Superconformal Copper Deposition in Through Silicon Vias by Suppression-Breakdown

Abstract

The evolution of superconformal Cu electrodeposition in high aspect ratio through silicon vias (TSVs) is examined as a function of polymer suppressor concentration, applied potential and hydrodynamics. Electroanalytical measurements in a CuSO₄-H₂SO₄-Cl electrolyte are used to explore and quantify the effect of such parameters on the metal deposition process. Hysteretic voltammetry due to suppressor breakdown reveals an S-shaped negative differential resistance that leads to non-linear spatial-temporal patterning during metal deposition. For the given hydrodynamic conditions, cyclic voltammetry reveals the potential regime over which positive-feedback gives rise to the superconformal feature filling dynamic. Breakdown of suppression is primarily related to polymer concentrations in the electrolyte while its reformation is dependent on its transport to the interface. Morphological evolution during the early stages of TSV filling reveals two distinct growth front geometries. For dilute polymer concentrations, an initial bifurcation into passive-active surfaces occurs on the side walls of the TSVs followed by bottom-up fill. The depth of the initial sidewall bifurcation within the via increases with polymer concentration. For higher polymer concentrations, i.e. ≥ 25 μmol/L, active metal deposition is rapidly confined to the bottom surface of the via followed by sustained bottom-up filling.

Suppressor adsorbates play a central role in superconformal electrodeposition of metals in through silicon vias (TSV). The relative impacts of transport, adsorption, deactivation, displacement and surface area change on suppressor adsorbate coverage depend on the length scale of the feature of interest. In microelectronics, such features range from deep sub-micrometer trenches and vias for chip circuitry to the much larger TSVs, micrometers in diameter and an order of magnitude larger in depth, that are used for chip stacking and even larger scale features in printed circuit boards. Presently, Cu is the metal of choice for interconnect technology, and void-free, bottom-up superconformal filling of TSV has been widely demonstrated using either current or potential regulation in electrolytes containing polyetherbased suppressors. While filling has often been reported and analyzed for multicomponent additive packages,^1–15 the demonstrated ability of suppressor-only systems to yield bottom-up filling is of particular interest for determining the essential processes that underlie this unusual superconformal filling evolution.^16–25 In the most extreme form of bottom-up filling, active deposition is localized to the bottom surface of recessed features while the immediately adjacent sidewall and free surface are maintained in the passive state by suppressor adsorption as shown for Cu deposition in the bottom row of Fig. 1.¹⁶ More recently, superconformal TSV filling studies with Au,^26,27 Zn,²⁸ Ni,^29–32 Ni(W) alloy³³ and Co³⁵ have revealed other growth front morphologies and dynamics related to bifurcation processes in single additive electrolytes due to the non-linear metal deposition kinetics. These systems exhibit a passive-active transition localized to the TSV sidewalls with uniform charge transfer kinetics in the recessed active region below and much slower deposition on the passivated surfaces above (analogous growth of Cu has previously been noted but only in much larger microvias³⁴). This geometry contrasts with the bottom-up filling that has been reported for Cu in TSVs, and an example of this alternative evolution of feature filling is shown in the upper row of Fig. 1 for Co grown in the presence of polyethyleneimine.³⁵ The location of the sidewall passive-active transition is a sensitive function of the suppressor concentration and applied potential. In further contrast to the sustained active deposition associated with extreme bottom-up Cu filling, the location of bifurcation on the side walls during Au,^26,27 Ni³² and Co³⁵ filling at a given applied potential is fixed in time. Thus, as captured in the upper row of Fig. 1, although the active deposit thickens, the growth front propagates no further up the sidewall. The increasing aspect ratio of the unfilled portion leads to increasing metal ion depletion and eventual void formation as the growing opposing sidewall deposits impinge. Importantly, such voids can be minimized or avoided through use of a ramped or staircase potential waveform that progressively advances the position of the passive-active transition up the sidewall to enable superconformal, void-free filling.^26,27,32,35

Figure 1.

(Top row) Deposition during Co deposition exhibits a passive-active transition along the sidewall that has also been detailed for a number of other metal systems (adapted from Ref. 35). (Bottom row) Bottom-up deposition of the type often reported for Cu TSV filling (adapted from Ref. 16).

The extent of metal ion depletion during feature filling provides another important distinction between the two characteristic deposition geometries. Analysis of bottom-up Cu filling indicates significant metal depletion occurs down the TSV.^16,18 In contrast, in the other systems, feature filling via bifurcation of the sidewalls into passive-active regions is marked by a uniform deposit thickness that indicates a minor role of metal depletion, at least for the first few minutes of deposition. A variety of activation modes have been considered for explaining both bottom-up and sidewall initiated filling. These include disruption of the polymer adsorption by the metal deposition reaction itself, its perturbation by the local state of interface hydration, as well as deactivation and consumption of the suppressor within the growing solid.

All the systems outlined above exhibit hysteretic voltammetric behavior that stems from breakdown of the suppressor layer that is stimulated and amplified by positive feedback associated with the metal deposition reaction. The non-linear response is characterized by suppressor derived S-shaped negative differential resistance (S-NDR). For planar electrodes operating in the presence of significant electrolyte resistance the voltammetric instability associated with the S-NDR can be obscured, and metal deposition on a planar surface at fixed applied potential bifurcates to form Turing patterns with active regions of localized deposition occurring on an otherwise passivated surface.^16,36,37,38 Models based on deactivation and consumption of the suppressor have yielded predictions of bottom-up filling evolution as well as accurate prediction of the position of the sidewall bifurcation as a function of suppressor concentration and potential for several systems.^{18,19,27,32,35} In the latter case, the depth of the passive-active transition is ascribed to the balance between transport limited suppressor flux and adsorbate consumption into the actively growing deposit. For some systems, it might be possible for the processes to operate sequentially whereby a transition from sidewall passive-active bifurcation to a bottom-up filling geometry occurs. For systems the use of a single inhibiting additive, and unambiguous evidence of microstructural changes associated with deposition in the passive and active states, provides justification for exploration of consumption based models. In contrast, suppression in the acid sulfate Cu systems is provided by the combination of polyether-halide (PEG-Cl), with some experiments and analysis suggesting minimal additive incorporation in Cu deposits.^25,39,40 That said, recent studies of the solderability of the Cu films deposited in the presence of the PEG-Cl clearly reveal significant deleterious microstructural effects related to additive operation as a function of the growth condition.^41–45

Inhibition of the Cu deposition requires co-adsorption of halide with the polyether “suppressor”.^{20,25,46–56} In conventional Damascene plating electrolytes, the Cl⁻ concentration is usually maintained near or slightly below 1 mmol/L while polyethers with molecular mass in the range (2000–6000) atomic mass units are held at concentrations near 100 μmol/L. For these conditions, smooth, monotonic and reversible voltammograms are observed on macroelectrodes where the apparent rate constant of metal deposition is suppressed by a factor 100 to 1000 relative to that in the polymer-free electrolyte. However, when the concentration of either component is decreased by an order of magnitude, i.e. halide to (10–100) μmol/L or the polymer to (1–10) μmol/L, critical phenomena become evident.^{20,25,46–56} Hysteretic voltammetry, marked by a rapid increase in metal deposition with breakdown of the halide-polyether passivated surface, is highly non-linear and is correlated with the unusual feature filling behavior noted above.

The present study continues the exploration of TSV filling with its dependence on poloxamine concentration and applied potential being examined while the Cl⁻ concentration is fixed at 1 mmol/L. The relationship between hysteretic S-NDR voltammetry, bifurcation of the electrode into active and passive regions and its impact on feature filling behavior are examined.

Experimental

All electrochemical experiments were conducted at room temperature in a cell containing 35 mL of 1 mol/L CuSO₄ and 0.5 mol/L H₂SO₄. Suppression was provided by co-adsorption of chloride and a poloxamine additive, Tetronic 701 (Aldrich¹). The Tetronic (TET) polyether contains an ethylenediamine core with four propoxylate-blocks capped with ethoxylate giving a molecular mass of 3600 g/mol. The Cl⁻ concentration was fixed with the addition of 1 mmol/L NaCl while micromolar additions of poloxamine were made from a master solution of 0.4 mmol/L Tetronic 701 dissolved in 18 MΩ · cm water. The electrolyte was sparged with argon between electrochemical measurements to reduce the amount of dissolved oxygen. All electrochemical measurements and filling experiments were performed using a Hg/Hg₂SO₄/saturated K₂SO₄ reference electrode (SSE) connected to the working electrode compartment via a fritted bridge filled with saturated solution of K₂SO₄; all potentials are quoted relative to the reference. A platinum counter electrode was held in a frit-separated cell immersed within the main cell.

Voltammetry was conducted on a 1.0 cm diameter (area 0.78 cm²) rotating disk electrode (RDE) machined from oxygen-free, high conductivity Cu that was freshly polished on 1200 grit silicon carbide paper prior to each experiment. Currents measured in the electroanalytical experiments were converted to current densities using the RDE area. However, the actual electroactive area in the additive-containing electrolytes was ill-defined beyond the onset of suppression breakdown, typically being restricted to numerous, localized circular deposits several tens of micrometers in diameter (see Ref. 16) distributed uniformly across the RDE surface.

Feature filling was performed using rectangular fragments of wafers patterned with ≈56 μm deep TSVs of annular cross-section (provided by IBM) with a 1 μm thick Cu seed in the field and a lesser amount on the sidewalls. To give definition to the metal ion and additive transport, the TSV patterned substrates were rotated about one end from a Pt spindle during deposition, like a helicopter blade, the patterned surface facing up. Based on the ≈1 cm distance between the features and rotational axis for most of the imaged TSVs, the 100 rpm (200π rad/min) rotation rate corresponds to an estimated 10 cm/s flow rate over the surface. Pre-wetting the wafer fragments with ethyl alcohol was used to displace air bubbles that were otherwise trapped in the TSVs during Cu deposition. Following immersion, the specimens were rotated for two minutes to mix the electrolyte with the alcohol in the TSVs; the potential was maintained at −0.40 V_SSE (≈−30 mV relative to the measured open circuit potential of ≈−0.37 V_SSE) to avoid corrosion of the copper seed layer. It is noteworthy that under such conditions significant Cu⁺ can be produced adjacent to the electrode that may influence subsequent metal deposition. Following the dwell period, the potential was stepped to a specific value for a fixed deposition time followed by rapid removal, rinsing and drying of the workpiece. Unless otherwise noted the applied potential was not corrected for uncompensated cell resistance.

Deposition on Rotating Disk Electrodes

Cyclic voltammetry shown in Fig. 2a reveals strong suppression of metal deposition by the polyether additive that forms a passivating layer following immersion and during the slow, negative-going potential scan. At more negative potentials, the sharp increase in current marks the breakdown of the passive state. The potential of suppression breakdown depends on the polymer concentration, shifting to more negative values with increasing poloxamine concentration. The incremental extension of the passive range is largest at the lowest polymer concentrations. Likewise, an increase in suppression associated with the passive state is evident with increasing polymer concentration as seen in the current density plotted on a logarithmic scale (Fig. 2b). The initial open circuit potential also shifts negative with poloxamine concentration, approaching a fixed value at higher polymer concentration. The passive current density increases with overpotential, exhibiting a slope close to 200 mV per decade. For a given potential, the passive current density decreases with an increase in polymer concentration to approach a saturation value above 6 μmol/L poloxamine. The passive (or leakage) current is ascribed to a combination of Cu²⁺ reduction to either Cu⁺ or Cu along with possible contributions from residual oxygen. Further insight might be provided by a high resolution gravimetric study.

Figure 2.

Voltammetry of Cu deposition in the copper sulfate electrolyte. Cyclic voltammograms for a) the indicated TET concentrations with RDE rotation rate 100 rpm and b) data replotted using a logarithmic scale for the current density to reveal the passive current density and open circuit potentials. Experimental currents are converted to current densities using the 0.78 cm² RDE area. The data were collected at a scan rate of 2 mV/s without compensation for iR potential drop across the measured cell resistance (R ≈3 Ω). The data is plotted against the applied potential.

At potentials negative of the critical potential, comparison of Figs. 2a and 2b makes clear that the current density increases in a linear, rather than exponential, manner. This constrained response is ascribed to the electrolyte resistance between the working and reference electrodes. Post-experiment correction of the applied potential for potential drop across the electrolyte reveals the presence of S-shaped NDR and critical behavior associated with suppressor breakdown as shown for the prototypical example with 12 μmol/L poloxamine in Fig. 3a. Speaking to the generality of this result, the slopes for all the uncorrected voltammograms shown in Fig. 2 are quite similar. An example of an impedance measurements used to obtain the value of the ≈3 Ω cell resistance is shown in Fig. 3c. A more accurate assessment of the impact of cell resistance on the voltammetry is revealed by using real time iR compensation for 70% of the measured cell resistance as shown Fig. 3b. Comparison with the post-experiment iR corrected results (Fig. 3a vs Fig. 3b) highlights the marked iR distortion of the voltammetric potential waveform itself, demonstrating the importance of using real time iR compensation to obtain an accurate assessment of metal deposition kinetics.

Figure 3.

a) Post experiment correction of the data for 6 μmol/L Tetronic 701 using the ≈3 Ω electrolyte resistance reveals the obscured S-NDR on the negative-going scan; the return scan exhibits S-NDR even without correction. b) Voltammograms acquired with and without 70% correction of the measured 3 Ω cell resistance plotted both as acquired and after full correction for cell resistance. c) Impedance data whose real axis intercept yields the cell resistance. The RDE was rotating at 100 rpm.

The influence of hydrodynamics on the voltammograms was examined by varying the RDE rotation rate. Negligible impact on the onset and acceleration of suppressor breakdown was evident as captured in Fig. 4 for 0.25 μmol/L to 25 μmol/L poloxamine. This indicates that neither polymer transport to the interface nor hydrodynamic shear play a significant role in determining the critical potential marked by the discontinuity of slope in the logarithmic plots. As only a fraction of the RDE surface area is seen to reach the active state for the given voltammetric parameters (similar to Fig. 6 in Ref. 16), the current densities calculated using the full RDE area necessarily represent a lower bound for the deposition rates on the activated surface segments. Positive feedback, coincident with deposition that stimulates breakdown of the suppressing polymer layer, is evident when the scan direction is reversed, the current continuing to increase even as the potential swings more positive. At more positive potentials, the deposition rate slows and the polymer suppressor layer reforms, closing the hysteretic voltammetric loop. In contrast to the breakdown process on the negative going sweep, reformation of the passive state on the return sweep is strongly dependent on rotation rate with enhanced quenching of active deposition occurring at higher rotation rates. This suggests that poloxamine transport to the interface plays a more important role than the disruptive forces associated with shearing of the interface. Nonetheless, the complex nature of shear on polymer adsorption and double layer structure remain topics of on-going research that may very well impact the current situation.^57,58 At yet more positive potentials the open circuit potential on the return sweep is systematically shifted toward more negative values with increasing rotation rate, and the deposition rate decreases below that observed on the original negative going sweep. The detailed interactions between the polymer, trace oxygen, Cl⁻, Cu²⁺ and Cu⁺ that lead to this shift remain to be determined. Prediction of the asymmetric effect of hydrodynamics on suppression breakdown versus reformation is an important criterion for any mechanistic model to meet.

Figure 4.

Cu deposition in electrolytes with increasing Tetronic 701 concentration exploring the effect of RDE rotation rate. a) 0.25 μmol/L, b) 1.6 μmol/L, c) 3 μmol/L, d) 6 μmol/L, e) 12 μmol/L and f) 25 μmol/L TET. Measured currents were converted to current densities using the 0.78 cm² RDE area. The data is replotted using a logarithmic scale in a′ – f′. The voltammetry was collected without compensation for iR potential drop across the measured cell resistance R ≈3 Ω.

Figure 6.

Cross-sectioned annular TSVs after Cu deposition in electrolyte containing 12 μmol/L TET for the indicated deposition times and potentials. The left and right halves of two representative TSVs are shown instead of a single TSV for 8 min and 12 min of deposition at −0.58 V to capture the bimodal filling observed at this potential. The patterned substrates were rotating at 100 rpm.

Deposition on TSV Patterned Substrates

Figure 5 shows Cu deposition on TSV-patterned substrates at applied potentials that lie within the hysteretic voltammetric region of electrolyte containing 6 μmol/L TET (Fig. 2). Consistent with the applied potentials being positive of the critical potential for suppressor breakdown, deposition is not observed on the field. For deposition at −0.66 V_SSE, immediately positive of the voltammetrically-determined critical potential, significant sub-conformal deposition within the TSV is evident, impingement of the sidewall deposits in the upper half of the TSV leaving a key-hole shaped void in the lower half after just 4 minutes of deposition. In contrast, early deposition at −0.64 V_SSE reveals a passive-active transition on the sidewall similar to that previously reported for Au,^26,27 Ni^29–32 and Co³⁵ filling of TSVs in the presence of a PEI suppressor. The passive-active sidewall transition is seen to lie near the midpoint of the vias, its location varying little with the applied potential over the range −0.64 V_SSE to −0.56 V_SSE. With time, the deposition process becomes (further) localized to the bottom surface so that bottom-up filling dominates by 8 min of deposition and thereafter; the change is captured nicely by the specimen at −0.64 V_SSE. A variation occurs at the most positive potential of −0.54 V_SSE where a brief increment of conformal sidewall deposition in the lower portion of the feature is followed by quenching of the reaction through delayed suppressor adsorption.

Figure 5.

Cross-sectioned annular TSVs after Cu deposition in electrolyte containing 6 μmol/L TET at the indicated deposition times and applied potentials. A pair of adjacent TSVs from via arrays of 50 μm pitch is shown for each condition. The patterned substrates were rotating at 100 rpm during deposition.

For deposition at −0.66 V_SSE and −0.64 V_SSE substantial deposition was noted at the edges of the rectangular specimens, giving rising currents and iR potential drop across the cell that approached ≈40 mV and 25 mV, respectively, based on measured electrolyte resistance of ≈4 Ω between the patterned specimens and the reference electrode. For specimens grown at more positive values no such extraneous deposition was observed.

Feature filling at applied potentials within the hysteretic region of the electrolyte containing 12 μmol/L Tetronic 701 is shown in Fig. 6. At −0.68 V_SSE, deposition is initially localized at the mouth of the TSV and quickly becomes sub−conformal, with sidewall pinch-off due to metal ion depletion analogous to that observed at −0.66 V_SSE with 6 μmol/L TET. At the less negative potential limit, complete suppression is evident at −0.56 V_SSE, a shift of −20 mV from the threshold observed in Fig. 5 for 6 μmol/L TET concentration. The shift is consistent with the voltammetric data in Fig. 2a. Close inspection of the images for 4 min deposition at −0.66 V_SSE, −0.64 V_SSE, −0.62V_SSE and −0.60 V_SSE reveals both passive-active sidewall activation and bottom-up filling, the sidewall bifurcation being visible approximately three-quarters of the way down the via, deeper than that observed in 6 μmol/L Tetronic (Fig. 5). For potentials of −0.66 V_SSE to −0.60 V_SSE, extreme bottom-up filling is manifest over the period 4 min to 12 min. However, at −0.58 V_SSE, which is just negative of the upper potential bound of the hysteretic voltammetric region, while the position of the initial sidewall passive-active transition at 4 min is the same, only a portion of the TSVs subsequently exhibit bottom-up filling; deposition in the others is quenched. The bimodal distribution between individual vias is summarized in Fig. 6 as a composite image of two half-vias, denoted by the solid line, after both 8 min and 12 min of deposition. Increased filling after 12 min deposition with more negative potential is captured from −0.60 V_SSE to −0.66 V_SSE in Fig. 6 as well as in Fig. 7, where the bimodal variation in feature filling at −0.58 V_SSE is also evident.

Figure 7.

Representative arrays of TSVs with 50 μm pitch after 12 minutes of Cu deposition at the indicated potentials in electrolyte containing 12 μmol/L TET. The images capture the consistency of filling observed at all potentials examined except −0.58 V. Suppression breakdown within the vias at this potential is seen to yield a bimodal distribution of filling heights. The patterned substrates were rotating at 100 rpm.

The potential dependence of feature filling was also examined using 25 μmol/L TET. As shown in Fig. 8, for −0.68 V_SSE, −0.66 V_SSE and −0.64 V_SSE, complete filling is obtained by 15 minutes. At more positive potentials, negligible deposition occurs at −0.60 V_SSE as it lies outside the hysteretic voltammetric window determined in the Cu RDE experiments while transient bottom-up deposition occurs at −0.62 V_SSE but is quenched after only a few micrometers of growth. Interestingly, behavior similar to that observed with 12 μmol/L TET at −0.58 V_SSE is observed with the 25 μmol/L TET, with variable filling observed at −0.64 V_SSE, −0.66 V_SSE and −0.68 V_SSE (not shown), the fraction of filled vias increasing at more negative potential. For the results detailed above, the TSV specimens were initially immersed with the potential held at −0.4 V_SSE for 2 minutes. Mixing between the ethanol wetted features and the fully comprised electrolyte occurs during this period along with Cu⁺ production that in combination influence the formation and nature of the suppressing poloxaminehalide overlayer. To gain more insight into the effect of this pretreatment step, its duration was varied from 30 s to 4 min prior to 4 min Cu deposition at two different potentials in an electrolyte containing 6 μmol/L poloxamine. The effect of this variation is shown in Fig. 9. For deposition at −0.54 V_SSE, the position of the passive-active transition shifts farther down in the TSV with increasing pretreatment time. In accord with Fig. 5, active deposition is entirely transient under these conditions, with negligible further deposition expected beyond 4 minutes. In contrast, deposition at −0.60 V_SSE, which is expected to support sustained feature filling, reveals a more dramatic effect of pretreatment time. The position of the passive-active transition shifts deeper down the via as dwell time increases from 0.5 min to 4 min as with the transient deposition at −0.54 V_SSE. For the two shorter dwell times conformal growth on the lower portion of the sidewall follows bifurcation into passive-active zones. The two longer pretreatment periods yield bottom-up feature filling, although a remnant of the transient sidewall deposition remains for the 4 minute pretreatment.

Figure 8.

Cross-sectioned annular TSVs after 15 minutes of Cu deposition in electrolyte containing 25 μmol/L TET at the indicated potentials. The fraction of fully filled TSVs decreases as the applied potential goes from −0.68 V to −0.64 V (not shown), with universal suppression obtained at −0.62 V and more positive values. The patterned substrates were rotating at 100 rpm.

Figure 9.

TSV filling after the indicated duration of the −0.4 V prehold and 4 minutes of Cu deposition at the indicated applied potential in electrolyte containing 6 μmol/L TET. The patterned substrates were rotating at 100 rpm.

Discussion

Developing the link between the S-NDR hysteretic voltammetry and inhomogeneous two-state (passive-active) deposition is required to quantitatively model superconformal feature filling. The present experiments demonstrate that the voltammetically determined hysteretic potential region, between suppressor breakdown and its reformation, defines the operational window that yields localized deposition on the recessed surfaces within the TSV. The similar shift of the voltammetrically determined suppression breakdown potential with the increase of suppressor concentration from 6 μmol/L to 12 μmol/L poloxamine (Fig. 2) and the potential range that yields superconformal deposition within the TSV is particularly noteworthy (Figs. 5 and 6). A similar potential dependence of superconformal filling on suppressor concentrations was captured by prior S-NDR models (i.e., Fig. 10 of Ref. 18 and Fig. 7 of Ref. 19).

In the present work two distinct growth morphologies related to the location of the initial bifurcation into active and passive regions were observed. Localized breakdown on the sidewalls is a transient state that subsequently converts to bottom-up filling like that previously reported for Cu systems.^16–24 The bottom-up filling process has been understood and modeled to be associated with growth rate dependent disruption of the polymer suppressor.^16–22 In contrast, a sustained sidewall bifurcation process has previously been observed and captured in PEI suppressed systems by models employing transport limited consumption or deactivation of the suppressor species at the passive-active boundary.^27,32,35 As suppression in the present system requires the co-adsorption of the polyether and chloride it is likely that the observed spatial-temporal variations in feature filling behavior reflect the adsorption and deactivation processes of each component as well as cross interactions related to co-adsorption. Further work is underway exploring these dynamics in polyether-halide Cu electroplating systems.

As captured in previous models of bottom-up feature filling, coupling between ohmic losses in the electrolyte, the metal deposition rate and suppressor adsorption dynamics underlie the localization of the bifurcation into active and passive regions. Uneven filling of the TSV in the present work shown for the 12 μmol/L TET at −0.58 V_SSE in Fig. 7, and noted for the 25 μmol/L TET over a wider potential range, reflect the impact of ohmic losses on process control particularly at potentials adjacent to the voltammetric thresholds (see Fig. 3b).

The transient nature of the passive-active transition on the sidewalls, evident in Fig. 5 over a range of applied potentials, even when initial conditions enable it to manifest substantially, is fundamentally different from the steady-state nature of this behavior in the S-NDR Au, Co and Ni systems. The differences in behavior are also evident in the rotation rate-invariant critical breakdown potential of the Cu-Cl-poloxamine system, shown in Fig. 4, in contrast to the smooth variation of suppressor breakdown potential with rotation rate reported for the PEI-(Ni, Co, Au) systems. The latter behavior was linked to the flux balance between additive transport and adsorbate deactivation congruent with substantial microstructural evidence of additive incorporation. In contrast, for the Cu-halide-polyether system the link between suppressor deactivation and microstructure disruption is more subtle and remains a topic of ongoing research.

Conclusions

For Cu²⁺ electrolytes containing 1 mmol/L Cl⁻ and a dilute concentration of a poloxamine suppressor additive, Cu electrodeposition occurs by bifurcation of the surface into passive and active zones. The location of the respective zones on TSV patterned wafer is dependent on polymer concentration, applied potential and time. For dilute polymer concentration, the passive-active transition initially occurs on the TSV sidewalls and is followed by a shift to bottom-up filling at longer times. For the given additive concentrations and hydrodynamics the potential thresholds associated with the hysteretic S-NDR voltammetry define the range within which successful feature filling is possible. For a given suppressor concentration the voltammetric breakdown potential is independent of hydrodynamics while reformation of the passive state, in contrast, is strongly influenced by hydrodynamic conditions. This asymmetric dependence of the hysteretic voltammetry on hydrodynamics stands in contrast to that reported for other S-NDR metal systems. Further work examining the interplay between polyether and halide co-adsorption and TSV feature filling is underway