00:00:44.410 - 00:00:53.060

The field in semiconductors is often inhomogeneous and influences the current through it. We will investigate this and have

00:00:53.060 - 00:01:02.150

chosen a model substance. This is the cadmium sulfide single crystal here shown as a platelet. To measure the current, we attach

00:01:02.150 - 00:01:09.600

two electrodes on opposite sides of the crystal. One measures the current as a function of the applied voltage.

00:01:09.600 - 00:01:15.970

But one needs to know the potential distribution within the crystal. And there are Orthogonal Trajectories

00:01:15.970 - 00:01:23.390

showing the corresponding current distribution. The space between the electrodes is too small for optical observations.

00:01:23.390 - 00:01:30.260

Therefore, we offset the electrodes sideways to investigate field and current between them easily.

00:01:30.260 - 00:01:38.570

In real crystals, these distributions are more complex. It is influenced by schottky barrier below and around

00:01:38.570 - 00:01:52.940

the cathode that compress the potential distributions here. Or by crystal defects here a small crack, the lower left, that causes

00:01:52.940 - 00:02:03.020

a more complex potential distribution. Or by inhomogeneous doping shown as light coloring the right, that causes further deformation of

00:02:03.020 - 00:02:12.770

both current and potential distributions. All this must be known before one understands the measured current voltage characteristics.

00:02:12.770 - 00:02:21.380

To obtain such a potential distribution, one probe to the surface of the crystal between the electrodes with a thin point contact

00:02:21.380 - 00:02:30.320

and measures the voltage with electrometer that shows such irregularities in an actual experiment. This is a very tedious procedure

00:02:30.320 - 00:02:38.730

by measuring this potential in two-dimensions along the entire surface of the crystal point for point. And even at each point,

00:02:38.730 - 00:02:44.680

the resistance and capacity of the metering prevents us from measuring fast changes in the distribution.

00:02:44.680 - 00:02:53.480

It would hide most kinetic effects. Only after hours of probing, we can now plot the complete potential distribution.

00:02:53.480 - 00:03:01.620

But we thought that there must be a better way to obtain directly the field distribution by looking through the crystal

00:03:01.620 - 00:03:11.210

and observing the changes in optical absorption as a function of the actual electric field within the crystal. See what we observe the very first

00:03:11.210 - 00:03:21.080

time we did this electro-optical observation. We have never seen such high field rings circling the cathode and disappearing when they hit the anode.

00:03:21.080 - 00:03:32.040

And the process repeated and repeated itself time and again. Up to this first observation in 1958, nobody has ever known such a phenomenon.

00:03:32.040 - 00:03:40.790

What kind of electro-optical effect permitted us to actually see this field distribution? Let us look at the conduction and

00:03:40.790 - 00:03:48.560

valence bands in a crystal. When electrical field is applied, these bands are tilted and the electrons run

00:03:48.560 - 00:03:57.830

against its upper edge of the valence band and are able to tunnel a bit into the forbidden zone. Light is absorbed by bringing electrons

00:03:57.830 - 00:04:06.730

from the valence into the conduction band. There is a sharp absorption edge for wavelengths longer than the corresponding photons

00:04:06.730 - 00:04:13.160

having this sufficient energy. However, when the bands are tilted and some of the electrons are already

00:04:13.160 - 00:04:22.950

slightly above the edge of the valence band. And these can also be excited into the conduction band. Consequently, they need less photon energy

00:04:22.950 - 00:04:33.370

here shown by an epsilon. The optical band edge is shifted to longer wavelengths. We coined this the Frans kel dish effect.

00:04:33.370 - 00:04:41.120

Hence, the shift of the absorption edge can be used as a direct measure of the actual electrical field.

00:04:41.120 - 00:04:52.160

Within a crystal. Higher field, the edge shifts to higher wavelength. When looking through the crystal

00:04:52.160 - 00:05:02.820

close to the absorption edge, one observes these fields as an increase of absorption, as a darkening. The more so, the higher the field is.

00:05:05.920 - 00:05:15.260

Since we already observed kinetic effects of the field distribution in our first experiment, we now use a movie camera to register

00:05:15.260 - 00:05:32.240

these effects in monochromatic green light at the absorption edge. Here we were indeed able to register the well-known Schottky barriers as

00:05:32.240 - 00:05:47.010

darkening under the transparent cathode. This becomes evident when the voltage is switched on or off, on or off, on or off.

00:05:47.110 - 00:05:56.330

In the animated part of the movie, we show such a Schottky barrier as darkening under the cathode, but also a bit of a barrier under

00:05:56.330 - 00:06:07.670

the anode as it was observed at some of the samples. The probing shows the steeper increase the voltage

00:06:07.670 - 00:06:28.420

before the cathode and a bit before the anode, then in the center of the crystal. However, the expanding rings we saw

00:06:28.420 - 00:06:35.960

before again appear here at higher applied voltages. But they remain stationary as a high field region that is well

00:06:35.960 - 00:06:48.230

separated from the cathode. This is a new effect that was not known previous to these first experiments. In probing, we see

00:06:48.230 - 00:07:10.400

this steep increase of the voltage where the ring appears that is well within the crystal. This voltage increase

00:07:10.400 - 00:07:33.490

reflects a field inhomogeneity that in turn requires a corresponding space charge double layer. It takes a few seconds after

00:07:33.490 - 00:07:49.310

switching on the voltage until the high field fully develops and after switching off to disappear. However, when the light is turned off,

00:07:49.310 - 00:07:59.410

before switching off the voltage, and without applied voltage, the light after 10 min is switched on again, the high field region darkening is still

00:07:59.410 - 00:08:11.360

visible for a second before it disappears. This again is demonstrated in the repetition observed for a moment, the dark area when the light is on again.

00:08:11.380 - 00:08:20.570

The time constant for changing the high field region with abruptly changing the applied voltage depends on the light intensity.

00:08:20.570 - 00:08:29.260

It takes longer for the field inhomogeneity to appear, the lower the light intensity is. In the extreme case of complete darkness,

00:08:29.260 - 00:08:39.140

this time constant is extremely long and it can take more than 10 min. These high field effects have some similarity to

00:08:39.140 - 00:08:49.820

a low pressure gas discharge. Hence, we coined this a plasma like field inhomogeneity. We show here two typical plasma effects.

00:08:49.820 - 00:08:59.060

The glow discharge of an old glow lamp here shown with DC applied, illuminating the cathode and a layered low pressure gas discharge,

00:08:59.060 - 00:09:08.200

these high field effects appear at higher applied voltage in a range of decreasing conductivity with increasing applied voltage.

00:09:08.200 - 00:09:31.960

In cadmium sulfide. This is caused by field quenching. Here the reaction kinetics between absorption of photons,

00:09:31.960 - 00:09:41.090

trapping and recombination is disturbed by electrically exciting holds from a whole traps into the valence band from where they reach

00:09:41.090 - 00:09:52.860

fast recombination centers and thereby decrease the photo conductivity. This is shown at reduced magnification at the right.

00:09:53.680 - 00:10:06.600

With light, the electrons are excited up into the conduction band. With a sufficient field applied partisan rain down again, they recombine.

00:10:15.960 - 00:10:29.770

Without field. Both the electron density is shown in yellow and hold entities shown in red are equal,

00:10:29.770 - 00:10:41.170

shown combined in orange. And both decrease when the field is high enough to initiate field quenching. When for statistical reasons,

00:10:41.170 - 00:10:48.640

the field at one point of the crystal is a little higher here in the middle the valley, and the carrier distribution gives rise

00:10:48.640 - 00:11:00.290

to a space charge double layer. This causes the field to increase further until it reaches the critical field strength for

00:11:00.290 - 00:11:09.670

initiation of the high field region. And then the high field layer becomes self sustained. When the applied voltage is insufficient,

00:11:09.670 - 00:11:16.880

the field inside the crystal can be increased to over critical values when placing a shadow in front of the crystal,

00:11:16.880 - 00:11:26.120

thereby decreasing the light and the shadow region and thereby decreases the photo conductivity. As a result, the field is

00:11:26.120 - 00:11:36.350

increased in the shadow band. When the field here has reached over critical values are high. Field domain is artificially initiated.

00:11:36.350 - 00:11:45.500

This is indicated in the graph by changing the shadow into the characteristic high field domain shape. When the applied voltage is high enough to

00:11:45.500 - 00:12:06.150

maintain such a high field domain, one sees it remaining when the shadow is removed. Once one domain is initiated,

00:12:06.150 - 00:12:17.970

one can initiate another one and another one, and so on. Observe that these artificial initiated domains

00:12:17.970 - 00:12:27.360

have a tendency to move towards the left, in contrast to the natural creative domains that all move toward the right, toward the anode.

00:12:28.390 - 00:12:35.030

In the animated part, such a domain is shown with the electrons shown as blue balls move from

00:12:35.030 - 00:12:46.100

the domain towards the anode. While the less mobile holes, shown as red balls are left behind. This causes the space charge region to move

00:12:46.100 - 00:12:55.530

slowly to the left as more holes are left behind. This is exactly as we have observed in the experiment before.

00:13:12.180 - 00:13:20.350

There was a lot of motion within each domain that is naturally created at the cathode. This is seen as a fast,

00:13:20.350 - 00:13:38.310

more chaotic bubbling within the domain. The domain by itself moves toward the anode. As the domain approaches the anode, a darker region becomes visible.

00:13:38.310 - 00:13:45.470

Here, a new anode adjacent part of the domain width, higher fields is created When a voltage is

00:13:45.470 - 00:13:53.570

switched and apply it again, the same features repeat themselves. In more homogeneously doped crystals, the subdomains have

00:13:53.570 - 00:14:01.130

a more coordinated motion. They clearly moved toward the left, that is in the opposite direction of the domain motion.

00:14:01.130 - 00:14:12.380

This phenomenon is only found in fully developed natural domains. This is suggested as a direct evidence that in a highly field

00:14:12.380 - 00:14:22.340

quenched region of the domain, the density of the hole can dominate the crystal here turns to be p type, which would then cause the sub-domain

00:14:22.340 - 00:14:32.770

to move in reverse. As an additional effect, one sees a darkening of the crystal when the domain comes close to the anode.

00:14:32.770 - 00:14:40.200

This is caused by a substantial increase of the current, hence by Jules warming, causing an additional shift

00:14:40.200 - 00:14:56.820

of the absorption edge. When the domain reaches the anode, the current increases much further and the crystal darkens totally.

00:14:58.780 - 00:15:09.780

When a new domain is formed at the cathode, the current decreases again and the warming stops. The crystal becomes transparent again.

00:15:19.800 - 00:15:30.410

The domain initiation can be observed in the characteristic when with increasing the voltage, the current first nearly linearly increases,

00:15:30.410 - 00:15:39.500

but turns sharply down when the domain appears. This is the beginning of the N shaped negative differential conductivity.

00:15:39.500 - 00:15:46.900

The statistical nature of the domain initiation can be seen when the voltage is increased with higher speed,

00:15:46.900 - 00:16:00.420

the initiation is delayed. And inversely, more time is given. For statistical initiation, the domain becomes visible at lower voltage.

00:16:05.290 - 00:16:13.960

It is well-known that the electric noise that is created in a crystal increases with increasing applied voltage.

00:16:13.960 - 00:16:21.650

Not known however, was the fact that parallel to the increase in low-frequency noise, the chaotic motion of

00:16:21.650 - 00:16:30.240

subdomains within and often extending beyond the domain increases substantially with increasing applied voltage.

00:16:38.190 - 00:17:02.170

In different cadmium sulfide crystals, this chaotic motion can assume different interesting patterns. When domains are artificially initiated by

00:17:02.170 - 00:17:13.790

a slanted shadow, they move, as discussed before, toward the cathode and slowly straighten out. They become normal looking.

00:17:13.790 - 00:17:45.480

Again. That even happens when a crooked shadow initiates a crooked domain. But it's straightens out again. In real crystals with more defects.

00:17:45.480 - 00:17:52.810

Here, a crystal with a small crack near the lower right edge, the domain motion is more substantially disturbed.

00:17:52.810 - 00:18:03.230

Domains in such crystals can assume a variety of shapes with a large variety of kinetics. Without electro-optical direct observation, We would simply be

00:18:03.230 - 00:18:13.220

blindfolded when we could only resort to the measured characteristics in part to the processes leading to the dielectric breakdown are shown.

00:18:13.220 - 00:18:23.850

Breakdown is the catastrophic end and the life of devices. It starts slowly weakening its structure, often unpredictable by life testing.

00:18:23.850 - 00:18:34.670

Minute current channels are often the beginning of a thermal break down. We will investigate this initial phase here. We show such thermal breakdown on

00:18:34.670 - 00:18:42.470

an insulator plate and the high voltage laboratory, Technical University of Dresden. Here this plate is immersed in

00:18:42.470 - 00:18:51.530

an oil bath where it will be exposed to high voltage. Asked her installation, the engineer's retracted into

00:18:51.530 - 00:19:22.810

a secure place where they can increase the voltage in the lab until the breakdown occurs. The insulator is taken out of

00:19:22.810 - 00:19:31.010

the bath and a burned holes through the plate is seen in its center as a result of the thermal break down.

00:19:37.020 - 00:19:45.490

To make the processes preceding such a breakdown visible, we again use the cadmium sulfide crystal and now

00:19:45.490 - 00:20:04.850

observe the heating in the crystal through the thermal shift of the absorption edge. These shifts of the absorption edge, We followed by moving

00:20:04.850 - 00:20:32.210

the crystal on its holder into the center of a furnace and raise the temperature. The crystal changes its color from

00:20:32.210 - 00:21:03.650

yellow to orange to red when the temperature is raised to 200 and then to 500 degrees centigrade. When cooled back to room temperature,

00:21:03.650 - 00:21:18.680

it turned again yellow. We want again, can use the now much larger shift of the absorption edge,

00:21:18.680 - 00:21:26.070

which causes a change of its color as a direct measure of the temperature, we can use it as a thermometer.

00:21:32.670 - 00:21:43.300

And inhomogeneous leading here and a small flame shows an inhomogeneous coloration of the crystal, which disappears again when

00:21:43.300 - 00:21:54.800

the crystal is removed from the flame. Now we produce the heating electrically. For this purpose, there are two electrodes on the cadmium sulfide crystal

00:21:54.800 - 00:22:04.360

deposited and that is mounted in a crystal holder. Then this holder is inserted into the light beam of

00:22:04.360 - 00:22:23.840

an optical bench where the changes due to joule heating can be photographed. When a sufficiently high

00:22:23.840 - 00:22:33.740

electric voltage is switched on, a current channel is formed between the two electrodes that is now visible as red channel.

00:22:33.740 - 00:22:42.230

This shows a substantial joules heating of the channel. From this temperature profile, we can now draw the current distribution

00:22:42.230 - 00:22:50.890

and its orthogonal trajectories. The equal potential lines with a set of color filters, one determines the extent

00:22:50.890 - 00:23:14.780

of a temperature range. What obtains the exact temperature profile of the current channel? This temperature profile is

00:23:14.780 - 00:23:32.520

shown here with temperatures in the center that extend up to 500 degrees centigrade. With increasing applied voltage, the current channel grows across the crystal

00:23:32.520 - 00:23:51.750

until it bridges the gap between the electrodes. The channel also moves from side to side and show some wave-like internal kinetics.

00:24:14.950 - 00:24:26.330

The current channel often begins with a node close to the cathode. This node elongates until it becomes a complete current channel

00:24:26.330 - 00:24:39.750

extending from the cathode to the anode. Current can be read from the ammeter that is mirrored into the lower part of the picture.

00:24:46.870 - 00:25:15.420

With increasing current, the channel broadens in width. It gets hotter with increasing current. And finally, heats almost the entire crystal.

00:25:22.060 - 00:25:34.250

Hi field domain, seen as an orange layering moving from left to right can briefly be observed. Here. This layer starts

00:25:34.250 - 00:25:55.210

the channel initiation at the anode. This again is seen in the replay. As soon as the current channel is formed, the current voltage characteristic shows

00:25:55.210 - 00:26:05.780

a negative differential resistance. One can see this by comparing the ammeter and the volt meter. Both of them are mirrored into the picture.

00:26:05.780 - 00:26:15.480

First, the voltage increases with slightly increasing current. When the current channel is formed, the current increases,

00:26:15.480 - 00:26:24.720

but the voltage decreases. When the electric power is decreased. Again, both instruments move in the opposite direction.

00:26:24.720 - 00:26:34.240

The current decreases, while first the voltage increases, until both decrease when the channel disappears.

00:26:34.240 - 00:26:49.450

The current voltage characteristic shows a typical s-shaped behavior. When the characteristic has reached its largest voltage point,

00:26:49.450 - 00:26:59.910

it changes its color from yellow to red, indicating the initiation of the current channel. In order to avoid a runaway current,

00:26:59.910 - 00:27:11.900

one employs a large series resistance. Here we are again in the high voltage laboratory and show a horn spark gap.

00:27:11.900 - 00:27:20.270

Spark initiates an arc between the nearest points of the electrodes. And the arc is pushed out by the magnetic field.

00:27:20.270 - 00:27:29.570

When it reaches the top of the horns. The distance between the electrodes becomes too large to support the arc. It is extinguished.

00:27:29.570 - 00:27:37.160

The interaction of the magnetic field with the arc forces it to the right until it can align with the field.

00:27:37.160 - 00:27:47.920

In a cadmium sulfide crystal with a wire like extension of the right electrode. The current channel is initiated at the closest range between the electrodes,

00:27:47.920 - 00:27:55.360

but it is rapidly pushed out to the electrode tip so that it continues in the direction of the wire like electrode.

00:27:55.360 - 00:28:20.890

This motion to the tip of the electrode is seen in several crystals. However, when the channel becomes too hot, the motion into a yet untreated part of

00:28:20.890 - 00:28:40.420

the crystal overcomes the force of the magnetic field. Here we see this heat induced motion again. In crystals that were recently

00:28:40.420 - 00:28:49.880

traversed by hot current channels. The final channel is pushed against one side where it becomes stationary. One can also observe

00:28:49.880 - 00:28:57.350

the destruction of part of the electrodes when the channel becomes rapidly too hot. Here, the hot nodes of

00:28:57.350 - 00:29:05.170

the current channel just peel part of the electrode off. At lower currents, we can follow more clearly the wave-like

00:29:05.170 - 00:29:22.770

motion within the channel. These are all part of the beginning processes of a thermal break down.

00:29:32.350 - 00:29:40.140

The final breakdown proceeds much more rapidly when the crystal is not protected by a series resistance.

00:29:46.600 - 00:30:02.270

The rapid heating produces mechanical stresses. As a result of the stress at higher power, the crystal fractures here

00:30:02.270 - 00:30:23.230

shown as small cracks in the field negative the thermal stresses visible by D forming part of the striations

00:30:23.230 - 00:30:42.260

near the initiation point of the channel. That's seen by switching the power on and off. And even a crystal injection

00:30:42.260 - 00:30:51.540

where the stress was too large. Such fractures can also result in the injection of an entire part of the heated channel.

00:30:55.990 - 00:31:18.540

In some crystals, the crack may occur across the crystal, thereby separating both electrodes. And arc may bridge the crack

00:31:18.540 - 00:31:29.410

and finally burns down most of the crystal. But someone who observes only the current through the crystal would hardly recognize the start

00:31:29.410 - 00:31:32.370

of this catastrophic event.