This
is the most elaborated conventional DLTS measurement mode and it provides the
great deal of information. It is based on a single temperature scan. During this
one scan even hundreds of rate windows (emission rates) can be obtained which
allows calculation of the defect energy from the Arrhenius plot with the large
accuracy. This temperature scan starts from the current temperature set point
and proceeds to the End temperature. For a given rate window setting a transient
with the preset acquisition parameters is measured. Next it is used for
calculation of the conventional DLTS signal and this signal is plotted on the
main graph. In parallel, the transient can be automatically analysed with a use
of the one-exponential function fitting procedure (red points) or the Fast_DC library (three
exponential function fitting, green points). These on-the-fly calculations have
to be allowed with check boxes on the Use for fitting panel. The emission
rates obtained by these routines are plotted on the Emission rate viewer window
opened with this button and the exponential component amplitudes are plotted on
the main graph together with the conventional DLTS spectrum. The blue horizontal
line on the Emission rate viewer shows the current value of the Rate Window. If
the one- and/or three-exponential analysis are allowed then the obtained results
will be stored in the result file.
Below the Rate window option buttons there are two other option buttons where one can chose between the lock-in versus box-car type of the conventional DLTS spectrum calculation. The first one sums up the first 40% of the transient points (between 10% and 50%) and subtracts it from the last 40%, while the latter methods takes only the first 12.5% of the transient length for the corresponding calculation. The first method is characterised by a better signal-to-noise ratio and by poorer resolution than the other one.
Due to a fact that for the fitting procedure the whole transient is used (similarly to the case of the Laplace spectrum calculations) it is essential that the beginning of the transient is properly assigned. If the Beginning removal is set to Auto then some numerical routine based on general criteria is used to find automatically the first point on each of the transient. This point is shown in the corresponding text box. If the Auto checkbox is unchecked then from each of the transient the same number of points is removed. This number can be set by the corresponding scroll bar on this panel. The validity of both ways of the beginning assignment should be visually inspected with the use of the Transient viewer described above.
The
measurement of the conventional DLTS spectrum in the Exponential Fitting mode
can be accompanied with a simultaneous measurement of the CT
(capacitance-temperature) curve for spectrum normalisation purposes. In order to
activate this function one has to (1) measure the sample capacitance at the
current bias (a command button on the Capacitance tab), and (2) activate the CT
normalisation curve measurement by pressing the corresponding button (with a
modified CT logo). If the sample bias capacitance has been measured but the CT
curve normalisation has not been activated then the given conventional DLTS
spectrum still can be normalised but in this case only with a use of one single
capacitance value. A visualisation of the spectrum normalisation procedure is
realised in the data base program. Any change in the sample bias results in
clearing the value of the latest sample bias capacitance. The measurement
of the CT normalisation curve creates the [ct] section in the conventional DLTS
data file. This data can be used as a standard CT curve at a given bias but it
does not go into the data base as a separate CT curve measured via the CT
measurement function. NOTE: the CT normalisation curve together with the
amplitude values obtained from the 1exp and 3exp fitting procedures can be
viewed in the data base program when the given spectrum is displayed in the
View/Show One mode (see the Data Base program section of this manual for
details).
In this measurement mode any change in the hardware setting is immediately passed to the hardware. This allows to modify the measurement parameters while the measurement in running. The end temperature and the temperature ramp can be also changed during the measurement.
-conventional vs Arrhenius-plot view of the main graph
The main graph can be drawn in either conventional mode, i.e., the DLTS spectrum overlaid with points representing the mono-exponential component amplitudes, or the Arrhenius-plot mode where the logarithms of component emission rates divided by the temperature square are plotted versus the
reciprocal temperature. The Arrhenius-plot is allowed if at least one
fitting procedure was during the measurement used. The conventional vs
Arrhenius-plot view can be changed by clicking either the Restore or Arrhenius
view button on the conventional DLTS form.
- mouse events and the Arrhenius-plot fitting
The mouse events are only effective for the main graph and, basically, they were designed to help manipulation with the results of the one- or three-exponential fitting procedures. In both types of the main graph view one can zoom and unzoom a group of points drawing a box around them with the left mouse button. In the Arrhenius-plot view using the right mouse button one can draw a corridor for a group of points which could represent the emission process from one defect. When the right mouse button is released the points in the corridor are taken for the Arrhenius-type calculations and the results are on the graph displayed. The points outside the corridor are from the graph removed and the fitted line is plotted. The width of the corridor can be changed in arbitrary graph units with the scroll bar on the General tab. The zooming and fitting actions can be repeated either on the zoomed or restored graph.
The results of the one- or/and three-exponential fitting procedures can be saved (exported in the “.pek” file format) to be later analysed with the TrapView ver. 3.0 software. Note, all points seen on the main graph are exported, so it is recommended to make first a selection of some of them by zooming or drawing a corridor. Some (LD) experience shows that TrapView ver 3.0 chocks if there are too many points in the “.pek” file.