Thermal wave physics is playing an ever increasing role in the on-line characterization of semiconductor materials and devices. This is especially true for thermal wave methods that employ laser beams for both the generation and detection of thermal waves. For the modulated reflectance method discussed here, the pump and probe beams are focused on the same spot. They therefore achieve the noncontact advantage of optical methods in addition to the optimum condition for high spatial resolution, a necessary condition for thermal wave measurements on product wafers.When a material is excited with an intensity-modulated laser beam or pump, a thermal wave is generated in the material and in the air above the sample. The material within this heated region will undergo a thermal expansion which can be detected with a probe beam interferometer or by deflecting the probe beam from the thermoelastic deformation of the surface. Since the complex refractive index of most materials depends on temperature, a modulated temperature will also induce a corresponding modulation in the refractive index and consequently a modulation on a probe beam passing anywhere near the thermal wave. A probe beam directed along the heated surface of the sample, for example, will be deflected as it passes through the heated region above the surface. This mirage effect can also be observed within the sample by directing a transmitting probe through the heated region beneath the surface. Likewise, using a probe beam directed onto the sample surface one can observe a modulation in reflection, transmission, or scattering. A related noncontact method is the photothermal measurement of infrared radiation emitted from the material's heated region. Note that with all these detection methods, thermal wave measurements can be, and most often are, done in air and at room temperature.

As indicated by the ITRS roadmap, obtaining accurate information on the electrically active dopant profile for sub-30-nm structures is a key issue. Presently, however, there is no conventional, probe-based (destructive) technique available satisfying the ITRS targeted depth (3%) and carrier level (5%−10%) reproducibility and accuracy. In this work, the authors explore the promising capabilities of nondestructive photomodulated optical reflectance (PMOR) techniques, based on the localized (micrometer beam size) detection of variations in the reflectivity of the sample, due to thermal and plasma (excess carrier) effects as can be generated by a modulated pump laser such as the Therma-Probe® (TP) system. Earlier and more recent work using low modulation (1 kHz) frequencies has shown that it is possible, but rather tedious, to extract the electric junction depth (at about 1018 cm−3) and carrier concentration of chemical vapor deposition grown (CVD) (boxlike) structures based on so-called power curves (where the reflected power of the probe laser is plotted versus the power of the pump laser). In this work the authors focus on high-frequency (1 MHz) PMOR, which gives two (instead of one) independent signals, i.e., the amplitude (A) and phase angle (ϕ) of the reflected probe beam. It has been proposed earlier and is confirmed in this work that a single simple measurement allows for the direct and easy extraction of the junction depth (Xj) and carrier concentration (N) of boxlike profiles. Furthermore, the shape of the so-called three-dimensional PMOR offset curves (A and ϕ versus offset), where the distance of the pump relative to the probe beam is varied over several micrometers, might help to obtain information on more complex profiles. The principles allowing for the extraction of arbitrary carrier profiles, with nanometer depth resolution and carrier concentrations between 1018 and 1021 cm−3, from offset curves will be discussed and evidence for the proposed ideas will be given for homogeneously doped material and CVD boxlike structures based on FSEM device simulations.

Shrinkage of device dimensions requires tighter lithography process control([1]). Current levels of Process Control leave less than 0.5 nm budget for CD metrology. An accurate and stable metrology solution requires measurement of CD and profile that are critically dependent on thin film material characterization at various earlier process stages.Opti-Probe (R) integrates five different technologies into a single platform to accurately characterize optical properties of 193 nm materials([2]). Real-time CD (RT/CD (R)) technology utilizes four independent spectra collected from the samples using a rotating-compensator spectroscopic ellipsometer [3] (RCSE) and analyzes the spectra with an innovative numerical solution-finding approach([4]) to construct detailed CD and profile of printed features in a 2- and ([5-8]) 3 Dimensional geometriesThe study presents a comparison ofi) Methodologies using an advanced combination of metrology techniques to characterize 193 nm materials (e.g. ARC)ii) Measured CD and profile variations using RCSE of Opti-Probe and RT/CD technology.iii) Correlation between measured CD variation and measured material characteristics.In order to achieve less than 0.3 nm accuracy and stability requirement for sub 65 nm process development and CD uniformity control, less than 0.003 variation and accuracy in optical dispersion (n&k) of critical material has to be ensured.

We demonstrate that a multi-technology approach enables the accurate characterization of the thickness and optical properties of Amorphous Carbon (α-C) films used in semiconductor manufacturing. Because the material is found to be highly birefringent, with its measured refractive index and extinction coefficient depending strongly upon the polarization and angle-of-incidence of the optical probe beam that is used, conventional single angle-of-incidence Spectroscopic Ellipsometry (SE) has insufficient information-content to detect the ordinary and extraordinary refractive indices with sufficient accuracy. On the other hand, Beam Profile Reflectometry® (BPR®) is particularly strong in this case because it measures the actual reflectances for both polarization components (rather than just the difference between them) at multiple angles-of-incidence. Furthermore, the fact that BPR is a single-wavelength technique means that no assumptions must be made about the optical dispersion of the film and an absolute measurement can be made. It is then possible to combine this with other, spectral technologies to obtain ordinary and extraordinary n and k at all wavelengths. We show that, to first order, the birefringence can be modeled by assuming a single “anisotropy parameter”, the ratio between the ordinary and extraordinary refractive indices (or, more correctly, the complex dielectric functions). Using this approach in combination with a simple Bruggeman effective-medium dispersion model enables robust characterization of these films for production.