Optical Inspection and Microscopy

There is a trend towards ever smaller component geometries, and hence there is an increasing need to inspect circuitry that is too small to be seen by the human eye. Surface mounted components are diminishing in size, making handling and rework of this technology increasingly dependant on having appropriate tools. Microscopic viewers, fine-tip irons and hot plates are now a necessity.

Beyond surface mount, there is an even more demanding technology called chip and wire bond. This is an advanced technique reserved for specialised applications, and requires a whole new level of tooling. Microscopes become mandatory, and the soldering iron is replaced by a wire bonding machine that attaches microscopic gold wires to the component die and any surrounding circuitry.

At Abex UK we are fortunate to have an extensive range of tooling and inspection equipment. What follows here is a brief introduction to the differences in scale that may be encountered, with some examples of our optical imaging capabilities.

Digital Camera, Macro

The module shown here is a 23GHz chip & wire bond RF front end for a point-point radio. It was manufactured by Nera, and is shown here as an example of our inspection capability.

The module base plate measures approximately 73mm x 48mm, is manufactured from milled brass and then gold plated. The image was captured by digital camera with a macro lens setting. Resolution is approximately 30 microns per pixel in the original 4M image. The finest details that can be discerned in this image are approximately 120 microns.

The circuitry uses mixed technologies on a duroid substrate which is bonded to the metal base. In this image is is possible to identify a variety of surface mounted components, which are mostly 0805 size. Printed microwave structures including hybrid-rings, radial stubs and filters are also visible.

It is also possible to see some of the bare-chip active devices, which are TRW (now Hittite) GaAs HEMT MMICs. There is little detail visible within these devices, and the bond wires are on the limit of visibility.

There are some even smaller devices present, including bare-chip single diodes, but these are not visible in this image. They are just too small.

Digital Camera, Super Close-up

By placing an additional close-up lens in front of the camera, the minimum focal distance can be reduced and hence the magnification is increased. This is a very useful technique, providing a worthy boost to the performance of the bare camera.

The image shown here has a resolution of approximately 10 microns per pixel, and represents the practical limit to this approach. The finest discernable details are perhaps 30-40 microns, and the MMIC is now clearly visible

The MMIC measures 2.5mm x 1.4mm. Some of the internal circuit structure can now be seen, and the bond wires are visible.

Achieving significantly higher levels of resolution will require a microscope.

 

Optical Microscope, x5

At Abex UK we are fortunate to have several of the very best optical microscopes. The individual instruments are discussed in detail on our microscope pages. Here we will just look at some imaging.

On the right can be seen a low power image of the above MMIC. The objective in use is x5, which is normally used with a x10 eyepiece to give x50 overall magnification. When projected into a camera this definition loses it's meaning. Here, the die which we know to have a diagonal of 2.9mm is nearly filling a 35mm frame, and the image magnification is approximately x11. However, when viewed on a computer screen the magnification is very much higher than this and also depends on the size of the screen. What becomes important is the resolution, and not the absolute value of the magnification.

Now filling the full frame, we can see the circuit detail clearly and it is a simple matter to follow the RF path from the input to the output. We can also see the bond wires quite clearly, and the pad identification text which is helpfully printed on the die.

The resolution of this image is approximately 2 microns per pixel, and the finest discernable details are around 8 microns.

 

Optical microscope, x10

This image is captured with a x10 objective. Normally this would be used with a x10 eyepiece to give an overall magnification of x100. As discussed above, this definition does not apply when a camera is attached.

Significantly more detail is now visible, and we have a resolution of approximately 1 pixel per micron, and finest discernable details around 4 microns.

In this image, the wire bonds are clearly visible.

Optical microscope, x20

This image is captured with a x20 objective, and as already mentioned, this would normally be used with a x10 eyepiece to give an overall magnification of x200. Again, this definition is not usefully applied to the situation when a camera is attached.

In this case, we now have a resolution of approximately 0.5 microns per pixel and the finest discernable details are around 2 microns.

In this image, the wire bond is shown with excellent clarity. This is one of the most useful aspects (for us) of a fine quality microscope.

Optical microscope, x40

These images are captured using a x40 objective. Once again, we would normally be using this objective with a x10 eyepiece to obtain an overall magnification of x400.

With the camera attached, we achieve a resolution of around 250nm per pixel and the finest discernable resolution is now just 1 micron.

At this resolution we can see the internal detail of the GaAs FET, and below, we can clearly see the way that a spiral inductor has been formed, and the bridge that has been formed to gain access to the internal connection.

Optical Microscope, The Limit

The optical microscope has a magnification limited not by the optics or quality of build, but by the wavelength of light. If we increase the magnification sufficiently, we reach a point where there is no improvement in image clarity, in other words, the resolution limit is reached. If we assume the wavelength of green light at approximately 550nm, under ideal circumstances the very best resolution that can be obtained is half of this value, i.e. 275nm.

In practice many things contrive against this, and it is generally not possible to actually achieve this figure. It is particularly difficult to obtain good photographic results due to additional camera related factors. A small improvement is possible if the illumination is filtered and wavelengths are restricted to the blue end of the spectrum.

The images shown here have a resolution of approximately 125nm per pixel, and the finest discernable detail is around 500nm. A pretty impressive performance, and the best photographic result that we can routinely achieve with our optical microscopes.

We also have a Cambridge Instruments Stereoscan SEM (Scanning Electron Microscope). That is a whole different story.