초록 ▼

A system, methods and apparatus for determining the properties of electrically active imperfections in semiconductor materials by Deep Level Transient Spectroscopy (DLTS) are disclosed. Source-drain resistance of a field-effect transistor (FET) is compared with a reference resistor and the differenc

A system, methods and apparatus for determining the properties of electrically active imperfections in semiconductor materials by Deep Level Transient Spectroscopy (DLTS) are disclosed. Source-drain resistance of a field-effect transistor (FET) is compared with a reference resistor and the difference is applied through a feedback circuit to the gate. The obtained voltage transient compensates for the threshold voltage changes resulting from the emission of charges trapped by the imperfections during pulses applied to the gate and alternating with the action of the feedback circuit. Knowledge of the free charge carrier mobility is not necessary, and the intensity of the DLTS signal and the sensitivity of the measurement are area independent. Thereof, the method is advantageous for measurement of deep submicron FETs. In the disclosed system, the DLTS signal is digitized and numerically processed by a combination of two averaging and data reduction methods that offers advantageous signal to noise ratio and convenience for use comparable to these usually found in analog DLTS systems, whereas the transient analysis can be executed by advantageous numerical analytical techniques. By utilizing the disclosed combination of numerical processing methods, a DLTS signal substantially free from noise can be displayed and the expected exponential behavior of the transient decay can be verified immediately. This combination of numerical processing methods is not limited to the disclosed system and can be employed in many other DLTS methods or in other experiments requiring analyses of transient data.

대표청구항 ▼

[ Wherefore, the following is claimed:] [1.] A combination of numerical methods for processing of a deep level transient spectroscopy (DLTS) transient signal with continuous distribution in time and in intensity, comprising the steps of:a. converting said DLTS transient signal from one period in a p

[ Wherefore, the following is claimed:] [1.] A combination of numerical methods for processing of a deep level transient spectroscopy (DLTS) transient signal with continuous distribution in time and in intensity, comprising the steps of:a. converting said DLTS transient signal from one period in a plurality of sampling points, each one of said sampling points representing an averaged intensity of said DLTS signal in a time segment with a constant length from said DLTS signal;b. averaging said sampling points over time intervals, said time intervals combined in time segments, with length of said time segments increasing by the power of two;c. assigning the results of said averaging to a plurality of data points, each data point with a distinct time delay measured from commencement of said one period;d. repeating steps (a) through (c) for a second period of said DLTS periodical signal;e. subtracting each said data point from the period preceding said second period from corresponding data point with same said time delay from said second period;f. dividing each result from said subtracting to a number, said number associated with periodicity of said DLTS periodical signal into a digital time constant of averaging;g. summing each result from said dividing with a corresponding data point with same said time delay from said preceding period; andh. assigning each result from said summing to a corresponding data point with same said time delay from said second period.