The downsizing of semiconductor structures predicted in 1965 by Moore's law has been ongoing ever since. This trend has lead semiconductor industry to many innovative solutions, involving reduced gate lengths, ``fins'' and ``surrounded gates'', and new materials such as high-k dielectrics and wide band-gap semiconductors, e.g., silicon carbide (SiC) and gallium nitride (GaN). In modern semiconductor devices the entire transistor is built within a few tens to hundreds of nanometers of depth. This drives the need for new characterization methods to image doped regions within one semiconductor material with high spatial resolution.
A method for the high resolution quantification of implant or charge carrier concentration was lacking for decades. This thesis aims at expanding the available scanning probe methods (SPM) to close this yet existing gap. A number of different carrier- and dopant profiling techniques are compared using a standard dopant calibration sample. Scanning Microwave Microscopy (SMM), on which the focus of this thesis lies, stands out by fulfilling the requirements in spatial resolution, and signal dependence on dopant density. Furthermore, the SMM signal contains the information to distinguish the dopant types, enabling the development of dopant quantification methods.
Since SMM combines the mechanical tip dimensions with the microwave excitation region, the effective scan resolution is not directly given by the cantilever parameters. The highly confined two dimensional electrongas (2DEG) in an aluminum gallium nitride (AlGaN)/GaN heterostructure was scanned and the obtained profile deconvoluted with the expected shape of the 2DEG profile to reveal the effective tip radius in SMM. A comparison between different cantilevers shows that the effective scan resolution is limited mainly by the cantilever radius and shape and only by a smaller extent by the microwave excitation region.
An established procedure is applied to calibrate the S11 data recorded in SMM into capacitance and resistance. Based on the resistance signal, a method to calculate resistivity and dopant density from the SMM resistance is employed. It enables the quantitative two-dimensional dopant profiling with an accuracy of better than 60%.
A second introduced method is based on the SMM capacitance data. It enables to distinguish the dopant type and shows a better repeatability. The difference between accumulation and depletion capacitance is used for dopant type distinction. The depletion capacitance is shown to present a reliable tool to investigate the relations between dopant densities. This enables the calculation of dopant densities deviating from the datasheet densities by a factor of 0.1 to 20. Further improvements are achieved by using parameter fits. The method enables the investigation of dopant types and densities in one single image with all data obtained in one single measurement.
The transfer of the dopant density calculation to an industrial application is demonstrated for an ESD diode. A further industrial application is the spatially resolved imaging of charge carriers accumulated by trapping effects, shown on the example of a Si diode.
«
The downsizing of semiconductor structures predicted in 1965 by Moore's law has been ongoing ever since. This trend has lead semiconductor industry to many innovative solutions, involving reduced gate lengths, ``fins'' and ``surrounded gates'', and new materials such as high-k dielectrics and wide band-gap semiconductors, e.g., silicon carbide (SiC) and gallium nitride (GaN). In modern semiconductor devices the entire transistor is built within a few tens to hundreds of nanometers of depth. Thi...
»