A model of copper chemical mechanical polishing ͑CMP͒ based on methods of chemical kinetics is presented which includes both chemical and mechanical processes. This CMP model explains observed patterns in removal rates for peroxide and nonperoxidebased slurries as a function of oxidizer concentration, polishing pressure and speed, etchant concentration, and pH.

The removal mechanism for interlevel dielectric ͑ILD͒ chemical mechanical polishing ͑CMP͒ with fumed silica abrasive slurry was studied by measuring silicon dioxide wafer removal rates as a function of abrasive concentration and pH and also by examining the surface charges of the abrasive particles at different pH's. The interfacial removal kinetics indicates that the wafer removal rate is consistent with a first-order reaction and is proportional to the concentration of surface silanolates on the abrasive particles. The actual removal is proposed to be achieved by a direct nucleophillic attack of the silica particle silanolates on the wafer Si-O bond. Pad surface texture provides the means to transport the slurry to the contact zone and the effective concentration of abrasive particles doing actual removal is greatly influenced by the pad surface macrotexture by grooving design and the microtexture by conditioning. The observed correlation between the pad surface parameters that are used to characterize the pad microtexture and the removal rate highlights the importance of the pad surface texture to ILD CMP.For the past 20 years, silica-based slurries have been used for chemical mechanical polishing ͑CMP͒ of interlevel dielectrics ͑ILDs͒. 1 These slurries achieve a near-perfect global planarization in manufacturing ultralarge scale integrated devices. 2 In a typical ILD CMP, the surface topography of the silicon oxide dielectric is planarized by polishing a rotating wafer that is pressed facedown on a rotating polyurethane pad in the presence of an abrasive slurry. CMP is an interfacial chemical mechanical phenomenon 3 and a better understanding of the interfacial interactions among the slurry, pad, and wafer is needed to improve the pad and the slurry performance.Most of the fundamental understanding of ILD CMP is derived from glass polishing. The widely accepted chemical mechanism is the "chemical tooth" model proposed by Cook in 1990 4 and this model was further refined by Osseo-Asare in 2002. 5 A high removal rate ͑RR͒ can be obtained when the slurry pH is close to the point of zero charge of the abrasive particles ͑e.g., ceria͒. In the model, the material removal during CMP is viewed as an adsorptive process involving release of substrate-derived species into a solution ͑disso-lution͒ followed by the adsorption of these species by abrasive particles. This model predicts that the highest oxide RR occurs at pH 9 for Al 2 O 3 . This has not been experimentally observed in normal CMP; further fundamental understanding of the interfacial chemical process in ILD CMP is needed.There is a large body of work exploring the mechanical contributions from the pad on ILD RRs. 6 The generic mechanistic framework for silicon oxide removal involves Preston's empirical equation, which states that RR is proportional to polishing down force and speed. 7 Unlike copper CMP, the material RR in ILD CMP follows Preston's equation fairly well, although some modifications have been proposed. 8 So, the CMP community views ILD polishin...

A 3M A3700 diamond disk was used to condition a Cabot Microelectronics Corporation D100 pad for 30 hours, and wear on its aggressive diamonds was analyzed. The top 20 aggressive diamonds for two perpendicular disk orientations were identified before wafer polishing, as well as after 15-hour and 30-hour polishing. Results showed that the original top 20 aggressive diamonds identified before polishing were subjected to wear after the first 15-hour polishing as the furrow surface area that they generated decreased dramatically by 47%. As these original aggressive diamonds were worn, seven new aggressive diamonds were "born" and joined the new top 20 list for both disk orientations. After the second 15-hour wafer polishing, the furrow surface area of these new top 20 aggressive diamonds did not change significantly. The furrow surface area created by all the active diamonds exhibited the same trend as the top 20 aggressive diamonds, confirming that most pad conditioning work was performed by these aggressive diamonds and that the disk lost its aggressiveness in the first 15 hours of polishing and then maintained its aggressiveness during the second 15 hours.Chemical and mechanical planarization (CMP) has been widely employed in the integrated circuit manufacturing industry to achieve local and global surface planarity. It is well known that diamond disks are commonly used for pad conditioning to prevent removal rate decay during CMP as the embedded diamonds cut across the pad surface under an applied load to regenerate pad asperities and remove used slurry and pad debris. 1-3 During pad conditioning, pad asperities and slurry abrasives make mechanical contact with the diamonds causing the diamonds to wear, 3 leading to the loss of disk effectiveness. A conventional diamond disk is typically replaced after dozens of hours of use. 4 Therefore, it is important to investigate diamond wear and loss of disk efficiency. There have been a few studies on diamond wear during CMP. For example, Liao 5 found that peripheral diamonds and the diamonds that originally protrude more highly with crest lines oriented upward wear faster. Borucki et al. 6,7 demonstrated that for a conventional diamond disk, typically less than 1 percent of the embedded diamonds (called active diamonds) create cutting furrows on the pad surface during pad conditioning. Other diamonds (referred as inactive diamonds) are not involved in pad cutting. Among the active diamonds, typically only 10 to 20 diamonds (named as aggressive diamonds) do the majority of the cutting and are therefore most susceptible to fracture or being pulled out. In a recent study, Meled et al. 3 conducted 24-hour wear tests using three types of diamond disks. Results showed the presence of micro-wear on the aggressive diamonds and no appreciable wear on the inactive diamonds. While diamond wear has been characterized to some extent, several questions remain unanswered. For example, do new aggressive diamonds appear as the original aggressive diamonds wear? How do these new aggressi...

CMP has been described qualitatively in terms of alternating cycles of chemical formation and mechanical removal of a thin layer on the wafer surface. A quantitative model of CMP has been developed2-7 which is based on mechanisms for surface kinetics, treating mechanical removal as one step in the mechanism. This model has been used successfully to explain removal rates for tungsten and thermal oxide CMP. In particular, for tungsten CMP the removal rate increases steeply with increasing oxidizer concentration at low concentrations, and then approaches an asymptotic maximum removal rate at high concentrations. The model explains this by starting with the assumption that mechanical abrasion removes only tungsten oxide but not tungsten metal. It then focuses on the fraction of wafer surface covered by a tungsten oxide layer. At low oxidizer concentrations, the oxide formation rate is small compared the removal rate, so only a small fraction of the surface is oxidized and the removal rate is small. At high oxidizer concentrations, the oxide formation rate is large compared to the removal rate, so most of the surface is oxidized and the removal rate is large. Increasing the oxidizer concentration in the high oxidizer concentration region does not significantly increase the surface fraction of tungsten oxide, and the removal rate approaches a constant value.

Precise understanding of the pad-wafer contact is needed for designing polishing pads and planarization processes. This paper presents a deterministic semi-analytical model for investigating the elastic contact between a rough bi-layer porous body (pad) and a rigid plane (wafer). Homogenized or equivalent material properties are obtained and utilized for modeling each layer. The frequency response functions (FRF) for contact involving a bi-layer material, based on the Papkovich-Neuber potentials, are used, and the model is solved with the conjugate gradient method (CGM) and a fast Fourier transform (FFT) approach. The simulated pad-wafer contact areas are compared with the results from optical contact measurements for model verification. The application region of the bi-layered model is determined, and a map for the use of the bi-layer contact model is generated. The impacts of materials and layer thicknesses on contact ratio are analyzed.