edfas.org 5 ELECTRONIC DEV ICE FA I LURE ANALYSIS | VOLUME 24 NO . 4 drastically at the edges of the spectral sensitivity of an InGaAs detector.[7] This means that even tiny inaccuracies in the evaluation of measured spectral information near the outer limits of the spectral sensitivitywill have amajor impact on the extracted spectral distribution. Evidence for exponential behavior of the full spectral range (for an FET in saturation) is given because the three PE detectors typically used overlap in their respective spectral sensitivity range. These are silicon (Si), indium gallium arsenide (InGaAs), and mercury cadmium telluride (MCT) detectors. PE with a Si detector is often used when themeasurement ismade from the front side of the device under test (DUT). Corresponding spectra showing exponential spectral curves of an FET in saturation over the full spectral range of a Si-CCD have been published.[2,8] The spectral range of a Si detector overlaps with the lower wavelength limits of an InGaAs detector. A similar exponential spectral behavior from an FET in saturation was measured with an MCT detector and published in Reference 4. The lower wavelength limits of anMCT detector overlaps with the spectral sensitivity on an InGaAs detector. Figure 1 illustrates why the overlap of spectral sensitivities for all three detectors serves to explain why the calibration of the outlined InGaAs system is still incomplete. The sensitization for a meticulous system calibration of a spectral photon emission microscopy (SPEM) system, as presented here, is the key to increase the number of trustworthy data points that contribute to a precise determination of Te. This precise calibration pays off even more when device properties like the band gap are to be determined from themaxima of a Gauss shaped PE spectra.[9] Erratic calibration might shift the maximum to an erroneous bandgap estimation. SETUP SPEM setups are usually realized as an extension of “normal” PE tools. Generally, there are twoways to realize such an extension. It is possible to insert different bandpass filters and combine the different measurements to one spectrum. A more convenient method is to insert a prism into the optical path and measure the whole spectrum at once. A schematic drawing of a PE setup that is extended with a prism is shown in Fig. 2. SYSTEM CALIBRATION This section gives a detailed step-by-step description for a meticulous calibration procedure of a SPEM system. The process is appliable to all available PE detectors. However, here the focus is on calibration for a systemwith an InGaAs detector as these systems arewidely used in the failure analysis community. CALIBRATION CONCEPT This calibration concept is based on the principle of substitution. As described in the “Setup” section, an SPEM system consists of many different optical components. Some of themare very well optically characterized but some of them are not. It turned out that bringing together the optical characteristics of all components is a complicated task as they are not naturally provided by Fig. 1 Spectral photon emission distribution of a FET in saturationmeasuredwithanMCTdetector,[4] a Si-CCD detector[2] and a not-properly calibrated InGaAs detector. Fig. 2 Schematic drawing of a PEM setup with movable prism to upgrade the setup with SPEM functionality.
RkJQdWJsaXNoZXIy MTMyMzg5NA==