occur, and that the description includes instances where the event occurs and instances where it does not. As used herein, “combination” is inclusive of blends, mixtures, alloys, reaction products, and the like. [00165] As used herein, “a combination thereof” refers to a combination comprising at least one of the named constituents, components, compounds, or elements, optionally together with one or more of the same class of constituents, components, compounds, or elements. [00166] All references are incorporated herein by reference. [00167] The use of the terms “a” and “an” and “the” and similar referents in the context of describing the invention (especially in the context of the following claims) are to be construed to cover both the singular and the plural, unless otherwise indicated herein or clearly contradicted by context. “Or” means “and/or.” Further, the conjunction “or” is used to link objects of a list or alternatives and is not disjunctive; rather the elements can be used separately or can be combined together under appropriate circumstances. It should further be noted that the terms “first,” “second,” “primary,” “secondary,” and the like herein do not denote any order, quantity, or importance, but rather are used to distinguish one element from another. The modifier “about” used in connection with a quantity is inclusive of the stated value and has the meaning dictated by the context (e.g., it includes the degree of error associated with measurement of the particular quantity).