Pull-Off Force Measurements—Deciphering Dispersant Molecule Adhesion Mechanisms
To design surfactant-dispersed slurries with appreciable polishing rates, the dominant mechanism leading to the presence of residual monomer at the solid-liquid interface needed to be determined. Pull-off force measurements— the force required to detach the colloidal probe from the planar substrate—were performed in an attempt to gain additional information on the origins and strength of adhesion of the residual surfactant layer. As with the lateral force microscopy measurements, the magnitude of the pull-off force is also strongly influenced by the existence of the intervening molecular films. In the presence of surfactant, the value of typical pull-off forces between the probe used in this investigation and the planar silica substrate increased by an order of magnitude in comparison to the pure state as indicated in Fig. 10. This additional adhesive component between the surfaces is attributed to the hydrophobic interactions between strongly bound surfactant molecules adsorbed to the opposing silica surfaces. However, whether these interaction forces represent the detachment of the adsorbed headgroups from the interface or the disengagement of the interacting hydrophobic surfactant groups themselves remains unknown. These two proposed scenarios are illustrated in Fig. 11. By monitoring the maximum pull-off force as a function of concentration, some clues are given. As indicated in Fig. 10, the adhesive force reaches a plateau value in the vicinity of the PZR and not at the maximum monolayer coverage, which occurs just prior to 8 mM C12TAB.[21] This suggests that force required to separate the engaged surfaces is largely electrostatic and not caused by the detachment of hydrophobic bonds. To verify this, the relative magnitude of the interfacial energies expected for the two scenarios depicted in
Fig. 11 Schematic representation of the two proposed scenarios for surfactant-mediated detachment of the silica surfaces in this study. (a) The surfactants remain in hydrophobic association and detach at the head-surface interface. (b) The surfactants remain adsorbed at the opposing interfaces and detach through breaking of hydrophobic chain-chain interactions.
Fig. 11 was investigated and compared with the values calculated from experimental results.
To estimate the energy required for electrostatic detachment, the contribution of the coulombic interaction to the interfacial separation energy gel was estimated by:[53,54]
where Gel represents the adsorption density of the electrostatically adsorbed surfactant monomers to the substrate, z is the valency of the headgroups, e is the electronic charge, and Ci, is the potential of the inner Helmholtz plane of the surface prior to surfactant adsorption. For simplicity, it was assumed that Ci~X, the zeta potential of the silica surface in pure water, and that Gel=Gpzr, the amount of surfactant adsorbed at the PZR. Taking typical values[25,51,52] of X = – 60 mV and Gpzr=1 mM/m2, the magnitude of gel was found to be approximately 3.8 mJ/m2.
To estimate the amount of energy required to separate the surfaces through purely hydrophobic interactions between the hydrocarbon tails (Fig. 11b), the interfacial separation energy, gh, was estimated as:
where Gh is the density of strongly bound surfactant molecules that mediate the separation event, n is the number of CH2 groups that effectively interact between the individual surfactant molecules, and 0h is the energy per CH2-CH2 group interaction. For this estimate, it was assumed that Gh=Gel (or the lower limiting case in which only the electrostatically adsorbed molecules are detached), and that full interpenetration of the hydrocarbon chains (n = 12) occurs. By employing a typical value[17] for 0h as 6.3 x 10-21 J, gh was calculated to be approximately 45 mJ/m2. Alternatively, the magnitude of the interfacial separation energy for hydrophobic detachment can be estimated to be close to the experimental values of the adhesion force between chemisorbed CH3 group-terminated surfactant monolayers interacting across an aqueous medium. From both force and contact angle measurements, the range of this interfacial energy gch is found to be between 40 and 100 mJ/m2,[55] which compares to thea above result and 1 order-of-magnitude greater that gel.
To further compare these values with the experimental pull-off force measurements in this study, the Johnson-Kendall-Roberts (JKR)[56] (upper limiting value) or [57] Derjaguin-Muller-Toporov (DMT)[ ] (lower limiting value) theories were employed. Accordingly, the effective experimental interfacial separation energies were given by:
By incorporating values of Fpull-off from Fig. 10, it was found that 3-4 mJ/m2, which is in the range of gei. This is an order-of-magnitude less than that speculated for purely hydrophobic-mediated detachment. Hence although the adsorption of surfactant at the solid-liquid interface mediates by both chain-chain and head-surface interactions, contributions from the adsorption affinity of the surfactant headgroup to the surface appear to dominate the overall resilience of the remaining films at the interface. For the surfactant system in this investigation, the chain-chain interactions appear much greater than the head-surface affinity (assumed purely electrostatic), causing the latter to dictate the extent of adhesion (pull-off force) and possibly the extent of lubrication. Accordingly, the scenario depicted in Fig. 11a was determined to be more plausible when the two surfaces are pulled apart. Hence to modulate the material removal rate in these surfactant-based slurries, the electrostatic attraction of the surfactant to the wafer and particle surfaces needs to be manipulated.
Although evidence in the preceding section suggested that the resilience of residual surfactant moieties depends largely on the extent of head-surface engagement, the utility of this for the control of the extent of particle-surface engagement mediated by physisorbed surfactant systems has not been illustrated. It is well established that quaternary amine surfactants adsorb to silica interfaces through a combination of electrostatic, hydrogen-bonding, and hydrophobic interactions.[58] However, the relative importance of these interactions in maintaining surface attachment and lubrication is not well understood.
If one assumes that the extent of lubrication is mediated by intervening molecular films attached to the solid-liquid interface primarily by electrostatic interactions, then there should be a large dependence on the lateral force with solution pH. It is well documented that the number of deprotonated (negatively charged) silanol groups at silica interfaces increases with pH causing an increase in the net surface charge. The isoelectric point of silica is near pH 2.5; the zeta potential is about — 50.0 mV at pH 5.6 and — 70 mV at pH 9.0.[59] Hence the extent of lubrication caused by electrostatically adsorbed surfactant structures should increase with pH. This is indeed the case, as Fig. 12 indicates. For all solution pH values, the lateral forces at low loading forces (100-300 nN) were found to be identical; however, a deviation in lubrication performance occurred at higher normal forces indicated by the variant of Amontons’ law and potentially a non-Prestonian polishing response.[3] This experimental evidence suggests that at a critical load, the adsorbed surfactant structures begin to detach from the surface as marked by the increase in the apparent friction coefficient. This ”activation energy” for surfactant detachment shifts to higher energies (loads) with pH. At a pH of 9.0, the transition was not observed for loads up to 1400 nN and was attributed to the strong electrostatic interactions between the surfactant and the silica surface at high pH.
Fig. 12 Lateral force as a function of the applied load for a 7.5-mm silica sphere interacting with a smooth silica surface in solutions of 32 mM C12TAB at (a) various pH values and (b) at various concentrations of NaCl given in logarithmic scale.
The above illustrates that electrostatic interactions could be used to modulate friction, which physically can be interpreted as the extent of silica-silica engagement; however, for silica CMP, a high pH is generally fixed to ensure the formation of a gel-like layer. This prompted investigations into an alternative approach for modulating the extent of particle engagement and material removal in CMP systems. The influence of co-ion competition with respect to the surfactant molecule for negatively charged surface sites was investigated to suit this purpose. The frictional force between silica surfaces in the presence of post-CMC surfactant solutions at a pH of 5.6 and various ionic strength NaCl solutions were measured and are depicted in Fig. 12b. A significant increase in frictional engagement occurs at high Na+ concentrations, further indicating that the electrostatic binding efficacy to the silica interface largely controls the extent of bare surface engagement in these surfactant systems. These findings also suggested for the first time that competitive adsorption strategies may be used to tailor the frictional properties and possibly the material removal rates in slurries dispersed by physisorbed surfactants. Figure 12b further shows that as the number of competing ions increases, the transition load at which higher friction coefficients appear becomes lower and the magnitude of the friction coefficient becomes higher.
Optimizing Slurry Formulation Through Systematic Atomic Force Microscopy Measurements
The strategies suggested above focused on diminishing the extent of surfactant surface affinity to allow for greater engagement of the bare particle with the substrate surfaces. It has not been clearly demonstrated, however, whether or not this loss in surfactant surface affinity will result in aggregation of the silica particulate system, which would be deleterious to CMP operations. Pull-off force and repulsive force barrier measurements were conducted as a function of pH and electrolyte concentration to further develop the correlations between stability, surfactant surface affinity, and lubrication. The pull-off force indicates the relative strength of attachment of surfactants to the solid-liquid interface, whereas the presence of a repulsive force barrier indicates that particle stability still exists.[21]
Figure 13a presents the pull-off and repulsive force barrier measurements as a function of pH for the same experimental conditions as in Fig. 12a. At twice the CMC surfactant concentrations, both the repulsive barrier and the pull-off force increase with increasing pH. These trends are attributed to both the greater electrostatic affinity and perhaps the greater localized concentration of surfactant at the silica-solution interface at higher pH values. Both the enhancement of the barrier and the pull-off forces correlate well with the increased lubrication observed with pH as demonstrated in Fig. 12a.
However, as indicated in Fig. 13b, the extent of the repulsive barrier does not primarily depend on the magnitude of surfactant surface affinity. Although both the lubrication and the pull-off force decrease by nearly an order of magnitude with a 1-M increase in ionic strength, the repulsive force barrier remains nearly constant. This counterintuitive resilience of the surfactant-mediated repulsion under high electrolyte concentrations was first demonstrated in a previous study,[21] in which the primary factor for establishing an effective force barrier was the extent of surfactant-surfactant cohesion in the three-dimensional surface structures. The relationship between the surfactant force barrier and the surfactant surface affinity has been inferred recently.[24] Figure 13b demonstrates that the relative number of strongly bound surfactant monomers at the solid-liquid interface has little bearing on the mechanical properties, or the extent of the repulsive force barrier, imparted by adsorbed micelles or bilayer-like structures. Hence to some extent, it is possible to control the degree of lubrication imbibed by adsorbed surfactants without significantly altering the suspension stability. The above discussion indicates that multifunctional and tailored pressure-sensitive surfactant-based lubricants are plausible through the use of competitive adsorption and surface affinity concepts.
Fig. 13 Force barrier and pull-off force between a 7.5-mm silica sphere and flat silica surface in solutions of 32 mM C12TAB as function of (a) solution pH and (b) solution electrolyte concentration.
In another study[8] utilizing an alternative colloidal probe, it was found that modifying the cohesion between surfactant chain length does not significantly induce a modification in the lubrication, or surface engagement, behavior in the absence of salt as depicted in Fig. 14a. However, with the addition of 0.6 M NaCl, the induced ion competition resolves the difference in surface affinity between C8TAB, C10TAB, and C12TAB surfactants as illustrated in Fig. 14b. A similar result has been previously shown in studies concerning lateral force measurements of monomolecular chemisorbed or deposited films.[60,61] These investigations depict a reduction in lubrication with a reduction of the hydrocarbon chain length for intervening films.
Fig. 14 Atomic force microscopy friction force measurements between a silica wafer and a 7.5-mm silica particle for solutions containing C12TAB, C10TAB, and C8TAB surfactants at post-CMC concentrations in the absence (a) and the presence (b) of 0.6 M NaCl.
As the extent of chain-chain cohesion increases with amphiphile chain length, a reduction in friction occurs because these molecules can more effectively mask the underlying surface because of their higher mutual affinity.
Although it is possible to modify surfactant chain length to decrease surfactant surface affinity and thereby induce surface engagement and material removal, maintaining slurry stability at the same time presents an issue. The normal force-distance interaction curves between a silica sphere and a silica wafer in the presence at post-CMC concentrations in the presence of 0.6 M NaCl were previously given in Fig. 4. Under pseudo-CMP conditions, the lower chain surfactant (C8TAB) is unable to provide a steric barrier; however, it is able to incur significant surface engagement. The benchtop polishing studies collaborate with these results illustrating a material removal rate of over 5000 A/min for C8TAB compared with removal rates of over 500 and approximately 50 A/ min, respectively, for the C10TAB and C12TAB slurries in the presence of 0.6 M electrolyte. However, as predicted by the normal force measurements, the C8TAB slurries result in an unacceptable surface finish with a root-mean-squared surface roughness near 1 nm and defect diameters on the order of 50 nm.
Because of the inadequacy of the above results to provide for both good surface quality and adequate material removal, an alternative approach was attempted. It was hypothesized that by modifying the strength of the competing ion from solution, the affinity of the surfactant to the surface could be modulated resulting in a potentially tunable material removal. Moreover, as shown with the lateral force measurements with alternative salt concentrations, the kink or the sudden increase in surface engagement after a certain load in the loading curves tends to shift toward lower loads with increasing salt and indicates that non-Prestonian polishing responses[3] may be achieved through ion competition strategies with the appropriate formulation.
Fig. 15 Atomic force microscopy friction force measurements between a silica wafer and a 7.5-mm silica particle in DI water at pH 10.5 (baseline), with 32 mM C12TAB with 0.6 M NaCl and 0.1 or 0.2 M CaCl2.
To test whether modifying the surfactant surface affinity via alternative competing species could be a valid approach for achieving both high material removal rates and good surface quality after polishing, CaCl2 was added to the silica slurries in the presence of 32 mM C12TAB. Equivalent ionic strengths were maintained in the solutions containing surfactant. The lateral force measurements are displayed in Fig. 15, which illustrates that the surface engagement between the silica surface and the silica particle significantly increased in comparison to the NaCl surfactant solutions and the baseline (pH 10.5, no salt) conditions. Approaching normal force measurements in all three cases (not shown) as well as direct particle sizing indicated that all three slurries were stable. The corresponding polishing results (200-nm sol-gel silica at 12 wt.%) confirmed the LFM experiments and showed that the CaCl2-based surfactant slurries could achieve higher removal rates 1000 A/min greater) and improved surface quality (nearly identical to the non-polishing surfactant slurries) when compared with the baseline slurry. Hence by changing the magnitude of interaction of competing ions in solution, the polishing behavior of surfactant-based slurries can be modulated, resulting in improved material removal, enhanced surface finish, and potentially tunable polishing responses with the competing ion concentration in solution.
CONCLUSION
The AFM is a powerful tool for simulating nanoscale CMP processes. Through a systematic study of both normal force (approaching and pull-off ) and lateral force interactions, fundamental mechanisms resulting in modified polishing behavior can be investigated. In CMP operations, the behavior of the slurry is often the defining parameter for overall polishing performance. The integration and effectiveness of dispersants in CMP processes can provide an improved nanoscale surface finish as well as an alternative means to control the extent of material removal.
In this article, the use of AFM to probe quaternary ammonium surfactant systems for the purpose of the integration and the optimization in silica dielectric polishing slurries was reviewed. Approaching AFM normal force measurements were shown to effectively predict slurry stability and subsequent surface defectivity from benchtop polishing tests. Pull-off force measurements in conjunction with systematic lateral force measurements were found to be effective for deciphering key mechanisms behind modulated polishing performance. The influence of intervening surfactant moieties on the frictional interactions between oxide surfaces was also examined in detail.
As new, softer, and more fragile materials are introduced into microelectronic device fabrication, more intricate slurry formulations will be needed to provide for efficient material removal while maintaining adequate levels of surface defectivity and selectivity in polishing. The AFM is expected to grow more important in future CMP slurry development investigations as well as in fundamental investigations into nanoscale CMP processes. The ion competition strategies and dispersant strategies discussed within may also play a significant role in the development of novel ”smart” slurries for future CMP processes.
