Page images
PDF
EPUB

system.

The lens L between the vertical and horizontal deflection mirrors refocus- (080) ses the vertical deflection, from V, onto the horizontal mirror, H [3]. The beam diverges from the horizontal deflection mirror, H, to form the scarning raster. Mirror H is located at the exit-pupil point of the microscope camera tube. The scan raster typically covers all of the field of view that can be seen with the eye when the microscope is used in its usual manner. For the microscope which we have, it is not possible to scan the specimen and observe the specimen with the eye at the same time. Switching between these two functions is performed by moving a prism within the microscope. Perhaps this is a good thing, for even though the lasers we use are low-powered there always is the possibility that under certain conditions the laser beam might be reflected directly into the eye and cause some discomfort. But it would be useful if we could always know which portion of a specimen is being scanned, and this is one of the main functions of the reflected-light circuit. The reflectedlight circuit makes use of a half-silvered mirror in the microscope, lers L2, and a photocell. The mirror comes as an integral part of the microscope vertical illuminator. Laser radiation reflected from the specimen is directed by the half-silvered mirror onto the lens and photocell. The photocell signal modulates the display screen to present a picture of the device surface topography. Used this way, we could call the apparatus a "flying-spot microscope". It is useful in its own right, and of course no electrical connections to the specimens are required. But the main purpose of the flying-spot scanner is to learn the specific details of device operation by monitoring the electrical effects of electron-hole pairs photogenerated within the device, and so the primary purpose of the reflected-light circuit is to pinpoint the device response with relation to surface features such as metallization areas. This is accomplished simply by mixing together the signals from the scanned specimen and the photocell.

Sawyer

But let us first finish
This is the schematic

We extended the capability of our scanner by adding a second dedicated laser. [Graphic 2] We can change between them by sliding a mirror which carries the second laser beam from the left in or out of the path of the first laser beam. I will tell you the reason for having two lasers in a moment. the description of the entire scanning system. [Graphic 3] of the complete system and shows a few items which did not appear before. The laser on the left operates at 0.633 um, in the visible red, and the right one at 1.15 μm, in the near infrared. They both are cw helium-neon lasers, one is equipped with mirrors to permit it to oscillate at 0.633 um, and the other with mirrors for 1.15 um operation. The system also has a rotatable analyzer in the 0.633-μm laser path on

[3] Beiser, L., Laser Beam Scan Enhancement through Periodic Aperture Transfer, Applied Optics 1, 647-650 (1968).

the left, a half-wave plate in the 1.15-um path, on the upper right, and a radiofrequency receiver in the specimen signal path, just to the left of the microscope specimen stage.

Now let us show you what the scanner actually looks like. [Graphic 4] This is a photograph of the scanner in our laboratory. A frame holding all of the optics for the system is rigidly attached to the microscope stand. One of the lasers is easily identified in the view. It is mounted on the right hand side of the frame. An 8 X 10 display screen, on the left, provides a large display that can easily be seen by several persons. The signal generators that provide the vertical and horizontal deflection signals for the scanning mirrors and the display can be seen above the display screen. A preamplifier

[blocks in formation]

and mixing amplifier are located
above the generators. The shelf above

the scanner holds some supporting
electronics for the system. These
include the radio receiver, scanning
mirror power amplifiers, laser power
supplies, a dc to 100 kHz laser mod-
ulator, and scme other power supplies
used for specimen excitation.

We added the second laser to the scanner to give us greater measurement flexability. This flexability comes from the difference in penetration depths of the radiation from the two lasers. Light and near-infrared radiation incident on silicon creates electron-hole pairs with a generation rate which exponentially trails off with distance into the material. The penetration depends on the wavelength of the incident radiation. For each wavelength we can associate a characteristic penetration length as that distance required to decrease the pair generation rate to 36.8% of its value at the surface. This 36.8% value is simply the reciprocal of e, the base of the natural logarithm system. The 0.633-um visible light from the first laser mentioned has a

(150)

LASER

ANALYZER

[graphic][merged small]

characteristic penetration depth in silicon of about 3 micrometers [4]. Because most modern silicon devices have their active regions within a few micrometers of the surface, the 0.633-um laser is quite effective in mapping active regions of operating devices. For this reason, this laser has been our "work horse" for device scanning. The light from the laser is already polarized, and so the rotatable analyzer I mentioned earlier provides a convenient way to adjust the illumination intensity. The intensity at the specimen can be varied to produce junction photocurrents over the range from about 10 picoamperes to about one-tenth of a milliampere.

Silicon at room temperature is almost transparent to 1.15 um infrared radiation from our second laser. The characteristic penetration depth of this radiation is about one centimeter [4]. Because of this weak absorption we need to utilize as much of the laser intensity as we can. There are several air-glass interfaces within the microscope which can attenuate the infrared radiation selectively, depending on the specific angle of polarization at each interface, and so it is helpful to be able to rotate the plane of polarization of the laser to find the optimun incident polarization angle through the system. This is the purpose of the half-wave plate I mentioned earlier. The half-wave plate is rotated to yield a maximum specimen. signal.

We use the infrared laser for two classes of measurements, both of which make use of the penetrating nature of the radiation. In the first, we can use the reflected light circuit to look through the silicon wafer and observe irregularities at the silicon-header interface. This application uses the scanner in the

(198)

[4]

Dash, W. C., and Newman, R., Intrinsic Optical Absorption in Single-Crystal
Germanium and Silicon at 77° and 300°K, Phys. Rev. 99, 1151-1155 (1955).

"flying-spot microscope" mode we mentioned earlier. The second application makes use (238) of the temperature sensitivity of the silicon absorption. It produces a larger signal on the display screen for those device portions which are warmer than others. Utilizing this sensitivity, we have an electronic technique for thermal mapping of devices which appears to have a number of advantages over the more traditional methods. We will show you an example of this application later on.

Since the scanner's optical system uses refractive optical elements, one has to refocus the scanning raster with the use of the reflected light circuit when one changes from one wavelength to the other. But this is not too bothersome, although a system using only reflecting optics would obviate this minor inconvenience.

During normal operation, the lasers produce optical radiation not at just one discrete wavelength but at a series of discrete wavelengths centered about the nominal wavelength. These wavelengths correspond to the individual allowed axial modes of the laser. Mixing of these wavelengths has the effect of modulating the light simultaneously at several frequencies. The modulation frequencies are multiples of a fundamental one, and for our visible laser, the light is self-modulated at 500 MHz and 1.0 GHz to a degree adequate for determining the response of devices to light modulated at these frequencies. The corresponding frequencies for our infrared laser are 385 and 770 MHz. We can determine the response of a specimen to optical radiation modulated at these frequencies. Looking again at the diagram of the scanning system, [Graphic 5] we determine this response by inserting a radio receiver between the specimen on the microscope stage and the mixing amplifier and tuning the receiver to the desired modulation frequency. This technique has been used to map the frequency dependence of device characteristics and also R-C effects in distributed passive structures.

[ocr errors][merged small][merged small][merged small][merged small][merged small][merged small][merged small]

Berning

The keynote of our scanner design was flexibility. We did not want the range of applications to be limited by the equipment, so we designed the scanner to be as versatile as possible. For example, the rate of the horizontal or vertical mirror deflection sweeps, either or both may be varied between 3 mHz and 1.1 kHz. The upper frequency is established by mechanical resonance of the mirrors. The lowfrequency range allows us to observe phenomena which are intrinsically slow, carrier trapping for example. But this is not a consideration in much of our work. We generally use a 1 kHz horizontal sweep frequency and a 3 Hz vertical sweep.

[blocks in formation]

This gives

(287)

The mixing amplifier was designed by us, and it is flat from dc to about 2 MHz. Filters are built-in which provide low-frequency roll-off when desired. The dc capability is a necessity when slow sweeps are employed. Its use is generally desirable at other times as well, since it yields a more accurate display; however, drifting and noisy specimens necessitate using the filters sometimes. The mixing amplifier contains three channels, which permit us to perform algebraic operations on the amplifier inputs. For example, we can obtain the specimen output for the unmodulated component of the laser radiation. We call this the specimen's video response. Simultaneously we can pick off the specimen output at the laser R.F. frequency component with the R.F. receiver, and can subtract the two to present on the screen only the difference to show the change in device behavior with frequency. At the same time, we may add the reflected light signal so that the net change in electrical response is superimposed on a picture of the surface of the device. The display mode also may be varied. One mode modulates the intensity of the electron beam on the 8 X 10 inch screen in the conventional manner while the other mode injects the signal into • the vertical deflection amplifier. This latter, "Y-axis", modulation mode produces pictures that appear to have a three-dimensional quality. One of the advantages of the vertical-deflection mode is that it allows the specimen response to be quantified, since the screen vertical deflection can be calibrated in terms of specimen signal current or voltage. The next chart [Graphic 6] demonstrates a mixture of both the intensity modulation and the vertical deflection display modes.

This is a photograph of the display screen where the specimen scanned is a chrome-on-glass pattern test target placed over a photodiode which generates the electrical signal monitored. You see only a very small portion of the test pattern. The display was generated using the visible laser. The finest pattern of lines shown here is on the upper right. This pattern consists of 2-um wide bars spaced 2 um apart which are clearly resolved. So we say that our optical resolution is somewhat better than 2 μm. The same measurements made using the infrared laser results in a slight degradation of detail. We associate with the infrared laser usage a resolution of better than 3 pm, referred to the specimen.

« PreviousContinue »